Update Linux to v5.10.109
Sourced from [1]
[1] https://cdn.kernel.org/pub/linux/kernel/v5.x/linux-5.10.109.tar.xz
Change-Id: I19bca9fc6762d4e63bcf3e4cba88bbe560d9c76c
Signed-off-by: Olivier Deprez <olivier.deprez@arm.com>
diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index 1671447..39b25a5 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -8,6 +8,7 @@
config IA64
bool
+ select ARCH_HAS_DMA_MARK_CLEAN
select ARCH_MIGHT_HAVE_PC_PARPORT
select ARCH_MIGHT_HAVE_PC_SERIO
select ACPI
@@ -31,10 +32,7 @@
select HAVE_FUNCTION_TRACER
select TTY
select HAVE_ARCH_TRACEHOOK
- select HAVE_MEMBLOCK_NODE_MAP
select HAVE_VIRT_CPU_ACCOUNTING
- select ARCH_HAS_DMA_COHERENT_TO_PFN
- select ARCH_HAS_SYNC_DMA_FOR_CPU
select VIRT_TO_BUS
select GENERIC_IRQ_PROBE
select GENERIC_PENDING_IRQ if SMP
@@ -57,6 +55,8 @@
select NEED_DMA_MAP_STATE
select NEED_SG_DMA_LENGTH
select NUMA if !FLATMEM
+ select PCI_MSI_ARCH_FALLBACKS if PCI_MSI
+ select SET_FS
default y
help
The Itanium Processor Family is Intel's 64-bit successor to
@@ -193,6 +193,7 @@
config IA64_HP_SBA_IOMMU
bool "HP SBA IOMMU support"
+ select DMA_OPS
default y
help
Say Y here to add support for the SBA IOMMU found on HP zx1 and
@@ -244,7 +245,7 @@
bool "Support for hot-pluggable CPUs"
depends on SMP
default n
- ---help---
+ help
Say Y here to experiment with turning CPUs off and on. CPUs
can be controlled through /sys/devices/system/cpu/cpu#.
Say N if you want to disable CPU hotplug.
@@ -267,7 +268,7 @@
bool "Support removal of Bootstrap Processor"
depends on HOTPLUG_CPU
default n
- ---help---
+ help
Say Y here if your platform SAL will support removal of BSP with HOTPLUG_CPU
support.
@@ -275,7 +276,7 @@
bool "Force assumption that CPEI can be re-targeted"
depends on PERMIT_BSP_REMOVE
default n
- ---help---
+ help
Say Y if you need to force the assumption that CPEI can be re-targeted to
any cpu in the system. This hint is available via ACPI 3.0 specifications.
Tiger4 systems are capable of re-directing CPEI to any CPU other than BSP.
@@ -362,15 +363,6 @@
config IA64_MCA_RECOVERY
tristate "MCA recovery from errors other than TLB."
-config PERFMON
- bool "Performance monitor support"
- depends on BROKEN
- help
- Selects whether support for the IA-64 performance monitor hardware
- is included in the kernel. This makes some kernel data-structures a
- little bigger and slows down execution a bit, but it is generally
- a good idea to turn this on. If you're unsure, say Y.
-
config IA64_PALINFO
tristate "/proc/pal support"
help
diff --git a/arch/ia64/Kconfig.debug b/arch/ia64/Kconfig.debug
index 40ca23b..2ce008e 100644
--- a/arch/ia64/Kconfig.debug
+++ b/arch/ia64/Kconfig.debug
@@ -39,7 +39,7 @@
config IA64_DEBUG_CMPXCHG
bool "Turn on compare-and-exchange bug checking (slow!)"
- depends on DEBUG_KERNEL
+ depends on DEBUG_KERNEL && PRINTK
help
Selecting this option turns on bug checking for the IA-64
compare-and-exchange instructions. This is slow! Itaniums
diff --git a/arch/ia64/Makefile b/arch/ia64/Makefile
index 2876a7d..703b1c4 100644
--- a/arch/ia64/Makefile
+++ b/arch/ia64/Makefile
@@ -20,7 +20,6 @@
OBJCOPYFLAGS := --strip-all
LDFLAGS_vmlinux := -static
-KBUILD_LDS_MODULE += $(srctree)/arch/ia64/module.lds
KBUILD_AFLAGS_KERNEL := -mconstant-gp
EXTRA :=
diff --git a/arch/ia64/configs/bigsur_defconfig b/arch/ia64/configs/bigsur_defconfig
index b630bd7..cfed5ed 100644
--- a/arch/ia64/configs/bigsur_defconfig
+++ b/arch/ia64/configs/bigsur_defconfig
@@ -11,7 +11,6 @@
CONFIG_NR_CPUS=2
CONFIG_PREEMPT=y
# CONFIG_VIRTUAL_MEM_MAP is not set
-CONFIG_PERFMON=y
CONFIG_IA64_PALINFO=y
CONFIG_EFI_VARS=y
CONFIG_BINFMT_MISC=m
@@ -27,10 +26,9 @@
CONFIG_BLK_DEV_CRYPTOLOOP=m
CONFIG_BLK_DEV_NBD=m
CONFIG_BLK_DEV_RAM=m
-CONFIG_IDE=m
-CONFIG_BLK_DEV_IDECD=m
-CONFIG_BLK_DEV_GENERIC=m
-CONFIG_BLK_DEV_PIIX=m
+CONFIG_ATA=m
+CONFIG_ATA_GENERIC=m
+CONFIG_ATA_PIIX=m
CONFIG_SCSI=y
CONFIG_BLK_DEV_SD=y
CONFIG_SCSI_CONSTANTS=y
@@ -57,7 +55,8 @@
CONFIG_SERIAL_8250_EXTENDED=y
CONFIG_SERIAL_8250_SHARE_IRQ=y
# CONFIG_HW_RANDOM is not set
-CONFIG_EFI_RTC=y
+CONFIG_RTC_CLASS=y
+CONFIG_RTC_DRV_EFI=y
CONFIG_I2C=y
CONFIG_I2C_CHARDEV=y
CONFIG_AGP=m
diff --git a/arch/ia64/configs/generic_defconfig b/arch/ia64/configs/generic_defconfig
index 661d90b..8916a28 100644
--- a/arch/ia64/configs/generic_defconfig
+++ b/arch/ia64/configs/generic_defconfig
@@ -18,7 +18,6 @@
CONFIG_SMP=y
CONFIG_HOTPLUG_CPU=y
CONFIG_IA64_MCA_RECOVERY=y
-CONFIG_PERFMON=y
CONFIG_IA64_PALINFO=y
CONFIG_KEXEC=y
CONFIG_CRASH_DUMP=y
@@ -45,11 +44,10 @@
CONFIG_BLK_DEV_NBD=m
CONFIG_BLK_DEV_RAM=y
CONFIG_SGI_XP=m
-CONFIG_IDE=y
-CONFIG_BLK_DEV_IDECD=y
-CONFIG_BLK_DEV_GENERIC=y
-CONFIG_BLK_DEV_CMD64X=y
-CONFIG_BLK_DEV_PIIX=y
+CONFIG_ATA=y
+CONFIG_ATA_GENERIC=y
+CONFIG_PATA_CMD64X=y
+CONFIG_ATA_PIIX=y
CONFIG_BLK_DEV_SD=y
CONFIG_CHR_DEV_ST=m
CONFIG_BLK_DEV_SR=m
@@ -57,8 +55,6 @@
CONFIG_SCSI_FC_ATTRS=y
CONFIG_SCSI_SYM53C8XX_2=y
CONFIG_SCSI_QLOGIC_1280=y
-CONFIG_ATA=y
-CONFIG_ATA_PIIX=y
CONFIG_SATA_VITESSE=y
CONFIG_MD=y
CONFIG_BLK_DEV_MD=m
@@ -94,7 +90,8 @@
CONFIG_SERIAL_8250_EXTENDED=y
CONFIG_SERIAL_8250_SHARE_IRQ=y
# CONFIG_HW_RANDOM is not set
-CONFIG_EFI_RTC=y
+CONFIG_RTC_CLASS=y
+CONFIG_RTC_DRV_EFI=y
CONFIG_RAW_DRIVER=m
CONFIG_HPET=y
CONFIG_AGP=m
diff --git a/arch/ia64/configs/gensparse_defconfig b/arch/ia64/configs/gensparse_defconfig
index 7844e6a..281eb9c 100644
--- a/arch/ia64/configs/gensparse_defconfig
+++ b/arch/ia64/configs/gensparse_defconfig
@@ -17,7 +17,6 @@
CONFIG_HOTPLUG_CPU=y
CONFIG_SPARSEMEM_MANUAL=y
CONFIG_IA64_MCA_RECOVERY=y
-CONFIG_PERFMON=y
CONFIG_IA64_PALINFO=y
CONFIG_EFI_VARS=y
CONFIG_BINFMT_MISC=m
@@ -36,12 +35,11 @@
CONFIG_BLK_DEV_CRYPTOLOOP=m
CONFIG_BLK_DEV_NBD=m
CONFIG_BLK_DEV_RAM=y
-CONFIG_IDE=y
+CONFIG_ATA=y
CONFIG_BLK_DEV_IDECD=y
-CONFIG_IDE_GENERIC=y
-CONFIG_BLK_DEV_GENERIC=y
-CONFIG_BLK_DEV_CMD64X=y
-CONFIG_BLK_DEV_PIIX=y
+CONFIG_ATA_GENERIC=y
+CONFIG_PATA_CMD64X=y
+CONFIG_ATA_PIIX=y
CONFIG_SCSI=y
CONFIG_BLK_DEV_SD=y
CONFIG_CHR_DEV_ST=m
@@ -82,7 +80,8 @@
CONFIG_SERIAL_8250_EXTENDED=y
CONFIG_SERIAL_8250_SHARE_IRQ=y
# CONFIG_HW_RANDOM is not set
-CONFIG_EFI_RTC=y
+CONFIG_RTC_CLASS=y
+CONFIG_RTC_DRV_EFI=y
CONFIG_RAW_DRIVER=m
CONFIG_HPET=y
CONFIG_AGP=m
diff --git a/arch/ia64/configs/tiger_defconfig b/arch/ia64/configs/tiger_defconfig
index 1d6e2a0..b4f9819 100644
--- a/arch/ia64/configs/tiger_defconfig
+++ b/arch/ia64/configs/tiger_defconfig
@@ -21,7 +21,6 @@
CONFIG_PERMIT_BSP_REMOVE=y
CONFIG_FORCE_CPEI_RETARGET=y
CONFIG_IA64_MCA_RECOVERY=y
-CONFIG_PERFMON=y
CONFIG_IA64_PALINFO=y
CONFIG_KEXEC=y
CONFIG_EFI_VARS=y
@@ -41,11 +40,10 @@
CONFIG_BLK_DEV_CRYPTOLOOP=m
CONFIG_BLK_DEV_NBD=m
CONFIG_BLK_DEV_RAM=y
-CONFIG_IDE=y
-CONFIG_BLK_DEV_IDECD=y
-CONFIG_BLK_DEV_GENERIC=y
-CONFIG_BLK_DEV_CMD64X=y
-CONFIG_BLK_DEV_PIIX=y
+CONFIG_ATA=y
+CONFIG_ATA_GENERIC=y
+CONFIG_PATA_CMD64X=y
+CONFIG_ATA_PIIX=y
CONFIG_SCSI=y
CONFIG_BLK_DEV_SD=y
CONFIG_CHR_DEV_ST=m
@@ -86,7 +84,8 @@
CONFIG_SERIAL_8250_EXTENDED=y
CONFIG_SERIAL_8250_SHARE_IRQ=y
# CONFIG_HW_RANDOM is not set
-CONFIG_EFI_RTC=y
+CONFIG_RTC_CLASS=y
+CONFIG_RTC_DRV_EFI=y
CONFIG_RAW_DRIVER=m
CONFIG_HPET=y
CONFIG_AGP=m
diff --git a/arch/ia64/configs/zx1_defconfig b/arch/ia64/configs/zx1_defconfig
index d42f79a..629cb9c 100644
--- a/arch/ia64/configs/zx1_defconfig
+++ b/arch/ia64/configs/zx1_defconfig
@@ -10,7 +10,6 @@
CONFIG_HOTPLUG_CPU=y
CONFIG_FLATMEM_MANUAL=y
CONFIG_IA64_MCA_RECOVERY=y
-CONFIG_PERFMON=y
CONFIG_IA64_PALINFO=y
CONFIG_CRASH_DUMP=y
CONFIG_EFI_VARS=y
@@ -26,10 +25,9 @@
CONFIG_NETFILTER=y
CONFIG_BLK_DEV_LOOP=y
CONFIG_BLK_DEV_RAM=y
-CONFIG_IDE=y
-CONFIG_BLK_DEV_IDECD=y
-CONFIG_BLK_DEV_GENERIC=y
-CONFIG_BLK_DEV_CMD64X=y
+CONFIG_ATA=y
+CONFIG_ATA_GENERIC=y
+CONFIG_PATA_CMD64X=y
CONFIG_SCSI=y
CONFIG_BLK_DEV_SD=y
CONFIG_CHR_DEV_ST=y
@@ -68,7 +66,8 @@
CONFIG_SERIAL_8250_EXTENDED=y
CONFIG_SERIAL_8250_SHARE_IRQ=y
# CONFIG_HW_RANDOM is not set
-CONFIG_EFI_RTC=y
+CONFIG_RTC_CLASS=y
+CONFIG_RTC_DRV_EFI=y
CONFIG_I2C_CHARDEV=y
CONFIG_AGP=y
CONFIG_AGP_HP_ZX1=y
diff --git a/arch/ia64/hp/common/sba_iommu.c b/arch/ia64/hp/common/sba_iommu.c
index a806227..9148ddb 100644
--- a/arch/ia64/hp/common/sba_iommu.c
+++ b/arch/ia64/hp/common/sba_iommu.c
@@ -33,7 +33,7 @@
#include <linux/bitops.h> /* hweight64() */
#include <linux/crash_dump.h>
#include <linux/iommu-helper.h>
-#include <linux/dma-mapping.h>
+#include <linux/dma-map-ops.h>
#include <linux/prefetch.h>
#include <linux/swiotlb.h>
@@ -485,8 +485,7 @@
ASSERT(((unsigned long) ioc->res_hint & (sizeof(unsigned long) - 1UL)) == 0);
ASSERT(res_ptr < res_end);
- boundary_size = (unsigned long long)dma_get_seg_boundary(dev) + 1;
- boundary_size = ALIGN(boundary_size, 1ULL << iovp_shift) >> iovp_shift;
+ boundary_size = dma_get_seg_boundary_nr_pages(dev, iovp_shift);
BUG_ON(ioc->ibase & ~iovp_mask);
shift = ioc->ibase >> iovp_shift;
@@ -907,7 +906,7 @@
* @dir: dma direction
* @attrs: optional dma attributes
*
- * See Documentation/DMA-API-HOWTO.txt
+ * See Documentation/core-api/dma-api-howto.rst
*/
static dma_addr_t sba_map_page(struct device *dev, struct page *page,
unsigned long poff, size_t size,
@@ -1028,7 +1027,7 @@
* @dir: R/W or both.
* @attrs: optional dma attributes
*
- * See Documentation/DMA-API-HOWTO.txt
+ * See Documentation/core-api/dma-api-howto.rst
*/
static void sba_unmap_page(struct device *dev, dma_addr_t iova, size_t size,
enum dma_data_direction dir, unsigned long attrs)
@@ -1105,7 +1104,7 @@
* @size: number of bytes mapped in driver buffer.
* @dma_handle: IOVA of new buffer.
*
- * See Documentation/DMA-API-HOWTO.txt
+ * See Documentation/core-api/dma-api-howto.rst
*/
static void *
sba_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_handle,
@@ -1162,7 +1161,7 @@
* @vaddr: virtual address IOVA of "consistent" buffer.
* @dma_handler: IO virtual address of "consistent" buffer.
*
- * See Documentation/DMA-API-HOWTO.txt
+ * See Documentation/core-api/dma-api-howto.rst
*/
static void sba_free_coherent(struct device *dev, size_t size, void *vaddr,
dma_addr_t dma_handle, unsigned long attrs)
@@ -1425,7 +1424,7 @@
* @dir: R/W or both.
* @attrs: optional dma attributes
*
- * See Documentation/DMA-API-HOWTO.txt
+ * See Documentation/core-api/dma-api-howto.rst
*/
static int sba_map_sg_attrs(struct device *dev, struct scatterlist *sglist,
int nents, enum dma_data_direction dir,
@@ -1524,7 +1523,7 @@
* @dir: R/W or both.
* @attrs: optional dma attributes
*
- * See Documentation/DMA-API-HOWTO.txt
+ * See Documentation/core-api/dma-api-howto.rst
*/
static void sba_unmap_sg_attrs(struct device *dev, struct scatterlist *sglist,
int nents, enum dma_data_direction dir,
@@ -2071,6 +2070,8 @@
.dma_supported = sba_dma_supported,
.mmap = dma_common_mmap,
.get_sgtable = dma_common_get_sgtable,
+ .alloc_pages = dma_common_alloc_pages,
+ .free_pages = dma_common_free_pages,
};
static int __init
diff --git a/arch/ia64/include/asm/Kbuild b/arch/ia64/include/asm/Kbuild
index 3903936..f994c1d 100644
--- a/arch/ia64/include/asm/Kbuild
+++ b/arch/ia64/include/asm/Kbuild
@@ -1,12 +1,5 @@
# SPDX-License-Identifier: GPL-2.0
generated-y += syscall_table.h
-generic-y += compat.h
-generic-y += exec.h
-generic-y += irq_work.h
generic-y += kvm_para.h
generic-y += mcs_spinlock.h
-generic-y += mm-arch-hooks.h
-generic-y += preempt.h
-generic-y += trace_clock.h
generic-y += vtime.h
-generic-y += word-at-a-time.h
diff --git a/arch/ia64/include/asm/acpi.h b/arch/ia64/include/asm/acpi.h
index f886d4d..87927eb 100644
--- a/arch/ia64/include/asm/acpi.h
+++ b/arch/ia64/include/asm/acpi.h
@@ -38,7 +38,10 @@
/* Low-level suspend routine. */
extern int acpi_suspend_lowlevel(void);
-extern unsigned long acpi_wakeup_address;
+static inline unsigned long acpi_get_wakeup_address(void)
+{
+ return 0;
+}
/*
* Record the cpei override flag and current logical cpu. This is
@@ -71,8 +74,6 @@
buf[2] |= ACPI_PDC_EST_CAPABILITY_SMP;
}
-#define acpi_unlazy_tlb(x)
-
#ifdef CONFIG_ACPI_NUMA
extern cpumask_t early_cpu_possible_map;
#define for_each_possible_early_cpu(cpu) \
diff --git a/arch/ia64/include/asm/agp.h b/arch/ia64/include/asm/agp.h
index 2b451c4..0261507 100644
--- a/arch/ia64/include/asm/agp.h
+++ b/arch/ia64/include/asm/agp.h
@@ -14,8 +14,8 @@
* in coherent mode, which lets us map the AGP memory as normal (write-back) memory
* (unlike x86, where it gets mapped "write-coalescing").
*/
-#define map_page_into_agp(page) /* nothing */
-#define unmap_page_from_agp(page) /* nothing */
+#define map_page_into_agp(page) do { } while (0)
+#define unmap_page_from_agp(page) do { } while (0)
#define flush_agp_cache() mb()
/* GATT allocation. Returns/accepts GATT kernel virtual address. */
diff --git a/arch/ia64/include/asm/atomic.h b/arch/ia64/include/asm/atomic.h
index 50440f3..f267d95 100644
--- a/arch/ia64/include/asm/atomic.h
+++ b/arch/ia64/include/asm/atomic.h
@@ -19,7 +19,6 @@
#include <asm/barrier.h>
-#define ATOMIC_INIT(i) { (i) }
#define ATOMIC64_INIT(i) { (i) }
#define atomic_read(v) READ_ONCE((v)->counter)
diff --git a/arch/ia64/include/asm/cache.h b/arch/ia64/include/asm/cache.h
index 4eb6f74..2f1c706 100644
--- a/arch/ia64/include/asm/cache.h
+++ b/arch/ia64/include/asm/cache.h
@@ -25,6 +25,6 @@
# define SMP_CACHE_BYTES (1 << 3)
#endif
-#define __read_mostly __attribute__((__section__(".data..read_mostly")))
+#define __read_mostly __section(".data..read_mostly")
#endif /* _ASM_IA64_CACHE_H */
diff --git a/arch/ia64/include/asm/cacheflush.h b/arch/ia64/include/asm/cacheflush.h
index 6d3478f..708c0fa 100644
--- a/arch/ia64/include/asm/cacheflush.h
+++ b/arch/ia64/include/asm/cacheflush.h
@@ -12,44 +12,22 @@
#include <asm/page.h>
-/*
- * Cache flushing routines. This is the kind of stuff that can be very expensive, so try
- * to avoid them whenever possible.
- */
-
-#define flush_cache_all() do { } while (0)
-#define flush_cache_mm(mm) do { } while (0)
-#define flush_cache_dup_mm(mm) do { } while (0)
-#define flush_cache_range(vma, start, end) do { } while (0)
-#define flush_cache_page(vma, vmaddr, pfn) do { } while (0)
-#define flush_icache_page(vma,page) do { } while (0)
-#define flush_cache_vmap(start, end) do { } while (0)
-#define flush_cache_vunmap(start, end) do { } while (0)
-
#define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE 1
#define flush_dcache_page(page) \
do { \
clear_bit(PG_arch_1, &(page)->flags); \
} while (0)
-#define flush_dcache_mmap_lock(mapping) do { } while (0)
-#define flush_dcache_mmap_unlock(mapping) do { } while (0)
-
-extern void flush_icache_range (unsigned long start, unsigned long end);
+extern void flush_icache_range(unsigned long start, unsigned long end);
+#define flush_icache_range flush_icache_range
extern void clflush_cache_range(void *addr, int size);
-
-#define flush_icache_user_range(vma, page, user_addr, len) \
+#define flush_icache_user_page(vma, page, user_addr, len) \
do { \
unsigned long _addr = (unsigned long) page_address(page) + ((user_addr) & ~PAGE_MASK); \
flush_icache_range(_addr, _addr + (len)); \
} while (0)
-#define copy_to_user_page(vma, page, vaddr, dst, src, len) \
-do { memcpy(dst, src, len); \
- flush_icache_user_range(vma, page, vaddr, len); \
-} while (0)
-#define copy_from_user_page(vma, page, vaddr, dst, src, len) \
- memcpy(dst, src, len)
+#include <asm-generic/cacheflush.h>
#endif /* _ASM_IA64_CACHEFLUSH_H */
diff --git a/arch/ia64/include/asm/checksum.h b/arch/ia64/include/asm/checksum.h
index 0ed18bc..f302621 100644
--- a/arch/ia64/include/asm/checksum.h
+++ b/arch/ia64/include/asm/checksum.h
@@ -38,19 +38,6 @@
extern __wsum csum_partial(const void *buff, int len, __wsum sum);
/*
- * Same as csum_partial, but copies from src while it checksums.
- *
- * Here it is even more important to align src and dst on a 32-bit (or
- * even better 64-bit) boundary.
- */
-extern __wsum csum_partial_copy_from_user(const void __user *src, void *dst,
- int len, __wsum sum,
- int *errp);
-
-extern __wsum csum_partial_copy_nocheck(const void *src, void *dst,
- int len, __wsum sum);
-
-/*
* This routine is used for miscellaneous IP-like checksums, mainly in
* icmp.c
*/
diff --git a/arch/ia64/include/asm/device.h b/arch/ia64/include/asm/device.h
index 410a769..918b198 100644
--- a/arch/ia64/include/asm/device.h
+++ b/arch/ia64/include/asm/device.h
@@ -6,9 +6,6 @@
#define _ASM_IA64_DEVICE_H
struct dev_archdata {
-#ifdef CONFIG_INTEL_IOMMU
- void *iommu; /* hook for IOMMU specific extension */
-#endif
};
struct pdev_archdata {
diff --git a/arch/ia64/include/asm/elf.h b/arch/ia64/include/asm/elf.h
index c70bb9c..6629301 100644
--- a/arch/ia64/include/asm/elf.h
+++ b/arch/ia64/include/asm/elf.h
@@ -179,8 +179,6 @@
#define ELF_AR_SSD_OFFSET (56 * sizeof(elf_greg_t))
#define ELF_AR_END_OFFSET (57 * sizeof(elf_greg_t))
-typedef unsigned long elf_fpxregset_t;
-
typedef unsigned long elf_greg_t;
typedef elf_greg_t elf_gregset_t[ELF_NGREG];
diff --git a/arch/ia64/include/asm/futex.h b/arch/ia64/include/asm/futex.h
index 2e106d4..1db26b4 100644
--- a/arch/ia64/include/asm/futex.h
+++ b/arch/ia64/include/asm/futex.h
@@ -50,7 +50,8 @@
{
int oldval = 0, ret;
- pagefault_disable();
+ if (!access_ok(uaddr, sizeof(u32)))
+ return -EFAULT;
switch (op) {
case FUTEX_OP_SET:
@@ -74,8 +75,6 @@
ret = -ENOSYS;
}
- pagefault_enable();
-
if (!ret)
*oval = oldval;
diff --git a/arch/ia64/include/asm/hugetlb.h b/arch/ia64/include/asm/hugetlb.h
index 36cc039..7e46ebd 100644
--- a/arch/ia64/include/asm/hugetlb.h
+++ b/arch/ia64/include/asm/hugetlb.h
@@ -20,6 +20,7 @@
return (REGION_NUMBER(addr) == RGN_HPAGE ||
REGION_NUMBER((addr)+(len)-1) == RGN_HPAGE);
}
+#define is_hugepage_only_range is_hugepage_only_range
#define __HAVE_ARCH_HUGE_PTEP_CLEAR_FLUSH
static inline void huge_ptep_clear_flush(struct vm_area_struct *vma,
@@ -27,10 +28,6 @@
{
}
-static inline void arch_clear_hugepage_flags(struct page *page)
-{
-}
-
#include <asm-generic/hugetlb.h>
#endif /* _ASM_IA64_HUGETLB_H */
diff --git a/arch/ia64/include/asm/hw_irq.h b/arch/ia64/include/asm/hw_irq.h
index e6385c7..f6ff95b 100644
--- a/arch/ia64/include/asm/hw_irq.h
+++ b/arch/ia64/include/asm/hw_irq.h
@@ -113,7 +113,6 @@
#define ia64_register_ipi ia64_native_register_ipi
#define assign_irq_vector ia64_native_assign_irq_vector
#define free_irq_vector ia64_native_free_irq_vector
-#define register_percpu_irq ia64_native_register_percpu_irq
#define ia64_resend_irq ia64_native_resend_irq
extern void ia64_native_register_ipi(void);
@@ -123,7 +122,6 @@
extern int reserve_irq_vector (int vector);
extern void __setup_vector_irq(int cpu);
extern void ia64_send_ipi (int cpu, int vector, int delivery_mode, int redirect);
-extern void ia64_native_register_percpu_irq (ia64_vector vec, struct irqaction *action);
extern void destroy_and_reserve_irq (unsigned int irq);
#ifdef CONFIG_SMP
diff --git a/arch/ia64/include/asm/io.h b/arch/ia64/include/asm/io.h
index 54e70c2..3d666a1 100644
--- a/arch/ia64/include/asm/io.h
+++ b/arch/ia64/include/asm/io.h
@@ -256,16 +256,15 @@
# ifdef __KERNEL__
extern void __iomem * ioremap(unsigned long offset, unsigned long size);
-extern void __iomem * ioremap_nocache (unsigned long offset, unsigned long size);
+extern void __iomem * ioremap_uc(unsigned long offset, unsigned long size);
extern void iounmap (volatile void __iomem *addr);
static inline void __iomem * ioremap_cache (unsigned long phys_addr, unsigned long size)
{
return ioremap(phys_addr, size);
}
#define ioremap ioremap
-#define ioremap_nocache ioremap_nocache
#define ioremap_cache ioremap_cache
-#define ioremap_uc ioremap_nocache
+#define ioremap_uc ioremap_uc
#define iounmap iounmap
/*
diff --git a/arch/ia64/include/asm/iommu.h b/arch/ia64/include/asm/iommu.h
index 7904f59..eb0db20 100644
--- a/arch/ia64/include/asm/iommu.h
+++ b/arch/ia64/include/asm/iommu.h
@@ -2,6 +2,8 @@
#ifndef _ASM_IA64_IOMMU_H
#define _ASM_IA64_IOMMU_H 1
+#include <linux/acpi.h>
+
/* 10 seconds */
#define DMAR_OPERATION_TIMEOUT (((cycles_t) local_cpu_data->itc_freq)*10)
@@ -9,6 +11,9 @@
#ifdef CONFIG_INTEL_IOMMU
extern int force_iommu, no_iommu;
extern int iommu_detected;
+
+static inline int __init
+arch_rmrr_sanity_check(struct acpi_dmar_reserved_memory *rmrr) { return 0; }
#else
#define no_iommu (1)
#define iommu_detected (0)
diff --git a/arch/ia64/include/asm/irqflags.h b/arch/ia64/include/asm/irqflags.h
index d97f843..1dc30f1 100644
--- a/arch/ia64/include/asm/irqflags.h
+++ b/arch/ia64/include/asm/irqflags.h
@@ -36,11 +36,7 @@
static inline unsigned long arch_local_save_flags(void)
{
ia64_stop();
-#ifdef CONFIG_PARAVIRT
- return ia64_get_psr_i();
-#else
return ia64_getreg(_IA64_REG_PSR);
-#endif
}
static inline unsigned long arch_local_irq_save(void)
diff --git a/arch/ia64/include/asm/local64.h b/arch/ia64/include/asm/local64.h
deleted file mode 100644
index 36c93b5..0000000
--- a/arch/ia64/include/asm/local64.h
+++ /dev/null
@@ -1 +0,0 @@
-#include <asm-generic/local64.h>
diff --git a/arch/ia64/include/asm/module.h b/arch/ia64/include/asm/module.h
index 9fbf32e..7271b9c 100644
--- a/arch/ia64/include/asm/module.h
+++ b/arch/ia64/include/asm/module.h
@@ -30,10 +30,6 @@
unsigned long opd_size;
};
-#define MODULE_PROC_FAMILY "ia64"
-#define MODULE_ARCH_VERMAGIC MODULE_PROC_FAMILY \
- "gcc-" __stringify(__GNUC__) "." __stringify(__GNUC_MINOR__)
-
#define ARCH_SHF_SMALL SHF_IA_64_SHORT
#endif /* _ASM_IA64_MODULE_H */
diff --git a/arch/ia64/module.lds b/arch/ia64/include/asm/module.lds.h
similarity index 100%
rename from arch/ia64/module.lds
rename to arch/ia64/include/asm/module.lds.h
diff --git a/arch/ia64/include/asm/page.h b/arch/ia64/include/asm/page.h
index 5798bd2..b69a549 100644
--- a/arch/ia64/include/asm/page.h
+++ b/arch/ia64/include/asm/page.h
@@ -218,10 +218,7 @@
#define PAGE_OFFSET RGN_BASE(RGN_KERNEL)
-#define VM_DATA_DEFAULT_FLAGS (VM_READ | VM_WRITE | \
- VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC | \
- (((current->personality & READ_IMPLIES_EXEC) != 0) \
- ? VM_EXEC : 0))
+#define VM_DATA_DEFAULT_FLAGS VM_DATA_FLAGS_TSK_EXEC
#define GATE_ADDR RGN_BASE(RGN_GATE)
diff --git a/arch/ia64/include/asm/pgalloc.h b/arch/ia64/include/asm/pgalloc.h
index f4c4910..9601cfe 100644
--- a/arch/ia64/include/asm/pgalloc.h
+++ b/arch/ia64/include/asm/pgalloc.h
@@ -29,27 +29,13 @@
return (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
}
-static inline void pgd_free(struct mm_struct *mm, pgd_t *pgd)
-{
- free_page((unsigned long)pgd);
-}
-
#if CONFIG_PGTABLE_LEVELS == 4
static inline void
-pgd_populate(struct mm_struct *mm, pgd_t * pgd_entry, pud_t * pud)
+p4d_populate(struct mm_struct *mm, p4d_t * p4d_entry, pud_t * pud)
{
- pgd_val(*pgd_entry) = __pa(pud);
+ p4d_val(*p4d_entry) = __pa(pud);
}
-static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long addr)
-{
- return (pud_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
-}
-
-static inline void pud_free(struct mm_struct *mm, pud_t *pud)
-{
- free_page((unsigned long)pud);
-}
#define __pud_free_tlb(tlb, pud, address) pud_free((tlb)->mm, pud)
#endif /* CONFIG_PGTABLE_LEVELS == 4 */
@@ -59,16 +45,6 @@
pud_val(*pud_entry) = __pa(pmd);
}
-static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long addr)
-{
- return (pmd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
-}
-
-static inline void pmd_free(struct mm_struct *mm, pmd_t *pmd)
-{
- free_page((unsigned long)pmd);
-}
-
#define __pmd_free_tlb(tlb, pmd, address) pmd_free((tlb)->mm, pmd)
static inline void
diff --git a/arch/ia64/include/asm/pgtable.h b/arch/ia64/include/asm/pgtable.h
index d602e7c..9f64fdf 100644
--- a/arch/ia64/include/asm/pgtable.h
+++ b/arch/ia64/include/asm/pgtable.h
@@ -283,12 +283,12 @@
#define pud_page(pud) virt_to_page((pud_val(pud) + PAGE_OFFSET))
#if CONFIG_PGTABLE_LEVELS == 4
-#define pgd_none(pgd) (!pgd_val(pgd))
-#define pgd_bad(pgd) (!ia64_phys_addr_valid(pgd_val(pgd)))
-#define pgd_present(pgd) (pgd_val(pgd) != 0UL)
-#define pgd_clear(pgdp) (pgd_val(*(pgdp)) = 0UL)
-#define pgd_page_vaddr(pgd) ((unsigned long) __va(pgd_val(pgd) & _PFN_MASK))
-#define pgd_page(pgd) virt_to_page((pgd_val(pgd) + PAGE_OFFSET))
+#define p4d_none(p4d) (!p4d_val(p4d))
+#define p4d_bad(p4d) (!ia64_phys_addr_valid(p4d_val(p4d)))
+#define p4d_present(p4d) (p4d_val(p4d) != 0UL)
+#define p4d_clear(p4dp) (p4d_val(*(p4dp)) = 0UL)
+#define p4d_page_vaddr(p4d) ((unsigned long) __va(p4d_val(p4d) & _PFN_MASK))
+#define p4d_page(p4d) virt_to_page((p4d_val(p4d) + PAGE_OFFSET))
#endif
/*
@@ -298,7 +298,6 @@
#define pte_exec(pte) ((pte_val(pte) & _PAGE_AR_RX) != 0)
#define pte_dirty(pte) ((pte_val(pte) & _PAGE_D) != 0)
#define pte_young(pte) ((pte_val(pte) & _PAGE_A) != 0)
-#define pte_special(pte) 0
/*
* Note: we convert AR_RWX to AR_RX and AR_RW to AR_R by clearing the 2nd bit in the
@@ -311,7 +310,6 @@
#define pte_mkclean(pte) (__pte(pte_val(pte) & ~_PAGE_D))
#define pte_mkdirty(pte) (__pte(pte_val(pte) | _PAGE_D))
#define pte_mkhuge(pte) (__pte(pte_val(pte)))
-#define pte_mkspecial(pte) (pte)
/*
* Because ia64's Icache and Dcache is not coherent (on a cpu), we need to
@@ -366,17 +364,14 @@
return (region << (PAGE_SHIFT - 6)) | l1index;
}
+#define pgd_index pgd_index
-/* The offset in the 1-level directory is given by the 3 region bits
- (61..63) and the level-1 bits. */
-static inline pgd_t*
-pgd_offset (const struct mm_struct *mm, unsigned long address)
-{
- return mm->pgd + pgd_index(address);
-}
-
-/* In the kernel's mapped region we completely ignore the region number
- (since we know it's in region number 5). */
+/*
+ * In the kernel's mapped region we know everything is in region number 5, so
+ * as an optimisation its PGD already points to the area for that region.
+ * However, this also means that we cannot use pgd_index() and we must
+ * never add the region here.
+ */
#define pgd_offset_k(addr) \
(init_mm.pgd + (((addr) >> PGDIR_SHIFT) & (PTRS_PER_PGD - 1)))
@@ -385,25 +380,6 @@
here. */
#define pgd_offset_gate(mm, addr) pgd_offset_k(addr)
-#if CONFIG_PGTABLE_LEVELS == 4
-/* Find an entry in the second-level page table.. */
-#define pud_offset(dir,addr) \
- ((pud_t *) pgd_page_vaddr(*(dir)) + (((addr) >> PUD_SHIFT) & (PTRS_PER_PUD - 1)))
-#endif
-
-/* Find an entry in the third-level page table.. */
-#define pmd_offset(dir,addr) \
- ((pmd_t *) pud_page_vaddr(*(dir)) + (((addr) >> PMD_SHIFT) & (PTRS_PER_PMD - 1)))
-
-/*
- * Find an entry in the third-level page table. This looks more complicated than it
- * should be because some platforms place page tables in high memory.
- */
-#define pte_index(addr) (((addr) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
-#define pte_offset_kernel(dir,addr) ((pte_t *) pmd_page_vaddr(*(dir)) + pte_index(addr))
-#define pte_offset_map(dir,addr) pte_offset_kernel(dir, addr)
-#define pte_unmap(pte) do { } while (0)
-
/* atomic versions of the some PTE manipulations: */
static inline int
@@ -544,8 +520,9 @@
# ifdef CONFIG_VIRTUAL_MEM_MAP
/* arch mem_map init routine is needed due to holes in a virtual mem_map */
- extern void memmap_init (unsigned long size, int nid, unsigned long zone,
- unsigned long start_pfn);
+void memmap_init(void);
+void arch_memmap_init(unsigned long size, int nid, unsigned long zone,
+ unsigned long start_pfn);
# endif /* CONFIG_VIRTUAL_MEM_MAP */
# endif /* !__ASSEMBLY__ */
@@ -582,10 +559,8 @@
#if CONFIG_PGTABLE_LEVELS == 3
-#define __ARCH_USE_5LEVEL_HACK
#include <asm-generic/pgtable-nopud.h>
#endif
-#include <asm-generic/5level-fixup.h>
-#include <asm-generic/pgtable.h>
+#include <asm-generic/pgtable-nop4d.h>
#endif /* _ASM_IA64_PGTABLE_H */
diff --git a/arch/ia64/include/asm/processor.h b/arch/ia64/include/asm/processor.h
index 95a2ec3..2d8bcdc 100644
--- a/arch/ia64/include/asm/processor.h
+++ b/arch/ia64/include/asm/processor.h
@@ -280,15 +280,6 @@
__u64 map_base; /* base address for get_unmapped_area() */
__u64 rbs_bot; /* the base address for the RBS */
int last_fph_cpu; /* CPU that may hold the contents of f32-f127 */
-
-#ifdef CONFIG_PERFMON
- void *pfm_context; /* pointer to detailed PMU context */
- unsigned long pfm_needs_checking; /* when >0, pending perfmon work on kernel exit */
-# define INIT_THREAD_PM .pfm_context = NULL, \
- .pfm_needs_checking = 0UL,
-#else
-# define INIT_THREAD_PM
-#endif
unsigned long dbr[IA64_NUM_DBG_REGS];
unsigned long ibr[IA64_NUM_DBG_REGS];
struct ia64_fpreg fph[96]; /* saved/loaded on demand */
@@ -301,7 +292,6 @@
.map_base = DEFAULT_MAP_BASE, \
.rbs_bot = STACK_TOP - DEFAULT_USER_STACK_SIZE, \
.last_fph_cpu = -1, \
- INIT_THREAD_PM \
.dbr = {0, }, \
.ibr = {0, }, \
.fph = {{{{0}}}, } \
diff --git a/arch/ia64/include/asm/ptrace.h b/arch/ia64/include/asm/ptrace.h
index f31e07f..0817913 100644
--- a/arch/ia64/include/asm/ptrace.h
+++ b/arch/ia64/include/asm/ptrace.h
@@ -108,7 +108,6 @@
struct task_struct; /* forward decl */
struct unw_frame_info; /* forward decl */
- extern void ia64_do_show_stack (struct unw_frame_info *, void *);
extern unsigned long ia64_get_user_rbs_end (struct task_struct *, struct pt_regs *,
unsigned long *);
extern long ia64_peek (struct task_struct *, struct switch_stack *, unsigned long,
diff --git a/arch/ia64/include/asm/sections.h b/arch/ia64/include/asm/sections.h
index cea15f2..3a033d2 100644
--- a/arch/ia64/include/asm/sections.h
+++ b/arch/ia64/include/asm/sections.h
@@ -35,7 +35,7 @@
struct fdesc *desc = ptr;
void *p;
- if (!probe_kernel_address(&desc->ip, p))
+ if (!get_kernel_nofault(p, (void *)&desc->ip))
ptr = p;
return ptr;
}
diff --git a/arch/ia64/include/asm/smp.h b/arch/ia64/include/asm/smp.h
index 7847ae4..aa92234 100644
--- a/arch/ia64/include/asm/smp.h
+++ b/arch/ia64/include/asm/smp.h
@@ -18,7 +18,6 @@
#include <linux/bitops.h>
#include <linux/irqreturn.h>
-#include <asm/io.h>
#include <asm/param.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
@@ -44,11 +43,6 @@
#ifdef CONFIG_SMP
-#define XTP_OFFSET 0x1e0008
-
-#define SMP_IRQ_REDIRECTION (1 << 0)
-#define SMP_IPI_REDIRECTION (1 << 1)
-
#define raw_smp_processor_id() (current_thread_info()->cpu)
extern struct smp_boot_data {
@@ -62,7 +56,6 @@
DECLARE_PER_CPU_SHARED_ALIGNED(cpumask_t, cpu_sibling_map);
extern int smp_num_siblings;
extern void __iomem *ipi_base_addr;
-extern unsigned char smp_int_redirect;
extern volatile int ia64_cpu_to_sapicid[];
#define cpu_physical_id(i) ia64_cpu_to_sapicid[i]
@@ -84,34 +77,6 @@
return i;
}
-/*
- * XTP control functions:
- * min_xtp : route all interrupts to this CPU
- * normal_xtp: nominal XTP value
- * max_xtp : never deliver interrupts to this CPU.
- */
-
-static inline void
-min_xtp (void)
-{
- if (smp_int_redirect & SMP_IRQ_REDIRECTION)
- writeb(0x00, ipi_base_addr + XTP_OFFSET); /* XTP to min */
-}
-
-static inline void
-normal_xtp (void)
-{
- if (smp_int_redirect & SMP_IRQ_REDIRECTION)
- writeb(0x08, ipi_base_addr + XTP_OFFSET); /* XTP normal */
-}
-
-static inline void
-max_xtp (void)
-{
- if (smp_int_redirect & SMP_IRQ_REDIRECTION)
- writeb(0x0f, ipi_base_addr + XTP_OFFSET); /* Set XTP to max */
-}
-
/* Upping and downing of CPUs */
extern int __cpu_disable (void);
extern void __cpu_die (unsigned int cpu);
diff --git a/arch/ia64/include/asm/sparsemem.h b/arch/ia64/include/asm/sparsemem.h
index 336d057..dd8c166 100644
--- a/arch/ia64/include/asm/sparsemem.h
+++ b/arch/ia64/include/asm/sparsemem.h
@@ -18,4 +18,10 @@
#endif
#endif /* CONFIG_SPARSEMEM */
+
+#ifdef CONFIG_MEMORY_HOTPLUG
+int memory_add_physaddr_to_nid(u64 addr);
+#define memory_add_physaddr_to_nid memory_add_physaddr_to_nid
+#endif
+
#endif /* _ASM_IA64_SPARSEMEM_H */
diff --git a/arch/ia64/include/asm/switch_to.h b/arch/ia64/include/asm/switch_to.h
index 9011e90..a5a4e09 100644
--- a/arch/ia64/include/asm/switch_to.h
+++ b/arch/ia64/include/asm/switch_to.h
@@ -31,16 +31,8 @@
extern void ia64_save_extra (struct task_struct *task);
extern void ia64_load_extra (struct task_struct *task);
-#ifdef CONFIG_PERFMON
- DECLARE_PER_CPU(unsigned long, pfm_syst_info);
-# define PERFMON_IS_SYSWIDE() (__this_cpu_read(pfm_syst_info) & 0x1)
-#else
-# define PERFMON_IS_SYSWIDE() (0)
-#endif
-
#define IA64_HAS_EXTRA_STATE(t) \
- ((t)->thread.flags & (IA64_THREAD_DBG_VALID|IA64_THREAD_PM_VALID) \
- || PERFMON_IS_SYSWIDE())
+ ((t)->thread.flags & (IA64_THREAD_DBG_VALID|IA64_THREAD_PM_VALID))
#define __switch_to(prev,next,last) do { \
if (IA64_HAS_EXTRA_STATE(prev)) \
diff --git a/arch/ia64/include/asm/tlb.h b/arch/ia64/include/asm/tlb.h
index f1f257d..8d9da6f 100644
--- a/arch/ia64/include/asm/tlb.h
+++ b/arch/ia64/include/asm/tlb.h
@@ -42,7 +42,6 @@
#include <linux/pagemap.h>
#include <linux/swap.h>
-#include <asm/pgalloc.h>
#include <asm/processor.h>
#include <asm/tlbflush.h>
diff --git a/arch/ia64/include/asm/uaccess.h b/arch/ia64/include/asm/uaccess.h
index 89782ad..179243c 100644
--- a/arch/ia64/include/asm/uaccess.h
+++ b/arch/ia64/include/asm/uaccess.h
@@ -35,10 +35,9 @@
#include <linux/compiler.h>
#include <linux/page-flags.h>
-#include <linux/mm.h>
#include <asm/intrinsics.h>
-#include <asm/pgtable.h>
+#include <linux/pgtable.h>
#include <asm/io.h>
#include <asm/extable.h>
@@ -51,7 +50,7 @@
#define get_fs() (current_thread_info()->addr_limit)
#define set_fs(x) (current_thread_info()->addr_limit = (x))
-#define segment_eq(a, b) ((a).seg == (b).seg)
+#define uaccess_kernel() (get_fs().seg == KERNEL_DS.seg)
/*
* When accessing user memory, we need to make sure the entire area really is in
diff --git a/arch/ia64/include/asm/vermagic.h b/arch/ia64/include/asm/vermagic.h
new file mode 100644
index 0000000..29c7424
--- /dev/null
+++ b/arch/ia64/include/asm/vermagic.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2003 Hewlett-Packard Co
+ * David Mosberger-Tang <davidm@hpl.hp.com>
+ */
+
+#ifndef _ASM_VERMAGIC_H
+#define _ASM_VERMAGIC_H
+
+#include <linux/stringify.h>
+
+#define MODULE_ARCH_VERMAGIC "ia64" \
+ "gcc-" __stringify(__GNUC__) "." __stringify(__GNUC_MINOR__)
+
+#endif /* _ASM_VERMAGIC_H */
diff --git a/arch/ia64/include/asm/vga.h b/arch/ia64/include/asm/vga.h
index 30cb373..64ce0b9 100644
--- a/arch/ia64/include/asm/vga.h
+++ b/arch/ia64/include/asm/vga.h
@@ -18,7 +18,7 @@
extern unsigned long vga_console_iobase;
extern unsigned long vga_console_membase;
-#define VGA_MAP_MEM(x,s) ((unsigned long) ioremap_nocache(vga_console_membase + (x), s))
+#define VGA_MAP_MEM(x,s) ((unsigned long) ioremap(vga_console_membase + (x), s))
#define vga_readb(x) (*(x))
#define vga_writeb(x,y) (*(y) = (x))
diff --git a/arch/ia64/include/asm/vmalloc.h b/arch/ia64/include/asm/vmalloc.h
new file mode 100644
index 0000000..a2b5114
--- /dev/null
+++ b/arch/ia64/include/asm/vmalloc.h
@@ -0,0 +1,4 @@
+#ifndef _ASM_IA64_VMALLOC_H
+#define _ASM_IA64_VMALLOC_H
+
+#endif /* _ASM_IA64_VMALLOC_H */
diff --git a/arch/ia64/include/asm/xtp.h b/arch/ia64/include/asm/xtp.h
new file mode 100644
index 0000000..5bf1d70
--- /dev/null
+++ b/arch/ia64/include/asm/xtp.h
@@ -0,0 +1,46 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_IA64_XTP_H
+#define _ASM_IA64_XTP_H
+
+#include <asm/io.h>
+
+#ifdef CONFIG_SMP
+
+#define XTP_OFFSET 0x1e0008
+
+#define SMP_IRQ_REDIRECTION (1 << 0)
+#define SMP_IPI_REDIRECTION (1 << 1)
+
+extern unsigned char smp_int_redirect;
+
+/*
+ * XTP control functions:
+ * min_xtp : route all interrupts to this CPU
+ * normal_xtp: nominal XTP value
+ * max_xtp : never deliver interrupts to this CPU.
+ */
+
+static inline void
+min_xtp (void)
+{
+ if (smp_int_redirect & SMP_IRQ_REDIRECTION)
+ writeb(0x00, ipi_base_addr + XTP_OFFSET); /* XTP to min */
+}
+
+static inline void
+normal_xtp (void)
+{
+ if (smp_int_redirect & SMP_IRQ_REDIRECTION)
+ writeb(0x08, ipi_base_addr + XTP_OFFSET); /* XTP normal */
+}
+
+static inline void
+max_xtp (void)
+{
+ if (smp_int_redirect & SMP_IRQ_REDIRECTION)
+ writeb(0x0f, ipi_base_addr + XTP_OFFSET); /* Set XTP to max */
+}
+
+#endif /* CONFIG_SMP */
+
+#endif /* _ASM_IA64_XTP_Hy */
diff --git a/arch/ia64/include/uapi/asm/errno.h b/arch/ia64/include/uapi/asm/errno.h
deleted file mode 100644
index 9addba5..0000000
--- a/arch/ia64/include/uapi/asm/errno.h
+++ /dev/null
@@ -1,2 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
-#include <asm-generic/errno.h>
diff --git a/arch/ia64/include/uapi/asm/gcc_intrin.h b/arch/ia64/include/uapi/asm/gcc_intrin.h
index c60696f..ecfa3ea 100644
--- a/arch/ia64/include/uapi/asm/gcc_intrin.h
+++ b/arch/ia64/include/uapi/asm/gcc_intrin.h
@@ -31,7 +31,7 @@
extern void ia64_bad_param_for_getreg (void);
-#define ia64_native_setreg(regnum, val) \
+#define ia64_setreg(regnum, val) \
({ \
switch (regnum) { \
case _IA64_REG_PSR_L: \
@@ -60,7 +60,7 @@
} \
})
-#define ia64_native_getreg(regnum) \
+#define ia64_getreg(regnum) \
({ \
__u64 ia64_intri_res; \
\
@@ -384,7 +384,7 @@
#define ia64_invala() asm volatile ("invala" ::: "memory")
-#define ia64_native_thash(addr) \
+#define ia64_thash(addr) \
({ \
unsigned long ia64_intri_res; \
asm volatile ("thash %0=%1" : "=r"(ia64_intri_res) : "r" (addr)); \
@@ -437,10 +437,10 @@
#define ia64_set_pmd(index, val) \
asm volatile ("mov pmd[%0]=%1" :: "r"(index), "r"(val) : "memory")
-#define ia64_native_set_rr(index, val) \
+#define ia64_set_rr(index, val) \
asm volatile ("mov rr[%0]=%1" :: "r"(index), "r"(val) : "memory");
-#define ia64_native_get_cpuid(index) \
+#define ia64_get_cpuid(index) \
({ \
unsigned long ia64_intri_res; \
asm volatile ("mov %0=cpuid[%r1]" : "=r"(ia64_intri_res) : "rO"(index)); \
@@ -476,33 +476,33 @@
})
-#define ia64_native_get_pmd(index) \
+#define ia64_get_pmd(index) \
({ \
unsigned long ia64_intri_res; \
asm volatile ("mov %0=pmd[%1]" : "=r"(ia64_intri_res) : "r"(index)); \
ia64_intri_res; \
})
-#define ia64_native_get_rr(index) \
+#define ia64_get_rr(index) \
({ \
unsigned long ia64_intri_res; \
asm volatile ("mov %0=rr[%1]" : "=r"(ia64_intri_res) : "r" (index)); \
ia64_intri_res; \
})
-#define ia64_native_fc(addr) asm volatile ("fc %0" :: "r"(addr) : "memory")
+#define ia64_fc(addr) asm volatile ("fc %0" :: "r"(addr) : "memory")
#define ia64_sync_i() asm volatile (";; sync.i" ::: "memory")
-#define ia64_native_ssm(mask) asm volatile ("ssm %0":: "i"((mask)) : "memory")
-#define ia64_native_rsm(mask) asm volatile ("rsm %0":: "i"((mask)) : "memory")
+#define ia64_ssm(mask) asm volatile ("ssm %0":: "i"((mask)) : "memory")
+#define ia64_rsm(mask) asm volatile ("rsm %0":: "i"((mask)) : "memory")
#define ia64_sum(mask) asm volatile ("sum %0":: "i"((mask)) : "memory")
#define ia64_rum(mask) asm volatile ("rum %0":: "i"((mask)) : "memory")
#define ia64_ptce(addr) asm volatile ("ptc.e %0" :: "r"(addr))
-#define ia64_native_ptcga(addr, size) \
+#define ia64_ptcga(addr, size) \
do { \
asm volatile ("ptc.ga %0,%1" :: "r"(addr), "r"(size) : "memory"); \
ia64_dv_serialize_data(); \
@@ -607,7 +607,7 @@
} \
})
-#define ia64_native_intrin_local_irq_restore(x) \
+#define ia64_intrin_local_irq_restore(x) \
do { \
asm volatile (";; cmp.ne p6,p7=%0,r0;;" \
"(p6) ssm psr.i;" \
diff --git a/arch/ia64/include/uapi/asm/intel_intrin.h b/arch/ia64/include/uapi/asm/intel_intrin.h
index ab64969..dc1884d 100644
--- a/arch/ia64/include/uapi/asm/intel_intrin.h
+++ b/arch/ia64/include/uapi/asm/intel_intrin.h
@@ -17,8 +17,8 @@
* intrinsic
*/
-#define ia64_native_getreg __getReg
-#define ia64_native_setreg __setReg
+#define ia64_getreg __getReg
+#define ia64_setreg __setReg
#define ia64_hint __hint
#define ia64_hint_pause __hint_pause
@@ -40,10 +40,10 @@
#define ia64_invala_fr __invala_fr
#define ia64_nop __nop
#define ia64_sum __sum
-#define ia64_native_ssm __ssm
+#define ia64_ssm __ssm
#define ia64_rum __rum
-#define ia64_native_rsm __rsm
-#define ia64_native_fc __fc
+#define ia64_rsm __rsm
+#define ia64_fc __fc
#define ia64_ldfs __ldfs
#define ia64_ldfd __ldfd
@@ -89,17 +89,17 @@
__setIndReg(_IA64_REG_INDR_PMC, index, val)
#define ia64_set_pmd(index, val) \
__setIndReg(_IA64_REG_INDR_PMD, index, val)
-#define ia64_native_set_rr(index, val) \
+#define ia64_set_rr(index, val) \
__setIndReg(_IA64_REG_INDR_RR, index, val)
-#define ia64_native_get_cpuid(index) \
+#define ia64_get_cpuid(index) \
__getIndReg(_IA64_REG_INDR_CPUID, index)
#define __ia64_get_dbr(index) __getIndReg(_IA64_REG_INDR_DBR, index)
#define ia64_get_ibr(index) __getIndReg(_IA64_REG_INDR_IBR, index)
#define ia64_get_pkr(index) __getIndReg(_IA64_REG_INDR_PKR, index)
#define ia64_get_pmc(index) __getIndReg(_IA64_REG_INDR_PMC, index)
-#define ia64_native_get_pmd(index) __getIndReg(_IA64_REG_INDR_PMD, index)
-#define ia64_native_get_rr(index) __getIndReg(_IA64_REG_INDR_RR, index)
+#define ia64_get_pmd(index) __getIndReg(_IA64_REG_INDR_PMD, index)
+#define ia64_get_rr(index) __getIndReg(_IA64_REG_INDR_RR, index)
#define ia64_srlz_d __dsrlz
#define ia64_srlz_i __isrlz
@@ -121,16 +121,16 @@
#define ia64_ld8_acq __ld8_acq
#define ia64_sync_i __synci
-#define ia64_native_thash __thash
-#define ia64_native_ttag __ttag
+#define ia64_thash __thash
+#define ia64_ttag __ttag
#define ia64_itcd __itcd
#define ia64_itci __itci
#define ia64_itrd __itrd
#define ia64_itri __itri
#define ia64_ptce __ptce
#define ia64_ptcl __ptcl
-#define ia64_native_ptcg __ptcg
-#define ia64_native_ptcga __ptcga
+#define ia64_ptcg __ptcg
+#define ia64_ptcga __ptcga
#define ia64_ptri __ptri
#define ia64_ptrd __ptrd
#define ia64_dep_mi _m64_dep_mi
@@ -147,13 +147,13 @@
#define ia64_lfetch_fault __lfetch_fault
#define ia64_lfetch_fault_excl __lfetch_fault_excl
-#define ia64_native_intrin_local_irq_restore(x) \
+#define ia64_intrin_local_irq_restore(x) \
do { \
if ((x) != 0) { \
- ia64_native_ssm(IA64_PSR_I); \
+ ia64_ssm(IA64_PSR_I); \
ia64_srlz_d(); \
} else { \
- ia64_native_rsm(IA64_PSR_I); \
+ ia64_rsm(IA64_PSR_I); \
} \
} while (0)
diff --git a/arch/ia64/include/uapi/asm/intrinsics.h b/arch/ia64/include/uapi/asm/intrinsics.h
index aecc217..a0e0a06 100644
--- a/arch/ia64/include/uapi/asm/intrinsics.h
+++ b/arch/ia64/include/uapi/asm/intrinsics.h
@@ -21,15 +21,13 @@
#endif
#include <asm/cmpxchg.h>
-#define ia64_native_get_psr_i() (ia64_native_getreg(_IA64_REG_PSR) & IA64_PSR_I)
-
-#define ia64_native_set_rr0_to_rr4(val0, val1, val2, val3, val4) \
+#define ia64_set_rr0_to_rr4(val0, val1, val2, val3, val4) \
do { \
- ia64_native_set_rr(0x0000000000000000UL, (val0)); \
- ia64_native_set_rr(0x2000000000000000UL, (val1)); \
- ia64_native_set_rr(0x4000000000000000UL, (val2)); \
- ia64_native_set_rr(0x6000000000000000UL, (val3)); \
- ia64_native_set_rr(0x8000000000000000UL, (val4)); \
+ ia64_set_rr(0x0000000000000000UL, (val0)); \
+ ia64_set_rr(0x2000000000000000UL, (val1)); \
+ ia64_set_rr(0x4000000000000000UL, (val2)); \
+ ia64_set_rr(0x6000000000000000UL, (val3)); \
+ ia64_set_rr(0x8000000000000000UL, (val4)); \
} while (0)
/*
@@ -85,41 +83,4 @@
#endif
-
-#ifndef __ASSEMBLY__
-
-#define IA64_INTRINSIC_API(name) ia64_native_ ## name
-#define IA64_INTRINSIC_MACRO(name) ia64_native_ ## name
-
-
-/************************************************/
-/* Instructions paravirtualized for correctness */
-/************************************************/
-/* fc, thash, get_cpuid, get_pmd, get_eflags, set_eflags */
-/* Note that "ttag" and "cover" are also privilege-sensitive; "ttag"
- * is not currently used (though it may be in a long-format VHPT system!)
- */
-#define ia64_fc IA64_INTRINSIC_API(fc)
-#define ia64_thash IA64_INTRINSIC_API(thash)
-#define ia64_get_cpuid IA64_INTRINSIC_API(get_cpuid)
-#define ia64_get_pmd IA64_INTRINSIC_API(get_pmd)
-
-
-/************************************************/
-/* Instructions paravirtualized for performance */
-/************************************************/
-#define ia64_ssm IA64_INTRINSIC_MACRO(ssm)
-#define ia64_rsm IA64_INTRINSIC_MACRO(rsm)
-#define ia64_getreg IA64_INTRINSIC_MACRO(getreg)
-#define ia64_setreg IA64_INTRINSIC_API(setreg)
-#define ia64_set_rr IA64_INTRINSIC_API(set_rr)
-#define ia64_get_rr IA64_INTRINSIC_API(get_rr)
-#define ia64_ptcga IA64_INTRINSIC_API(ptcga)
-#define ia64_get_psr_i IA64_INTRINSIC_API(get_psr_i)
-#define ia64_intrin_local_irq_restore \
- IA64_INTRINSIC_API(intrin_local_irq_restore)
-#define ia64_set_rr0_to_rr4 IA64_INTRINSIC_API(set_rr0_to_rr4)
-
-#endif /* !__ASSEMBLY__ */
-
#endif /* _UAPI_ASM_IA64_INTRINSICS_H */
diff --git a/arch/ia64/include/uapi/asm/ioctl.h b/arch/ia64/include/uapi/asm/ioctl.h
deleted file mode 100644
index b809c45..0000000
--- a/arch/ia64/include/uapi/asm/ioctl.h
+++ /dev/null
@@ -1,2 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
-#include <asm-generic/ioctl.h>
diff --git a/arch/ia64/include/uapi/asm/ioctls.h b/arch/ia64/include/uapi/asm/ioctls.h
deleted file mode 100644
index b860019..0000000
--- a/arch/ia64/include/uapi/asm/ioctls.h
+++ /dev/null
@@ -1,7 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
-#ifndef _ASM_IA64_IOCTLS_H
-#define _ASM_IA64_IOCTLS_H
-
-#include <asm-generic/ioctls.h>
-
-#endif /* _ASM_IA64_IOCTLS_H */
diff --git a/arch/ia64/kernel/.gitignore b/arch/ia64/kernel/.gitignore
index 21cb0da..0374827 100644
--- a/arch/ia64/kernel/.gitignore
+++ b/arch/ia64/kernel/.gitignore
@@ -1,2 +1,3 @@
+# SPDX-License-Identifier: GPL-2.0-only
gate.lds
vmlinux.lds
diff --git a/arch/ia64/kernel/Makefile b/arch/ia64/kernel/Makefile
index 18d6008..c89bd5f 100644
--- a/arch/ia64/kernel/Makefile
+++ b/arch/ia64/kernel/Makefile
@@ -10,7 +10,7 @@
extra-y := head.o vmlinux.lds
obj-y := entry.o efi.o efi_stub.o gate-data.o fsys.o ia64_ksyms.o irq.o irq_ia64.o \
- irq_lsapic.o ivt.o pal.o patch.o process.o perfmon.o ptrace.o sal.o \
+ irq_lsapic.o ivt.o pal.o patch.o process.o ptrace.o sal.o \
salinfo.o setup.o signal.o sys_ia64.o time.o traps.o unaligned.o \
unwind.o mca.o mca_asm.o topology.o dma-mapping.o iosapic.o acpi.o \
acpi-ext.o
@@ -21,7 +21,6 @@
obj-$(CONFIG_MODULES) += module.o
obj-$(CONFIG_SMP) += smp.o smpboot.o
obj-$(CONFIG_NUMA) += numa.o
-obj-$(CONFIG_PERFMON) += perfmon_default_smpl.o
obj-$(CONFIG_IA64_CYCLONE) += cyclone.o
obj-$(CONFIG_IA64_MCA_RECOVERY) += mca_recovery.o
obj-$(CONFIG_KPROBES) += kprobes.o
diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c
index 70d1587..e2af6b1 100644
--- a/arch/ia64/kernel/acpi.c
+++ b/arch/ia64/kernel/acpi.c
@@ -42,8 +42,6 @@
unsigned int acpi_cpei_override;
unsigned int acpi_cpei_phys_cpuid;
-unsigned long acpi_wakeup_address = 0;
-
#define ACPI_MAX_PLATFORM_INTERRUPTS 256
/* Array to record platform interrupt vectors for generic interrupt routing. */
@@ -448,7 +446,8 @@
if (srat_num_cpus == 0) {
node_set_online(0);
node_cpuid[0].phys_id = hard_smp_processor_id();
- return;
+ slit_distance(0, 0) = LOCAL_DISTANCE;
+ goto out;
}
/*
@@ -491,7 +490,7 @@
for (j = 0; j < MAX_NUMNODES; j++)
slit_distance(i, j) = i == j ?
LOCAL_DISTANCE : REMOTE_DISTANCE;
- return;
+ goto out;
}
memset(numa_slit, -1, sizeof(numa_slit));
@@ -516,6 +515,8 @@
printk("\n");
}
#endif
+out:
+ node_possible_map = node_online_map;
}
#endif /* CONFIG_ACPI_NUMA */
diff --git a/arch/ia64/kernel/asm-offsets.c b/arch/ia64/kernel/asm-offsets.c
index 00e8e2a..fb0deb8 100644
--- a/arch/ia64/kernel/asm-offsets.c
+++ b/arch/ia64/kernel/asm-offsets.c
@@ -211,7 +211,7 @@
offsetof (struct cpuinfo_ia64, ptce_stride));
BLANK();
DEFINE(IA64_TIMESPEC_TV_NSEC_OFFSET,
- offsetof (struct timespec, tv_nsec));
+ offsetof (struct __kernel_old_timespec, tv_nsec));
DEFINE(IA64_TIME_SN_SPEC_SNSEC_OFFSET,
offsetof (struct time_sn_spec, snsec));
diff --git a/arch/ia64/kernel/crash.c b/arch/ia64/kernel/crash.c
index bec762a..fec70d6 100644
--- a/arch/ia64/kernel/crash.c
+++ b/arch/ia64/kernel/crash.c
@@ -163,7 +163,7 @@
case DIE_INIT_MONARCH_LEAVE:
if (!kdump_freeze_monarch)
break;
- /* fall through */
+ fallthrough;
case DIE_INIT_SLAVE_LEAVE:
case DIE_INIT_MONARCH_ENTER:
case DIE_MCA_RENDZVOUS_LEAVE:
diff --git a/arch/ia64/kernel/cyclone.c b/arch/ia64/kernel/cyclone.c
index f80eb7f..258d7b7 100644
--- a/arch/ia64/kernel/cyclone.c
+++ b/arch/ia64/kernel/cyclone.c
@@ -50,7 +50,7 @@
/* find base address */
offset = (CYCLONE_CBAR_ADDR);
- reg = ioremap_nocache(offset, sizeof(u64));
+ reg = ioremap(offset, sizeof(u64));
if(!reg){
printk(KERN_ERR "Summit chipset: Could not find valid CBAR"
" register.\n");
@@ -68,7 +68,7 @@
/* setup PMCC */
offset = (base + CYCLONE_PMCC_OFFSET);
- reg = ioremap_nocache(offset, sizeof(u64));
+ reg = ioremap(offset, sizeof(u64));
if(!reg){
printk(KERN_ERR "Summit chipset: Could not find valid PMCC"
" register.\n");
@@ -80,7 +80,7 @@
/* setup MPCS */
offset = (base + CYCLONE_MPCS_OFFSET);
- reg = ioremap_nocache(offset, sizeof(u64));
+ reg = ioremap(offset, sizeof(u64));
if(!reg){
printk(KERN_ERR "Summit chipset: Could not find valid MPCS"
" register.\n");
@@ -92,7 +92,7 @@
/* map in cyclone_timer */
offset = (base + CYCLONE_MPMC_OFFSET);
- cyclone_timer = ioremap_nocache(offset, sizeof(u32));
+ cyclone_timer = ioremap(offset, sizeof(u32));
if(!cyclone_timer){
printk(KERN_ERR "Summit chipset: Could not find valid MPMC"
" register.\n");
diff --git a/arch/ia64/kernel/dma-mapping.c b/arch/ia64/kernel/dma-mapping.c
index 4a32627..cd0c166 100644
--- a/arch/ia64/kernel/dma-mapping.c
+++ b/arch/ia64/kernel/dma-mapping.c
@@ -1,5 +1,5 @@
// SPDX-License-Identifier: GPL-2.0
-#include <linux/dma-direct.h>
+#include <linux/dma-map-ops.h>
#include <linux/export.h>
/* Set this to 1 if there is a HW IOMMU in the system */
@@ -7,21 +7,3 @@
const struct dma_map_ops *dma_ops;
EXPORT_SYMBOL(dma_ops);
-
-void *arch_dma_alloc(struct device *dev, size_t size,
- dma_addr_t *dma_handle, gfp_t gfp, unsigned long attrs)
-{
- return dma_direct_alloc_pages(dev, size, dma_handle, gfp, attrs);
-}
-
-void arch_dma_free(struct device *dev, size_t size, void *cpu_addr,
- dma_addr_t dma_addr, unsigned long attrs)
-{
- dma_direct_free_pages(dev, size, cpu_addr, dma_addr, attrs);
-}
-
-long arch_dma_coherent_to_pfn(struct device *dev, void *cpu_addr,
- dma_addr_t dma_addr)
-{
- return page_to_pfn(virt_to_page(cpu_addr));
-}
diff --git a/arch/ia64/kernel/efi.c b/arch/ia64/kernel/efi.c
index 0a34dcc..33282f3 100644
--- a/arch/ia64/kernel/efi.c
+++ b/arch/ia64/kernel/efi.c
@@ -37,7 +37,6 @@
#include <asm/io.h>
#include <asm/kregs.h>
#include <asm/meminit.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/mca.h>
#include <asm/setup.h>
@@ -45,14 +44,24 @@
#define EFI_DEBUG 0
+#define ESI_TABLE_GUID \
+ EFI_GUID(0x43EA58DC, 0xCF28, 0x4b06, 0xB3, \
+ 0x91, 0xB7, 0x50, 0x59, 0x34, 0x2B, 0xD4)
+
+static unsigned long mps_phys = EFI_INVALID_TABLE_ADDR;
static __initdata unsigned long palo_phys;
+unsigned long __initdata esi_phys = EFI_INVALID_TABLE_ADDR;
+unsigned long hcdp_phys = EFI_INVALID_TABLE_ADDR;
unsigned long sal_systab_phys = EFI_INVALID_TABLE_ADDR;
-static __initdata efi_config_table_type_t arch_tables[] = {
- {PROCESSOR_ABSTRACTION_LAYER_OVERWRITE_GUID, "PALO", &palo_phys},
- {SAL_SYSTEM_TABLE_GUID, "SALsystab", &sal_systab_phys},
- {NULL_GUID, NULL, 0},
+static const efi_config_table_type_t arch_tables[] __initconst = {
+ {ESI_TABLE_GUID, &esi_phys, "ESI" },
+ {HCDP_TABLE_GUID, &hcdp_phys, "HCDP" },
+ {MPS_TABLE_GUID, &mps_phys, "MPS" },
+ {PROCESSOR_ABSTRACTION_LAYER_OVERWRITE_GUID, &palo_phys, "PALO" },
+ {SAL_SYSTEM_TABLE_GUID, &sal_systab_phys, "SALsystab" },
+ {},
};
extern efi_status_t efi_call_phys (void *, ...);
@@ -404,10 +413,10 @@
mask = ~((1 << IA64_GRANULE_SHIFT) - 1);
printk(KERN_INFO "CPU %d: mapping PAL code "
- "[0x%lx-0x%lx) into [0x%lx-0x%lx)\n",
- smp_processor_id(), md->phys_addr,
- md->phys_addr + efi_md_size(md),
- vaddr & mask, (vaddr & mask) + IA64_GRANULE_SIZE);
+ "[0x%llx-0x%llx) into [0x%llx-0x%llx)\n",
+ smp_processor_id(), md->phys_addr,
+ md->phys_addr + efi_md_size(md),
+ vaddr & mask, (vaddr & mask) + IA64_GRANULE_SIZE);
#endif
return __va(md->phys_addr);
}
@@ -474,11 +483,10 @@
void __init
efi_init (void)
{
+ const efi_system_table_t *efi_systab;
void *efi_map_start, *efi_map_end;
- efi_char16_t *c16;
u64 efi_desc_size;
- char *cp, vendor[100] = "unknown";
- int i;
+ char *cp;
set_bit(EFI_BOOT, &efi.flags);
set_bit(EFI_64BIT, &efi.flags);
@@ -508,42 +516,29 @@
printk(KERN_INFO "Ignoring memory above %lluMB\n",
max_addr >> 20);
- efi.systab = __va(ia64_boot_param->efi_systab);
+ efi_systab = __va(ia64_boot_param->efi_systab);
/*
* Verify the EFI Table
*/
- if (efi.systab == NULL)
+ if (efi_systab == NULL)
panic("Whoa! Can't find EFI system table.\n");
- if (efi.systab->hdr.signature != EFI_SYSTEM_TABLE_SIGNATURE)
+ if (efi_systab_check_header(&efi_systab->hdr, 1))
panic("Whoa! EFI system table signature incorrect\n");
- if ((efi.systab->hdr.revision >> 16) == 0)
- printk(KERN_WARNING "Warning: EFI system table version "
- "%d.%02d, expected 1.00 or greater\n",
- efi.systab->hdr.revision >> 16,
- efi.systab->hdr.revision & 0xffff);
- /* Show what we know for posterity */
- c16 = __va(efi.systab->fw_vendor);
- if (c16) {
- for (i = 0;i < (int) sizeof(vendor) - 1 && *c16; ++i)
- vendor[i] = *c16++;
- vendor[i] = '\0';
- }
-
- printk(KERN_INFO "EFI v%u.%.02u by %s:",
- efi.systab->hdr.revision >> 16,
- efi.systab->hdr.revision & 0xffff, vendor);
+ efi_systab_report_header(&efi_systab->hdr, efi_systab->fw_vendor);
palo_phys = EFI_INVALID_TABLE_ADDR;
- if (efi_config_init(arch_tables) != 0)
+ if (efi_config_parse_tables(__va(efi_systab->tables),
+ efi_systab->nr_tables,
+ arch_tables) != 0)
return;
if (palo_phys != EFI_INVALID_TABLE_ADDR)
handle_palo(palo_phys);
- runtime = __va(efi.systab->runtime);
+ runtime = __va(efi_systab->runtime);
efi.get_time = phys_get_time;
efi.set_time = phys_set_time;
efi.get_wakeup_time = phys_get_wakeup_time;
@@ -563,6 +558,7 @@
{
efi_memory_desc_t *md;
void *p;
+ unsigned int i;
for (i = 0, p = efi_map_start; p < efi_map_end;
++i, p += efi_desc_size)
@@ -589,7 +585,7 @@
}
printk("mem%02d: %s "
- "range=[0x%016lx-0x%016lx) (%4lu%s)\n",
+ "range=[0x%016llx-0x%016llx) (%4lu%s)\n",
i, efi_md_typeattr_format(buf, sizeof(buf), md),
md->phys_addr,
md->phys_addr + efi_md_size(md), size, unit);
@@ -1351,3 +1347,12 @@
return ret;
}
#endif
+
+char *efi_systab_show_arch(char *str)
+{
+ if (mps_phys != EFI_INVALID_TABLE_ADDR)
+ str += sprintf(str, "MPS=0x%lx\n", mps_phys);
+ if (hcdp_phys != EFI_INVALID_TABLE_ADDR)
+ str += sprintf(str, "HCDP=0x%lx\n", hcdp_phys);
+ return str;
+}
diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S
index a9992be..e98e3da 100644
--- a/arch/ia64/kernel/entry.S
+++ b/arch/ia64/kernel/entry.S
@@ -38,12 +38,12 @@
*/
+#include <linux/pgtable.h>
#include <asm/asmmacro.h>
#include <asm/cache.h>
#include <asm/errno.h>
#include <asm/kregs.h>
#include <asm/asm-offsets.h>
-#include <asm/pgtable.h>
#include <asm/percpu.h>
#include <asm/processor.h>
#include <asm/thread_info.h>
@@ -112,19 +112,16 @@
.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
alloc r16=ar.pfs,8,2,6,0
DO_SAVE_SWITCH_STACK
- adds r2=PT(R16)+IA64_SWITCH_STACK_SIZE+16,sp
mov loc0=rp
- mov loc1=r16 // save ar.pfs across do_fork
+ mov loc1=r16 // save ar.pfs across ia64_clone
.body
+ mov out0=in0
mov out1=in1
mov out2=in2
- tbit.nz p6,p0=in0,CLONE_SETTLS_BIT
- mov out3=in3 // parent_tidptr: valid only w/CLONE_PARENT_SETTID
- ;;
-(p6) st8 [r2]=in5 // store TLS in r16 for copy_thread()
- mov out4=in4 // child_tidptr: valid only w/CLONE_CHILD_SETTID or CLONE_CHILD_CLEARTID
- mov out0=in0 // out0 = clone_flags
- br.call.sptk.many rp=do_fork
+ mov out3=in3
+ mov out4=in4
+ mov out5=in5
+ br.call.sptk.many rp=ia64_clone
.ret1: .restore sp
adds sp=IA64_SWITCH_STACK_SIZE,sp // pop the switch stack
mov ar.pfs=loc1
@@ -143,19 +140,16 @@
.prologue ASM_UNW_PRLG_RP|ASM_UNW_PRLG_PFS, ASM_UNW_PRLG_GRSAVE(8)
alloc r16=ar.pfs,8,2,6,0
DO_SAVE_SWITCH_STACK
- adds r2=PT(R16)+IA64_SWITCH_STACK_SIZE+16,sp
mov loc0=rp
- mov loc1=r16 // save ar.pfs across do_fork
+ mov loc1=r16 // save ar.pfs across ia64_clone
.body
+ mov out0=in0
mov out1=in1
mov out2=16 // stacksize (compensates for 16-byte scratch area)
- tbit.nz p6,p0=in0,CLONE_SETTLS_BIT
- mov out3=in2 // parent_tidptr: valid only w/CLONE_PARENT_SETTID
- ;;
-(p6) st8 [r2]=in4 // store TLS in r13 (tp)
- mov out4=in3 // child_tidptr: valid only w/CLONE_CHILD_SETTID or CLONE_CHILD_CLEARTID
- mov out0=in0 // out0 = clone_flags
- br.call.sptk.many rp=do_fork
+ mov out3=in3
+ mov out4=in4
+ mov out5=in5
+ br.call.sptk.many rp=ia64_clone
.ret2: .restore sp
adds sp=IA64_SWITCH_STACK_SIZE,sp // pop the switch stack
mov ar.pfs=loc1
@@ -590,7 +584,7 @@
nop.i 0
/*
* We need to call schedule_tail() to complete the scheduling process.
- * Called by ia64_switch_to() after do_fork()->copy_thread(). r8 contains the
+ * Called by ia64_switch_to() after ia64_clone()->copy_thread(). r8 contains the
* address of the previously executing task.
*/
br.call.sptk.many rp=ia64_invoke_schedule_tail
@@ -670,12 +664,12 @@
*
* p6 controls whether current_thread_info()->flags needs to be check for
* extra work. We always check for extra work when returning to user-level.
- * With CONFIG_PREEMPT, we also check for extra work when the preempt_count
+ * With CONFIG_PREEMPTION, we also check for extra work when the preempt_count
* is 0. After extra work processing has been completed, execution
* resumes at ia64_work_processed_syscall with p6 set to 1 if the extra-work-check
* needs to be redone.
*/
-#ifdef CONFIG_PREEMPT
+#ifdef CONFIG_PREEMPTION
RSM_PSR_I(p0, r2, r18) // disable interrupts
cmp.eq pLvSys,p0=r0,r0 // pLvSys=1: leave from syscall
(pKStk) adds r20=TI_PRE_COUNT+IA64_TASK_SIZE,r13
@@ -685,7 +679,7 @@
(pUStk) mov r21=0 // r21 <- 0
;;
cmp.eq p6,p0=r21,r0 // p6 <- pUStk || (preempt_count == 0)
-#else /* !CONFIG_PREEMPT */
+#else /* !CONFIG_PREEMPTION */
RSM_PSR_I(pUStk, r2, r18)
cmp.eq pLvSys,p0=r0,r0 // pLvSys=1: leave from syscall
(pUStk) cmp.eq.unc p6,p0=r0,r0 // p6 <- pUStk
@@ -814,12 +808,12 @@
*
* p6 controls whether current_thread_info()->flags needs to be check for
* extra work. We always check for extra work when returning to user-level.
- * With CONFIG_PREEMPT, we also check for extra work when the preempt_count
+ * With CONFIG_PREEMPTION, we also check for extra work when the preempt_count
* is 0. After extra work processing has been completed, execution
* resumes at .work_processed_syscall with p6 set to 1 if the extra-work-check
* needs to be redone.
*/
-#ifdef CONFIG_PREEMPT
+#ifdef CONFIG_PREEMPTION
RSM_PSR_I(p0, r17, r31) // disable interrupts
cmp.eq p0,pLvSys=r0,r0 // pLvSys=0: leave from kernel
(pKStk) adds r20=TI_PRE_COUNT+IA64_TASK_SIZE,r13
@@ -1120,7 +1114,7 @@
/*
* On entry:
- * r20 = ¤t->thread_info->pre_count (if CONFIG_PREEMPT)
+ * r20 = ¤t->thread_info->pre_count (if CONFIG_PREEMPTION)
* r31 = current->thread_info->flags
* On exit:
* p6 = TRUE if work-pending-check needs to be redone
diff --git a/arch/ia64/kernel/esi.c b/arch/ia64/kernel/esi.c
index cb51412..4df57c9 100644
--- a/arch/ia64/kernel/esi.c
+++ b/arch/ia64/kernel/esi.c
@@ -19,10 +19,6 @@
#define MODULE_NAME "esi"
-#define ESI_TABLE_GUID \
- EFI_GUID(0x43EA58DC, 0xCF28, 0x4b06, 0xB3, \
- 0x91, 0xB7, 0x50, 0x59, 0x34, 0x2B, 0xD4)
-
enum esi_systab_entry_type {
ESI_DESC_ENTRY_POINT = 0
};
@@ -48,27 +44,18 @@
static struct ia64_sal_systab *esi_systab;
+extern unsigned long esi_phys;
+
static int __init esi_init (void)
{
- efi_config_table_t *config_tables;
struct ia64_sal_systab *systab;
- unsigned long esi = 0;
char *p;
int i;
- config_tables = __va(efi.systab->tables);
-
- for (i = 0; i < (int) efi.systab->nr_tables; ++i) {
- if (efi_guidcmp(config_tables[i].guid, ESI_TABLE_GUID) == 0) {
- esi = config_tables[i].table;
- break;
- }
- }
-
- if (!esi)
+ if (esi_phys == EFI_INVALID_TABLE_ADDR)
return -ENODEV;
- systab = __va(esi);
+ systab = __va(esi_phys);
if (strncmp(systab->signature, "ESIT", 4) != 0) {
printk(KERN_ERR "bad signature in ESI system table!");
diff --git a/arch/ia64/kernel/ftrace.c b/arch/ia64/kernel/ftrace.c
index cee411e..b2ab2d5 100644
--- a/arch/ia64/kernel/ftrace.c
+++ b/arch/ia64/kernel/ftrace.c
@@ -108,7 +108,7 @@
goto skip_check;
/* read the text we want to modify */
- if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
+ if (copy_from_kernel_nofault(replaced, (void *)ip, MCOUNT_INSN_SIZE))
return -EFAULT;
/* Make sure it is what we expect it to be */
@@ -117,7 +117,7 @@
skip_check:
/* replace the text with the new text */
- if (probe_kernel_write(((void *)ip), new_code, MCOUNT_INSN_SIZE))
+ if (copy_to_kernel_nofault(((void *)ip), new_code, MCOUNT_INSN_SIZE))
return -EPERM;
flush_icache_range(ip, ip + MCOUNT_INSN_SIZE);
@@ -129,7 +129,7 @@
unsigned char __attribute__((aligned(8))) replaced[MCOUNT_INSN_SIZE];
unsigned long ip = rec->ip;
- if (probe_kernel_read(replaced, (void *)ip, MCOUNT_INSN_SIZE))
+ if (copy_from_kernel_nofault(replaced, (void *)ip, MCOUNT_INSN_SIZE))
return -EFAULT;
if (rec->flags & FTRACE_FL_CONVERTED) {
struct ftrace_call_insn *call_insn, *tmp_call;
diff --git a/arch/ia64/kernel/head.S b/arch/ia64/kernel/head.S
index e6f4517..30f1ef7 100644
--- a/arch/ia64/kernel/head.S
+++ b/arch/ia64/kernel/head.S
@@ -21,18 +21,19 @@
*/
+#include <linux/pgtable.h>
#include <asm/asmmacro.h>
#include <asm/fpu.h>
#include <asm/kregs.h>
#include <asm/mmu_context.h>
#include <asm/asm-offsets.h>
#include <asm/pal.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/mca_asm.h>
#include <linux/init.h>
#include <linux/linkage.h>
+#include <linux/pgtable.h>
#include <asm/export.h>
#ifdef CONFIG_HOTPLUG_CPU
diff --git a/arch/ia64/kernel/ia64_ksyms.c b/arch/ia64/kernel/ia64_ksyms.c
index b49fe6f..f8150ee 100644
--- a/arch/ia64/kernel/ia64_ksyms.c
+++ b/arch/ia64/kernel/ia64_ksyms.c
@@ -3,7 +3,7 @@
* Architecture-specific kernel symbols
*/
-#ifdef CONFIG_VIRTUAL_MEM_MAP
+#if defined(CONFIG_VIRTUAL_MEM_MAP) || defined(CONFIG_DISCONTIGMEM)
#include <linux/compiler.h>
#include <linux/export.h>
#include <linux/memblock.h>
diff --git a/arch/ia64/kernel/iosapic.c b/arch/ia64/kernel/iosapic.c
index fad4db2..35adcf8 100644
--- a/arch/ia64/kernel/iosapic.c
+++ b/arch/ia64/kernel/iosapic.c
@@ -95,6 +95,7 @@
#include <asm/iosapic.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
+#include <asm/xtp.h>
#undef DEBUG_INTERRUPT_ROUTING
diff --git a/arch/ia64/kernel/irq.c b/arch/ia64/kernel/irq.c
index 0a8e5e5..ecef17c 100644
--- a/arch/ia64/kernel/irq.c
+++ b/arch/ia64/kernel/irq.c
@@ -25,6 +25,7 @@
#include <linux/kernel_stat.h>
#include <asm/mca.h>
+#include <asm/xtp.h>
/*
* 'what should we do if we get a hw irq event on an illegal vector'.
diff --git a/arch/ia64/kernel/irq.h b/arch/ia64/kernel/irq.h
new file mode 100644
index 0000000..4d16f3c
--- /dev/null
+++ b/arch/ia64/kernel/irq.h
@@ -0,0 +1,3 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+extern void register_percpu_irq(ia64_vector vec, irq_handler_t handler,
+ unsigned long flags, const char *name);
diff --git a/arch/ia64/kernel/irq_ia64.c b/arch/ia64/kernel/irq_ia64.c
index 8e91c86..46e33c5 100644
--- a/arch/ia64/kernel/irq_ia64.c
+++ b/arch/ia64/kernel/irq_ia64.c
@@ -16,6 +16,7 @@
*/
#include <linux/module.h>
+#include <linux/pgtable.h>
#include <linux/jiffies.h>
#include <linux/errno.h>
@@ -37,13 +38,8 @@
#include <asm/intrinsics.h>
#include <asm/io.h>
#include <asm/hw_irq.h>
-#include <asm/pgtable.h>
#include <asm/tlbflush.h>
-#ifdef CONFIG_PERFMON
-# include <asm/perfmon.h>
-#endif
-
#define IRQ_DEBUG 0
#define IRQ_VECTOR_UNASSIGNED (0)
@@ -351,11 +347,6 @@
return IRQ_HANDLED;
}
-static struct irqaction irq_move_irqaction = {
- .handler = smp_irq_move_cleanup_interrupt,
- .name = "irq_move"
-};
-
static int __init parse_vector_domain(char *arg)
{
if (!arg)
@@ -586,28 +577,15 @@
return IRQ_NONE;
}
-static struct irqaction ipi_irqaction = {
- .handler = handle_IPI,
- .name = "IPI"
-};
-
/*
* KVM uses this interrupt to force a cpu out of guest mode
*/
-static struct irqaction resched_irqaction = {
- .handler = dummy_handler,
- .name = "resched"
-};
-
-static struct irqaction tlb_irqaction = {
- .handler = dummy_handler,
- .name = "tlb_flush"
-};
#endif
void
-ia64_native_register_percpu_irq (ia64_vector vec, struct irqaction *action)
+register_percpu_irq(ia64_vector vec, irq_handler_t handler, unsigned long flags,
+ const char *name)
{
unsigned int irq;
@@ -615,8 +593,9 @@
BUG_ON(bind_irq_vector(irq, vec, CPU_MASK_ALL));
irq_set_status_flags(irq, IRQ_PER_CPU);
irq_set_chip(irq, &irq_type_ia64_lsapic);
- if (action)
- setup_irq(irq, action);
+ if (handler)
+ if (request_irq(irq, handler, flags, name, NULL))
+ pr_err("Failed to request irq %u (%s)\n", irq, name);
irq_set_handler(irq, handle_percpu_irq);
}
@@ -624,9 +603,10 @@
ia64_native_register_ipi(void)
{
#ifdef CONFIG_SMP
- register_percpu_irq(IA64_IPI_VECTOR, &ipi_irqaction);
- register_percpu_irq(IA64_IPI_RESCHEDULE, &resched_irqaction);
- register_percpu_irq(IA64_IPI_LOCAL_TLB_FLUSH, &tlb_irqaction);
+ register_percpu_irq(IA64_IPI_VECTOR, handle_IPI, 0, "IPI");
+ register_percpu_irq(IA64_IPI_RESCHEDULE, dummy_handler, 0, "resched");
+ register_percpu_irq(IA64_IPI_LOCAL_TLB_FLUSH, dummy_handler, 0,
+ "tlb_flush");
#endif
}
@@ -635,13 +615,13 @@
{
acpi_boot_init();
ia64_register_ipi();
- register_percpu_irq(IA64_SPURIOUS_INT_VECTOR, NULL);
+ register_percpu_irq(IA64_SPURIOUS_INT_VECTOR, NULL, 0, NULL);
#ifdef CONFIG_SMP
- if (vector_domain_type != VECTOR_DOMAIN_NONE)
- register_percpu_irq(IA64_IRQ_MOVE_VECTOR, &irq_move_irqaction);
-#endif
-#ifdef CONFIG_PERFMON
- pfm_init_percpu();
+ if (vector_domain_type != VECTOR_DOMAIN_NONE) {
+ register_percpu_irq(IA64_IRQ_MOVE_VECTOR,
+ smp_irq_move_cleanup_interrupt, 0,
+ "irq_move");
+ }
#endif
}
diff --git a/arch/ia64/kernel/ivt.S b/arch/ia64/kernel/ivt.S
index 1efcbe5..d6d4229 100644
--- a/arch/ia64/kernel/ivt.S
+++ b/arch/ia64/kernel/ivt.S
@@ -48,11 +48,11 @@
*/
+#include <linux/pgtable.h>
#include <asm/asmmacro.h>
#include <asm/break.h>
#include <asm/kregs.h>
#include <asm/asm-offsets.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/thread_info.h>
diff --git a/arch/ia64/kernel/kprobes.c b/arch/ia64/kernel/kprobes.c
index b3dc390..ca4b4fa 100644
--- a/arch/ia64/kernel/kprobes.c
+++ b/arch/ia64/kernel/kprobes.c
@@ -17,8 +17,8 @@
#include <linux/preempt.h>
#include <linux/extable.h>
#include <linux/kdebug.h>
+#include <linux/pgtable.h>
-#include <asm/pgtable.h>
#include <asm/sections.h>
#include <asm/exception.h>
@@ -398,7 +398,8 @@
int __kprobes trampoline_probe_handler(struct kprobe *p, struct pt_regs *regs)
{
- regs->cr_iip = __kretprobe_trampoline_handler(regs, kretprobe_trampoline, NULL);
+ regs->cr_iip = __kretprobe_trampoline_handler(regs,
+ dereference_function_descriptor(kretprobe_trampoline), NULL);
/*
* By returning a non-zero value, we are telling
* kprobe_handler() that we don't want the post_handler
@@ -414,7 +415,7 @@
ri->fp = NULL;
/* Replace the return addr with trampoline addr */
- regs->b0 = ((struct fnptr *)kretprobe_trampoline)->ip;
+ regs->b0 = (unsigned long)dereference_function_descriptor(kretprobe_trampoline);
}
/* Check the instruction in the slot is break */
@@ -768,7 +769,7 @@
return 1;
}
-#if !defined(CONFIG_PREEMPT)
+#if !defined(CONFIG_PREEMPTION)
if (p->ainsn.inst_flag == INST_FLAG_BOOSTABLE && !p->post_handler) {
/* Boost up -- we can execute copied instructions directly */
ia64_psr(regs)->ri = p->ainsn.slot;
@@ -918,14 +919,14 @@
int __init arch_init_kprobes(void)
{
trampoline_p.addr =
- (kprobe_opcode_t *)((struct fnptr *)kretprobe_trampoline)->ip;
+ dereference_function_descriptor(kretprobe_trampoline);
return register_kprobe(&trampoline_p);
}
int __kprobes arch_trampoline_kprobe(struct kprobe *p)
{
if (p->addr ==
- (kprobe_opcode_t *)((struct fnptr *)kretprobe_trampoline)->ip)
+ dereference_function_descriptor(kretprobe_trampoline))
return 1;
return 0;
diff --git a/arch/ia64/kernel/mca.c b/arch/ia64/kernel/mca.c
index d96c681..bd0a51d 100644
--- a/arch/ia64/kernel/mca.c
+++ b/arch/ia64/kernel/mca.c
@@ -104,6 +104,7 @@
#include "mca_drv.h"
#include "entry.h"
+#include "irq.h"
#if defined(IA64_MCA_DEBUG_INFO)
# define IA64_MCA_DEBUG(fmt...) printk(fmt)
@@ -1630,7 +1631,7 @@
if (read_trylock(&tasklist_lock)) {
do_each_thread (g, t) {
printk("\nBacktrace of pid %d (%s)\n", t->pid, t->comm);
- show_stack(t, NULL);
+ show_stack(t, NULL, KERN_DEFAULT);
} while_each_thread (g, t);
read_unlock(&tasklist_lock);
}
@@ -1766,36 +1767,6 @@
__setup("disable_cpe_poll", ia64_mca_disable_cpe_polling);
-static struct irqaction cmci_irqaction = {
- .handler = ia64_mca_cmc_int_handler,
- .name = "cmc_hndlr"
-};
-
-static struct irqaction cmcp_irqaction = {
- .handler = ia64_mca_cmc_int_caller,
- .name = "cmc_poll"
-};
-
-static struct irqaction mca_rdzv_irqaction = {
- .handler = ia64_mca_rendez_int_handler,
- .name = "mca_rdzv"
-};
-
-static struct irqaction mca_wkup_irqaction = {
- .handler = ia64_mca_wakeup_int_handler,
- .name = "mca_wkup"
-};
-
-static struct irqaction mca_cpe_irqaction = {
- .handler = ia64_mca_cpe_int_handler,
- .name = "cpe_hndlr"
-};
-
-static struct irqaction mca_cpep_irqaction = {
- .handler = ia64_mca_cpe_int_caller,
- .name = "cpe_poll"
-};
-
/* Minimal format of the MCA/INIT stacks. The pseudo processes that run on
* these stacks can never sleep, they cannot return from the kernel to user
* space, they do not appear in a normal ps listing. So there is no need to
@@ -2056,18 +2027,23 @@
* Configure the CMCI/P vector and handler. Interrupts for CMC are
* per-processor, so AP CMC interrupts are setup in smp_callin() (smpboot.c).
*/
- register_percpu_irq(IA64_CMC_VECTOR, &cmci_irqaction);
- register_percpu_irq(IA64_CMCP_VECTOR, &cmcp_irqaction);
+ register_percpu_irq(IA64_CMC_VECTOR, ia64_mca_cmc_int_handler, 0,
+ "cmc_hndlr");
+ register_percpu_irq(IA64_CMCP_VECTOR, ia64_mca_cmc_int_caller, 0,
+ "cmc_poll");
ia64_mca_cmc_vector_setup(); /* Setup vector on BSP */
/* Setup the MCA rendezvous interrupt vector */
- register_percpu_irq(IA64_MCA_RENDEZ_VECTOR, &mca_rdzv_irqaction);
+ register_percpu_irq(IA64_MCA_RENDEZ_VECTOR, ia64_mca_rendez_int_handler,
+ 0, "mca_rdzv");
/* Setup the MCA wakeup interrupt vector */
- register_percpu_irq(IA64_MCA_WAKEUP_VECTOR, &mca_wkup_irqaction);
+ register_percpu_irq(IA64_MCA_WAKEUP_VECTOR, ia64_mca_wakeup_int_handler,
+ 0, "mca_wkup");
/* Setup the CPEI/P handler */
- register_percpu_irq(IA64_CPEP_VECTOR, &mca_cpep_irqaction);
+ register_percpu_irq(IA64_CPEP_VECTOR, ia64_mca_cpe_int_caller, 0,
+ "cpe_poll");
}
/*
@@ -2108,7 +2084,9 @@
if (irq > 0) {
cpe_poll_enabled = 0;
irq_set_status_flags(irq, IRQ_PER_CPU);
- setup_irq(irq, &mca_cpe_irqaction);
+ if (request_irq(irq, ia64_mca_cpe_int_handler,
+ 0, "cpe_hndlr", NULL))
+ pr_err("Failed to register cpe_hndlr interrupt\n");
ia64_cpe_irq = irq;
ia64_mca_register_cpev(cpe_vector);
IA64_MCA_DEBUG("%s: CPEI/P setup and enabled.\n",
diff --git a/arch/ia64/kernel/mca_asm.S b/arch/ia64/kernel/mca_asm.S
index 086cfa4..0d6b8cf 100644
--- a/arch/ia64/kernel/mca_asm.S
+++ b/arch/ia64/kernel/mca_asm.S
@@ -25,9 +25,9 @@
* Use per cpu MCA/INIT stacks for all data.
*/
#include <linux/threads.h>
+#include <linux/pgtable.h>
#include <asm/asmmacro.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/mca_asm.h>
#include <asm/mca.h>
diff --git a/arch/ia64/kernel/module.c b/arch/ia64/kernel/module.c
index ee693c8..2cba53c 100644
--- a/arch/ia64/kernel/module.c
+++ b/arch/ia64/kernel/module.c
@@ -654,7 +654,7 @@
}
} else if (!is_internal(mod, val))
val = get_plt(mod, location, val, &ok);
- /* FALL THROUGH */
+ fallthrough;
default:
val -= bundle(location);
break;
diff --git a/arch/ia64/kernel/perfmon.c b/arch/ia64/kernel/perfmon.c
deleted file mode 100644
index a23c393..0000000
--- a/arch/ia64/kernel/perfmon.c
+++ /dev/null
@@ -1,6705 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-only
-/*
- * This file implements the perfmon-2 subsystem which is used
- * to program the IA-64 Performance Monitoring Unit (PMU).
- *
- * The initial version of perfmon.c was written by
- * Ganesh Venkitachalam, IBM Corp.
- *
- * Then it was modified for perfmon-1.x by Stephane Eranian and
- * David Mosberger, Hewlett Packard Co.
- *
- * Version Perfmon-2.x is a rewrite of perfmon-1.x
- * by Stephane Eranian, Hewlett Packard Co.
- *
- * Copyright (C) 1999-2005 Hewlett Packard Co
- * Stephane Eranian <eranian@hpl.hp.com>
- * David Mosberger-Tang <davidm@hpl.hp.com>
- *
- * More information about perfmon available at:
- * http://www.hpl.hp.com/research/linux/perfmon
- */
-
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/sched.h>
-#include <linux/sched/task.h>
-#include <linux/sched/task_stack.h>
-#include <linux/interrupt.h>
-#include <linux/proc_fs.h>
-#include <linux/seq_file.h>
-#include <linux/init.h>
-#include <linux/vmalloc.h>
-#include <linux/mm.h>
-#include <linux/sysctl.h>
-#include <linux/list.h>
-#include <linux/file.h>
-#include <linux/poll.h>
-#include <linux/vfs.h>
-#include <linux/smp.h>
-#include <linux/pagemap.h>
-#include <linux/mount.h>
-#include <linux/pseudo_fs.h>
-#include <linux/bitops.h>
-#include <linux/capability.h>
-#include <linux/rcupdate.h>
-#include <linux/completion.h>
-#include <linux/tracehook.h>
-#include <linux/slab.h>
-#include <linux/cpu.h>
-
-#include <asm/errno.h>
-#include <asm/intrinsics.h>
-#include <asm/page.h>
-#include <asm/perfmon.h>
-#include <asm/processor.h>
-#include <asm/signal.h>
-#include <linux/uaccess.h>
-#include <asm/delay.h>
-
-#ifdef CONFIG_PERFMON
-/*
- * perfmon context state
- */
-#define PFM_CTX_UNLOADED 1 /* context is not loaded onto any task */
-#define PFM_CTX_LOADED 2 /* context is loaded onto a task */
-#define PFM_CTX_MASKED 3 /* context is loaded but monitoring is masked due to overflow */
-#define PFM_CTX_ZOMBIE 4 /* owner of the context is closing it */
-
-#define PFM_INVALID_ACTIVATION (~0UL)
-
-#define PFM_NUM_PMC_REGS 64 /* PMC save area for ctxsw */
-#define PFM_NUM_PMD_REGS 64 /* PMD save area for ctxsw */
-
-/*
- * depth of message queue
- */
-#define PFM_MAX_MSGS 32
-#define PFM_CTXQ_EMPTY(g) ((g)->ctx_msgq_head == (g)->ctx_msgq_tail)
-
-/*
- * type of a PMU register (bitmask).
- * bitmask structure:
- * bit0 : register implemented
- * bit1 : end marker
- * bit2-3 : reserved
- * bit4 : pmc has pmc.pm
- * bit5 : pmc controls a counter (has pmc.oi), pmd is used as counter
- * bit6-7 : register type
- * bit8-31: reserved
- */
-#define PFM_REG_NOTIMPL 0x0 /* not implemented at all */
-#define PFM_REG_IMPL 0x1 /* register implemented */
-#define PFM_REG_END 0x2 /* end marker */
-#define PFM_REG_MONITOR (0x1<<4|PFM_REG_IMPL) /* a PMC with a pmc.pm field only */
-#define PFM_REG_COUNTING (0x2<<4|PFM_REG_MONITOR) /* a monitor + pmc.oi+ PMD used as a counter */
-#define PFM_REG_CONTROL (0x4<<4|PFM_REG_IMPL) /* PMU control register */
-#define PFM_REG_CONFIG (0x8<<4|PFM_REG_IMPL) /* configuration register */
-#define PFM_REG_BUFFER (0xc<<4|PFM_REG_IMPL) /* PMD used as buffer */
-
-#define PMC_IS_LAST(i) (pmu_conf->pmc_desc[i].type & PFM_REG_END)
-#define PMD_IS_LAST(i) (pmu_conf->pmd_desc[i].type & PFM_REG_END)
-
-#define PMC_OVFL_NOTIFY(ctx, i) ((ctx)->ctx_pmds[i].flags & PFM_REGFL_OVFL_NOTIFY)
-
-/* i assumed unsigned */
-#define PMC_IS_IMPL(i) (i< PMU_MAX_PMCS && (pmu_conf->pmc_desc[i].type & PFM_REG_IMPL))
-#define PMD_IS_IMPL(i) (i< PMU_MAX_PMDS && (pmu_conf->pmd_desc[i].type & PFM_REG_IMPL))
-
-/* XXX: these assume that register i is implemented */
-#define PMD_IS_COUNTING(i) ((pmu_conf->pmd_desc[i].type & PFM_REG_COUNTING) == PFM_REG_COUNTING)
-#define PMC_IS_COUNTING(i) ((pmu_conf->pmc_desc[i].type & PFM_REG_COUNTING) == PFM_REG_COUNTING)
-#define PMC_IS_MONITOR(i) ((pmu_conf->pmc_desc[i].type & PFM_REG_MONITOR) == PFM_REG_MONITOR)
-#define PMC_IS_CONTROL(i) ((pmu_conf->pmc_desc[i].type & PFM_REG_CONTROL) == PFM_REG_CONTROL)
-
-#define PMC_DFL_VAL(i) pmu_conf->pmc_desc[i].default_value
-#define PMC_RSVD_MASK(i) pmu_conf->pmc_desc[i].reserved_mask
-#define PMD_PMD_DEP(i) pmu_conf->pmd_desc[i].dep_pmd[0]
-#define PMC_PMD_DEP(i) pmu_conf->pmc_desc[i].dep_pmd[0]
-
-#define PFM_NUM_IBRS IA64_NUM_DBG_REGS
-#define PFM_NUM_DBRS IA64_NUM_DBG_REGS
-
-#define CTX_OVFL_NOBLOCK(c) ((c)->ctx_fl_block == 0)
-#define CTX_HAS_SMPL(c) ((c)->ctx_fl_is_sampling)
-#define PFM_CTX_TASK(h) (h)->ctx_task
-
-#define PMU_PMC_OI 5 /* position of pmc.oi bit */
-
-/* XXX: does not support more than 64 PMDs */
-#define CTX_USED_PMD(ctx, mask) (ctx)->ctx_used_pmds[0] |= (mask)
-#define CTX_IS_USED_PMD(ctx, c) (((ctx)->ctx_used_pmds[0] & (1UL << (c))) != 0UL)
-
-#define CTX_USED_MONITOR(ctx, mask) (ctx)->ctx_used_monitors[0] |= (mask)
-
-#define CTX_USED_IBR(ctx,n) (ctx)->ctx_used_ibrs[(n)>>6] |= 1UL<< ((n) % 64)
-#define CTX_USED_DBR(ctx,n) (ctx)->ctx_used_dbrs[(n)>>6] |= 1UL<< ((n) % 64)
-#define CTX_USES_DBREGS(ctx) (((pfm_context_t *)(ctx))->ctx_fl_using_dbreg==1)
-#define PFM_CODE_RR 0 /* requesting code range restriction */
-#define PFM_DATA_RR 1 /* requestion data range restriction */
-
-#define PFM_CPUINFO_CLEAR(v) pfm_get_cpu_var(pfm_syst_info) &= ~(v)
-#define PFM_CPUINFO_SET(v) pfm_get_cpu_var(pfm_syst_info) |= (v)
-#define PFM_CPUINFO_GET() pfm_get_cpu_var(pfm_syst_info)
-
-#define RDEP(x) (1UL<<(x))
-
-/*
- * context protection macros
- * in SMP:
- * - we need to protect against CPU concurrency (spin_lock)
- * - we need to protect against PMU overflow interrupts (local_irq_disable)
- * in UP:
- * - we need to protect against PMU overflow interrupts (local_irq_disable)
- *
- * spin_lock_irqsave()/spin_unlock_irqrestore():
- * in SMP: local_irq_disable + spin_lock
- * in UP : local_irq_disable
- *
- * spin_lock()/spin_lock():
- * in UP : removed automatically
- * in SMP: protect against context accesses from other CPU. interrupts
- * are not masked. This is useful for the PMU interrupt handler
- * because we know we will not get PMU concurrency in that code.
- */
-#define PROTECT_CTX(c, f) \
- do { \
- DPRINT(("spinlock_irq_save ctx %p by [%d]\n", c, task_pid_nr(current))); \
- spin_lock_irqsave(&(c)->ctx_lock, f); \
- DPRINT(("spinlocked ctx %p by [%d]\n", c, task_pid_nr(current))); \
- } while(0)
-
-#define UNPROTECT_CTX(c, f) \
- do { \
- DPRINT(("spinlock_irq_restore ctx %p by [%d]\n", c, task_pid_nr(current))); \
- spin_unlock_irqrestore(&(c)->ctx_lock, f); \
- } while(0)
-
-#define PROTECT_CTX_NOPRINT(c, f) \
- do { \
- spin_lock_irqsave(&(c)->ctx_lock, f); \
- } while(0)
-
-
-#define UNPROTECT_CTX_NOPRINT(c, f) \
- do { \
- spin_unlock_irqrestore(&(c)->ctx_lock, f); \
- } while(0)
-
-
-#define PROTECT_CTX_NOIRQ(c) \
- do { \
- spin_lock(&(c)->ctx_lock); \
- } while(0)
-
-#define UNPROTECT_CTX_NOIRQ(c) \
- do { \
- spin_unlock(&(c)->ctx_lock); \
- } while(0)
-
-
-#ifdef CONFIG_SMP
-
-#define GET_ACTIVATION() pfm_get_cpu_var(pmu_activation_number)
-#define INC_ACTIVATION() pfm_get_cpu_var(pmu_activation_number)++
-#define SET_ACTIVATION(c) (c)->ctx_last_activation = GET_ACTIVATION()
-
-#else /* !CONFIG_SMP */
-#define SET_ACTIVATION(t) do {} while(0)
-#define GET_ACTIVATION(t) do {} while(0)
-#define INC_ACTIVATION(t) do {} while(0)
-#endif /* CONFIG_SMP */
-
-#define SET_PMU_OWNER(t, c) do { pfm_get_cpu_var(pmu_owner) = (t); pfm_get_cpu_var(pmu_ctx) = (c); } while(0)
-#define GET_PMU_OWNER() pfm_get_cpu_var(pmu_owner)
-#define GET_PMU_CTX() pfm_get_cpu_var(pmu_ctx)
-
-#define LOCK_PFS(g) spin_lock_irqsave(&pfm_sessions.pfs_lock, g)
-#define UNLOCK_PFS(g) spin_unlock_irqrestore(&pfm_sessions.pfs_lock, g)
-
-#define PFM_REG_RETFLAG_SET(flags, val) do { flags &= ~PFM_REG_RETFL_MASK; flags |= (val); } while(0)
-
-/*
- * cmp0 must be the value of pmc0
- */
-#define PMC0_HAS_OVFL(cmp0) (cmp0 & ~0x1UL)
-
-#define PFMFS_MAGIC 0xa0b4d889
-
-/*
- * debugging
- */
-#define PFM_DEBUGGING 1
-#ifdef PFM_DEBUGGING
-#define DPRINT(a) \
- do { \
- if (unlikely(pfm_sysctl.debug >0)) { printk("%s.%d: CPU%d [%d] ", __func__, __LINE__, smp_processor_id(), task_pid_nr(current)); printk a; } \
- } while (0)
-
-#define DPRINT_ovfl(a) \
- do { \
- if (unlikely(pfm_sysctl.debug > 0 && pfm_sysctl.debug_ovfl >0)) { printk("%s.%d: CPU%d [%d] ", __func__, __LINE__, smp_processor_id(), task_pid_nr(current)); printk a; } \
- } while (0)
-#endif
-
-/*
- * 64-bit software counter structure
- *
- * the next_reset_type is applied to the next call to pfm_reset_regs()
- */
-typedef struct {
- unsigned long val; /* virtual 64bit counter value */
- unsigned long lval; /* last reset value */
- unsigned long long_reset; /* reset value on sampling overflow */
- unsigned long short_reset; /* reset value on overflow */
- unsigned long reset_pmds[4]; /* which other pmds to reset when this counter overflows */
- unsigned long smpl_pmds[4]; /* which pmds are accessed when counter overflow */
- unsigned long seed; /* seed for random-number generator */
- unsigned long mask; /* mask for random-number generator */
- unsigned int flags; /* notify/do not notify */
- unsigned long eventid; /* overflow event identifier */
-} pfm_counter_t;
-
-/*
- * context flags
- */
-typedef struct {
- unsigned int block:1; /* when 1, task will blocked on user notifications */
- unsigned int system:1; /* do system wide monitoring */
- unsigned int using_dbreg:1; /* using range restrictions (debug registers) */
- unsigned int is_sampling:1; /* true if using a custom format */
- unsigned int excl_idle:1; /* exclude idle task in system wide session */
- unsigned int going_zombie:1; /* context is zombie (MASKED+blocking) */
- unsigned int trap_reason:2; /* reason for going into pfm_handle_work() */
- unsigned int no_msg:1; /* no message sent on overflow */
- unsigned int can_restart:1; /* allowed to issue a PFM_RESTART */
- unsigned int reserved:22;
-} pfm_context_flags_t;
-
-#define PFM_TRAP_REASON_NONE 0x0 /* default value */
-#define PFM_TRAP_REASON_BLOCK 0x1 /* we need to block on overflow */
-#define PFM_TRAP_REASON_RESET 0x2 /* we need to reset PMDs */
-
-
-/*
- * perfmon context: encapsulates all the state of a monitoring session
- */
-
-typedef struct pfm_context {
- spinlock_t ctx_lock; /* context protection */
-
- pfm_context_flags_t ctx_flags; /* bitmask of flags (block reason incl.) */
- unsigned int ctx_state; /* state: active/inactive (no bitfield) */
-
- struct task_struct *ctx_task; /* task to which context is attached */
-
- unsigned long ctx_ovfl_regs[4]; /* which registers overflowed (notification) */
-
- struct completion ctx_restart_done; /* use for blocking notification mode */
-
- unsigned long ctx_used_pmds[4]; /* bitmask of PMD used */
- unsigned long ctx_all_pmds[4]; /* bitmask of all accessible PMDs */
- unsigned long ctx_reload_pmds[4]; /* bitmask of force reload PMD on ctxsw in */
-
- unsigned long ctx_all_pmcs[4]; /* bitmask of all accessible PMCs */
- unsigned long ctx_reload_pmcs[4]; /* bitmask of force reload PMC on ctxsw in */
- unsigned long ctx_used_monitors[4]; /* bitmask of monitor PMC being used */
-
- unsigned long ctx_pmcs[PFM_NUM_PMC_REGS]; /* saved copies of PMC values */
-
- unsigned int ctx_used_ibrs[1]; /* bitmask of used IBR (speedup ctxsw in) */
- unsigned int ctx_used_dbrs[1]; /* bitmask of used DBR (speedup ctxsw in) */
- unsigned long ctx_dbrs[IA64_NUM_DBG_REGS]; /* DBR values (cache) when not loaded */
- unsigned long ctx_ibrs[IA64_NUM_DBG_REGS]; /* IBR values (cache) when not loaded */
-
- pfm_counter_t ctx_pmds[PFM_NUM_PMD_REGS]; /* software state for PMDS */
-
- unsigned long th_pmcs[PFM_NUM_PMC_REGS]; /* PMC thread save state */
- unsigned long th_pmds[PFM_NUM_PMD_REGS]; /* PMD thread save state */
-
- unsigned long ctx_saved_psr_up; /* only contains psr.up value */
-
- unsigned long ctx_last_activation; /* context last activation number for last_cpu */
- unsigned int ctx_last_cpu; /* CPU id of current or last CPU used (SMP only) */
- unsigned int ctx_cpu; /* cpu to which perfmon is applied (system wide) */
-
- int ctx_fd; /* file descriptor used my this context */
- pfm_ovfl_arg_t ctx_ovfl_arg; /* argument to custom buffer format handler */
-
- pfm_buffer_fmt_t *ctx_buf_fmt; /* buffer format callbacks */
- void *ctx_smpl_hdr; /* points to sampling buffer header kernel vaddr */
- unsigned long ctx_smpl_size; /* size of sampling buffer */
- void *ctx_smpl_vaddr; /* user level virtual address of smpl buffer */
-
- wait_queue_head_t ctx_msgq_wait;
- pfm_msg_t ctx_msgq[PFM_MAX_MSGS];
- int ctx_msgq_head;
- int ctx_msgq_tail;
- struct fasync_struct *ctx_async_queue;
-
- wait_queue_head_t ctx_zombieq; /* termination cleanup wait queue */
-} pfm_context_t;
-
-/*
- * magic number used to verify that structure is really
- * a perfmon context
- */
-#define PFM_IS_FILE(f) ((f)->f_op == &pfm_file_ops)
-
-#define PFM_GET_CTX(t) ((pfm_context_t *)(t)->thread.pfm_context)
-
-#ifdef CONFIG_SMP
-#define SET_LAST_CPU(ctx, v) (ctx)->ctx_last_cpu = (v)
-#define GET_LAST_CPU(ctx) (ctx)->ctx_last_cpu
-#else
-#define SET_LAST_CPU(ctx, v) do {} while(0)
-#define GET_LAST_CPU(ctx) do {} while(0)
-#endif
-
-
-#define ctx_fl_block ctx_flags.block
-#define ctx_fl_system ctx_flags.system
-#define ctx_fl_using_dbreg ctx_flags.using_dbreg
-#define ctx_fl_is_sampling ctx_flags.is_sampling
-#define ctx_fl_excl_idle ctx_flags.excl_idle
-#define ctx_fl_going_zombie ctx_flags.going_zombie
-#define ctx_fl_trap_reason ctx_flags.trap_reason
-#define ctx_fl_no_msg ctx_flags.no_msg
-#define ctx_fl_can_restart ctx_flags.can_restart
-
-#define PFM_SET_WORK_PENDING(t, v) do { (t)->thread.pfm_needs_checking = v; } while(0);
-#define PFM_GET_WORK_PENDING(t) (t)->thread.pfm_needs_checking
-
-/*
- * global information about all sessions
- * mostly used to synchronize between system wide and per-process
- */
-typedef struct {
- spinlock_t pfs_lock; /* lock the structure */
-
- unsigned int pfs_task_sessions; /* number of per task sessions */
- unsigned int pfs_sys_sessions; /* number of per system wide sessions */
- unsigned int pfs_sys_use_dbregs; /* incremented when a system wide session uses debug regs */
- unsigned int pfs_ptrace_use_dbregs; /* incremented when a process uses debug regs */
- struct task_struct *pfs_sys_session[NR_CPUS]; /* point to task owning a system-wide session */
-} pfm_session_t;
-
-/*
- * information about a PMC or PMD.
- * dep_pmd[]: a bitmask of dependent PMD registers
- * dep_pmc[]: a bitmask of dependent PMC registers
- */
-typedef int (*pfm_reg_check_t)(struct task_struct *task, pfm_context_t *ctx, unsigned int cnum, unsigned long *val, struct pt_regs *regs);
-typedef struct {
- unsigned int type;
- int pm_pos;
- unsigned long default_value; /* power-on default value */
- unsigned long reserved_mask; /* bitmask of reserved bits */
- pfm_reg_check_t read_check;
- pfm_reg_check_t write_check;
- unsigned long dep_pmd[4];
- unsigned long dep_pmc[4];
-} pfm_reg_desc_t;
-
-/* assume cnum is a valid monitor */
-#define PMC_PM(cnum, val) (((val) >> (pmu_conf->pmc_desc[cnum].pm_pos)) & 0x1)
-
-/*
- * This structure is initialized at boot time and contains
- * a description of the PMU main characteristics.
- *
- * If the probe function is defined, detection is based
- * on its return value:
- * - 0 means recognized PMU
- * - anything else means not supported
- * When the probe function is not defined, then the pmu_family field
- * is used and it must match the host CPU family such that:
- * - cpu->family & config->pmu_family != 0
- */
-typedef struct {
- unsigned long ovfl_val; /* overflow value for counters */
-
- pfm_reg_desc_t *pmc_desc; /* detailed PMC register dependencies descriptions */
- pfm_reg_desc_t *pmd_desc; /* detailed PMD register dependencies descriptions */
-
- unsigned int num_pmcs; /* number of PMCS: computed at init time */
- unsigned int num_pmds; /* number of PMDS: computed at init time */
- unsigned long impl_pmcs[4]; /* bitmask of implemented PMCS */
- unsigned long impl_pmds[4]; /* bitmask of implemented PMDS */
-
- char *pmu_name; /* PMU family name */
- unsigned int pmu_family; /* cpuid family pattern used to identify pmu */
- unsigned int flags; /* pmu specific flags */
- unsigned int num_ibrs; /* number of IBRS: computed at init time */
- unsigned int num_dbrs; /* number of DBRS: computed at init time */
- unsigned int num_counters; /* PMC/PMD counting pairs : computed at init time */
- int (*probe)(void); /* customized probe routine */
- unsigned int use_rr_dbregs:1; /* set if debug registers used for range restriction */
-} pmu_config_t;
-/*
- * PMU specific flags
- */
-#define PFM_PMU_IRQ_RESEND 1 /* PMU needs explicit IRQ resend */
-
-/*
- * debug register related type definitions
- */
-typedef struct {
- unsigned long ibr_mask:56;
- unsigned long ibr_plm:4;
- unsigned long ibr_ig:3;
- unsigned long ibr_x:1;
-} ibr_mask_reg_t;
-
-typedef struct {
- unsigned long dbr_mask:56;
- unsigned long dbr_plm:4;
- unsigned long dbr_ig:2;
- unsigned long dbr_w:1;
- unsigned long dbr_r:1;
-} dbr_mask_reg_t;
-
-typedef union {
- unsigned long val;
- ibr_mask_reg_t ibr;
- dbr_mask_reg_t dbr;
-} dbreg_t;
-
-
-/*
- * perfmon command descriptions
- */
-typedef struct {
- int (*cmd_func)(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs);
- char *cmd_name;
- int cmd_flags;
- unsigned int cmd_narg;
- size_t cmd_argsize;
- int (*cmd_getsize)(void *arg, size_t *sz);
-} pfm_cmd_desc_t;
-
-#define PFM_CMD_FD 0x01 /* command requires a file descriptor */
-#define PFM_CMD_ARG_READ 0x02 /* command must read argument(s) */
-#define PFM_CMD_ARG_RW 0x04 /* command must read/write argument(s) */
-#define PFM_CMD_STOP 0x08 /* command does not work on zombie context */
-
-
-#define PFM_CMD_NAME(cmd) pfm_cmd_tab[(cmd)].cmd_name
-#define PFM_CMD_READ_ARG(cmd) (pfm_cmd_tab[(cmd)].cmd_flags & PFM_CMD_ARG_READ)
-#define PFM_CMD_RW_ARG(cmd) (pfm_cmd_tab[(cmd)].cmd_flags & PFM_CMD_ARG_RW)
-#define PFM_CMD_USE_FD(cmd) (pfm_cmd_tab[(cmd)].cmd_flags & PFM_CMD_FD)
-#define PFM_CMD_STOPPED(cmd) (pfm_cmd_tab[(cmd)].cmd_flags & PFM_CMD_STOP)
-
-#define PFM_CMD_ARG_MANY -1 /* cannot be zero */
-
-typedef struct {
- unsigned long pfm_spurious_ovfl_intr_count; /* keep track of spurious ovfl interrupts */
- unsigned long pfm_replay_ovfl_intr_count; /* keep track of replayed ovfl interrupts */
- unsigned long pfm_ovfl_intr_count; /* keep track of ovfl interrupts */
- unsigned long pfm_ovfl_intr_cycles; /* cycles spent processing ovfl interrupts */
- unsigned long pfm_ovfl_intr_cycles_min; /* min cycles spent processing ovfl interrupts */
- unsigned long pfm_ovfl_intr_cycles_max; /* max cycles spent processing ovfl interrupts */
- unsigned long pfm_smpl_handler_calls;
- unsigned long pfm_smpl_handler_cycles;
- char pad[SMP_CACHE_BYTES] ____cacheline_aligned;
-} pfm_stats_t;
-
-/*
- * perfmon internal variables
- */
-static pfm_stats_t pfm_stats[NR_CPUS];
-static pfm_session_t pfm_sessions; /* global sessions information */
-
-static DEFINE_SPINLOCK(pfm_alt_install_check);
-static pfm_intr_handler_desc_t *pfm_alt_intr_handler;
-
-static struct proc_dir_entry *perfmon_dir;
-static pfm_uuid_t pfm_null_uuid = {0,};
-
-static spinlock_t pfm_buffer_fmt_lock;
-static LIST_HEAD(pfm_buffer_fmt_list);
-
-static pmu_config_t *pmu_conf;
-
-/* sysctl() controls */
-pfm_sysctl_t pfm_sysctl;
-EXPORT_SYMBOL(pfm_sysctl);
-
-static struct ctl_table pfm_ctl_table[] = {
- {
- .procname = "debug",
- .data = &pfm_sysctl.debug,
- .maxlen = sizeof(int),
- .mode = 0666,
- .proc_handler = proc_dointvec,
- },
- {
- .procname = "debug_ovfl",
- .data = &pfm_sysctl.debug_ovfl,
- .maxlen = sizeof(int),
- .mode = 0666,
- .proc_handler = proc_dointvec,
- },
- {
- .procname = "fastctxsw",
- .data = &pfm_sysctl.fastctxsw,
- .maxlen = sizeof(int),
- .mode = 0600,
- .proc_handler = proc_dointvec,
- },
- {
- .procname = "expert_mode",
- .data = &pfm_sysctl.expert_mode,
- .maxlen = sizeof(int),
- .mode = 0600,
- .proc_handler = proc_dointvec,
- },
- {}
-};
-static struct ctl_table pfm_sysctl_dir[] = {
- {
- .procname = "perfmon",
- .mode = 0555,
- .child = pfm_ctl_table,
- },
- {}
-};
-static struct ctl_table pfm_sysctl_root[] = {
- {
- .procname = "kernel",
- .mode = 0555,
- .child = pfm_sysctl_dir,
- },
- {}
-};
-static struct ctl_table_header *pfm_sysctl_header;
-
-static int pfm_context_unload(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs);
-
-#define pfm_get_cpu_var(v) __ia64_per_cpu_var(v)
-#define pfm_get_cpu_data(a,b) per_cpu(a, b)
-
-static inline void
-pfm_put_task(struct task_struct *task)
-{
- if (task != current) put_task_struct(task);
-}
-
-static inline unsigned long
-pfm_protect_ctx_ctxsw(pfm_context_t *x)
-{
- spin_lock(&(x)->ctx_lock);
- return 0UL;
-}
-
-static inline void
-pfm_unprotect_ctx_ctxsw(pfm_context_t *x, unsigned long f)
-{
- spin_unlock(&(x)->ctx_lock);
-}
-
-/* forward declaration */
-static const struct dentry_operations pfmfs_dentry_operations;
-
-static int pfmfs_init_fs_context(struct fs_context *fc)
-{
- struct pseudo_fs_context *ctx = init_pseudo(fc, PFMFS_MAGIC);
- if (!ctx)
- return -ENOMEM;
- ctx->dops = &pfmfs_dentry_operations;
- return 0;
-}
-
-static struct file_system_type pfm_fs_type = {
- .name = "pfmfs",
- .init_fs_context = pfmfs_init_fs_context,
- .kill_sb = kill_anon_super,
-};
-MODULE_ALIAS_FS("pfmfs");
-
-DEFINE_PER_CPU(unsigned long, pfm_syst_info);
-DEFINE_PER_CPU(struct task_struct *, pmu_owner);
-DEFINE_PER_CPU(pfm_context_t *, pmu_ctx);
-DEFINE_PER_CPU(unsigned long, pmu_activation_number);
-EXPORT_PER_CPU_SYMBOL_GPL(pfm_syst_info);
-
-
-/* forward declaration */
-static const struct file_operations pfm_file_ops;
-
-/*
- * forward declarations
- */
-#ifndef CONFIG_SMP
-static void pfm_lazy_save_regs (struct task_struct *ta);
-#endif
-
-void dump_pmu_state(const char *);
-static int pfm_write_ibr_dbr(int mode, pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs);
-
-#include "perfmon_itanium.h"
-#include "perfmon_mckinley.h"
-#include "perfmon_montecito.h"
-#include "perfmon_generic.h"
-
-static pmu_config_t *pmu_confs[]={
- &pmu_conf_mont,
- &pmu_conf_mck,
- &pmu_conf_ita,
- &pmu_conf_gen, /* must be last */
- NULL
-};
-
-
-static int pfm_end_notify_user(pfm_context_t *ctx);
-
-static inline void
-pfm_clear_psr_pp(void)
-{
- ia64_rsm(IA64_PSR_PP);
- ia64_srlz_i();
-}
-
-static inline void
-pfm_set_psr_pp(void)
-{
- ia64_ssm(IA64_PSR_PP);
- ia64_srlz_i();
-}
-
-static inline void
-pfm_clear_psr_up(void)
-{
- ia64_rsm(IA64_PSR_UP);
- ia64_srlz_i();
-}
-
-static inline void
-pfm_set_psr_up(void)
-{
- ia64_ssm(IA64_PSR_UP);
- ia64_srlz_i();
-}
-
-static inline unsigned long
-pfm_get_psr(void)
-{
- unsigned long tmp;
- tmp = ia64_getreg(_IA64_REG_PSR);
- ia64_srlz_i();
- return tmp;
-}
-
-static inline void
-pfm_set_psr_l(unsigned long val)
-{
- ia64_setreg(_IA64_REG_PSR_L, val);
- ia64_srlz_i();
-}
-
-static inline void
-pfm_freeze_pmu(void)
-{
- ia64_set_pmc(0,1UL);
- ia64_srlz_d();
-}
-
-static inline void
-pfm_unfreeze_pmu(void)
-{
- ia64_set_pmc(0,0UL);
- ia64_srlz_d();
-}
-
-static inline void
-pfm_restore_ibrs(unsigned long *ibrs, unsigned int nibrs)
-{
- int i;
-
- for (i=0; i < nibrs; i++) {
- ia64_set_ibr(i, ibrs[i]);
- ia64_dv_serialize_instruction();
- }
- ia64_srlz_i();
-}
-
-static inline void
-pfm_restore_dbrs(unsigned long *dbrs, unsigned int ndbrs)
-{
- int i;
-
- for (i=0; i < ndbrs; i++) {
- ia64_set_dbr(i, dbrs[i]);
- ia64_dv_serialize_data();
- }
- ia64_srlz_d();
-}
-
-/*
- * PMD[i] must be a counter. no check is made
- */
-static inline unsigned long
-pfm_read_soft_counter(pfm_context_t *ctx, int i)
-{
- return ctx->ctx_pmds[i].val + (ia64_get_pmd(i) & pmu_conf->ovfl_val);
-}
-
-/*
- * PMD[i] must be a counter. no check is made
- */
-static inline void
-pfm_write_soft_counter(pfm_context_t *ctx, int i, unsigned long val)
-{
- unsigned long ovfl_val = pmu_conf->ovfl_val;
-
- ctx->ctx_pmds[i].val = val & ~ovfl_val;
- /*
- * writing to unimplemented part is ignore, so we do not need to
- * mask off top part
- */
- ia64_set_pmd(i, val & ovfl_val);
-}
-
-static pfm_msg_t *
-pfm_get_new_msg(pfm_context_t *ctx)
-{
- int idx, next;
-
- next = (ctx->ctx_msgq_tail+1) % PFM_MAX_MSGS;
-
- DPRINT(("ctx_fd=%p head=%d tail=%d\n", ctx, ctx->ctx_msgq_head, ctx->ctx_msgq_tail));
- if (next == ctx->ctx_msgq_head) return NULL;
-
- idx = ctx->ctx_msgq_tail;
- ctx->ctx_msgq_tail = next;
-
- DPRINT(("ctx=%p head=%d tail=%d msg=%d\n", ctx, ctx->ctx_msgq_head, ctx->ctx_msgq_tail, idx));
-
- return ctx->ctx_msgq+idx;
-}
-
-static pfm_msg_t *
-pfm_get_next_msg(pfm_context_t *ctx)
-{
- pfm_msg_t *msg;
-
- DPRINT(("ctx=%p head=%d tail=%d\n", ctx, ctx->ctx_msgq_head, ctx->ctx_msgq_tail));
-
- if (PFM_CTXQ_EMPTY(ctx)) return NULL;
-
- /*
- * get oldest message
- */
- msg = ctx->ctx_msgq+ctx->ctx_msgq_head;
-
- /*
- * and move forward
- */
- ctx->ctx_msgq_head = (ctx->ctx_msgq_head+1) % PFM_MAX_MSGS;
-
- DPRINT(("ctx=%p head=%d tail=%d type=%d\n", ctx, ctx->ctx_msgq_head, ctx->ctx_msgq_tail, msg->pfm_gen_msg.msg_type));
-
- return msg;
-}
-
-static void
-pfm_reset_msgq(pfm_context_t *ctx)
-{
- ctx->ctx_msgq_head = ctx->ctx_msgq_tail = 0;
- DPRINT(("ctx=%p msgq reset\n", ctx));
-}
-
-static pfm_context_t *
-pfm_context_alloc(int ctx_flags)
-{
- pfm_context_t *ctx;
-
- /*
- * allocate context descriptor
- * must be able to free with interrupts disabled
- */
- ctx = kzalloc(sizeof(pfm_context_t), GFP_KERNEL);
- if (ctx) {
- DPRINT(("alloc ctx @%p\n", ctx));
-
- /*
- * init context protection lock
- */
- spin_lock_init(&ctx->ctx_lock);
-
- /*
- * context is unloaded
- */
- ctx->ctx_state = PFM_CTX_UNLOADED;
-
- /*
- * initialization of context's flags
- */
- ctx->ctx_fl_block = (ctx_flags & PFM_FL_NOTIFY_BLOCK) ? 1 : 0;
- ctx->ctx_fl_system = (ctx_flags & PFM_FL_SYSTEM_WIDE) ? 1: 0;
- ctx->ctx_fl_no_msg = (ctx_flags & PFM_FL_OVFL_NO_MSG) ? 1: 0;
- /*
- * will move to set properties
- * ctx->ctx_fl_excl_idle = (ctx_flags & PFM_FL_EXCL_IDLE) ? 1: 0;
- */
-
- /*
- * init restart semaphore to locked
- */
- init_completion(&ctx->ctx_restart_done);
-
- /*
- * activation is used in SMP only
- */
- ctx->ctx_last_activation = PFM_INVALID_ACTIVATION;
- SET_LAST_CPU(ctx, -1);
-
- /*
- * initialize notification message queue
- */
- ctx->ctx_msgq_head = ctx->ctx_msgq_tail = 0;
- init_waitqueue_head(&ctx->ctx_msgq_wait);
- init_waitqueue_head(&ctx->ctx_zombieq);
-
- }
- return ctx;
-}
-
-static void
-pfm_context_free(pfm_context_t *ctx)
-{
- if (ctx) {
- DPRINT(("free ctx @%p\n", ctx));
- kfree(ctx);
- }
-}
-
-static void
-pfm_mask_monitoring(struct task_struct *task)
-{
- pfm_context_t *ctx = PFM_GET_CTX(task);
- unsigned long mask, val, ovfl_mask;
- int i;
-
- DPRINT_ovfl(("masking monitoring for [%d]\n", task_pid_nr(task)));
-
- ovfl_mask = pmu_conf->ovfl_val;
- /*
- * monitoring can only be masked as a result of a valid
- * counter overflow. In UP, it means that the PMU still
- * has an owner. Note that the owner can be different
- * from the current task. However the PMU state belongs
- * to the owner.
- * In SMP, a valid overflow only happens when task is
- * current. Therefore if we come here, we know that
- * the PMU state belongs to the current task, therefore
- * we can access the live registers.
- *
- * So in both cases, the live register contains the owner's
- * state. We can ONLY touch the PMU registers and NOT the PSR.
- *
- * As a consequence to this call, the ctx->th_pmds[] array
- * contains stale information which must be ignored
- * when context is reloaded AND monitoring is active (see
- * pfm_restart).
- */
- mask = ctx->ctx_used_pmds[0];
- for (i = 0; mask; i++, mask>>=1) {
- /* skip non used pmds */
- if ((mask & 0x1) == 0) continue;
- val = ia64_get_pmd(i);
-
- if (PMD_IS_COUNTING(i)) {
- /*
- * we rebuild the full 64 bit value of the counter
- */
- ctx->ctx_pmds[i].val += (val & ovfl_mask);
- } else {
- ctx->ctx_pmds[i].val = val;
- }
- DPRINT_ovfl(("pmd[%d]=0x%lx hw_pmd=0x%lx\n",
- i,
- ctx->ctx_pmds[i].val,
- val & ovfl_mask));
- }
- /*
- * mask monitoring by setting the privilege level to 0
- * we cannot use psr.pp/psr.up for this, it is controlled by
- * the user
- *
- * if task is current, modify actual registers, otherwise modify
- * thread save state, i.e., what will be restored in pfm_load_regs()
- */
- mask = ctx->ctx_used_monitors[0] >> PMU_FIRST_COUNTER;
- for(i= PMU_FIRST_COUNTER; mask; i++, mask>>=1) {
- if ((mask & 0x1) == 0UL) continue;
- ia64_set_pmc(i, ctx->th_pmcs[i] & ~0xfUL);
- ctx->th_pmcs[i] &= ~0xfUL;
- DPRINT_ovfl(("pmc[%d]=0x%lx\n", i, ctx->th_pmcs[i]));
- }
- /*
- * make all of this visible
- */
- ia64_srlz_d();
-}
-
-/*
- * must always be done with task == current
- *
- * context must be in MASKED state when calling
- */
-static void
-pfm_restore_monitoring(struct task_struct *task)
-{
- pfm_context_t *ctx = PFM_GET_CTX(task);
- unsigned long mask, ovfl_mask;
- unsigned long psr, val;
- int i, is_system;
-
- is_system = ctx->ctx_fl_system;
- ovfl_mask = pmu_conf->ovfl_val;
-
- if (task != current) {
- printk(KERN_ERR "perfmon.%d: invalid task[%d] current[%d]\n", __LINE__, task_pid_nr(task), task_pid_nr(current));
- return;
- }
- if (ctx->ctx_state != PFM_CTX_MASKED) {
- printk(KERN_ERR "perfmon.%d: task[%d] current[%d] invalid state=%d\n", __LINE__,
- task_pid_nr(task), task_pid_nr(current), ctx->ctx_state);
- return;
- }
- psr = pfm_get_psr();
- /*
- * monitoring is masked via the PMC.
- * As we restore their value, we do not want each counter to
- * restart right away. We stop monitoring using the PSR,
- * restore the PMC (and PMD) and then re-establish the psr
- * as it was. Note that there can be no pending overflow at
- * this point, because monitoring was MASKED.
- *
- * system-wide session are pinned and self-monitoring
- */
- if (is_system && (PFM_CPUINFO_GET() & PFM_CPUINFO_DCR_PP)) {
- /* disable dcr pp */
- ia64_setreg(_IA64_REG_CR_DCR, ia64_getreg(_IA64_REG_CR_DCR) & ~IA64_DCR_PP);
- pfm_clear_psr_pp();
- } else {
- pfm_clear_psr_up();
- }
- /*
- * first, we restore the PMD
- */
- mask = ctx->ctx_used_pmds[0];
- for (i = 0; mask; i++, mask>>=1) {
- /* skip non used pmds */
- if ((mask & 0x1) == 0) continue;
-
- if (PMD_IS_COUNTING(i)) {
- /*
- * we split the 64bit value according to
- * counter width
- */
- val = ctx->ctx_pmds[i].val & ovfl_mask;
- ctx->ctx_pmds[i].val &= ~ovfl_mask;
- } else {
- val = ctx->ctx_pmds[i].val;
- }
- ia64_set_pmd(i, val);
-
- DPRINT(("pmd[%d]=0x%lx hw_pmd=0x%lx\n",
- i,
- ctx->ctx_pmds[i].val,
- val));
- }
- /*
- * restore the PMCs
- */
- mask = ctx->ctx_used_monitors[0] >> PMU_FIRST_COUNTER;
- for(i= PMU_FIRST_COUNTER; mask; i++, mask>>=1) {
- if ((mask & 0x1) == 0UL) continue;
- ctx->th_pmcs[i] = ctx->ctx_pmcs[i];
- ia64_set_pmc(i, ctx->th_pmcs[i]);
- DPRINT(("[%d] pmc[%d]=0x%lx\n",
- task_pid_nr(task), i, ctx->th_pmcs[i]));
- }
- ia64_srlz_d();
-
- /*
- * must restore DBR/IBR because could be modified while masked
- * XXX: need to optimize
- */
- if (ctx->ctx_fl_using_dbreg) {
- pfm_restore_ibrs(ctx->ctx_ibrs, pmu_conf->num_ibrs);
- pfm_restore_dbrs(ctx->ctx_dbrs, pmu_conf->num_dbrs);
- }
-
- /*
- * now restore PSR
- */
- if (is_system && (PFM_CPUINFO_GET() & PFM_CPUINFO_DCR_PP)) {
- /* enable dcr pp */
- ia64_setreg(_IA64_REG_CR_DCR, ia64_getreg(_IA64_REG_CR_DCR) | IA64_DCR_PP);
- ia64_srlz_i();
- }
- pfm_set_psr_l(psr);
-}
-
-static inline void
-pfm_save_pmds(unsigned long *pmds, unsigned long mask)
-{
- int i;
-
- ia64_srlz_d();
-
- for (i=0; mask; i++, mask>>=1) {
- if (mask & 0x1) pmds[i] = ia64_get_pmd(i);
- }
-}
-
-/*
- * reload from thread state (used for ctxw only)
- */
-static inline void
-pfm_restore_pmds(unsigned long *pmds, unsigned long mask)
-{
- int i;
- unsigned long val, ovfl_val = pmu_conf->ovfl_val;
-
- for (i=0; mask; i++, mask>>=1) {
- if ((mask & 0x1) == 0) continue;
- val = PMD_IS_COUNTING(i) ? pmds[i] & ovfl_val : pmds[i];
- ia64_set_pmd(i, val);
- }
- ia64_srlz_d();
-}
-
-/*
- * propagate PMD from context to thread-state
- */
-static inline void
-pfm_copy_pmds(struct task_struct *task, pfm_context_t *ctx)
-{
- unsigned long ovfl_val = pmu_conf->ovfl_val;
- unsigned long mask = ctx->ctx_all_pmds[0];
- unsigned long val;
- int i;
-
- DPRINT(("mask=0x%lx\n", mask));
-
- for (i=0; mask; i++, mask>>=1) {
-
- val = ctx->ctx_pmds[i].val;
-
- /*
- * We break up the 64 bit value into 2 pieces
- * the lower bits go to the machine state in the
- * thread (will be reloaded on ctxsw in).
- * The upper part stays in the soft-counter.
- */
- if (PMD_IS_COUNTING(i)) {
- ctx->ctx_pmds[i].val = val & ~ovfl_val;
- val &= ovfl_val;
- }
- ctx->th_pmds[i] = val;
-
- DPRINT(("pmd[%d]=0x%lx soft_val=0x%lx\n",
- i,
- ctx->th_pmds[i],
- ctx->ctx_pmds[i].val));
- }
-}
-
-/*
- * propagate PMC from context to thread-state
- */
-static inline void
-pfm_copy_pmcs(struct task_struct *task, pfm_context_t *ctx)
-{
- unsigned long mask = ctx->ctx_all_pmcs[0];
- int i;
-
- DPRINT(("mask=0x%lx\n", mask));
-
- for (i=0; mask; i++, mask>>=1) {
- /* masking 0 with ovfl_val yields 0 */
- ctx->th_pmcs[i] = ctx->ctx_pmcs[i];
- DPRINT(("pmc[%d]=0x%lx\n", i, ctx->th_pmcs[i]));
- }
-}
-
-
-
-static inline void
-pfm_restore_pmcs(unsigned long *pmcs, unsigned long mask)
-{
- int i;
-
- for (i=0; mask; i++, mask>>=1) {
- if ((mask & 0x1) == 0) continue;
- ia64_set_pmc(i, pmcs[i]);
- }
- ia64_srlz_d();
-}
-
-static inline int
-pfm_uuid_cmp(pfm_uuid_t a, pfm_uuid_t b)
-{
- return memcmp(a, b, sizeof(pfm_uuid_t));
-}
-
-static inline int
-pfm_buf_fmt_exit(pfm_buffer_fmt_t *fmt, struct task_struct *task, void *buf, struct pt_regs *regs)
-{
- int ret = 0;
- if (fmt->fmt_exit) ret = (*fmt->fmt_exit)(task, buf, regs);
- return ret;
-}
-
-static inline int
-pfm_buf_fmt_getsize(pfm_buffer_fmt_t *fmt, struct task_struct *task, unsigned int flags, int cpu, void *arg, unsigned long *size)
-{
- int ret = 0;
- if (fmt->fmt_getsize) ret = (*fmt->fmt_getsize)(task, flags, cpu, arg, size);
- return ret;
-}
-
-
-static inline int
-pfm_buf_fmt_validate(pfm_buffer_fmt_t *fmt, struct task_struct *task, unsigned int flags,
- int cpu, void *arg)
-{
- int ret = 0;
- if (fmt->fmt_validate) ret = (*fmt->fmt_validate)(task, flags, cpu, arg);
- return ret;
-}
-
-static inline int
-pfm_buf_fmt_init(pfm_buffer_fmt_t *fmt, struct task_struct *task, void *buf, unsigned int flags,
- int cpu, void *arg)
-{
- int ret = 0;
- if (fmt->fmt_init) ret = (*fmt->fmt_init)(task, buf, flags, cpu, arg);
- return ret;
-}
-
-static inline int
-pfm_buf_fmt_restart(pfm_buffer_fmt_t *fmt, struct task_struct *task, pfm_ovfl_ctrl_t *ctrl, void *buf, struct pt_regs *regs)
-{
- int ret = 0;
- if (fmt->fmt_restart) ret = (*fmt->fmt_restart)(task, ctrl, buf, regs);
- return ret;
-}
-
-static inline int
-pfm_buf_fmt_restart_active(pfm_buffer_fmt_t *fmt, struct task_struct *task, pfm_ovfl_ctrl_t *ctrl, void *buf, struct pt_regs *regs)
-{
- int ret = 0;
- if (fmt->fmt_restart_active) ret = (*fmt->fmt_restart_active)(task, ctrl, buf, regs);
- return ret;
-}
-
-static pfm_buffer_fmt_t *
-__pfm_find_buffer_fmt(pfm_uuid_t uuid)
-{
- struct list_head * pos;
- pfm_buffer_fmt_t * entry;
-
- list_for_each(pos, &pfm_buffer_fmt_list) {
- entry = list_entry(pos, pfm_buffer_fmt_t, fmt_list);
- if (pfm_uuid_cmp(uuid, entry->fmt_uuid) == 0)
- return entry;
- }
- return NULL;
-}
-
-/*
- * find a buffer format based on its uuid
- */
-static pfm_buffer_fmt_t *
-pfm_find_buffer_fmt(pfm_uuid_t uuid)
-{
- pfm_buffer_fmt_t * fmt;
- spin_lock(&pfm_buffer_fmt_lock);
- fmt = __pfm_find_buffer_fmt(uuid);
- spin_unlock(&pfm_buffer_fmt_lock);
- return fmt;
-}
-
-int
-pfm_register_buffer_fmt(pfm_buffer_fmt_t *fmt)
-{
- int ret = 0;
-
- /* some sanity checks */
- if (fmt == NULL || fmt->fmt_name == NULL) return -EINVAL;
-
- /* we need at least a handler */
- if (fmt->fmt_handler == NULL) return -EINVAL;
-
- /*
- * XXX: need check validity of fmt_arg_size
- */
-
- spin_lock(&pfm_buffer_fmt_lock);
-
- if (__pfm_find_buffer_fmt(fmt->fmt_uuid)) {
- printk(KERN_ERR "perfmon: duplicate sampling format: %s\n", fmt->fmt_name);
- ret = -EBUSY;
- goto out;
- }
- list_add(&fmt->fmt_list, &pfm_buffer_fmt_list);
- printk(KERN_INFO "perfmon: added sampling format %s\n", fmt->fmt_name);
-
-out:
- spin_unlock(&pfm_buffer_fmt_lock);
- return ret;
-}
-EXPORT_SYMBOL(pfm_register_buffer_fmt);
-
-int
-pfm_unregister_buffer_fmt(pfm_uuid_t uuid)
-{
- pfm_buffer_fmt_t *fmt;
- int ret = 0;
-
- spin_lock(&pfm_buffer_fmt_lock);
-
- fmt = __pfm_find_buffer_fmt(uuid);
- if (!fmt) {
- printk(KERN_ERR "perfmon: cannot unregister format, not found\n");
- ret = -EINVAL;
- goto out;
- }
- list_del_init(&fmt->fmt_list);
- printk(KERN_INFO "perfmon: removed sampling format: %s\n", fmt->fmt_name);
-
-out:
- spin_unlock(&pfm_buffer_fmt_lock);
- return ret;
-
-}
-EXPORT_SYMBOL(pfm_unregister_buffer_fmt);
-
-static int
-pfm_reserve_session(struct task_struct *task, int is_syswide, unsigned int cpu)
-{
- unsigned long flags;
- /*
- * validity checks on cpu_mask have been done upstream
- */
- LOCK_PFS(flags);
-
- DPRINT(("in sys_sessions=%u task_sessions=%u dbregs=%u syswide=%d cpu=%u\n",
- pfm_sessions.pfs_sys_sessions,
- pfm_sessions.pfs_task_sessions,
- pfm_sessions.pfs_sys_use_dbregs,
- is_syswide,
- cpu));
-
- if (is_syswide) {
- /*
- * cannot mix system wide and per-task sessions
- */
- if (pfm_sessions.pfs_task_sessions > 0UL) {
- DPRINT(("system wide not possible, %u conflicting task_sessions\n",
- pfm_sessions.pfs_task_sessions));
- goto abort;
- }
-
- if (pfm_sessions.pfs_sys_session[cpu]) goto error_conflict;
-
- DPRINT(("reserving system wide session on CPU%u currently on CPU%u\n", cpu, smp_processor_id()));
-
- pfm_sessions.pfs_sys_session[cpu] = task;
-
- pfm_sessions.pfs_sys_sessions++ ;
-
- } else {
- if (pfm_sessions.pfs_sys_sessions) goto abort;
- pfm_sessions.pfs_task_sessions++;
- }
-
- DPRINT(("out sys_sessions=%u task_sessions=%u dbregs=%u syswide=%d cpu=%u\n",
- pfm_sessions.pfs_sys_sessions,
- pfm_sessions.pfs_task_sessions,
- pfm_sessions.pfs_sys_use_dbregs,
- is_syswide,
- cpu));
-
- /*
- * Force idle() into poll mode
- */
- cpu_idle_poll_ctrl(true);
-
- UNLOCK_PFS(flags);
-
- return 0;
-
-error_conflict:
- DPRINT(("system wide not possible, conflicting session [%d] on CPU%d\n",
- task_pid_nr(pfm_sessions.pfs_sys_session[cpu]),
- cpu));
-abort:
- UNLOCK_PFS(flags);
-
- return -EBUSY;
-
-}
-
-static int
-pfm_unreserve_session(pfm_context_t *ctx, int is_syswide, unsigned int cpu)
-{
- unsigned long flags;
- /*
- * validity checks on cpu_mask have been done upstream
- */
- LOCK_PFS(flags);
-
- DPRINT(("in sys_sessions=%u task_sessions=%u dbregs=%u syswide=%d cpu=%u\n",
- pfm_sessions.pfs_sys_sessions,
- pfm_sessions.pfs_task_sessions,
- pfm_sessions.pfs_sys_use_dbregs,
- is_syswide,
- cpu));
-
-
- if (is_syswide) {
- pfm_sessions.pfs_sys_session[cpu] = NULL;
- /*
- * would not work with perfmon+more than one bit in cpu_mask
- */
- if (ctx && ctx->ctx_fl_using_dbreg) {
- if (pfm_sessions.pfs_sys_use_dbregs == 0) {
- printk(KERN_ERR "perfmon: invalid release for ctx %p sys_use_dbregs=0\n", ctx);
- } else {
- pfm_sessions.pfs_sys_use_dbregs--;
- }
- }
- pfm_sessions.pfs_sys_sessions--;
- } else {
- pfm_sessions.pfs_task_sessions--;
- }
- DPRINT(("out sys_sessions=%u task_sessions=%u dbregs=%u syswide=%d cpu=%u\n",
- pfm_sessions.pfs_sys_sessions,
- pfm_sessions.pfs_task_sessions,
- pfm_sessions.pfs_sys_use_dbregs,
- is_syswide,
- cpu));
-
- /* Undo forced polling. Last session reenables pal_halt */
- cpu_idle_poll_ctrl(false);
-
- UNLOCK_PFS(flags);
-
- return 0;
-}
-
-/*
- * removes virtual mapping of the sampling buffer.
- * IMPORTANT: cannot be called with interrupts disable, e.g. inside
- * a PROTECT_CTX() section.
- */
-static int
-pfm_remove_smpl_mapping(void *vaddr, unsigned long size)
-{
- struct task_struct *task = current;
- int r;
-
- /* sanity checks */
- if (task->mm == NULL || size == 0UL || vaddr == NULL) {
- printk(KERN_ERR "perfmon: pfm_remove_smpl_mapping [%d] invalid context mm=%p\n", task_pid_nr(task), task->mm);
- return -EINVAL;
- }
-
- DPRINT(("smpl_vaddr=%p size=%lu\n", vaddr, size));
-
- /*
- * does the actual unmapping
- */
- r = vm_munmap((unsigned long)vaddr, size);
-
- if (r !=0) {
- printk(KERN_ERR "perfmon: [%d] unable to unmap sampling buffer @%p size=%lu\n", task_pid_nr(task), vaddr, size);
- }
-
- DPRINT(("do_unmap(%p, %lu)=%d\n", vaddr, size, r));
-
- return 0;
-}
-
-/*
- * free actual physical storage used by sampling buffer
- */
-#if 0
-static int
-pfm_free_smpl_buffer(pfm_context_t *ctx)
-{
- pfm_buffer_fmt_t *fmt;
-
- if (ctx->ctx_smpl_hdr == NULL) goto invalid_free;
-
- /*
- * we won't use the buffer format anymore
- */
- fmt = ctx->ctx_buf_fmt;
-
- DPRINT(("sampling buffer @%p size %lu vaddr=%p\n",
- ctx->ctx_smpl_hdr,
- ctx->ctx_smpl_size,
- ctx->ctx_smpl_vaddr));
-
- pfm_buf_fmt_exit(fmt, current, NULL, NULL);
-
- /*
- * free the buffer
- */
- vfree(ctx->ctx_smpl_hdr);
-
- ctx->ctx_smpl_hdr = NULL;
- ctx->ctx_smpl_size = 0UL;
-
- return 0;
-
-invalid_free:
- printk(KERN_ERR "perfmon: pfm_free_smpl_buffer [%d] no buffer\n", task_pid_nr(current));
- return -EINVAL;
-}
-#endif
-
-static inline void
-pfm_exit_smpl_buffer(pfm_buffer_fmt_t *fmt)
-{
- if (fmt == NULL) return;
-
- pfm_buf_fmt_exit(fmt, current, NULL, NULL);
-
-}
-
-/*
- * pfmfs should _never_ be mounted by userland - too much of security hassle,
- * no real gain from having the whole whorehouse mounted. So we don't need
- * any operations on the root directory. However, we need a non-trivial
- * d_name - pfm: will go nicely and kill the special-casing in procfs.
- */
-static struct vfsmount *pfmfs_mnt __read_mostly;
-
-static int __init
-init_pfm_fs(void)
-{
- int err = register_filesystem(&pfm_fs_type);
- if (!err) {
- pfmfs_mnt = kern_mount(&pfm_fs_type);
- err = PTR_ERR(pfmfs_mnt);
- if (IS_ERR(pfmfs_mnt))
- unregister_filesystem(&pfm_fs_type);
- else
- err = 0;
- }
- return err;
-}
-
-static ssize_t
-pfm_read(struct file *filp, char __user *buf, size_t size, loff_t *ppos)
-{
- pfm_context_t *ctx;
- pfm_msg_t *msg;
- ssize_t ret;
- unsigned long flags;
- DECLARE_WAITQUEUE(wait, current);
- if (PFM_IS_FILE(filp) == 0) {
- printk(KERN_ERR "perfmon: pfm_poll: bad magic [%d]\n", task_pid_nr(current));
- return -EINVAL;
- }
-
- ctx = filp->private_data;
- if (ctx == NULL) {
- printk(KERN_ERR "perfmon: pfm_read: NULL ctx [%d]\n", task_pid_nr(current));
- return -EINVAL;
- }
-
- /*
- * check even when there is no message
- */
- if (size < sizeof(pfm_msg_t)) {
- DPRINT(("message is too small ctx=%p (>=%ld)\n", ctx, sizeof(pfm_msg_t)));
- return -EINVAL;
- }
-
- PROTECT_CTX(ctx, flags);
-
- /*
- * put ourselves on the wait queue
- */
- add_wait_queue(&ctx->ctx_msgq_wait, &wait);
-
-
- for(;;) {
- /*
- * check wait queue
- */
-
- set_current_state(TASK_INTERRUPTIBLE);
-
- DPRINT(("head=%d tail=%d\n", ctx->ctx_msgq_head, ctx->ctx_msgq_tail));
-
- ret = 0;
- if(PFM_CTXQ_EMPTY(ctx) == 0) break;
-
- UNPROTECT_CTX(ctx, flags);
-
- /*
- * check non-blocking read
- */
- ret = -EAGAIN;
- if(filp->f_flags & O_NONBLOCK) break;
-
- /*
- * check pending signals
- */
- if(signal_pending(current)) {
- ret = -EINTR;
- break;
- }
- /*
- * no message, so wait
- */
- schedule();
-
- PROTECT_CTX(ctx, flags);
- }
- DPRINT(("[%d] back to running ret=%ld\n", task_pid_nr(current), ret));
- set_current_state(TASK_RUNNING);
- remove_wait_queue(&ctx->ctx_msgq_wait, &wait);
-
- if (ret < 0) goto abort;
-
- ret = -EINVAL;
- msg = pfm_get_next_msg(ctx);
- if (msg == NULL) {
- printk(KERN_ERR "perfmon: pfm_read no msg for ctx=%p [%d]\n", ctx, task_pid_nr(current));
- goto abort_locked;
- }
-
- DPRINT(("fd=%d type=%d\n", msg->pfm_gen_msg.msg_ctx_fd, msg->pfm_gen_msg.msg_type));
-
- ret = -EFAULT;
- if(copy_to_user(buf, msg, sizeof(pfm_msg_t)) == 0) ret = sizeof(pfm_msg_t);
-
-abort_locked:
- UNPROTECT_CTX(ctx, flags);
-abort:
- return ret;
-}
-
-static ssize_t
-pfm_write(struct file *file, const char __user *ubuf,
- size_t size, loff_t *ppos)
-{
- DPRINT(("pfm_write called\n"));
- return -EINVAL;
-}
-
-static __poll_t
-pfm_poll(struct file *filp, poll_table * wait)
-{
- pfm_context_t *ctx;
- unsigned long flags;
- __poll_t mask = 0;
-
- if (PFM_IS_FILE(filp) == 0) {
- printk(KERN_ERR "perfmon: pfm_poll: bad magic [%d]\n", task_pid_nr(current));
- return 0;
- }
-
- ctx = filp->private_data;
- if (ctx == NULL) {
- printk(KERN_ERR "perfmon: pfm_poll: NULL ctx [%d]\n", task_pid_nr(current));
- return 0;
- }
-
-
- DPRINT(("pfm_poll ctx_fd=%d before poll_wait\n", ctx->ctx_fd));
-
- poll_wait(filp, &ctx->ctx_msgq_wait, wait);
-
- PROTECT_CTX(ctx, flags);
-
- if (PFM_CTXQ_EMPTY(ctx) == 0)
- mask = EPOLLIN | EPOLLRDNORM;
-
- UNPROTECT_CTX(ctx, flags);
-
- DPRINT(("pfm_poll ctx_fd=%d mask=0x%x\n", ctx->ctx_fd, mask));
-
- return mask;
-}
-
-static long
-pfm_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
-{
- DPRINT(("pfm_ioctl called\n"));
- return -EINVAL;
-}
-
-/*
- * interrupt cannot be masked when coming here
- */
-static inline int
-pfm_do_fasync(int fd, struct file *filp, pfm_context_t *ctx, int on)
-{
- int ret;
-
- ret = fasync_helper (fd, filp, on, &ctx->ctx_async_queue);
-
- DPRINT(("pfm_fasync called by [%d] on ctx_fd=%d on=%d async_queue=%p ret=%d\n",
- task_pid_nr(current),
- fd,
- on,
- ctx->ctx_async_queue, ret));
-
- return ret;
-}
-
-static int
-pfm_fasync(int fd, struct file *filp, int on)
-{
- pfm_context_t *ctx;
- int ret;
-
- if (PFM_IS_FILE(filp) == 0) {
- printk(KERN_ERR "perfmon: pfm_fasync bad magic [%d]\n", task_pid_nr(current));
- return -EBADF;
- }
-
- ctx = filp->private_data;
- if (ctx == NULL) {
- printk(KERN_ERR "perfmon: pfm_fasync NULL ctx [%d]\n", task_pid_nr(current));
- return -EBADF;
- }
- /*
- * we cannot mask interrupts during this call because this may
- * may go to sleep if memory is not readily avalaible.
- *
- * We are protected from the conetxt disappearing by the get_fd()/put_fd()
- * done in caller. Serialization of this function is ensured by caller.
- */
- ret = pfm_do_fasync(fd, filp, ctx, on);
-
-
- DPRINT(("pfm_fasync called on ctx_fd=%d on=%d async_queue=%p ret=%d\n",
- fd,
- on,
- ctx->ctx_async_queue, ret));
-
- return ret;
-}
-
-#ifdef CONFIG_SMP
-/*
- * this function is exclusively called from pfm_close().
- * The context is not protected at that time, nor are interrupts
- * on the remote CPU. That's necessary to avoid deadlocks.
- */
-static void
-pfm_syswide_force_stop(void *info)
-{
- pfm_context_t *ctx = (pfm_context_t *)info;
- struct pt_regs *regs = task_pt_regs(current);
- struct task_struct *owner;
- unsigned long flags;
- int ret;
-
- if (ctx->ctx_cpu != smp_processor_id()) {
- printk(KERN_ERR "perfmon: pfm_syswide_force_stop for CPU%d but on CPU%d\n",
- ctx->ctx_cpu,
- smp_processor_id());
- return;
- }
- owner = GET_PMU_OWNER();
- if (owner != ctx->ctx_task) {
- printk(KERN_ERR "perfmon: pfm_syswide_force_stop CPU%d unexpected owner [%d] instead of [%d]\n",
- smp_processor_id(),
- task_pid_nr(owner), task_pid_nr(ctx->ctx_task));
- return;
- }
- if (GET_PMU_CTX() != ctx) {
- printk(KERN_ERR "perfmon: pfm_syswide_force_stop CPU%d unexpected ctx %p instead of %p\n",
- smp_processor_id(),
- GET_PMU_CTX(), ctx);
- return;
- }
-
- DPRINT(("on CPU%d forcing system wide stop for [%d]\n", smp_processor_id(), task_pid_nr(ctx->ctx_task)));
- /*
- * the context is already protected in pfm_close(), we simply
- * need to mask interrupts to avoid a PMU interrupt race on
- * this CPU
- */
- local_irq_save(flags);
-
- ret = pfm_context_unload(ctx, NULL, 0, regs);
- if (ret) {
- DPRINT(("context_unload returned %d\n", ret));
- }
-
- /*
- * unmask interrupts, PMU interrupts are now spurious here
- */
- local_irq_restore(flags);
-}
-
-static void
-pfm_syswide_cleanup_other_cpu(pfm_context_t *ctx)
-{
- int ret;
-
- DPRINT(("calling CPU%d for cleanup\n", ctx->ctx_cpu));
- ret = smp_call_function_single(ctx->ctx_cpu, pfm_syswide_force_stop, ctx, 1);
- DPRINT(("called CPU%d for cleanup ret=%d\n", ctx->ctx_cpu, ret));
-}
-#endif /* CONFIG_SMP */
-
-/*
- * called for each close(). Partially free resources.
- * When caller is self-monitoring, the context is unloaded.
- */
-static int
-pfm_flush(struct file *filp, fl_owner_t id)
-{
- pfm_context_t *ctx;
- struct task_struct *task;
- struct pt_regs *regs;
- unsigned long flags;
- unsigned long smpl_buf_size = 0UL;
- void *smpl_buf_vaddr = NULL;
- int state, is_system;
-
- if (PFM_IS_FILE(filp) == 0) {
- DPRINT(("bad magic for\n"));
- return -EBADF;
- }
-
- ctx = filp->private_data;
- if (ctx == NULL) {
- printk(KERN_ERR "perfmon: pfm_flush: NULL ctx [%d]\n", task_pid_nr(current));
- return -EBADF;
- }
-
- /*
- * remove our file from the async queue, if we use this mode.
- * This can be done without the context being protected. We come
- * here when the context has become unreachable by other tasks.
- *
- * We may still have active monitoring at this point and we may
- * end up in pfm_overflow_handler(). However, fasync_helper()
- * operates with interrupts disabled and it cleans up the
- * queue. If the PMU handler is called prior to entering
- * fasync_helper() then it will send a signal. If it is
- * invoked after, it will find an empty queue and no
- * signal will be sent. In both case, we are safe
- */
- PROTECT_CTX(ctx, flags);
-
- state = ctx->ctx_state;
- is_system = ctx->ctx_fl_system;
-
- task = PFM_CTX_TASK(ctx);
- regs = task_pt_regs(task);
-
- DPRINT(("ctx_state=%d is_current=%d\n",
- state,
- task == current ? 1 : 0));
-
- /*
- * if state == UNLOADED, then task is NULL
- */
-
- /*
- * we must stop and unload because we are losing access to the context.
- */
- if (task == current) {
-#ifdef CONFIG_SMP
- /*
- * the task IS the owner but it migrated to another CPU: that's bad
- * but we must handle this cleanly. Unfortunately, the kernel does
- * not provide a mechanism to block migration (while the context is loaded).
- *
- * We need to release the resource on the ORIGINAL cpu.
- */
- if (is_system && ctx->ctx_cpu != smp_processor_id()) {
-
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- /*
- * keep context protected but unmask interrupt for IPI
- */
- local_irq_restore(flags);
-
- pfm_syswide_cleanup_other_cpu(ctx);
-
- /*
- * restore interrupt masking
- */
- local_irq_save(flags);
-
- /*
- * context is unloaded at this point
- */
- } else
-#endif /* CONFIG_SMP */
- {
-
- DPRINT(("forcing unload\n"));
- /*
- * stop and unload, returning with state UNLOADED
- * and session unreserved.
- */
- pfm_context_unload(ctx, NULL, 0, regs);
-
- DPRINT(("ctx_state=%d\n", ctx->ctx_state));
- }
- }
-
- /*
- * remove virtual mapping, if any, for the calling task.
- * cannot reset ctx field until last user is calling close().
- *
- * ctx_smpl_vaddr must never be cleared because it is needed
- * by every task with access to the context
- *
- * When called from do_exit(), the mm context is gone already, therefore
- * mm is NULL, i.e., the VMA is already gone and we do not have to
- * do anything here
- */
- if (ctx->ctx_smpl_vaddr && current->mm) {
- smpl_buf_vaddr = ctx->ctx_smpl_vaddr;
- smpl_buf_size = ctx->ctx_smpl_size;
- }
-
- UNPROTECT_CTX(ctx, flags);
-
- /*
- * if there was a mapping, then we systematically remove it
- * at this point. Cannot be done inside critical section
- * because some VM function reenables interrupts.
- *
- */
- if (smpl_buf_vaddr) pfm_remove_smpl_mapping(smpl_buf_vaddr, smpl_buf_size);
-
- return 0;
-}
-/*
- * called either on explicit close() or from exit_files().
- * Only the LAST user of the file gets to this point, i.e., it is
- * called only ONCE.
- *
- * IMPORTANT: we get called ONLY when the refcnt on the file gets to zero
- * (fput()),i.e, last task to access the file. Nobody else can access the
- * file at this point.
- *
- * When called from exit_files(), the VMA has been freed because exit_mm()
- * is executed before exit_files().
- *
- * When called from exit_files(), the current task is not yet ZOMBIE but we
- * flush the PMU state to the context.
- */
-static int
-pfm_close(struct inode *inode, struct file *filp)
-{
- pfm_context_t *ctx;
- struct task_struct *task;
- struct pt_regs *regs;
- DECLARE_WAITQUEUE(wait, current);
- unsigned long flags;
- unsigned long smpl_buf_size = 0UL;
- void *smpl_buf_addr = NULL;
- int free_possible = 1;
- int state, is_system;
-
- DPRINT(("pfm_close called private=%p\n", filp->private_data));
-
- if (PFM_IS_FILE(filp) == 0) {
- DPRINT(("bad magic\n"));
- return -EBADF;
- }
-
- ctx = filp->private_data;
- if (ctx == NULL) {
- printk(KERN_ERR "perfmon: pfm_close: NULL ctx [%d]\n", task_pid_nr(current));
- return -EBADF;
- }
-
- PROTECT_CTX(ctx, flags);
-
- state = ctx->ctx_state;
- is_system = ctx->ctx_fl_system;
-
- task = PFM_CTX_TASK(ctx);
- regs = task_pt_regs(task);
-
- DPRINT(("ctx_state=%d is_current=%d\n",
- state,
- task == current ? 1 : 0));
-
- /*
- * if task == current, then pfm_flush() unloaded the context
- */
- if (state == PFM_CTX_UNLOADED) goto doit;
-
- /*
- * context is loaded/masked and task != current, we need to
- * either force an unload or go zombie
- */
-
- /*
- * The task is currently blocked or will block after an overflow.
- * we must force it to wakeup to get out of the
- * MASKED state and transition to the unloaded state by itself.
- *
- * This situation is only possible for per-task mode
- */
- if (state == PFM_CTX_MASKED && CTX_OVFL_NOBLOCK(ctx) == 0) {
-
- /*
- * set a "partial" zombie state to be checked
- * upon return from down() in pfm_handle_work().
- *
- * We cannot use the ZOMBIE state, because it is checked
- * by pfm_load_regs() which is called upon wakeup from down().
- * In such case, it would free the context and then we would
- * return to pfm_handle_work() which would access the
- * stale context. Instead, we set a flag invisible to pfm_load_regs()
- * but visible to pfm_handle_work().
- *
- * For some window of time, we have a zombie context with
- * ctx_state = MASKED and not ZOMBIE
- */
- ctx->ctx_fl_going_zombie = 1;
-
- /*
- * force task to wake up from MASKED state
- */
- complete(&ctx->ctx_restart_done);
-
- DPRINT(("waking up ctx_state=%d\n", state));
-
- /*
- * put ourself to sleep waiting for the other
- * task to report completion
- *
- * the context is protected by mutex, therefore there
- * is no risk of being notified of completion before
- * begin actually on the waitq.
- */
- set_current_state(TASK_INTERRUPTIBLE);
- add_wait_queue(&ctx->ctx_zombieq, &wait);
-
- UNPROTECT_CTX(ctx, flags);
-
- /*
- * XXX: check for signals :
- * - ok for explicit close
- * - not ok when coming from exit_files()
- */
- schedule();
-
-
- PROTECT_CTX(ctx, flags);
-
-
- remove_wait_queue(&ctx->ctx_zombieq, &wait);
- set_current_state(TASK_RUNNING);
-
- /*
- * context is unloaded at this point
- */
- DPRINT(("after zombie wakeup ctx_state=%d for\n", state));
- }
- else if (task != current) {
-#ifdef CONFIG_SMP
- /*
- * switch context to zombie state
- */
- ctx->ctx_state = PFM_CTX_ZOMBIE;
-
- DPRINT(("zombie ctx for [%d]\n", task_pid_nr(task)));
- /*
- * cannot free the context on the spot. deferred until
- * the task notices the ZOMBIE state
- */
- free_possible = 0;
-#else
- pfm_context_unload(ctx, NULL, 0, regs);
-#endif
- }
-
-doit:
- /* reload state, may have changed during opening of critical section */
- state = ctx->ctx_state;
-
- /*
- * the context is still attached to a task (possibly current)
- * we cannot destroy it right now
- */
-
- /*
- * we must free the sampling buffer right here because
- * we cannot rely on it being cleaned up later by the
- * monitored task. It is not possible to free vmalloc'ed
- * memory in pfm_load_regs(). Instead, we remove the buffer
- * now. should there be subsequent PMU overflow originally
- * meant for sampling, the will be converted to spurious
- * and that's fine because the monitoring tools is gone anyway.
- */
- if (ctx->ctx_smpl_hdr) {
- smpl_buf_addr = ctx->ctx_smpl_hdr;
- smpl_buf_size = ctx->ctx_smpl_size;
- /* no more sampling */
- ctx->ctx_smpl_hdr = NULL;
- ctx->ctx_fl_is_sampling = 0;
- }
-
- DPRINT(("ctx_state=%d free_possible=%d addr=%p size=%lu\n",
- state,
- free_possible,
- smpl_buf_addr,
- smpl_buf_size));
-
- if (smpl_buf_addr) pfm_exit_smpl_buffer(ctx->ctx_buf_fmt);
-
- /*
- * UNLOADED that the session has already been unreserved.
- */
- if (state == PFM_CTX_ZOMBIE) {
- pfm_unreserve_session(ctx, ctx->ctx_fl_system , ctx->ctx_cpu);
- }
-
- /*
- * disconnect file descriptor from context must be done
- * before we unlock.
- */
- filp->private_data = NULL;
-
- /*
- * if we free on the spot, the context is now completely unreachable
- * from the callers side. The monitored task side is also cut, so we
- * can freely cut.
- *
- * If we have a deferred free, only the caller side is disconnected.
- */
- UNPROTECT_CTX(ctx, flags);
-
- /*
- * All memory free operations (especially for vmalloc'ed memory)
- * MUST be done with interrupts ENABLED.
- */
- vfree(smpl_buf_addr);
-
- /*
- * return the memory used by the context
- */
- if (free_possible) pfm_context_free(ctx);
-
- return 0;
-}
-
-static const struct file_operations pfm_file_ops = {
- .llseek = no_llseek,
- .read = pfm_read,
- .write = pfm_write,
- .poll = pfm_poll,
- .unlocked_ioctl = pfm_ioctl,
- .fasync = pfm_fasync,
- .release = pfm_close,
- .flush = pfm_flush
-};
-
-static char *pfmfs_dname(struct dentry *dentry, char *buffer, int buflen)
-{
- return dynamic_dname(dentry, buffer, buflen, "pfm:[%lu]",
- d_inode(dentry)->i_ino);
-}
-
-static const struct dentry_operations pfmfs_dentry_operations = {
- .d_delete = always_delete_dentry,
- .d_dname = pfmfs_dname,
-};
-
-
-static struct file *
-pfm_alloc_file(pfm_context_t *ctx)
-{
- struct file *file;
- struct inode *inode;
- struct path path;
- struct qstr this = { .name = "" };
-
- /*
- * allocate a new inode
- */
- inode = new_inode(pfmfs_mnt->mnt_sb);
- if (!inode)
- return ERR_PTR(-ENOMEM);
-
- DPRINT(("new inode ino=%ld @%p\n", inode->i_ino, inode));
-
- inode->i_mode = S_IFCHR|S_IRUGO;
- inode->i_uid = current_fsuid();
- inode->i_gid = current_fsgid();
-
- /*
- * allocate a new dcache entry
- */
- path.dentry = d_alloc(pfmfs_mnt->mnt_root, &this);
- if (!path.dentry) {
- iput(inode);
- return ERR_PTR(-ENOMEM);
- }
- path.mnt = mntget(pfmfs_mnt);
-
- d_add(path.dentry, inode);
-
- file = alloc_file(&path, FMODE_READ, &pfm_file_ops);
- if (IS_ERR(file)) {
- path_put(&path);
- return file;
- }
-
- file->f_flags = O_RDONLY;
- file->private_data = ctx;
-
- return file;
-}
-
-static int
-pfm_remap_buffer(struct vm_area_struct *vma, unsigned long buf, unsigned long addr, unsigned long size)
-{
- DPRINT(("CPU%d buf=0x%lx addr=0x%lx size=%ld\n", smp_processor_id(), buf, addr, size));
-
- while (size > 0) {
- unsigned long pfn = ia64_tpa(buf) >> PAGE_SHIFT;
-
-
- if (remap_pfn_range(vma, addr, pfn, PAGE_SIZE, PAGE_READONLY))
- return -ENOMEM;
-
- addr += PAGE_SIZE;
- buf += PAGE_SIZE;
- size -= PAGE_SIZE;
- }
- return 0;
-}
-
-/*
- * allocate a sampling buffer and remaps it into the user address space of the task
- */
-static int
-pfm_smpl_buffer_alloc(struct task_struct *task, struct file *filp, pfm_context_t *ctx, unsigned long rsize, void **user_vaddr)
-{
- struct mm_struct *mm = task->mm;
- struct vm_area_struct *vma = NULL;
- unsigned long size;
- void *smpl_buf;
-
-
- /*
- * the fixed header + requested size and align to page boundary
- */
- size = PAGE_ALIGN(rsize);
-
- DPRINT(("sampling buffer rsize=%lu size=%lu bytes\n", rsize, size));
-
- /*
- * check requested size to avoid Denial-of-service attacks
- * XXX: may have to refine this test
- * Check against address space limit.
- *
- * if ((mm->total_vm << PAGE_SHIFT) + len> task->rlim[RLIMIT_AS].rlim_cur)
- * return -ENOMEM;
- */
- if (size > task_rlimit(task, RLIMIT_MEMLOCK))
- return -ENOMEM;
-
- /*
- * We do the easy to undo allocations first.
- */
- smpl_buf = vzalloc(size);
- if (smpl_buf == NULL) {
- DPRINT(("Can't allocate sampling buffer\n"));
- return -ENOMEM;
- }
-
- DPRINT(("smpl_buf @%p\n", smpl_buf));
-
- /* allocate vma */
- vma = vm_area_alloc(mm);
- if (!vma) {
- DPRINT(("Cannot allocate vma\n"));
- goto error_kmem;
- }
-
- /*
- * partially initialize the vma for the sampling buffer
- */
- vma->vm_file = get_file(filp);
- vma->vm_flags = VM_READ|VM_MAYREAD|VM_DONTEXPAND|VM_DONTDUMP;
- vma->vm_page_prot = PAGE_READONLY; /* XXX may need to change */
-
- /*
- * Now we have everything we need and we can initialize
- * and connect all the data structures
- */
-
- ctx->ctx_smpl_hdr = smpl_buf;
- ctx->ctx_smpl_size = size; /* aligned size */
-
- /*
- * Let's do the difficult operations next.
- *
- * now we atomically find some area in the address space and
- * remap the buffer in it.
- */
- down_write(&task->mm->mmap_sem);
-
- /* find some free area in address space, must have mmap sem held */
- vma->vm_start = get_unmapped_area(NULL, 0, size, 0, MAP_PRIVATE|MAP_ANONYMOUS);
- if (IS_ERR_VALUE(vma->vm_start)) {
- DPRINT(("Cannot find unmapped area for size %ld\n", size));
- up_write(&task->mm->mmap_sem);
- goto error;
- }
- vma->vm_end = vma->vm_start + size;
- vma->vm_pgoff = vma->vm_start >> PAGE_SHIFT;
-
- DPRINT(("aligned size=%ld, hdr=%p mapped @0x%lx\n", size, ctx->ctx_smpl_hdr, vma->vm_start));
-
- /* can only be applied to current task, need to have the mm semaphore held when called */
- if (pfm_remap_buffer(vma, (unsigned long)smpl_buf, vma->vm_start, size)) {
- DPRINT(("Can't remap buffer\n"));
- up_write(&task->mm->mmap_sem);
- goto error;
- }
-
- /*
- * now insert the vma in the vm list for the process, must be
- * done with mmap lock held
- */
- insert_vm_struct(mm, vma);
-
- vm_stat_account(vma->vm_mm, vma->vm_flags, vma_pages(vma));
- up_write(&task->mm->mmap_sem);
-
- /*
- * keep track of user level virtual address
- */
- ctx->ctx_smpl_vaddr = (void *)vma->vm_start;
- *(unsigned long *)user_vaddr = vma->vm_start;
-
- return 0;
-
-error:
- vm_area_free(vma);
-error_kmem:
- vfree(smpl_buf);
-
- return -ENOMEM;
-}
-
-/*
- * XXX: do something better here
- */
-static int
-pfm_bad_permissions(struct task_struct *task)
-{
- const struct cred *tcred;
- kuid_t uid = current_uid();
- kgid_t gid = current_gid();
- int ret;
-
- rcu_read_lock();
- tcred = __task_cred(task);
-
- /* inspired by ptrace_attach() */
- DPRINT(("cur: uid=%d gid=%d task: euid=%d suid=%d uid=%d egid=%d sgid=%d\n",
- from_kuid(&init_user_ns, uid),
- from_kgid(&init_user_ns, gid),
- from_kuid(&init_user_ns, tcred->euid),
- from_kuid(&init_user_ns, tcred->suid),
- from_kuid(&init_user_ns, tcred->uid),
- from_kgid(&init_user_ns, tcred->egid),
- from_kgid(&init_user_ns, tcred->sgid)));
-
- ret = ((!uid_eq(uid, tcred->euid))
- || (!uid_eq(uid, tcred->suid))
- || (!uid_eq(uid, tcred->uid))
- || (!gid_eq(gid, tcred->egid))
- || (!gid_eq(gid, tcred->sgid))
- || (!gid_eq(gid, tcred->gid))) && !capable(CAP_SYS_PTRACE);
-
- rcu_read_unlock();
- return ret;
-}
-
-static int
-pfarg_is_sane(struct task_struct *task, pfarg_context_t *pfx)
-{
- int ctx_flags;
-
- /* valid signal */
-
- ctx_flags = pfx->ctx_flags;
-
- if (ctx_flags & PFM_FL_SYSTEM_WIDE) {
-
- /*
- * cannot block in this mode
- */
- if (ctx_flags & PFM_FL_NOTIFY_BLOCK) {
- DPRINT(("cannot use blocking mode when in system wide monitoring\n"));
- return -EINVAL;
- }
- } else {
- }
- /* probably more to add here */
-
- return 0;
-}
-
-static int
-pfm_setup_buffer_fmt(struct task_struct *task, struct file *filp, pfm_context_t *ctx, unsigned int ctx_flags,
- unsigned int cpu, pfarg_context_t *arg)
-{
- pfm_buffer_fmt_t *fmt = NULL;
- unsigned long size = 0UL;
- void *uaddr = NULL;
- void *fmt_arg = NULL;
- int ret = 0;
-#define PFM_CTXARG_BUF_ARG(a) (pfm_buffer_fmt_t *)(a+1)
-
- /* invoke and lock buffer format, if found */
- fmt = pfm_find_buffer_fmt(arg->ctx_smpl_buf_id);
- if (fmt == NULL) {
- DPRINT(("[%d] cannot find buffer format\n", task_pid_nr(task)));
- return -EINVAL;
- }
-
- /*
- * buffer argument MUST be contiguous to pfarg_context_t
- */
- if (fmt->fmt_arg_size) fmt_arg = PFM_CTXARG_BUF_ARG(arg);
-
- ret = pfm_buf_fmt_validate(fmt, task, ctx_flags, cpu, fmt_arg);
-
- DPRINT(("[%d] after validate(0x%x,%d,%p)=%d\n", task_pid_nr(task), ctx_flags, cpu, fmt_arg, ret));
-
- if (ret) goto error;
-
- /* link buffer format and context */
- ctx->ctx_buf_fmt = fmt;
- ctx->ctx_fl_is_sampling = 1; /* assume record() is defined */
-
- /*
- * check if buffer format wants to use perfmon buffer allocation/mapping service
- */
- ret = pfm_buf_fmt_getsize(fmt, task, ctx_flags, cpu, fmt_arg, &size);
- if (ret) goto error;
-
- if (size) {
- /*
- * buffer is always remapped into the caller's address space
- */
- ret = pfm_smpl_buffer_alloc(current, filp, ctx, size, &uaddr);
- if (ret) goto error;
-
- /* keep track of user address of buffer */
- arg->ctx_smpl_vaddr = uaddr;
- }
- ret = pfm_buf_fmt_init(fmt, task, ctx->ctx_smpl_hdr, ctx_flags, cpu, fmt_arg);
-
-error:
- return ret;
-}
-
-static void
-pfm_reset_pmu_state(pfm_context_t *ctx)
-{
- int i;
-
- /*
- * install reset values for PMC.
- */
- for (i=1; PMC_IS_LAST(i) == 0; i++) {
- if (PMC_IS_IMPL(i) == 0) continue;
- ctx->ctx_pmcs[i] = PMC_DFL_VAL(i);
- DPRINT(("pmc[%d]=0x%lx\n", i, ctx->ctx_pmcs[i]));
- }
- /*
- * PMD registers are set to 0UL when the context in memset()
- */
-
- /*
- * On context switched restore, we must restore ALL pmc and ALL pmd even
- * when they are not actively used by the task. In UP, the incoming process
- * may otherwise pick up left over PMC, PMD state from the previous process.
- * As opposed to PMD, stale PMC can cause harm to the incoming
- * process because they may change what is being measured.
- * Therefore, we must systematically reinstall the entire
- * PMC state. In SMP, the same thing is possible on the
- * same CPU but also on between 2 CPUs.
- *
- * The problem with PMD is information leaking especially
- * to user level when psr.sp=0
- *
- * There is unfortunately no easy way to avoid this problem
- * on either UP or SMP. This definitively slows down the
- * pfm_load_regs() function.
- */
-
- /*
- * bitmask of all PMCs accessible to this context
- *
- * PMC0 is treated differently.
- */
- ctx->ctx_all_pmcs[0] = pmu_conf->impl_pmcs[0] & ~0x1;
-
- /*
- * bitmask of all PMDs that are accessible to this context
- */
- ctx->ctx_all_pmds[0] = pmu_conf->impl_pmds[0];
-
- DPRINT(("<%d> all_pmcs=0x%lx all_pmds=0x%lx\n", ctx->ctx_fd, ctx->ctx_all_pmcs[0],ctx->ctx_all_pmds[0]));
-
- /*
- * useful in case of re-enable after disable
- */
- ctx->ctx_used_ibrs[0] = 0UL;
- ctx->ctx_used_dbrs[0] = 0UL;
-}
-
-static int
-pfm_ctx_getsize(void *arg, size_t *sz)
-{
- pfarg_context_t *req = (pfarg_context_t *)arg;
- pfm_buffer_fmt_t *fmt;
-
- *sz = 0;
-
- if (!pfm_uuid_cmp(req->ctx_smpl_buf_id, pfm_null_uuid)) return 0;
-
- fmt = pfm_find_buffer_fmt(req->ctx_smpl_buf_id);
- if (fmt == NULL) {
- DPRINT(("cannot find buffer format\n"));
- return -EINVAL;
- }
- /* get just enough to copy in user parameters */
- *sz = fmt->fmt_arg_size;
- DPRINT(("arg_size=%lu\n", *sz));
-
- return 0;
-}
-
-
-
-/*
- * cannot attach if :
- * - kernel task
- * - task not owned by caller
- * - task incompatible with context mode
- */
-static int
-pfm_task_incompatible(pfm_context_t *ctx, struct task_struct *task)
-{
- /*
- * no kernel task or task not owner by caller
- */
- if (task->mm == NULL) {
- DPRINT(("task [%d] has not memory context (kernel thread)\n", task_pid_nr(task)));
- return -EPERM;
- }
- if (pfm_bad_permissions(task)) {
- DPRINT(("no permission to attach to [%d]\n", task_pid_nr(task)));
- return -EPERM;
- }
- /*
- * cannot block in self-monitoring mode
- */
- if (CTX_OVFL_NOBLOCK(ctx) == 0 && task == current) {
- DPRINT(("cannot load a blocking context on self for [%d]\n", task_pid_nr(task)));
- return -EINVAL;
- }
-
- if (task->exit_state == EXIT_ZOMBIE) {
- DPRINT(("cannot attach to zombie task [%d]\n", task_pid_nr(task)));
- return -EBUSY;
- }
-
- /*
- * always ok for self
- */
- if (task == current) return 0;
-
- if (!task_is_stopped_or_traced(task)) {
- DPRINT(("cannot attach to non-stopped task [%d] state=%ld\n", task_pid_nr(task), task->state));
- return -EBUSY;
- }
- /*
- * make sure the task is off any CPU
- */
- wait_task_inactive(task, 0);
-
- /* more to come... */
-
- return 0;
-}
-
-static int
-pfm_get_task(pfm_context_t *ctx, pid_t pid, struct task_struct **task)
-{
- struct task_struct *p = current;
- int ret;
-
- /* XXX: need to add more checks here */
- if (pid < 2) return -EPERM;
-
- if (pid != task_pid_vnr(current)) {
- /* make sure task cannot go away while we operate on it */
- p = find_get_task_by_vpid(pid);
- if (!p)
- return -ESRCH;
- }
-
- ret = pfm_task_incompatible(ctx, p);
- if (ret == 0) {
- *task = p;
- } else if (p != current) {
- pfm_put_task(p);
- }
- return ret;
-}
-
-
-
-static int
-pfm_context_create(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- pfarg_context_t *req = (pfarg_context_t *)arg;
- struct file *filp;
- struct path path;
- int ctx_flags;
- int fd;
- int ret;
-
- /* let's check the arguments first */
- ret = pfarg_is_sane(current, req);
- if (ret < 0)
- return ret;
-
- ctx_flags = req->ctx_flags;
-
- ret = -ENOMEM;
-
- fd = get_unused_fd_flags(0);
- if (fd < 0)
- return fd;
-
- ctx = pfm_context_alloc(ctx_flags);
- if (!ctx)
- goto error;
-
- filp = pfm_alloc_file(ctx);
- if (IS_ERR(filp)) {
- ret = PTR_ERR(filp);
- goto error_file;
- }
-
- req->ctx_fd = ctx->ctx_fd = fd;
-
- /*
- * does the user want to sample?
- */
- if (pfm_uuid_cmp(req->ctx_smpl_buf_id, pfm_null_uuid)) {
- ret = pfm_setup_buffer_fmt(current, filp, ctx, ctx_flags, 0, req);
- if (ret)
- goto buffer_error;
- }
-
- DPRINT(("ctx=%p flags=0x%x system=%d notify_block=%d excl_idle=%d no_msg=%d ctx_fd=%d\n",
- ctx,
- ctx_flags,
- ctx->ctx_fl_system,
- ctx->ctx_fl_block,
- ctx->ctx_fl_excl_idle,
- ctx->ctx_fl_no_msg,
- ctx->ctx_fd));
-
- /*
- * initialize soft PMU state
- */
- pfm_reset_pmu_state(ctx);
-
- fd_install(fd, filp);
-
- return 0;
-
-buffer_error:
- path = filp->f_path;
- put_filp(filp);
- path_put(&path);
-
- if (ctx->ctx_buf_fmt) {
- pfm_buf_fmt_exit(ctx->ctx_buf_fmt, current, NULL, regs);
- }
-error_file:
- pfm_context_free(ctx);
-
-error:
- put_unused_fd(fd);
- return ret;
-}
-
-static inline unsigned long
-pfm_new_counter_value (pfm_counter_t *reg, int is_long_reset)
-{
- unsigned long val = is_long_reset ? reg->long_reset : reg->short_reset;
- unsigned long new_seed, old_seed = reg->seed, mask = reg->mask;
- extern unsigned long carta_random32 (unsigned long seed);
-
- if (reg->flags & PFM_REGFL_RANDOM) {
- new_seed = carta_random32(old_seed);
- val -= (old_seed & mask); /* counter values are negative numbers! */
- if ((mask >> 32) != 0)
- /* construct a full 64-bit random value: */
- new_seed |= carta_random32(old_seed >> 32) << 32;
- reg->seed = new_seed;
- }
- reg->lval = val;
- return val;
-}
-
-static void
-pfm_reset_regs_masked(pfm_context_t *ctx, unsigned long *ovfl_regs, int is_long_reset)
-{
- unsigned long mask = ovfl_regs[0];
- unsigned long reset_others = 0UL;
- unsigned long val;
- int i;
-
- /*
- * now restore reset value on sampling overflowed counters
- */
- mask >>= PMU_FIRST_COUNTER;
- for(i = PMU_FIRST_COUNTER; mask; i++, mask >>= 1) {
-
- if ((mask & 0x1UL) == 0UL) continue;
-
- ctx->ctx_pmds[i].val = val = pfm_new_counter_value(ctx->ctx_pmds+ i, is_long_reset);
- reset_others |= ctx->ctx_pmds[i].reset_pmds[0];
-
- DPRINT_ovfl((" %s reset ctx_pmds[%d]=%lx\n", is_long_reset ? "long" : "short", i, val));
- }
-
- /*
- * Now take care of resetting the other registers
- */
- for(i = 0; reset_others; i++, reset_others >>= 1) {
-
- if ((reset_others & 0x1) == 0) continue;
-
- ctx->ctx_pmds[i].val = val = pfm_new_counter_value(ctx->ctx_pmds + i, is_long_reset);
-
- DPRINT_ovfl(("%s reset_others pmd[%d]=%lx\n",
- is_long_reset ? "long" : "short", i, val));
- }
-}
-
-static void
-pfm_reset_regs(pfm_context_t *ctx, unsigned long *ovfl_regs, int is_long_reset)
-{
- unsigned long mask = ovfl_regs[0];
- unsigned long reset_others = 0UL;
- unsigned long val;
- int i;
-
- DPRINT_ovfl(("ovfl_regs=0x%lx is_long_reset=%d\n", ovfl_regs[0], is_long_reset));
-
- if (ctx->ctx_state == PFM_CTX_MASKED) {
- pfm_reset_regs_masked(ctx, ovfl_regs, is_long_reset);
- return;
- }
-
- /*
- * now restore reset value on sampling overflowed counters
- */
- mask >>= PMU_FIRST_COUNTER;
- for(i = PMU_FIRST_COUNTER; mask; i++, mask >>= 1) {
-
- if ((mask & 0x1UL) == 0UL) continue;
-
- val = pfm_new_counter_value(ctx->ctx_pmds+ i, is_long_reset);
- reset_others |= ctx->ctx_pmds[i].reset_pmds[0];
-
- DPRINT_ovfl((" %s reset ctx_pmds[%d]=%lx\n", is_long_reset ? "long" : "short", i, val));
-
- pfm_write_soft_counter(ctx, i, val);
- }
-
- /*
- * Now take care of resetting the other registers
- */
- for(i = 0; reset_others; i++, reset_others >>= 1) {
-
- if ((reset_others & 0x1) == 0) continue;
-
- val = pfm_new_counter_value(ctx->ctx_pmds + i, is_long_reset);
-
- if (PMD_IS_COUNTING(i)) {
- pfm_write_soft_counter(ctx, i, val);
- } else {
- ia64_set_pmd(i, val);
- }
- DPRINT_ovfl(("%s reset_others pmd[%d]=%lx\n",
- is_long_reset ? "long" : "short", i, val));
- }
- ia64_srlz_d();
-}
-
-static int
-pfm_write_pmcs(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct task_struct *task;
- pfarg_reg_t *req = (pfarg_reg_t *)arg;
- unsigned long value, pmc_pm;
- unsigned long smpl_pmds, reset_pmds, impl_pmds;
- unsigned int cnum, reg_flags, flags, pmc_type;
- int i, can_access_pmu = 0, is_loaded, is_system, expert_mode;
- int is_monitor, is_counting, state;
- int ret = -EINVAL;
- pfm_reg_check_t wr_func;
-#define PFM_CHECK_PMC_PM(x, y, z) ((x)->ctx_fl_system ^ PMC_PM(y, z))
-
- state = ctx->ctx_state;
- is_loaded = state == PFM_CTX_LOADED ? 1 : 0;
- is_system = ctx->ctx_fl_system;
- task = ctx->ctx_task;
- impl_pmds = pmu_conf->impl_pmds[0];
-
- if (state == PFM_CTX_ZOMBIE) return -EINVAL;
-
- if (is_loaded) {
- /*
- * In system wide and when the context is loaded, access can only happen
- * when the caller is running on the CPU being monitored by the session.
- * It does not have to be the owner (ctx_task) of the context per se.
- */
- if (is_system && ctx->ctx_cpu != smp_processor_id()) {
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- return -EBUSY;
- }
- can_access_pmu = GET_PMU_OWNER() == task || is_system ? 1 : 0;
- }
- expert_mode = pfm_sysctl.expert_mode;
-
- for (i = 0; i < count; i++, req++) {
-
- cnum = req->reg_num;
- reg_flags = req->reg_flags;
- value = req->reg_value;
- smpl_pmds = req->reg_smpl_pmds[0];
- reset_pmds = req->reg_reset_pmds[0];
- flags = 0;
-
-
- if (cnum >= PMU_MAX_PMCS) {
- DPRINT(("pmc%u is invalid\n", cnum));
- goto error;
- }
-
- pmc_type = pmu_conf->pmc_desc[cnum].type;
- pmc_pm = (value >> pmu_conf->pmc_desc[cnum].pm_pos) & 0x1;
- is_counting = (pmc_type & PFM_REG_COUNTING) == PFM_REG_COUNTING ? 1 : 0;
- is_monitor = (pmc_type & PFM_REG_MONITOR) == PFM_REG_MONITOR ? 1 : 0;
-
- /*
- * we reject all non implemented PMC as well
- * as attempts to modify PMC[0-3] which are used
- * as status registers by the PMU
- */
- if ((pmc_type & PFM_REG_IMPL) == 0 || (pmc_type & PFM_REG_CONTROL) == PFM_REG_CONTROL) {
- DPRINT(("pmc%u is unimplemented or no-access pmc_type=%x\n", cnum, pmc_type));
- goto error;
- }
- wr_func = pmu_conf->pmc_desc[cnum].write_check;
- /*
- * If the PMC is a monitor, then if the value is not the default:
- * - system-wide session: PMCx.pm=1 (privileged monitor)
- * - per-task : PMCx.pm=0 (user monitor)
- */
- if (is_monitor && value != PMC_DFL_VAL(cnum) && is_system ^ pmc_pm) {
- DPRINT(("pmc%u pmc_pm=%lu is_system=%d\n",
- cnum,
- pmc_pm,
- is_system));
- goto error;
- }
-
- if (is_counting) {
- /*
- * enforce generation of overflow interrupt. Necessary on all
- * CPUs.
- */
- value |= 1 << PMU_PMC_OI;
-
- if (reg_flags & PFM_REGFL_OVFL_NOTIFY) {
- flags |= PFM_REGFL_OVFL_NOTIFY;
- }
-
- if (reg_flags & PFM_REGFL_RANDOM) flags |= PFM_REGFL_RANDOM;
-
- /* verify validity of smpl_pmds */
- if ((smpl_pmds & impl_pmds) != smpl_pmds) {
- DPRINT(("invalid smpl_pmds 0x%lx for pmc%u\n", smpl_pmds, cnum));
- goto error;
- }
-
- /* verify validity of reset_pmds */
- if ((reset_pmds & impl_pmds) != reset_pmds) {
- DPRINT(("invalid reset_pmds 0x%lx for pmc%u\n", reset_pmds, cnum));
- goto error;
- }
- } else {
- if (reg_flags & (PFM_REGFL_OVFL_NOTIFY|PFM_REGFL_RANDOM)) {
- DPRINT(("cannot set ovfl_notify or random on pmc%u\n", cnum));
- goto error;
- }
- /* eventid on non-counting monitors are ignored */
- }
-
- /*
- * execute write checker, if any
- */
- if (likely(expert_mode == 0 && wr_func)) {
- ret = (*wr_func)(task, ctx, cnum, &value, regs);
- if (ret) goto error;
- ret = -EINVAL;
- }
-
- /*
- * no error on this register
- */
- PFM_REG_RETFLAG_SET(req->reg_flags, 0);
-
- /*
- * Now we commit the changes to the software state
- */
-
- /*
- * update overflow information
- */
- if (is_counting) {
- /*
- * full flag update each time a register is programmed
- */
- ctx->ctx_pmds[cnum].flags = flags;
-
- ctx->ctx_pmds[cnum].reset_pmds[0] = reset_pmds;
- ctx->ctx_pmds[cnum].smpl_pmds[0] = smpl_pmds;
- ctx->ctx_pmds[cnum].eventid = req->reg_smpl_eventid;
-
- /*
- * Mark all PMDS to be accessed as used.
- *
- * We do not keep track of PMC because we have to
- * systematically restore ALL of them.
- *
- * We do not update the used_monitors mask, because
- * if we have not programmed them, then will be in
- * a quiescent state, therefore we will not need to
- * mask/restore then when context is MASKED.
- */
- CTX_USED_PMD(ctx, reset_pmds);
- CTX_USED_PMD(ctx, smpl_pmds);
- /*
- * make sure we do not try to reset on
- * restart because we have established new values
- */
- if (state == PFM_CTX_MASKED) ctx->ctx_ovfl_regs[0] &= ~1UL << cnum;
- }
- /*
- * Needed in case the user does not initialize the equivalent
- * PMD. Clearing is done indirectly via pfm_reset_pmu_state() so there is no
- * possible leak here.
- */
- CTX_USED_PMD(ctx, pmu_conf->pmc_desc[cnum].dep_pmd[0]);
-
- /*
- * keep track of the monitor PMC that we are using.
- * we save the value of the pmc in ctx_pmcs[] and if
- * the monitoring is not stopped for the context we also
- * place it in the saved state area so that it will be
- * picked up later by the context switch code.
- *
- * The value in ctx_pmcs[] can only be changed in pfm_write_pmcs().
- *
- * The value in th_pmcs[] may be modified on overflow, i.e., when
- * monitoring needs to be stopped.
- */
- if (is_monitor) CTX_USED_MONITOR(ctx, 1UL << cnum);
-
- /*
- * update context state
- */
- ctx->ctx_pmcs[cnum] = value;
-
- if (is_loaded) {
- /*
- * write thread state
- */
- if (is_system == 0) ctx->th_pmcs[cnum] = value;
-
- /*
- * write hardware register if we can
- */
- if (can_access_pmu) {
- ia64_set_pmc(cnum, value);
- }
-#ifdef CONFIG_SMP
- else {
- /*
- * per-task SMP only here
- *
- * we are guaranteed that the task is not running on the other CPU,
- * we indicate that this PMD will need to be reloaded if the task
- * is rescheduled on the CPU it ran last on.
- */
- ctx->ctx_reload_pmcs[0] |= 1UL << cnum;
- }
-#endif
- }
-
- DPRINT(("pmc[%u]=0x%lx ld=%d apmu=%d flags=0x%x all_pmcs=0x%lx used_pmds=0x%lx eventid=%ld smpl_pmds=0x%lx reset_pmds=0x%lx reloads_pmcs=0x%lx used_monitors=0x%lx ovfl_regs=0x%lx\n",
- cnum,
- value,
- is_loaded,
- can_access_pmu,
- flags,
- ctx->ctx_all_pmcs[0],
- ctx->ctx_used_pmds[0],
- ctx->ctx_pmds[cnum].eventid,
- smpl_pmds,
- reset_pmds,
- ctx->ctx_reload_pmcs[0],
- ctx->ctx_used_monitors[0],
- ctx->ctx_ovfl_regs[0]));
- }
-
- /*
- * make sure the changes are visible
- */
- if (can_access_pmu) ia64_srlz_d();
-
- return 0;
-error:
- PFM_REG_RETFLAG_SET(req->reg_flags, PFM_REG_RETFL_EINVAL);
- return ret;
-}
-
-static int
-pfm_write_pmds(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct task_struct *task;
- pfarg_reg_t *req = (pfarg_reg_t *)arg;
- unsigned long value, hw_value, ovfl_mask;
- unsigned int cnum;
- int i, can_access_pmu = 0, state;
- int is_counting, is_loaded, is_system, expert_mode;
- int ret = -EINVAL;
- pfm_reg_check_t wr_func;
-
-
- state = ctx->ctx_state;
- is_loaded = state == PFM_CTX_LOADED ? 1 : 0;
- is_system = ctx->ctx_fl_system;
- ovfl_mask = pmu_conf->ovfl_val;
- task = ctx->ctx_task;
-
- if (unlikely(state == PFM_CTX_ZOMBIE)) return -EINVAL;
-
- /*
- * on both UP and SMP, we can only write to the PMC when the task is
- * the owner of the local PMU.
- */
- if (likely(is_loaded)) {
- /*
- * In system wide and when the context is loaded, access can only happen
- * when the caller is running on the CPU being monitored by the session.
- * It does not have to be the owner (ctx_task) of the context per se.
- */
- if (unlikely(is_system && ctx->ctx_cpu != smp_processor_id())) {
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- return -EBUSY;
- }
- can_access_pmu = GET_PMU_OWNER() == task || is_system ? 1 : 0;
- }
- expert_mode = pfm_sysctl.expert_mode;
-
- for (i = 0; i < count; i++, req++) {
-
- cnum = req->reg_num;
- value = req->reg_value;
-
- if (!PMD_IS_IMPL(cnum)) {
- DPRINT(("pmd[%u] is unimplemented or invalid\n", cnum));
- goto abort_mission;
- }
- is_counting = PMD_IS_COUNTING(cnum);
- wr_func = pmu_conf->pmd_desc[cnum].write_check;
-
- /*
- * execute write checker, if any
- */
- if (unlikely(expert_mode == 0 && wr_func)) {
- unsigned long v = value;
-
- ret = (*wr_func)(task, ctx, cnum, &v, regs);
- if (ret) goto abort_mission;
-
- value = v;
- ret = -EINVAL;
- }
-
- /*
- * no error on this register
- */
- PFM_REG_RETFLAG_SET(req->reg_flags, 0);
-
- /*
- * now commit changes to software state
- */
- hw_value = value;
-
- /*
- * update virtualized (64bits) counter
- */
- if (is_counting) {
- /*
- * write context state
- */
- ctx->ctx_pmds[cnum].lval = value;
-
- /*
- * when context is load we use the split value
- */
- if (is_loaded) {
- hw_value = value & ovfl_mask;
- value = value & ~ovfl_mask;
- }
- }
- /*
- * update reset values (not just for counters)
- */
- ctx->ctx_pmds[cnum].long_reset = req->reg_long_reset;
- ctx->ctx_pmds[cnum].short_reset = req->reg_short_reset;
-
- /*
- * update randomization parameters (not just for counters)
- */
- ctx->ctx_pmds[cnum].seed = req->reg_random_seed;
- ctx->ctx_pmds[cnum].mask = req->reg_random_mask;
-
- /*
- * update context value
- */
- ctx->ctx_pmds[cnum].val = value;
-
- /*
- * Keep track of what we use
- *
- * We do not keep track of PMC because we have to
- * systematically restore ALL of them.
- */
- CTX_USED_PMD(ctx, PMD_PMD_DEP(cnum));
-
- /*
- * mark this PMD register used as well
- */
- CTX_USED_PMD(ctx, RDEP(cnum));
-
- /*
- * make sure we do not try to reset on
- * restart because we have established new values
- */
- if (is_counting && state == PFM_CTX_MASKED) {
- ctx->ctx_ovfl_regs[0] &= ~1UL << cnum;
- }
-
- if (is_loaded) {
- /*
- * write thread state
- */
- if (is_system == 0) ctx->th_pmds[cnum] = hw_value;
-
- /*
- * write hardware register if we can
- */
- if (can_access_pmu) {
- ia64_set_pmd(cnum, hw_value);
- } else {
-#ifdef CONFIG_SMP
- /*
- * we are guaranteed that the task is not running on the other CPU,
- * we indicate that this PMD will need to be reloaded if the task
- * is rescheduled on the CPU it ran last on.
- */
- ctx->ctx_reload_pmds[0] |= 1UL << cnum;
-#endif
- }
- }
-
- DPRINT(("pmd[%u]=0x%lx ld=%d apmu=%d, hw_value=0x%lx ctx_pmd=0x%lx short_reset=0x%lx "
- "long_reset=0x%lx notify=%c seed=0x%lx mask=0x%lx used_pmds=0x%lx reset_pmds=0x%lx reload_pmds=0x%lx all_pmds=0x%lx ovfl_regs=0x%lx\n",
- cnum,
- value,
- is_loaded,
- can_access_pmu,
- hw_value,
- ctx->ctx_pmds[cnum].val,
- ctx->ctx_pmds[cnum].short_reset,
- ctx->ctx_pmds[cnum].long_reset,
- PMC_OVFL_NOTIFY(ctx, cnum) ? 'Y':'N',
- ctx->ctx_pmds[cnum].seed,
- ctx->ctx_pmds[cnum].mask,
- ctx->ctx_used_pmds[0],
- ctx->ctx_pmds[cnum].reset_pmds[0],
- ctx->ctx_reload_pmds[0],
- ctx->ctx_all_pmds[0],
- ctx->ctx_ovfl_regs[0]));
- }
-
- /*
- * make changes visible
- */
- if (can_access_pmu) ia64_srlz_d();
-
- return 0;
-
-abort_mission:
- /*
- * for now, we have only one possibility for error
- */
- PFM_REG_RETFLAG_SET(req->reg_flags, PFM_REG_RETFL_EINVAL);
- return ret;
-}
-
-/*
- * By the way of PROTECT_CONTEXT(), interrupts are masked while we are in this function.
- * Therefore we know, we do not have to worry about the PMU overflow interrupt. If an
- * interrupt is delivered during the call, it will be kept pending until we leave, making
- * it appears as if it had been generated at the UNPROTECT_CONTEXT(). At least we are
- * guaranteed to return consistent data to the user, it may simply be old. It is not
- * trivial to treat the overflow while inside the call because you may end up in
- * some module sampling buffer code causing deadlocks.
- */
-static int
-pfm_read_pmds(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct task_struct *task;
- unsigned long val = 0UL, lval, ovfl_mask, sval;
- pfarg_reg_t *req = (pfarg_reg_t *)arg;
- unsigned int cnum, reg_flags = 0;
- int i, can_access_pmu = 0, state;
- int is_loaded, is_system, is_counting, expert_mode;
- int ret = -EINVAL;
- pfm_reg_check_t rd_func;
-
- /*
- * access is possible when loaded only for
- * self-monitoring tasks or in UP mode
- */
-
- state = ctx->ctx_state;
- is_loaded = state == PFM_CTX_LOADED ? 1 : 0;
- is_system = ctx->ctx_fl_system;
- ovfl_mask = pmu_conf->ovfl_val;
- task = ctx->ctx_task;
-
- if (state == PFM_CTX_ZOMBIE) return -EINVAL;
-
- if (likely(is_loaded)) {
- /*
- * In system wide and when the context is loaded, access can only happen
- * when the caller is running on the CPU being monitored by the session.
- * It does not have to be the owner (ctx_task) of the context per se.
- */
- if (unlikely(is_system && ctx->ctx_cpu != smp_processor_id())) {
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- return -EBUSY;
- }
- /*
- * this can be true when not self-monitoring only in UP
- */
- can_access_pmu = GET_PMU_OWNER() == task || is_system ? 1 : 0;
-
- if (can_access_pmu) ia64_srlz_d();
- }
- expert_mode = pfm_sysctl.expert_mode;
-
- DPRINT(("ld=%d apmu=%d ctx_state=%d\n",
- is_loaded,
- can_access_pmu,
- state));
-
- /*
- * on both UP and SMP, we can only read the PMD from the hardware register when
- * the task is the owner of the local PMU.
- */
-
- for (i = 0; i < count; i++, req++) {
-
- cnum = req->reg_num;
- reg_flags = req->reg_flags;
-
- if (unlikely(!PMD_IS_IMPL(cnum))) goto error;
- /*
- * we can only read the register that we use. That includes
- * the one we explicitly initialize AND the one we want included
- * in the sampling buffer (smpl_regs).
- *
- * Having this restriction allows optimization in the ctxsw routine
- * without compromising security (leaks)
- */
- if (unlikely(!CTX_IS_USED_PMD(ctx, cnum))) goto error;
-
- sval = ctx->ctx_pmds[cnum].val;
- lval = ctx->ctx_pmds[cnum].lval;
- is_counting = PMD_IS_COUNTING(cnum);
-
- /*
- * If the task is not the current one, then we check if the
- * PMU state is still in the local live register due to lazy ctxsw.
- * If true, then we read directly from the registers.
- */
- if (can_access_pmu){
- val = ia64_get_pmd(cnum);
- } else {
- /*
- * context has been saved
- * if context is zombie, then task does not exist anymore.
- * In this case, we use the full value saved in the context (pfm_flush_regs()).
- */
- val = is_loaded ? ctx->th_pmds[cnum] : 0UL;
- }
- rd_func = pmu_conf->pmd_desc[cnum].read_check;
-
- if (is_counting) {
- /*
- * XXX: need to check for overflow when loaded
- */
- val &= ovfl_mask;
- val += sval;
- }
-
- /*
- * execute read checker, if any
- */
- if (unlikely(expert_mode == 0 && rd_func)) {
- unsigned long v = val;
- ret = (*rd_func)(ctx->ctx_task, ctx, cnum, &v, regs);
- if (ret) goto error;
- val = v;
- ret = -EINVAL;
- }
-
- PFM_REG_RETFLAG_SET(reg_flags, 0);
-
- DPRINT(("pmd[%u]=0x%lx\n", cnum, val));
-
- /*
- * update register return value, abort all if problem during copy.
- * we only modify the reg_flags field. no check mode is fine because
- * access has been verified upfront in sys_perfmonctl().
- */
- req->reg_value = val;
- req->reg_flags = reg_flags;
- req->reg_last_reset_val = lval;
- }
-
- return 0;
-
-error:
- PFM_REG_RETFLAG_SET(req->reg_flags, PFM_REG_RETFL_EINVAL);
- return ret;
-}
-
-int
-pfm_mod_write_pmcs(struct task_struct *task, void *req, unsigned int nreq, struct pt_regs *regs)
-{
- pfm_context_t *ctx;
-
- if (req == NULL) return -EINVAL;
-
- ctx = GET_PMU_CTX();
-
- if (ctx == NULL) return -EINVAL;
-
- /*
- * for now limit to current task, which is enough when calling
- * from overflow handler
- */
- if (task != current && ctx->ctx_fl_system == 0) return -EBUSY;
-
- return pfm_write_pmcs(ctx, req, nreq, regs);
-}
-EXPORT_SYMBOL(pfm_mod_write_pmcs);
-
-int
-pfm_mod_read_pmds(struct task_struct *task, void *req, unsigned int nreq, struct pt_regs *regs)
-{
- pfm_context_t *ctx;
-
- if (req == NULL) return -EINVAL;
-
- ctx = GET_PMU_CTX();
-
- if (ctx == NULL) return -EINVAL;
-
- /*
- * for now limit to current task, which is enough when calling
- * from overflow handler
- */
- if (task != current && ctx->ctx_fl_system == 0) return -EBUSY;
-
- return pfm_read_pmds(ctx, req, nreq, regs);
-}
-EXPORT_SYMBOL(pfm_mod_read_pmds);
-
-/*
- * Only call this function when a process it trying to
- * write the debug registers (reading is always allowed)
- */
-int
-pfm_use_debug_registers(struct task_struct *task)
-{
- pfm_context_t *ctx = task->thread.pfm_context;
- unsigned long flags;
- int ret = 0;
-
- if (pmu_conf->use_rr_dbregs == 0) return 0;
-
- DPRINT(("called for [%d]\n", task_pid_nr(task)));
-
- /*
- * do it only once
- */
- if (task->thread.flags & IA64_THREAD_DBG_VALID) return 0;
-
- /*
- * Even on SMP, we do not need to use an atomic here because
- * the only way in is via ptrace() and this is possible only when the
- * process is stopped. Even in the case where the ctxsw out is not totally
- * completed by the time we come here, there is no way the 'stopped' process
- * could be in the middle of fiddling with the pfm_write_ibr_dbr() routine.
- * So this is always safe.
- */
- if (ctx && ctx->ctx_fl_using_dbreg == 1) return -1;
-
- LOCK_PFS(flags);
-
- /*
- * We cannot allow setting breakpoints when system wide monitoring
- * sessions are using the debug registers.
- */
- if (pfm_sessions.pfs_sys_use_dbregs> 0)
- ret = -1;
- else
- pfm_sessions.pfs_ptrace_use_dbregs++;
-
- DPRINT(("ptrace_use_dbregs=%u sys_use_dbregs=%u by [%d] ret = %d\n",
- pfm_sessions.pfs_ptrace_use_dbregs,
- pfm_sessions.pfs_sys_use_dbregs,
- task_pid_nr(task), ret));
-
- UNLOCK_PFS(flags);
-
- return ret;
-}
-
-/*
- * This function is called for every task that exits with the
- * IA64_THREAD_DBG_VALID set. This indicates a task which was
- * able to use the debug registers for debugging purposes via
- * ptrace(). Therefore we know it was not using them for
- * performance monitoring, so we only decrement the number
- * of "ptraced" debug register users to keep the count up to date
- */
-int
-pfm_release_debug_registers(struct task_struct *task)
-{
- unsigned long flags;
- int ret;
-
- if (pmu_conf->use_rr_dbregs == 0) return 0;
-
- LOCK_PFS(flags);
- if (pfm_sessions.pfs_ptrace_use_dbregs == 0) {
- printk(KERN_ERR "perfmon: invalid release for [%d] ptrace_use_dbregs=0\n", task_pid_nr(task));
- ret = -1;
- } else {
- pfm_sessions.pfs_ptrace_use_dbregs--;
- ret = 0;
- }
- UNLOCK_PFS(flags);
-
- return ret;
-}
-
-static int
-pfm_restart(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct task_struct *task;
- pfm_buffer_fmt_t *fmt;
- pfm_ovfl_ctrl_t rst_ctrl;
- int state, is_system;
- int ret = 0;
-
- state = ctx->ctx_state;
- fmt = ctx->ctx_buf_fmt;
- is_system = ctx->ctx_fl_system;
- task = PFM_CTX_TASK(ctx);
-
- switch(state) {
- case PFM_CTX_MASKED:
- break;
- case PFM_CTX_LOADED:
- if (CTX_HAS_SMPL(ctx) && fmt->fmt_restart_active) break;
- /* fall through */
- case PFM_CTX_UNLOADED:
- case PFM_CTX_ZOMBIE:
- DPRINT(("invalid state=%d\n", state));
- return -EBUSY;
- default:
- DPRINT(("state=%d, cannot operate (no active_restart handler)\n", state));
- return -EINVAL;
- }
-
- /*
- * In system wide and when the context is loaded, access can only happen
- * when the caller is running on the CPU being monitored by the session.
- * It does not have to be the owner (ctx_task) of the context per se.
- */
- if (is_system && ctx->ctx_cpu != smp_processor_id()) {
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- return -EBUSY;
- }
-
- /* sanity check */
- if (unlikely(task == NULL)) {
- printk(KERN_ERR "perfmon: [%d] pfm_restart no task\n", task_pid_nr(current));
- return -EINVAL;
- }
-
- if (task == current || is_system) {
-
- fmt = ctx->ctx_buf_fmt;
-
- DPRINT(("restarting self %d ovfl=0x%lx\n",
- task_pid_nr(task),
- ctx->ctx_ovfl_regs[0]));
-
- if (CTX_HAS_SMPL(ctx)) {
-
- prefetch(ctx->ctx_smpl_hdr);
-
- rst_ctrl.bits.mask_monitoring = 0;
- rst_ctrl.bits.reset_ovfl_pmds = 0;
-
- if (state == PFM_CTX_LOADED)
- ret = pfm_buf_fmt_restart_active(fmt, task, &rst_ctrl, ctx->ctx_smpl_hdr, regs);
- else
- ret = pfm_buf_fmt_restart(fmt, task, &rst_ctrl, ctx->ctx_smpl_hdr, regs);
- } else {
- rst_ctrl.bits.mask_monitoring = 0;
- rst_ctrl.bits.reset_ovfl_pmds = 1;
- }
-
- if (ret == 0) {
- if (rst_ctrl.bits.reset_ovfl_pmds)
- pfm_reset_regs(ctx, ctx->ctx_ovfl_regs, PFM_PMD_LONG_RESET);
-
- if (rst_ctrl.bits.mask_monitoring == 0) {
- DPRINT(("resuming monitoring for [%d]\n", task_pid_nr(task)));
-
- if (state == PFM_CTX_MASKED) pfm_restore_monitoring(task);
- } else {
- DPRINT(("keeping monitoring stopped for [%d]\n", task_pid_nr(task)));
-
- // cannot use pfm_stop_monitoring(task, regs);
- }
- }
- /*
- * clear overflowed PMD mask to remove any stale information
- */
- ctx->ctx_ovfl_regs[0] = 0UL;
-
- /*
- * back to LOADED state
- */
- ctx->ctx_state = PFM_CTX_LOADED;
-
- /*
- * XXX: not really useful for self monitoring
- */
- ctx->ctx_fl_can_restart = 0;
-
- return 0;
- }
-
- /*
- * restart another task
- */
-
- /*
- * When PFM_CTX_MASKED, we cannot issue a restart before the previous
- * one is seen by the task.
- */
- if (state == PFM_CTX_MASKED) {
- if (ctx->ctx_fl_can_restart == 0) return -EINVAL;
- /*
- * will prevent subsequent restart before this one is
- * seen by other task
- */
- ctx->ctx_fl_can_restart = 0;
- }
-
- /*
- * if blocking, then post the semaphore is PFM_CTX_MASKED, i.e.
- * the task is blocked or on its way to block. That's the normal
- * restart path. If the monitoring is not masked, then the task
- * can be actively monitoring and we cannot directly intervene.
- * Therefore we use the trap mechanism to catch the task and
- * force it to reset the buffer/reset PMDs.
- *
- * if non-blocking, then we ensure that the task will go into
- * pfm_handle_work() before returning to user mode.
- *
- * We cannot explicitly reset another task, it MUST always
- * be done by the task itself. This works for system wide because
- * the tool that is controlling the session is logically doing
- * "self-monitoring".
- */
- if (CTX_OVFL_NOBLOCK(ctx) == 0 && state == PFM_CTX_MASKED) {
- DPRINT(("unblocking [%d]\n", task_pid_nr(task)));
- complete(&ctx->ctx_restart_done);
- } else {
- DPRINT(("[%d] armed exit trap\n", task_pid_nr(task)));
-
- ctx->ctx_fl_trap_reason = PFM_TRAP_REASON_RESET;
-
- PFM_SET_WORK_PENDING(task, 1);
-
- set_notify_resume(task);
-
- /*
- * XXX: send reschedule if task runs on another CPU
- */
- }
- return 0;
-}
-
-static int
-pfm_debug(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- unsigned int m = *(unsigned int *)arg;
-
- pfm_sysctl.debug = m == 0 ? 0 : 1;
-
- printk(KERN_INFO "perfmon debugging %s (timing reset)\n", pfm_sysctl.debug ? "on" : "off");
-
- if (m == 0) {
- memset(pfm_stats, 0, sizeof(pfm_stats));
- for(m=0; m < NR_CPUS; m++) pfm_stats[m].pfm_ovfl_intr_cycles_min = ~0UL;
- }
- return 0;
-}
-
-/*
- * arg can be NULL and count can be zero for this function
- */
-static int
-pfm_write_ibr_dbr(int mode, pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct thread_struct *thread = NULL;
- struct task_struct *task;
- pfarg_dbreg_t *req = (pfarg_dbreg_t *)arg;
- unsigned long flags;
- dbreg_t dbreg;
- unsigned int rnum;
- int first_time;
- int ret = 0, state;
- int i, can_access_pmu = 0;
- int is_system, is_loaded;
-
- if (pmu_conf->use_rr_dbregs == 0) return -EINVAL;
-
- state = ctx->ctx_state;
- is_loaded = state == PFM_CTX_LOADED ? 1 : 0;
- is_system = ctx->ctx_fl_system;
- task = ctx->ctx_task;
-
- if (state == PFM_CTX_ZOMBIE) return -EINVAL;
-
- /*
- * on both UP and SMP, we can only write to the PMC when the task is
- * the owner of the local PMU.
- */
- if (is_loaded) {
- thread = &task->thread;
- /*
- * In system wide and when the context is loaded, access can only happen
- * when the caller is running on the CPU being monitored by the session.
- * It does not have to be the owner (ctx_task) of the context per se.
- */
- if (unlikely(is_system && ctx->ctx_cpu != smp_processor_id())) {
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- return -EBUSY;
- }
- can_access_pmu = GET_PMU_OWNER() == task || is_system ? 1 : 0;
- }
-
- /*
- * we do not need to check for ipsr.db because we do clear ibr.x, dbr.r, and dbr.w
- * ensuring that no real breakpoint can be installed via this call.
- *
- * IMPORTANT: regs can be NULL in this function
- */
-
- first_time = ctx->ctx_fl_using_dbreg == 0;
-
- /*
- * don't bother if we are loaded and task is being debugged
- */
- if (is_loaded && (thread->flags & IA64_THREAD_DBG_VALID) != 0) {
- DPRINT(("debug registers already in use for [%d]\n", task_pid_nr(task)));
- return -EBUSY;
- }
-
- /*
- * check for debug registers in system wide mode
- *
- * If though a check is done in pfm_context_load(),
- * we must repeat it here, in case the registers are
- * written after the context is loaded
- */
- if (is_loaded) {
- LOCK_PFS(flags);
-
- if (first_time && is_system) {
- if (pfm_sessions.pfs_ptrace_use_dbregs)
- ret = -EBUSY;
- else
- pfm_sessions.pfs_sys_use_dbregs++;
- }
- UNLOCK_PFS(flags);
- }
-
- if (ret != 0) return ret;
-
- /*
- * mark ourself as user of the debug registers for
- * perfmon purposes.
- */
- ctx->ctx_fl_using_dbreg = 1;
-
- /*
- * clear hardware registers to make sure we don't
- * pick up stale state.
- *
- * for a system wide session, we do not use
- * thread.dbr, thread.ibr because this process
- * never leaves the current CPU and the state
- * is shared by all processes running on it
- */
- if (first_time && can_access_pmu) {
- DPRINT(("[%d] clearing ibrs, dbrs\n", task_pid_nr(task)));
- for (i=0; i < pmu_conf->num_ibrs; i++) {
- ia64_set_ibr(i, 0UL);
- ia64_dv_serialize_instruction();
- }
- ia64_srlz_i();
- for (i=0; i < pmu_conf->num_dbrs; i++) {
- ia64_set_dbr(i, 0UL);
- ia64_dv_serialize_data();
- }
- ia64_srlz_d();
- }
-
- /*
- * Now install the values into the registers
- */
- for (i = 0; i < count; i++, req++) {
-
- rnum = req->dbreg_num;
- dbreg.val = req->dbreg_value;
-
- ret = -EINVAL;
-
- if ((mode == PFM_CODE_RR && rnum >= PFM_NUM_IBRS) || ((mode == PFM_DATA_RR) && rnum >= PFM_NUM_DBRS)) {
- DPRINT(("invalid register %u val=0x%lx mode=%d i=%d count=%d\n",
- rnum, dbreg.val, mode, i, count));
-
- goto abort_mission;
- }
-
- /*
- * make sure we do not install enabled breakpoint
- */
- if (rnum & 0x1) {
- if (mode == PFM_CODE_RR)
- dbreg.ibr.ibr_x = 0;
- else
- dbreg.dbr.dbr_r = dbreg.dbr.dbr_w = 0;
- }
-
- PFM_REG_RETFLAG_SET(req->dbreg_flags, 0);
-
- /*
- * Debug registers, just like PMC, can only be modified
- * by a kernel call. Moreover, perfmon() access to those
- * registers are centralized in this routine. The hardware
- * does not modify the value of these registers, therefore,
- * if we save them as they are written, we can avoid having
- * to save them on context switch out. This is made possible
- * by the fact that when perfmon uses debug registers, ptrace()
- * won't be able to modify them concurrently.
- */
- if (mode == PFM_CODE_RR) {
- CTX_USED_IBR(ctx, rnum);
-
- if (can_access_pmu) {
- ia64_set_ibr(rnum, dbreg.val);
- ia64_dv_serialize_instruction();
- }
-
- ctx->ctx_ibrs[rnum] = dbreg.val;
-
- DPRINT(("write ibr%u=0x%lx used_ibrs=0x%x ld=%d apmu=%d\n",
- rnum, dbreg.val, ctx->ctx_used_ibrs[0], is_loaded, can_access_pmu));
- } else {
- CTX_USED_DBR(ctx, rnum);
-
- if (can_access_pmu) {
- ia64_set_dbr(rnum, dbreg.val);
- ia64_dv_serialize_data();
- }
- ctx->ctx_dbrs[rnum] = dbreg.val;
-
- DPRINT(("write dbr%u=0x%lx used_dbrs=0x%x ld=%d apmu=%d\n",
- rnum, dbreg.val, ctx->ctx_used_dbrs[0], is_loaded, can_access_pmu));
- }
- }
-
- return 0;
-
-abort_mission:
- /*
- * in case it was our first attempt, we undo the global modifications
- */
- if (first_time) {
- LOCK_PFS(flags);
- if (ctx->ctx_fl_system) {
- pfm_sessions.pfs_sys_use_dbregs--;
- }
- UNLOCK_PFS(flags);
- ctx->ctx_fl_using_dbreg = 0;
- }
- /*
- * install error return flag
- */
- PFM_REG_RETFLAG_SET(req->dbreg_flags, PFM_REG_RETFL_EINVAL);
-
- return ret;
-}
-
-static int
-pfm_write_ibrs(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- return pfm_write_ibr_dbr(PFM_CODE_RR, ctx, arg, count, regs);
-}
-
-static int
-pfm_write_dbrs(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- return pfm_write_ibr_dbr(PFM_DATA_RR, ctx, arg, count, regs);
-}
-
-int
-pfm_mod_write_ibrs(struct task_struct *task, void *req, unsigned int nreq, struct pt_regs *regs)
-{
- pfm_context_t *ctx;
-
- if (req == NULL) return -EINVAL;
-
- ctx = GET_PMU_CTX();
-
- if (ctx == NULL) return -EINVAL;
-
- /*
- * for now limit to current task, which is enough when calling
- * from overflow handler
- */
- if (task != current && ctx->ctx_fl_system == 0) return -EBUSY;
-
- return pfm_write_ibrs(ctx, req, nreq, regs);
-}
-EXPORT_SYMBOL(pfm_mod_write_ibrs);
-
-int
-pfm_mod_write_dbrs(struct task_struct *task, void *req, unsigned int nreq, struct pt_regs *regs)
-{
- pfm_context_t *ctx;
-
- if (req == NULL) return -EINVAL;
-
- ctx = GET_PMU_CTX();
-
- if (ctx == NULL) return -EINVAL;
-
- /*
- * for now limit to current task, which is enough when calling
- * from overflow handler
- */
- if (task != current && ctx->ctx_fl_system == 0) return -EBUSY;
-
- return pfm_write_dbrs(ctx, req, nreq, regs);
-}
-EXPORT_SYMBOL(pfm_mod_write_dbrs);
-
-
-static int
-pfm_get_features(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- pfarg_features_t *req = (pfarg_features_t *)arg;
-
- req->ft_version = PFM_VERSION;
- return 0;
-}
-
-static int
-pfm_stop(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct pt_regs *tregs;
- struct task_struct *task = PFM_CTX_TASK(ctx);
- int state, is_system;
-
- state = ctx->ctx_state;
- is_system = ctx->ctx_fl_system;
-
- /*
- * context must be attached to issue the stop command (includes LOADED,MASKED,ZOMBIE)
- */
- if (state == PFM_CTX_UNLOADED) return -EINVAL;
-
- /*
- * In system wide and when the context is loaded, access can only happen
- * when the caller is running on the CPU being monitored by the session.
- * It does not have to be the owner (ctx_task) of the context per se.
- */
- if (is_system && ctx->ctx_cpu != smp_processor_id()) {
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- return -EBUSY;
- }
- DPRINT(("task [%d] ctx_state=%d is_system=%d\n",
- task_pid_nr(PFM_CTX_TASK(ctx)),
- state,
- is_system));
- /*
- * in system mode, we need to update the PMU directly
- * and the user level state of the caller, which may not
- * necessarily be the creator of the context.
- */
- if (is_system) {
- /*
- * Update local PMU first
- *
- * disable dcr pp
- */
- ia64_setreg(_IA64_REG_CR_DCR, ia64_getreg(_IA64_REG_CR_DCR) & ~IA64_DCR_PP);
- ia64_srlz_i();
-
- /*
- * update local cpuinfo
- */
- PFM_CPUINFO_CLEAR(PFM_CPUINFO_DCR_PP);
-
- /*
- * stop monitoring, does srlz.i
- */
- pfm_clear_psr_pp();
-
- /*
- * stop monitoring in the caller
- */
- ia64_psr(regs)->pp = 0;
-
- return 0;
- }
- /*
- * per-task mode
- */
-
- if (task == current) {
- /* stop monitoring at kernel level */
- pfm_clear_psr_up();
-
- /*
- * stop monitoring at the user level
- */
- ia64_psr(regs)->up = 0;
- } else {
- tregs = task_pt_regs(task);
-
- /*
- * stop monitoring at the user level
- */
- ia64_psr(tregs)->up = 0;
-
- /*
- * monitoring disabled in kernel at next reschedule
- */
- ctx->ctx_saved_psr_up = 0;
- DPRINT(("task=[%d]\n", task_pid_nr(task)));
- }
- return 0;
-}
-
-
-static int
-pfm_start(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct pt_regs *tregs;
- int state, is_system;
-
- state = ctx->ctx_state;
- is_system = ctx->ctx_fl_system;
-
- if (state != PFM_CTX_LOADED) return -EINVAL;
-
- /*
- * In system wide and when the context is loaded, access can only happen
- * when the caller is running on the CPU being monitored by the session.
- * It does not have to be the owner (ctx_task) of the context per se.
- */
- if (is_system && ctx->ctx_cpu != smp_processor_id()) {
- DPRINT(("should be running on CPU%d\n", ctx->ctx_cpu));
- return -EBUSY;
- }
-
- /*
- * in system mode, we need to update the PMU directly
- * and the user level state of the caller, which may not
- * necessarily be the creator of the context.
- */
- if (is_system) {
-
- /*
- * set user level psr.pp for the caller
- */
- ia64_psr(regs)->pp = 1;
-
- /*
- * now update the local PMU and cpuinfo
- */
- PFM_CPUINFO_SET(PFM_CPUINFO_DCR_PP);
-
- /*
- * start monitoring at kernel level
- */
- pfm_set_psr_pp();
-
- /* enable dcr pp */
- ia64_setreg(_IA64_REG_CR_DCR, ia64_getreg(_IA64_REG_CR_DCR) | IA64_DCR_PP);
- ia64_srlz_i();
-
- return 0;
- }
-
- /*
- * per-process mode
- */
-
- if (ctx->ctx_task == current) {
-
- /* start monitoring at kernel level */
- pfm_set_psr_up();
-
- /*
- * activate monitoring at user level
- */
- ia64_psr(regs)->up = 1;
-
- } else {
- tregs = task_pt_regs(ctx->ctx_task);
-
- /*
- * start monitoring at the kernel level the next
- * time the task is scheduled
- */
- ctx->ctx_saved_psr_up = IA64_PSR_UP;
-
- /*
- * activate monitoring at user level
- */
- ia64_psr(tregs)->up = 1;
- }
- return 0;
-}
-
-static int
-pfm_get_pmc_reset(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- pfarg_reg_t *req = (pfarg_reg_t *)arg;
- unsigned int cnum;
- int i;
- int ret = -EINVAL;
-
- for (i = 0; i < count; i++, req++) {
-
- cnum = req->reg_num;
-
- if (!PMC_IS_IMPL(cnum)) goto abort_mission;
-
- req->reg_value = PMC_DFL_VAL(cnum);
-
- PFM_REG_RETFLAG_SET(req->reg_flags, 0);
-
- DPRINT(("pmc_reset_val pmc[%u]=0x%lx\n", cnum, req->reg_value));
- }
- return 0;
-
-abort_mission:
- PFM_REG_RETFLAG_SET(req->reg_flags, PFM_REG_RETFL_EINVAL);
- return ret;
-}
-
-static int
-pfm_check_task_exist(pfm_context_t *ctx)
-{
- struct task_struct *g, *t;
- int ret = -ESRCH;
-
- read_lock(&tasklist_lock);
-
- do_each_thread (g, t) {
- if (t->thread.pfm_context == ctx) {
- ret = 0;
- goto out;
- }
- } while_each_thread (g, t);
-out:
- read_unlock(&tasklist_lock);
-
- DPRINT(("pfm_check_task_exist: ret=%d ctx=%p\n", ret, ctx));
-
- return ret;
-}
-
-static int
-pfm_context_load(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct task_struct *task;
- struct thread_struct *thread;
- struct pfm_context_t *old;
- unsigned long flags;
-#ifndef CONFIG_SMP
- struct task_struct *owner_task = NULL;
-#endif
- pfarg_load_t *req = (pfarg_load_t *)arg;
- unsigned long *pmcs_source, *pmds_source;
- int the_cpu;
- int ret = 0;
- int state, is_system, set_dbregs = 0;
-
- state = ctx->ctx_state;
- is_system = ctx->ctx_fl_system;
- /*
- * can only load from unloaded or terminated state
- */
- if (state != PFM_CTX_UNLOADED) {
- DPRINT(("cannot load to [%d], invalid ctx_state=%d\n",
- req->load_pid,
- ctx->ctx_state));
- return -EBUSY;
- }
-
- DPRINT(("load_pid [%d] using_dbreg=%d\n", req->load_pid, ctx->ctx_fl_using_dbreg));
-
- if (CTX_OVFL_NOBLOCK(ctx) == 0 && req->load_pid == current->pid) {
- DPRINT(("cannot use blocking mode on self\n"));
- return -EINVAL;
- }
-
- ret = pfm_get_task(ctx, req->load_pid, &task);
- if (ret) {
- DPRINT(("load_pid [%d] get_task=%d\n", req->load_pid, ret));
- return ret;
- }
-
- ret = -EINVAL;
-
- /*
- * system wide is self monitoring only
- */
- if (is_system && task != current) {
- DPRINT(("system wide is self monitoring only load_pid=%d\n",
- req->load_pid));
- goto error;
- }
-
- thread = &task->thread;
-
- ret = 0;
- /*
- * cannot load a context which is using range restrictions,
- * into a task that is being debugged.
- */
- if (ctx->ctx_fl_using_dbreg) {
- if (thread->flags & IA64_THREAD_DBG_VALID) {
- ret = -EBUSY;
- DPRINT(("load_pid [%d] task is debugged, cannot load range restrictions\n", req->load_pid));
- goto error;
- }
- LOCK_PFS(flags);
-
- if (is_system) {
- if (pfm_sessions.pfs_ptrace_use_dbregs) {
- DPRINT(("cannot load [%d] dbregs in use\n",
- task_pid_nr(task)));
- ret = -EBUSY;
- } else {
- pfm_sessions.pfs_sys_use_dbregs++;
- DPRINT(("load [%d] increased sys_use_dbreg=%u\n", task_pid_nr(task), pfm_sessions.pfs_sys_use_dbregs));
- set_dbregs = 1;
- }
- }
-
- UNLOCK_PFS(flags);
-
- if (ret) goto error;
- }
-
- /*
- * SMP system-wide monitoring implies self-monitoring.
- *
- * The programming model expects the task to
- * be pinned on a CPU throughout the session.
- * Here we take note of the current CPU at the
- * time the context is loaded. No call from
- * another CPU will be allowed.
- *
- * The pinning via shed_setaffinity()
- * must be done by the calling task prior
- * to this call.
- *
- * systemwide: keep track of CPU this session is supposed to run on
- */
- the_cpu = ctx->ctx_cpu = smp_processor_id();
-
- ret = -EBUSY;
- /*
- * now reserve the session
- */
- ret = pfm_reserve_session(current, is_system, the_cpu);
- if (ret) goto error;
-
- /*
- * task is necessarily stopped at this point.
- *
- * If the previous context was zombie, then it got removed in
- * pfm_save_regs(). Therefore we should not see it here.
- * If we see a context, then this is an active context
- *
- * XXX: needs to be atomic
- */
- DPRINT(("before cmpxchg() old_ctx=%p new_ctx=%p\n",
- thread->pfm_context, ctx));
-
- ret = -EBUSY;
- old = ia64_cmpxchg(acq, &thread->pfm_context, NULL, ctx, sizeof(pfm_context_t *));
- if (old != NULL) {
- DPRINT(("load_pid [%d] already has a context\n", req->load_pid));
- goto error_unres;
- }
-
- pfm_reset_msgq(ctx);
-
- ctx->ctx_state = PFM_CTX_LOADED;
-
- /*
- * link context to task
- */
- ctx->ctx_task = task;
-
- if (is_system) {
- /*
- * we load as stopped
- */
- PFM_CPUINFO_SET(PFM_CPUINFO_SYST_WIDE);
- PFM_CPUINFO_CLEAR(PFM_CPUINFO_DCR_PP);
-
- if (ctx->ctx_fl_excl_idle) PFM_CPUINFO_SET(PFM_CPUINFO_EXCL_IDLE);
- } else {
- thread->flags |= IA64_THREAD_PM_VALID;
- }
-
- /*
- * propagate into thread-state
- */
- pfm_copy_pmds(task, ctx);
- pfm_copy_pmcs(task, ctx);
-
- pmcs_source = ctx->th_pmcs;
- pmds_source = ctx->th_pmds;
-
- /*
- * always the case for system-wide
- */
- if (task == current) {
-
- if (is_system == 0) {
-
- /* allow user level control */
- ia64_psr(regs)->sp = 0;
- DPRINT(("clearing psr.sp for [%d]\n", task_pid_nr(task)));
-
- SET_LAST_CPU(ctx, smp_processor_id());
- INC_ACTIVATION();
- SET_ACTIVATION(ctx);
-#ifndef CONFIG_SMP
- /*
- * push the other task out, if any
- */
- owner_task = GET_PMU_OWNER();
- if (owner_task) pfm_lazy_save_regs(owner_task);
-#endif
- }
- /*
- * load all PMD from ctx to PMU (as opposed to thread state)
- * restore all PMC from ctx to PMU
- */
- pfm_restore_pmds(pmds_source, ctx->ctx_all_pmds[0]);
- pfm_restore_pmcs(pmcs_source, ctx->ctx_all_pmcs[0]);
-
- ctx->ctx_reload_pmcs[0] = 0UL;
- ctx->ctx_reload_pmds[0] = 0UL;
-
- /*
- * guaranteed safe by earlier check against DBG_VALID
- */
- if (ctx->ctx_fl_using_dbreg) {
- pfm_restore_ibrs(ctx->ctx_ibrs, pmu_conf->num_ibrs);
- pfm_restore_dbrs(ctx->ctx_dbrs, pmu_conf->num_dbrs);
- }
- /*
- * set new ownership
- */
- SET_PMU_OWNER(task, ctx);
-
- DPRINT(("context loaded on PMU for [%d]\n", task_pid_nr(task)));
- } else {
- /*
- * when not current, task MUST be stopped, so this is safe
- */
- regs = task_pt_regs(task);
-
- /* force a full reload */
- ctx->ctx_last_activation = PFM_INVALID_ACTIVATION;
- SET_LAST_CPU(ctx, -1);
-
- /* initial saved psr (stopped) */
- ctx->ctx_saved_psr_up = 0UL;
- ia64_psr(regs)->up = ia64_psr(regs)->pp = 0;
- }
-
- ret = 0;
-
-error_unres:
- if (ret) pfm_unreserve_session(ctx, ctx->ctx_fl_system, the_cpu);
-error:
- /*
- * we must undo the dbregs setting (for system-wide)
- */
- if (ret && set_dbregs) {
- LOCK_PFS(flags);
- pfm_sessions.pfs_sys_use_dbregs--;
- UNLOCK_PFS(flags);
- }
- /*
- * release task, there is now a link with the context
- */
- if (is_system == 0 && task != current) {
- pfm_put_task(task);
-
- if (ret == 0) {
- ret = pfm_check_task_exist(ctx);
- if (ret) {
- ctx->ctx_state = PFM_CTX_UNLOADED;
- ctx->ctx_task = NULL;
- }
- }
- }
- return ret;
-}
-
-/*
- * in this function, we do not need to increase the use count
- * for the task via get_task_struct(), because we hold the
- * context lock. If the task were to disappear while having
- * a context attached, it would go through pfm_exit_thread()
- * which also grabs the context lock and would therefore be blocked
- * until we are here.
- */
-static void pfm_flush_pmds(struct task_struct *, pfm_context_t *ctx);
-
-static int
-pfm_context_unload(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs)
-{
- struct task_struct *task = PFM_CTX_TASK(ctx);
- struct pt_regs *tregs;
- int prev_state, is_system;
- int ret;
-
- DPRINT(("ctx_state=%d task [%d]\n", ctx->ctx_state, task ? task_pid_nr(task) : -1));
-
- prev_state = ctx->ctx_state;
- is_system = ctx->ctx_fl_system;
-
- /*
- * unload only when necessary
- */
- if (prev_state == PFM_CTX_UNLOADED) {
- DPRINT(("ctx_state=%d, nothing to do\n", prev_state));
- return 0;
- }
-
- /*
- * clear psr and dcr bits
- */
- ret = pfm_stop(ctx, NULL, 0, regs);
- if (ret) return ret;
-
- ctx->ctx_state = PFM_CTX_UNLOADED;
-
- /*
- * in system mode, we need to update the PMU directly
- * and the user level state of the caller, which may not
- * necessarily be the creator of the context.
- */
- if (is_system) {
-
- /*
- * Update cpuinfo
- *
- * local PMU is taken care of in pfm_stop()
- */
- PFM_CPUINFO_CLEAR(PFM_CPUINFO_SYST_WIDE);
- PFM_CPUINFO_CLEAR(PFM_CPUINFO_EXCL_IDLE);
-
- /*
- * save PMDs in context
- * release ownership
- */
- pfm_flush_pmds(current, ctx);
-
- /*
- * at this point we are done with the PMU
- * so we can unreserve the resource.
- */
- if (prev_state != PFM_CTX_ZOMBIE)
- pfm_unreserve_session(ctx, 1 , ctx->ctx_cpu);
-
- /*
- * disconnect context from task
- */
- task->thread.pfm_context = NULL;
- /*
- * disconnect task from context
- */
- ctx->ctx_task = NULL;
-
- /*
- * There is nothing more to cleanup here.
- */
- return 0;
- }
-
- /*
- * per-task mode
- */
- tregs = task == current ? regs : task_pt_regs(task);
-
- if (task == current) {
- /*
- * cancel user level control
- */
- ia64_psr(regs)->sp = 1;
-
- DPRINT(("setting psr.sp for [%d]\n", task_pid_nr(task)));
- }
- /*
- * save PMDs to context
- * release ownership
- */
- pfm_flush_pmds(task, ctx);
-
- /*
- * at this point we are done with the PMU
- * so we can unreserve the resource.
- *
- * when state was ZOMBIE, we have already unreserved.
- */
- if (prev_state != PFM_CTX_ZOMBIE)
- pfm_unreserve_session(ctx, 0 , ctx->ctx_cpu);
-
- /*
- * reset activation counter and psr
- */
- ctx->ctx_last_activation = PFM_INVALID_ACTIVATION;
- SET_LAST_CPU(ctx, -1);
-
- /*
- * PMU state will not be restored
- */
- task->thread.flags &= ~IA64_THREAD_PM_VALID;
-
- /*
- * break links between context and task
- */
- task->thread.pfm_context = NULL;
- ctx->ctx_task = NULL;
-
- PFM_SET_WORK_PENDING(task, 0);
-
- ctx->ctx_fl_trap_reason = PFM_TRAP_REASON_NONE;
- ctx->ctx_fl_can_restart = 0;
- ctx->ctx_fl_going_zombie = 0;
-
- DPRINT(("disconnected [%d] from context\n", task_pid_nr(task)));
-
- return 0;
-}
-
-
-/*
- * called only from exit_thread()
- * we come here only if the task has a context attached (loaded or masked)
- */
-void
-pfm_exit_thread(struct task_struct *task)
-{
- pfm_context_t *ctx;
- unsigned long flags;
- struct pt_regs *regs = task_pt_regs(task);
- int ret, state;
- int free_ok = 0;
-
- ctx = PFM_GET_CTX(task);
-
- PROTECT_CTX(ctx, flags);
-
- DPRINT(("state=%d task [%d]\n", ctx->ctx_state, task_pid_nr(task)));
-
- state = ctx->ctx_state;
- switch(state) {
- case PFM_CTX_UNLOADED:
- /*
- * only comes to this function if pfm_context is not NULL, i.e., cannot
- * be in unloaded state
- */
- printk(KERN_ERR "perfmon: pfm_exit_thread [%d] ctx unloaded\n", task_pid_nr(task));
- break;
- case PFM_CTX_LOADED:
- case PFM_CTX_MASKED:
- ret = pfm_context_unload(ctx, NULL, 0, regs);
- if (ret) {
- printk(KERN_ERR "perfmon: pfm_exit_thread [%d] state=%d unload failed %d\n", task_pid_nr(task), state, ret);
- }
- DPRINT(("ctx unloaded for current state was %d\n", state));
-
- pfm_end_notify_user(ctx);
- break;
- case PFM_CTX_ZOMBIE:
- ret = pfm_context_unload(ctx, NULL, 0, regs);
- if (ret) {
- printk(KERN_ERR "perfmon: pfm_exit_thread [%d] state=%d unload failed %d\n", task_pid_nr(task), state, ret);
- }
- free_ok = 1;
- break;
- default:
- printk(KERN_ERR "perfmon: pfm_exit_thread [%d] unexpected state=%d\n", task_pid_nr(task), state);
- break;
- }
- UNPROTECT_CTX(ctx, flags);
-
- { u64 psr = pfm_get_psr();
- BUG_ON(psr & (IA64_PSR_UP|IA64_PSR_PP));
- BUG_ON(GET_PMU_OWNER());
- BUG_ON(ia64_psr(regs)->up);
- BUG_ON(ia64_psr(regs)->pp);
- }
-
- /*
- * All memory free operations (especially for vmalloc'ed memory)
- * MUST be done with interrupts ENABLED.
- */
- if (free_ok) pfm_context_free(ctx);
-}
-
-/*
- * functions MUST be listed in the increasing order of their index (see permfon.h)
- */
-#define PFM_CMD(name, flags, arg_count, arg_type, getsz) { name, #name, flags, arg_count, sizeof(arg_type), getsz }
-#define PFM_CMD_S(name, flags) { name, #name, flags, 0, 0, NULL }
-#define PFM_CMD_PCLRWS (PFM_CMD_FD|PFM_CMD_ARG_RW|PFM_CMD_STOP)
-#define PFM_CMD_PCLRW (PFM_CMD_FD|PFM_CMD_ARG_RW)
-#define PFM_CMD_NONE { NULL, "no-cmd", 0, 0, 0, NULL}
-
-static pfm_cmd_desc_t pfm_cmd_tab[]={
-/* 0 */PFM_CMD_NONE,
-/* 1 */PFM_CMD(pfm_write_pmcs, PFM_CMD_PCLRWS, PFM_CMD_ARG_MANY, pfarg_reg_t, NULL),
-/* 2 */PFM_CMD(pfm_write_pmds, PFM_CMD_PCLRWS, PFM_CMD_ARG_MANY, pfarg_reg_t, NULL),
-/* 3 */PFM_CMD(pfm_read_pmds, PFM_CMD_PCLRWS, PFM_CMD_ARG_MANY, pfarg_reg_t, NULL),
-/* 4 */PFM_CMD_S(pfm_stop, PFM_CMD_PCLRWS),
-/* 5 */PFM_CMD_S(pfm_start, PFM_CMD_PCLRWS),
-/* 6 */PFM_CMD_NONE,
-/* 7 */PFM_CMD_NONE,
-/* 8 */PFM_CMD(pfm_context_create, PFM_CMD_ARG_RW, 1, pfarg_context_t, pfm_ctx_getsize),
-/* 9 */PFM_CMD_NONE,
-/* 10 */PFM_CMD_S(pfm_restart, PFM_CMD_PCLRW),
-/* 11 */PFM_CMD_NONE,
-/* 12 */PFM_CMD(pfm_get_features, PFM_CMD_ARG_RW, 1, pfarg_features_t, NULL),
-/* 13 */PFM_CMD(pfm_debug, 0, 1, unsigned int, NULL),
-/* 14 */PFM_CMD_NONE,
-/* 15 */PFM_CMD(pfm_get_pmc_reset, PFM_CMD_ARG_RW, PFM_CMD_ARG_MANY, pfarg_reg_t, NULL),
-/* 16 */PFM_CMD(pfm_context_load, PFM_CMD_PCLRWS, 1, pfarg_load_t, NULL),
-/* 17 */PFM_CMD_S(pfm_context_unload, PFM_CMD_PCLRWS),
-/* 18 */PFM_CMD_NONE,
-/* 19 */PFM_CMD_NONE,
-/* 20 */PFM_CMD_NONE,
-/* 21 */PFM_CMD_NONE,
-/* 22 */PFM_CMD_NONE,
-/* 23 */PFM_CMD_NONE,
-/* 24 */PFM_CMD_NONE,
-/* 25 */PFM_CMD_NONE,
-/* 26 */PFM_CMD_NONE,
-/* 27 */PFM_CMD_NONE,
-/* 28 */PFM_CMD_NONE,
-/* 29 */PFM_CMD_NONE,
-/* 30 */PFM_CMD_NONE,
-/* 31 */PFM_CMD_NONE,
-/* 32 */PFM_CMD(pfm_write_ibrs, PFM_CMD_PCLRWS, PFM_CMD_ARG_MANY, pfarg_dbreg_t, NULL),
-/* 33 */PFM_CMD(pfm_write_dbrs, PFM_CMD_PCLRWS, PFM_CMD_ARG_MANY, pfarg_dbreg_t, NULL)
-};
-#define PFM_CMD_COUNT (sizeof(pfm_cmd_tab)/sizeof(pfm_cmd_desc_t))
-
-static int
-pfm_check_task_state(pfm_context_t *ctx, int cmd, unsigned long flags)
-{
- struct task_struct *task;
- int state, old_state;
-
-recheck:
- state = ctx->ctx_state;
- task = ctx->ctx_task;
-
- if (task == NULL) {
- DPRINT(("context %d no task, state=%d\n", ctx->ctx_fd, state));
- return 0;
- }
-
- DPRINT(("context %d state=%d [%d] task_state=%ld must_stop=%d\n",
- ctx->ctx_fd,
- state,
- task_pid_nr(task),
- task->state, PFM_CMD_STOPPED(cmd)));
-
- /*
- * self-monitoring always ok.
- *
- * for system-wide the caller can either be the creator of the
- * context (to one to which the context is attached to) OR
- * a task running on the same CPU as the session.
- */
- if (task == current || ctx->ctx_fl_system) return 0;
-
- /*
- * we are monitoring another thread
- */
- switch(state) {
- case PFM_CTX_UNLOADED:
- /*
- * if context is UNLOADED we are safe to go
- */
- return 0;
- case PFM_CTX_ZOMBIE:
- /*
- * no command can operate on a zombie context
- */
- DPRINT(("cmd %d state zombie cannot operate on context\n", cmd));
- return -EINVAL;
- case PFM_CTX_MASKED:
- /*
- * PMU state has been saved to software even though
- * the thread may still be running.
- */
- if (cmd != PFM_UNLOAD_CONTEXT) return 0;
- }
-
- /*
- * context is LOADED or MASKED. Some commands may need to have
- * the task stopped.
- *
- * We could lift this restriction for UP but it would mean that
- * the user has no guarantee the task would not run between
- * two successive calls to perfmonctl(). That's probably OK.
- * If this user wants to ensure the task does not run, then
- * the task must be stopped.
- */
- if (PFM_CMD_STOPPED(cmd)) {
- if (!task_is_stopped_or_traced(task)) {
- DPRINT(("[%d] task not in stopped state\n", task_pid_nr(task)));
- return -EBUSY;
- }
- /*
- * task is now stopped, wait for ctxsw out
- *
- * This is an interesting point in the code.
- * We need to unprotect the context because
- * the pfm_save_regs() routines needs to grab
- * the same lock. There are danger in doing
- * this because it leaves a window open for
- * another task to get access to the context
- * and possibly change its state. The one thing
- * that is not possible is for the context to disappear
- * because we are protected by the VFS layer, i.e.,
- * get_fd()/put_fd().
- */
- old_state = state;
-
- UNPROTECT_CTX(ctx, flags);
-
- wait_task_inactive(task, 0);
-
- PROTECT_CTX(ctx, flags);
-
- /*
- * we must recheck to verify if state has changed
- */
- if (ctx->ctx_state != old_state) {
- DPRINT(("old_state=%d new_state=%d\n", old_state, ctx->ctx_state));
- goto recheck;
- }
- }
- return 0;
-}
-
-/*
- * system-call entry point (must return long)
- */
-asmlinkage long
-sys_perfmonctl (int fd, int cmd, void __user *arg, int count)
-{
- struct fd f = {NULL, 0};
- pfm_context_t *ctx = NULL;
- unsigned long flags = 0UL;
- void *args_k = NULL;
- long ret; /* will expand int return types */
- size_t base_sz, sz, xtra_sz = 0;
- int narg, completed_args = 0, call_made = 0, cmd_flags;
- int (*func)(pfm_context_t *ctx, void *arg, int count, struct pt_regs *regs);
- int (*getsize)(void *arg, size_t *sz);
-#define PFM_MAX_ARGSIZE 4096
-
- /*
- * reject any call if perfmon was disabled at initialization
- */
- if (unlikely(pmu_conf == NULL)) return -ENOSYS;
-
- if (unlikely(cmd < 0 || cmd >= PFM_CMD_COUNT)) {
- DPRINT(("invalid cmd=%d\n", cmd));
- return -EINVAL;
- }
-
- func = pfm_cmd_tab[cmd].cmd_func;
- narg = pfm_cmd_tab[cmd].cmd_narg;
- base_sz = pfm_cmd_tab[cmd].cmd_argsize;
- getsize = pfm_cmd_tab[cmd].cmd_getsize;
- cmd_flags = pfm_cmd_tab[cmd].cmd_flags;
-
- if (unlikely(func == NULL)) {
- DPRINT(("invalid cmd=%d\n", cmd));
- return -EINVAL;
- }
-
- DPRINT(("cmd=%s idx=%d narg=0x%x argsz=%lu count=%d\n",
- PFM_CMD_NAME(cmd),
- cmd,
- narg,
- base_sz,
- count));
-
- /*
- * check if number of arguments matches what the command expects
- */
- if (unlikely((narg == PFM_CMD_ARG_MANY && count <= 0) || (narg > 0 && narg != count)))
- return -EINVAL;
-
-restart_args:
- sz = xtra_sz + base_sz*count;
- /*
- * limit abuse to min page size
- */
- if (unlikely(sz > PFM_MAX_ARGSIZE)) {
- printk(KERN_ERR "perfmon: [%d] argument too big %lu\n", task_pid_nr(current), sz);
- return -E2BIG;
- }
-
- /*
- * allocate default-sized argument buffer
- */
- if (likely(count && args_k == NULL)) {
- args_k = kmalloc(PFM_MAX_ARGSIZE, GFP_KERNEL);
- if (args_k == NULL) return -ENOMEM;
- }
-
- ret = -EFAULT;
-
- /*
- * copy arguments
- *
- * assume sz = 0 for command without parameters
- */
- if (sz && copy_from_user(args_k, arg, sz)) {
- DPRINT(("cannot copy_from_user %lu bytes @%p\n", sz, arg));
- goto error_args;
- }
-
- /*
- * check if command supports extra parameters
- */
- if (completed_args == 0 && getsize) {
- /*
- * get extra parameters size (based on main argument)
- */
- ret = (*getsize)(args_k, &xtra_sz);
- if (ret) goto error_args;
-
- completed_args = 1;
-
- DPRINT(("restart_args sz=%lu xtra_sz=%lu\n", sz, xtra_sz));
-
- /* retry if necessary */
- if (likely(xtra_sz)) goto restart_args;
- }
-
- if (unlikely((cmd_flags & PFM_CMD_FD) == 0)) goto skip_fd;
-
- ret = -EBADF;
-
- f = fdget(fd);
- if (unlikely(f.file == NULL)) {
- DPRINT(("invalid fd %d\n", fd));
- goto error_args;
- }
- if (unlikely(PFM_IS_FILE(f.file) == 0)) {
- DPRINT(("fd %d not related to perfmon\n", fd));
- goto error_args;
- }
-
- ctx = f.file->private_data;
- if (unlikely(ctx == NULL)) {
- DPRINT(("no context for fd %d\n", fd));
- goto error_args;
- }
- prefetch(&ctx->ctx_state);
-
- PROTECT_CTX(ctx, flags);
-
- /*
- * check task is stopped
- */
- ret = pfm_check_task_state(ctx, cmd, flags);
- if (unlikely(ret)) goto abort_locked;
-
-skip_fd:
- ret = (*func)(ctx, args_k, count, task_pt_regs(current));
-
- call_made = 1;
-
-abort_locked:
- if (likely(ctx)) {
- DPRINT(("context unlocked\n"));
- UNPROTECT_CTX(ctx, flags);
- }
-
- /* copy argument back to user, if needed */
- if (call_made && PFM_CMD_RW_ARG(cmd) && copy_to_user(arg, args_k, base_sz*count)) ret = -EFAULT;
-
-error_args:
- if (f.file)
- fdput(f);
-
- kfree(args_k);
-
- DPRINT(("cmd=%s ret=%ld\n", PFM_CMD_NAME(cmd), ret));
-
- return ret;
-}
-
-static void
-pfm_resume_after_ovfl(pfm_context_t *ctx, unsigned long ovfl_regs, struct pt_regs *regs)
-{
- pfm_buffer_fmt_t *fmt = ctx->ctx_buf_fmt;
- pfm_ovfl_ctrl_t rst_ctrl;
- int state;
- int ret = 0;
-
- state = ctx->ctx_state;
- /*
- * Unlock sampling buffer and reset index atomically
- * XXX: not really needed when blocking
- */
- if (CTX_HAS_SMPL(ctx)) {
-
- rst_ctrl.bits.mask_monitoring = 0;
- rst_ctrl.bits.reset_ovfl_pmds = 0;
-
- if (state == PFM_CTX_LOADED)
- ret = pfm_buf_fmt_restart_active(fmt, current, &rst_ctrl, ctx->ctx_smpl_hdr, regs);
- else
- ret = pfm_buf_fmt_restart(fmt, current, &rst_ctrl, ctx->ctx_smpl_hdr, regs);
- } else {
- rst_ctrl.bits.mask_monitoring = 0;
- rst_ctrl.bits.reset_ovfl_pmds = 1;
- }
-
- if (ret == 0) {
- if (rst_ctrl.bits.reset_ovfl_pmds) {
- pfm_reset_regs(ctx, &ovfl_regs, PFM_PMD_LONG_RESET);
- }
- if (rst_ctrl.bits.mask_monitoring == 0) {
- DPRINT(("resuming monitoring\n"));
- if (ctx->ctx_state == PFM_CTX_MASKED) pfm_restore_monitoring(current);
- } else {
- DPRINT(("stopping monitoring\n"));
- //pfm_stop_monitoring(current, regs);
- }
- ctx->ctx_state = PFM_CTX_LOADED;
- }
-}
-
-/*
- * context MUST BE LOCKED when calling
- * can only be called for current
- */
-static void
-pfm_context_force_terminate(pfm_context_t *ctx, struct pt_regs *regs)
-{
- int ret;
-
- DPRINT(("entering for [%d]\n", task_pid_nr(current)));
-
- ret = pfm_context_unload(ctx, NULL, 0, regs);
- if (ret) {
- printk(KERN_ERR "pfm_context_force_terminate: [%d] unloaded failed with %d\n", task_pid_nr(current), ret);
- }
-
- /*
- * and wakeup controlling task, indicating we are now disconnected
- */
- wake_up_interruptible(&ctx->ctx_zombieq);
-
- /*
- * given that context is still locked, the controlling
- * task will only get access when we return from
- * pfm_handle_work().
- */
-}
-
-static int pfm_ovfl_notify_user(pfm_context_t *ctx, unsigned long ovfl_pmds);
-
- /*
- * pfm_handle_work() can be called with interrupts enabled
- * (TIF_NEED_RESCHED) or disabled. The down_interruptible
- * call may sleep, therefore we must re-enable interrupts
- * to avoid deadlocks. It is safe to do so because this function
- * is called ONLY when returning to user level (pUStk=1), in which case
- * there is no risk of kernel stack overflow due to deep
- * interrupt nesting.
- */
-void
-pfm_handle_work(void)
-{
- pfm_context_t *ctx;
- struct pt_regs *regs;
- unsigned long flags, dummy_flags;
- unsigned long ovfl_regs;
- unsigned int reason;
- int ret;
-
- ctx = PFM_GET_CTX(current);
- if (ctx == NULL) {
- printk(KERN_ERR "perfmon: [%d] has no PFM context\n",
- task_pid_nr(current));
- return;
- }
-
- PROTECT_CTX(ctx, flags);
-
- PFM_SET_WORK_PENDING(current, 0);
-
- regs = task_pt_regs(current);
-
- /*
- * extract reason for being here and clear
- */
- reason = ctx->ctx_fl_trap_reason;
- ctx->ctx_fl_trap_reason = PFM_TRAP_REASON_NONE;
- ovfl_regs = ctx->ctx_ovfl_regs[0];
-
- DPRINT(("reason=%d state=%d\n", reason, ctx->ctx_state));
-
- /*
- * must be done before we check for simple-reset mode
- */
- if (ctx->ctx_fl_going_zombie || ctx->ctx_state == PFM_CTX_ZOMBIE)
- goto do_zombie;
-
- //if (CTX_OVFL_NOBLOCK(ctx)) goto skip_blocking;
- if (reason == PFM_TRAP_REASON_RESET)
- goto skip_blocking;
-
- /*
- * restore interrupt mask to what it was on entry.
- * Could be enabled/diasbled.
- */
- UNPROTECT_CTX(ctx, flags);
-
- /*
- * force interrupt enable because of down_interruptible()
- */
- local_irq_enable();
-
- DPRINT(("before block sleeping\n"));
-
- /*
- * may go through without blocking on SMP systems
- * if restart has been received already by the time we call down()
- */
- ret = wait_for_completion_interruptible(&ctx->ctx_restart_done);
-
- DPRINT(("after block sleeping ret=%d\n", ret));
-
- /*
- * lock context and mask interrupts again
- * We save flags into a dummy because we may have
- * altered interrupts mask compared to entry in this
- * function.
- */
- PROTECT_CTX(ctx, dummy_flags);
-
- /*
- * we need to read the ovfl_regs only after wake-up
- * because we may have had pfm_write_pmds() in between
- * and that can changed PMD values and therefore
- * ovfl_regs is reset for these new PMD values.
- */
- ovfl_regs = ctx->ctx_ovfl_regs[0];
-
- if (ctx->ctx_fl_going_zombie) {
-do_zombie:
- DPRINT(("context is zombie, bailing out\n"));
- pfm_context_force_terminate(ctx, regs);
- goto nothing_to_do;
- }
- /*
- * in case of interruption of down() we don't restart anything
- */
- if (ret < 0)
- goto nothing_to_do;
-
-skip_blocking:
- pfm_resume_after_ovfl(ctx, ovfl_regs, regs);
- ctx->ctx_ovfl_regs[0] = 0UL;
-
-nothing_to_do:
- /*
- * restore flags as they were upon entry
- */
- UNPROTECT_CTX(ctx, flags);
-}
-
-static int
-pfm_notify_user(pfm_context_t *ctx, pfm_msg_t *msg)
-{
- if (ctx->ctx_state == PFM_CTX_ZOMBIE) {
- DPRINT(("ignoring overflow notification, owner is zombie\n"));
- return 0;
- }
-
- DPRINT(("waking up somebody\n"));
-
- if (msg) wake_up_interruptible(&ctx->ctx_msgq_wait);
-
- /*
- * safe, we are not in intr handler, nor in ctxsw when
- * we come here
- */
- kill_fasync (&ctx->ctx_async_queue, SIGIO, POLL_IN);
-
- return 0;
-}
-
-static int
-pfm_ovfl_notify_user(pfm_context_t *ctx, unsigned long ovfl_pmds)
-{
- pfm_msg_t *msg = NULL;
-
- if (ctx->ctx_fl_no_msg == 0) {
- msg = pfm_get_new_msg(ctx);
- if (msg == NULL) {
- printk(KERN_ERR "perfmon: pfm_ovfl_notify_user no more notification msgs\n");
- return -1;
- }
-
- msg->pfm_ovfl_msg.msg_type = PFM_MSG_OVFL;
- msg->pfm_ovfl_msg.msg_ctx_fd = ctx->ctx_fd;
- msg->pfm_ovfl_msg.msg_active_set = 0;
- msg->pfm_ovfl_msg.msg_ovfl_pmds[0] = ovfl_pmds;
- msg->pfm_ovfl_msg.msg_ovfl_pmds[1] = 0UL;
- msg->pfm_ovfl_msg.msg_ovfl_pmds[2] = 0UL;
- msg->pfm_ovfl_msg.msg_ovfl_pmds[3] = 0UL;
- msg->pfm_ovfl_msg.msg_tstamp = 0UL;
- }
-
- DPRINT(("ovfl msg: msg=%p no_msg=%d fd=%d ovfl_pmds=0x%lx\n",
- msg,
- ctx->ctx_fl_no_msg,
- ctx->ctx_fd,
- ovfl_pmds));
-
- return pfm_notify_user(ctx, msg);
-}
-
-static int
-pfm_end_notify_user(pfm_context_t *ctx)
-{
- pfm_msg_t *msg;
-
- msg = pfm_get_new_msg(ctx);
- if (msg == NULL) {
- printk(KERN_ERR "perfmon: pfm_end_notify_user no more notification msgs\n");
- return -1;
- }
- /* no leak */
- memset(msg, 0, sizeof(*msg));
-
- msg->pfm_end_msg.msg_type = PFM_MSG_END;
- msg->pfm_end_msg.msg_ctx_fd = ctx->ctx_fd;
- msg->pfm_ovfl_msg.msg_tstamp = 0UL;
-
- DPRINT(("end msg: msg=%p no_msg=%d ctx_fd=%d\n",
- msg,
- ctx->ctx_fl_no_msg,
- ctx->ctx_fd));
-
- return pfm_notify_user(ctx, msg);
-}
-
-/*
- * main overflow processing routine.
- * it can be called from the interrupt path or explicitly during the context switch code
- */
-static void pfm_overflow_handler(struct task_struct *task, pfm_context_t *ctx,
- unsigned long pmc0, struct pt_regs *regs)
-{
- pfm_ovfl_arg_t *ovfl_arg;
- unsigned long mask;
- unsigned long old_val, ovfl_val, new_val;
- unsigned long ovfl_notify = 0UL, ovfl_pmds = 0UL, smpl_pmds = 0UL, reset_pmds;
- unsigned long tstamp;
- pfm_ovfl_ctrl_t ovfl_ctrl;
- unsigned int i, has_smpl;
- int must_notify = 0;
-
- if (unlikely(ctx->ctx_state == PFM_CTX_ZOMBIE)) goto stop_monitoring;
-
- /*
- * sanity test. Should never happen
- */
- if (unlikely((pmc0 & 0x1) == 0)) goto sanity_check;
-
- tstamp = ia64_get_itc();
- mask = pmc0 >> PMU_FIRST_COUNTER;
- ovfl_val = pmu_conf->ovfl_val;
- has_smpl = CTX_HAS_SMPL(ctx);
-
- DPRINT_ovfl(("pmc0=0x%lx pid=%d iip=0x%lx, %s "
- "used_pmds=0x%lx\n",
- pmc0,
- task ? task_pid_nr(task): -1,
- (regs ? regs->cr_iip : 0),
- CTX_OVFL_NOBLOCK(ctx) ? "nonblocking" : "blocking",
- ctx->ctx_used_pmds[0]));
-
-
- /*
- * first we update the virtual counters
- * assume there was a prior ia64_srlz_d() issued
- */
- for (i = PMU_FIRST_COUNTER; mask ; i++, mask >>= 1) {
-
- /* skip pmd which did not overflow */
- if ((mask & 0x1) == 0) continue;
-
- /*
- * Note that the pmd is not necessarily 0 at this point as qualified events
- * may have happened before the PMU was frozen. The residual count is not
- * taken into consideration here but will be with any read of the pmd via
- * pfm_read_pmds().
- */
- old_val = new_val = ctx->ctx_pmds[i].val;
- new_val += 1 + ovfl_val;
- ctx->ctx_pmds[i].val = new_val;
-
- /*
- * check for overflow condition
- */
- if (likely(old_val > new_val)) {
- ovfl_pmds |= 1UL << i;
- if (PMC_OVFL_NOTIFY(ctx, i)) ovfl_notify |= 1UL << i;
- }
-
- DPRINT_ovfl(("ctx_pmd[%d].val=0x%lx old_val=0x%lx pmd=0x%lx ovfl_pmds=0x%lx ovfl_notify=0x%lx\n",
- i,
- new_val,
- old_val,
- ia64_get_pmd(i) & ovfl_val,
- ovfl_pmds,
- ovfl_notify));
- }
-
- /*
- * there was no 64-bit overflow, nothing else to do
- */
- if (ovfl_pmds == 0UL) return;
-
- /*
- * reset all control bits
- */
- ovfl_ctrl.val = 0;
- reset_pmds = 0UL;
-
- /*
- * if a sampling format module exists, then we "cache" the overflow by
- * calling the module's handler() routine.
- */
- if (has_smpl) {
- unsigned long start_cycles, end_cycles;
- unsigned long pmd_mask;
- int j, k, ret = 0;
- int this_cpu = smp_processor_id();
-
- pmd_mask = ovfl_pmds >> PMU_FIRST_COUNTER;
- ovfl_arg = &ctx->ctx_ovfl_arg;
-
- prefetch(ctx->ctx_smpl_hdr);
-
- for(i=PMU_FIRST_COUNTER; pmd_mask && ret == 0; i++, pmd_mask >>=1) {
-
- mask = 1UL << i;
-
- if ((pmd_mask & 0x1) == 0) continue;
-
- ovfl_arg->ovfl_pmd = (unsigned char )i;
- ovfl_arg->ovfl_notify = ovfl_notify & mask ? 1 : 0;
- ovfl_arg->active_set = 0;
- ovfl_arg->ovfl_ctrl.val = 0; /* module must fill in all fields */
- ovfl_arg->smpl_pmds[0] = smpl_pmds = ctx->ctx_pmds[i].smpl_pmds[0];
-
- ovfl_arg->pmd_value = ctx->ctx_pmds[i].val;
- ovfl_arg->pmd_last_reset = ctx->ctx_pmds[i].lval;
- ovfl_arg->pmd_eventid = ctx->ctx_pmds[i].eventid;
-
- /*
- * copy values of pmds of interest. Sampling format may copy them
- * into sampling buffer.
- */
- if (smpl_pmds) {
- for(j=0, k=0; smpl_pmds; j++, smpl_pmds >>=1) {
- if ((smpl_pmds & 0x1) == 0) continue;
- ovfl_arg->smpl_pmds_values[k++] = PMD_IS_COUNTING(j) ? pfm_read_soft_counter(ctx, j) : ia64_get_pmd(j);
- DPRINT_ovfl(("smpl_pmd[%d]=pmd%u=0x%lx\n", k-1, j, ovfl_arg->smpl_pmds_values[k-1]));
- }
- }
-
- pfm_stats[this_cpu].pfm_smpl_handler_calls++;
-
- start_cycles = ia64_get_itc();
-
- /*
- * call custom buffer format record (handler) routine
- */
- ret = (*ctx->ctx_buf_fmt->fmt_handler)(task, ctx->ctx_smpl_hdr, ovfl_arg, regs, tstamp);
-
- end_cycles = ia64_get_itc();
-
- /*
- * For those controls, we take the union because they have
- * an all or nothing behavior.
- */
- ovfl_ctrl.bits.notify_user |= ovfl_arg->ovfl_ctrl.bits.notify_user;
- ovfl_ctrl.bits.block_task |= ovfl_arg->ovfl_ctrl.bits.block_task;
- ovfl_ctrl.bits.mask_monitoring |= ovfl_arg->ovfl_ctrl.bits.mask_monitoring;
- /*
- * build the bitmask of pmds to reset now
- */
- if (ovfl_arg->ovfl_ctrl.bits.reset_ovfl_pmds) reset_pmds |= mask;
-
- pfm_stats[this_cpu].pfm_smpl_handler_cycles += end_cycles - start_cycles;
- }
- /*
- * when the module cannot handle the rest of the overflows, we abort right here
- */
- if (ret && pmd_mask) {
- DPRINT(("handler aborts leftover ovfl_pmds=0x%lx\n",
- pmd_mask<<PMU_FIRST_COUNTER));
- }
- /*
- * remove the pmds we reset now from the set of pmds to reset in pfm_restart()
- */
- ovfl_pmds &= ~reset_pmds;
- } else {
- /*
- * when no sampling module is used, then the default
- * is to notify on overflow if requested by user
- */
- ovfl_ctrl.bits.notify_user = ovfl_notify ? 1 : 0;
- ovfl_ctrl.bits.block_task = ovfl_notify ? 1 : 0;
- ovfl_ctrl.bits.mask_monitoring = ovfl_notify ? 1 : 0; /* XXX: change for saturation */
- ovfl_ctrl.bits.reset_ovfl_pmds = ovfl_notify ? 0 : 1;
- /*
- * if needed, we reset all overflowed pmds
- */
- if (ovfl_notify == 0) reset_pmds = ovfl_pmds;
- }
-
- DPRINT_ovfl(("ovfl_pmds=0x%lx reset_pmds=0x%lx\n", ovfl_pmds, reset_pmds));
-
- /*
- * reset the requested PMD registers using the short reset values
- */
- if (reset_pmds) {
- unsigned long bm = reset_pmds;
- pfm_reset_regs(ctx, &bm, PFM_PMD_SHORT_RESET);
- }
-
- if (ovfl_notify && ovfl_ctrl.bits.notify_user) {
- /*
- * keep track of what to reset when unblocking
- */
- ctx->ctx_ovfl_regs[0] = ovfl_pmds;
-
- /*
- * check for blocking context
- */
- if (CTX_OVFL_NOBLOCK(ctx) == 0 && ovfl_ctrl.bits.block_task) {
-
- ctx->ctx_fl_trap_reason = PFM_TRAP_REASON_BLOCK;
-
- /*
- * set the perfmon specific checking pending work for the task
- */
- PFM_SET_WORK_PENDING(task, 1);
-
- /*
- * when coming from ctxsw, current still points to the
- * previous task, therefore we must work with task and not current.
- */
- set_notify_resume(task);
- }
- /*
- * defer until state is changed (shorten spin window). the context is locked
- * anyway, so the signal receiver would come spin for nothing.
- */
- must_notify = 1;
- }
-
- DPRINT_ovfl(("owner [%d] pending=%ld reason=%u ovfl_pmds=0x%lx ovfl_notify=0x%lx masked=%d\n",
- GET_PMU_OWNER() ? task_pid_nr(GET_PMU_OWNER()) : -1,
- PFM_GET_WORK_PENDING(task),
- ctx->ctx_fl_trap_reason,
- ovfl_pmds,
- ovfl_notify,
- ovfl_ctrl.bits.mask_monitoring ? 1 : 0));
- /*
- * in case monitoring must be stopped, we toggle the psr bits
- */
- if (ovfl_ctrl.bits.mask_monitoring) {
- pfm_mask_monitoring(task);
- ctx->ctx_state = PFM_CTX_MASKED;
- ctx->ctx_fl_can_restart = 1;
- }
-
- /*
- * send notification now
- */
- if (must_notify) pfm_ovfl_notify_user(ctx, ovfl_notify);
-
- return;
-
-sanity_check:
- printk(KERN_ERR "perfmon: CPU%d overflow handler [%d] pmc0=0x%lx\n",
- smp_processor_id(),
- task ? task_pid_nr(task) : -1,
- pmc0);
- return;
-
-stop_monitoring:
- /*
- * in SMP, zombie context is never restored but reclaimed in pfm_load_regs().
- * Moreover, zombies are also reclaimed in pfm_save_regs(). Therefore we can
- * come here as zombie only if the task is the current task. In which case, we
- * can access the PMU hardware directly.
- *
- * Note that zombies do have PM_VALID set. So here we do the minimal.
- *
- * In case the context was zombified it could not be reclaimed at the time
- * the monitoring program exited. At this point, the PMU reservation has been
- * returned, the sampiing buffer has been freed. We must convert this call
- * into a spurious interrupt. However, we must also avoid infinite overflows
- * by stopping monitoring for this task. We can only come here for a per-task
- * context. All we need to do is to stop monitoring using the psr bits which
- * are always task private. By re-enabling secure montioring, we ensure that
- * the monitored task will not be able to re-activate monitoring.
- * The task will eventually be context switched out, at which point the context
- * will be reclaimed (that includes releasing ownership of the PMU).
- *
- * So there might be a window of time where the number of per-task session is zero
- * yet one PMU might have a owner and get at most one overflow interrupt for a zombie
- * context. This is safe because if a per-task session comes in, it will push this one
- * out and by the virtue on pfm_save_regs(), this one will disappear. If a system wide
- * session is force on that CPU, given that we use task pinning, pfm_save_regs() will
- * also push our zombie context out.
- *
- * Overall pretty hairy stuff....
- */
- DPRINT(("ctx is zombie for [%d], converted to spurious\n", task ? task_pid_nr(task): -1));
- pfm_clear_psr_up();
- ia64_psr(regs)->up = 0;
- ia64_psr(regs)->sp = 1;
- return;
-}
-
-static int
-pfm_do_interrupt_handler(void *arg, struct pt_regs *regs)
-{
- struct task_struct *task;
- pfm_context_t *ctx;
- unsigned long flags;
- u64 pmc0;
- int this_cpu = smp_processor_id();
- int retval = 0;
-
- pfm_stats[this_cpu].pfm_ovfl_intr_count++;
-
- /*
- * srlz.d done before arriving here
- */
- pmc0 = ia64_get_pmc(0);
-
- task = GET_PMU_OWNER();
- ctx = GET_PMU_CTX();
-
- /*
- * if we have some pending bits set
- * assumes : if any PMC0.bit[63-1] is set, then PMC0.fr = 1
- */
- if (PMC0_HAS_OVFL(pmc0) && task) {
- /*
- * we assume that pmc0.fr is always set here
- */
-
- /* sanity check */
- if (!ctx) goto report_spurious1;
-
- if (ctx->ctx_fl_system == 0 && (task->thread.flags & IA64_THREAD_PM_VALID) == 0)
- goto report_spurious2;
-
- PROTECT_CTX_NOPRINT(ctx, flags);
-
- pfm_overflow_handler(task, ctx, pmc0, regs);
-
- UNPROTECT_CTX_NOPRINT(ctx, flags);
-
- } else {
- pfm_stats[this_cpu].pfm_spurious_ovfl_intr_count++;
- retval = -1;
- }
- /*
- * keep it unfrozen at all times
- */
- pfm_unfreeze_pmu();
-
- return retval;
-
-report_spurious1:
- printk(KERN_INFO "perfmon: spurious overflow interrupt on CPU%d: process %d has no PFM context\n",
- this_cpu, task_pid_nr(task));
- pfm_unfreeze_pmu();
- return -1;
-report_spurious2:
- printk(KERN_INFO "perfmon: spurious overflow interrupt on CPU%d: process %d, invalid flag\n",
- this_cpu,
- task_pid_nr(task));
- pfm_unfreeze_pmu();
- return -1;
-}
-
-static irqreturn_t
-pfm_interrupt_handler(int irq, void *arg)
-{
- unsigned long start_cycles, total_cycles;
- unsigned long min, max;
- int this_cpu;
- int ret;
- struct pt_regs *regs = get_irq_regs();
-
- this_cpu = get_cpu();
- if (likely(!pfm_alt_intr_handler)) {
- min = pfm_stats[this_cpu].pfm_ovfl_intr_cycles_min;
- max = pfm_stats[this_cpu].pfm_ovfl_intr_cycles_max;
-
- start_cycles = ia64_get_itc();
-
- ret = pfm_do_interrupt_handler(arg, regs);
-
- total_cycles = ia64_get_itc();
-
- /*
- * don't measure spurious interrupts
- */
- if (likely(ret == 0)) {
- total_cycles -= start_cycles;
-
- if (total_cycles < min) pfm_stats[this_cpu].pfm_ovfl_intr_cycles_min = total_cycles;
- if (total_cycles > max) pfm_stats[this_cpu].pfm_ovfl_intr_cycles_max = total_cycles;
-
- pfm_stats[this_cpu].pfm_ovfl_intr_cycles += total_cycles;
- }
- }
- else {
- (*pfm_alt_intr_handler->handler)(irq, arg, regs);
- }
-
- put_cpu();
- return IRQ_HANDLED;
-}
-
-/*
- * /proc/perfmon interface, for debug only
- */
-
-#define PFM_PROC_SHOW_HEADER ((void *)(long)nr_cpu_ids+1)
-
-static void *
-pfm_proc_start(struct seq_file *m, loff_t *pos)
-{
- if (*pos == 0) {
- return PFM_PROC_SHOW_HEADER;
- }
-
- while (*pos <= nr_cpu_ids) {
- if (cpu_online(*pos - 1)) {
- return (void *)*pos;
- }
- ++*pos;
- }
- return NULL;
-}
-
-static void *
-pfm_proc_next(struct seq_file *m, void *v, loff_t *pos)
-{
- ++*pos;
- return pfm_proc_start(m, pos);
-}
-
-static void
-pfm_proc_stop(struct seq_file *m, void *v)
-{
-}
-
-static void
-pfm_proc_show_header(struct seq_file *m)
-{
- struct list_head * pos;
- pfm_buffer_fmt_t * entry;
- unsigned long flags;
-
- seq_printf(m,
- "perfmon version : %u.%u\n"
- "model : %s\n"
- "fastctxsw : %s\n"
- "expert mode : %s\n"
- "ovfl_mask : 0x%lx\n"
- "PMU flags : 0x%x\n",
- PFM_VERSION_MAJ, PFM_VERSION_MIN,
- pmu_conf->pmu_name,
- pfm_sysctl.fastctxsw > 0 ? "Yes": "No",
- pfm_sysctl.expert_mode > 0 ? "Yes": "No",
- pmu_conf->ovfl_val,
- pmu_conf->flags);
-
- LOCK_PFS(flags);
-
- seq_printf(m,
- "proc_sessions : %u\n"
- "sys_sessions : %u\n"
- "sys_use_dbregs : %u\n"
- "ptrace_use_dbregs : %u\n",
- pfm_sessions.pfs_task_sessions,
- pfm_sessions.pfs_sys_sessions,
- pfm_sessions.pfs_sys_use_dbregs,
- pfm_sessions.pfs_ptrace_use_dbregs);
-
- UNLOCK_PFS(flags);
-
- spin_lock(&pfm_buffer_fmt_lock);
-
- list_for_each(pos, &pfm_buffer_fmt_list) {
- entry = list_entry(pos, pfm_buffer_fmt_t, fmt_list);
- seq_printf(m, "format : %16phD %s\n",
- entry->fmt_uuid, entry->fmt_name);
- }
- spin_unlock(&pfm_buffer_fmt_lock);
-
-}
-
-static int
-pfm_proc_show(struct seq_file *m, void *v)
-{
- unsigned long psr;
- unsigned int i;
- int cpu;
-
- if (v == PFM_PROC_SHOW_HEADER) {
- pfm_proc_show_header(m);
- return 0;
- }
-
- /* show info for CPU (v - 1) */
-
- cpu = (long)v - 1;
- seq_printf(m,
- "CPU%-2d overflow intrs : %lu\n"
- "CPU%-2d overflow cycles : %lu\n"
- "CPU%-2d overflow min : %lu\n"
- "CPU%-2d overflow max : %lu\n"
- "CPU%-2d smpl handler calls : %lu\n"
- "CPU%-2d smpl handler cycles : %lu\n"
- "CPU%-2d spurious intrs : %lu\n"
- "CPU%-2d replay intrs : %lu\n"
- "CPU%-2d syst_wide : %d\n"
- "CPU%-2d dcr_pp : %d\n"
- "CPU%-2d exclude idle : %d\n"
- "CPU%-2d owner : %d\n"
- "CPU%-2d context : %p\n"
- "CPU%-2d activations : %lu\n",
- cpu, pfm_stats[cpu].pfm_ovfl_intr_count,
- cpu, pfm_stats[cpu].pfm_ovfl_intr_cycles,
- cpu, pfm_stats[cpu].pfm_ovfl_intr_cycles_min,
- cpu, pfm_stats[cpu].pfm_ovfl_intr_cycles_max,
- cpu, pfm_stats[cpu].pfm_smpl_handler_calls,
- cpu, pfm_stats[cpu].pfm_smpl_handler_cycles,
- cpu, pfm_stats[cpu].pfm_spurious_ovfl_intr_count,
- cpu, pfm_stats[cpu].pfm_replay_ovfl_intr_count,
- cpu, pfm_get_cpu_data(pfm_syst_info, cpu) & PFM_CPUINFO_SYST_WIDE ? 1 : 0,
- cpu, pfm_get_cpu_data(pfm_syst_info, cpu) & PFM_CPUINFO_DCR_PP ? 1 : 0,
- cpu, pfm_get_cpu_data(pfm_syst_info, cpu) & PFM_CPUINFO_EXCL_IDLE ? 1 : 0,
- cpu, pfm_get_cpu_data(pmu_owner, cpu) ? pfm_get_cpu_data(pmu_owner, cpu)->pid: -1,
- cpu, pfm_get_cpu_data(pmu_ctx, cpu),
- cpu, pfm_get_cpu_data(pmu_activation_number, cpu));
-
- if (num_online_cpus() == 1 && pfm_sysctl.debug > 0) {
-
- psr = pfm_get_psr();
-
- ia64_srlz_d();
-
- seq_printf(m,
- "CPU%-2d psr : 0x%lx\n"
- "CPU%-2d pmc0 : 0x%lx\n",
- cpu, psr,
- cpu, ia64_get_pmc(0));
-
- for (i=0; PMC_IS_LAST(i) == 0; i++) {
- if (PMC_IS_COUNTING(i) == 0) continue;
- seq_printf(m,
- "CPU%-2d pmc%u : 0x%lx\n"
- "CPU%-2d pmd%u : 0x%lx\n",
- cpu, i, ia64_get_pmc(i),
- cpu, i, ia64_get_pmd(i));
- }
- }
- return 0;
-}
-
-const struct seq_operations pfm_seq_ops = {
- .start = pfm_proc_start,
- .next = pfm_proc_next,
- .stop = pfm_proc_stop,
- .show = pfm_proc_show
-};
-
-/*
- * we come here as soon as local_cpu_data->pfm_syst_wide is set. this happens
- * during pfm_enable() hence before pfm_start(). We cannot assume monitoring
- * is active or inactive based on mode. We must rely on the value in
- * local_cpu_data->pfm_syst_info
- */
-void
-pfm_syst_wide_update_task(struct task_struct *task, unsigned long info, int is_ctxswin)
-{
- struct pt_regs *regs;
- unsigned long dcr;
- unsigned long dcr_pp;
-
- dcr_pp = info & PFM_CPUINFO_DCR_PP ? 1 : 0;
-
- /*
- * pid 0 is guaranteed to be the idle task. There is one such task with pid 0
- * on every CPU, so we can rely on the pid to identify the idle task.
- */
- if ((info & PFM_CPUINFO_EXCL_IDLE) == 0 || task->pid) {
- regs = task_pt_regs(task);
- ia64_psr(regs)->pp = is_ctxswin ? dcr_pp : 0;
- return;
- }
- /*
- * if monitoring has started
- */
- if (dcr_pp) {
- dcr = ia64_getreg(_IA64_REG_CR_DCR);
- /*
- * context switching in?
- */
- if (is_ctxswin) {
- /* mask monitoring for the idle task */
- ia64_setreg(_IA64_REG_CR_DCR, dcr & ~IA64_DCR_PP);
- pfm_clear_psr_pp();
- ia64_srlz_i();
- return;
- }
- /*
- * context switching out
- * restore monitoring for next task
- *
- * Due to inlining this odd if-then-else construction generates
- * better code.
- */
- ia64_setreg(_IA64_REG_CR_DCR, dcr |IA64_DCR_PP);
- pfm_set_psr_pp();
- ia64_srlz_i();
- }
-}
-
-#ifdef CONFIG_SMP
-
-static void
-pfm_force_cleanup(pfm_context_t *ctx, struct pt_regs *regs)
-{
- struct task_struct *task = ctx->ctx_task;
-
- ia64_psr(regs)->up = 0;
- ia64_psr(regs)->sp = 1;
-
- if (GET_PMU_OWNER() == task) {
- DPRINT(("cleared ownership for [%d]\n",
- task_pid_nr(ctx->ctx_task)));
- SET_PMU_OWNER(NULL, NULL);
- }
-
- /*
- * disconnect the task from the context and vice-versa
- */
- PFM_SET_WORK_PENDING(task, 0);
-
- task->thread.pfm_context = NULL;
- task->thread.flags &= ~IA64_THREAD_PM_VALID;
-
- DPRINT(("force cleanup for [%d]\n", task_pid_nr(task)));
-}
-
-
-/*
- * in 2.6, interrupts are masked when we come here and the runqueue lock is held
- */
-void
-pfm_save_regs(struct task_struct *task)
-{
- pfm_context_t *ctx;
- unsigned long flags;
- u64 psr;
-
-
- ctx = PFM_GET_CTX(task);
- if (ctx == NULL) return;
-
- /*
- * we always come here with interrupts ALREADY disabled by
- * the scheduler. So we simply need to protect against concurrent
- * access, not CPU concurrency.
- */
- flags = pfm_protect_ctx_ctxsw(ctx);
-
- if (ctx->ctx_state == PFM_CTX_ZOMBIE) {
- struct pt_regs *regs = task_pt_regs(task);
-
- pfm_clear_psr_up();
-
- pfm_force_cleanup(ctx, regs);
-
- BUG_ON(ctx->ctx_smpl_hdr);
-
- pfm_unprotect_ctx_ctxsw(ctx, flags);
-
- pfm_context_free(ctx);
- return;
- }
-
- /*
- * save current PSR: needed because we modify it
- */
- ia64_srlz_d();
- psr = pfm_get_psr();
-
- BUG_ON(psr & (IA64_PSR_I));
-
- /*
- * stop monitoring:
- * This is the last instruction which may generate an overflow
- *
- * We do not need to set psr.sp because, it is irrelevant in kernel.
- * It will be restored from ipsr when going back to user level
- */
- pfm_clear_psr_up();
-
- /*
- * keep a copy of psr.up (for reload)
- */
- ctx->ctx_saved_psr_up = psr & IA64_PSR_UP;
-
- /*
- * release ownership of this PMU.
- * PM interrupts are masked, so nothing
- * can happen.
- */
- SET_PMU_OWNER(NULL, NULL);
-
- /*
- * we systematically save the PMD as we have no
- * guarantee we will be schedule at that same
- * CPU again.
- */
- pfm_save_pmds(ctx->th_pmds, ctx->ctx_used_pmds[0]);
-
- /*
- * save pmc0 ia64_srlz_d() done in pfm_save_pmds()
- * we will need it on the restore path to check
- * for pending overflow.
- */
- ctx->th_pmcs[0] = ia64_get_pmc(0);
-
- /*
- * unfreeze PMU if had pending overflows
- */
- if (ctx->th_pmcs[0] & ~0x1UL) pfm_unfreeze_pmu();
-
- /*
- * finally, allow context access.
- * interrupts will still be masked after this call.
- */
- pfm_unprotect_ctx_ctxsw(ctx, flags);
-}
-
-#else /* !CONFIG_SMP */
-void
-pfm_save_regs(struct task_struct *task)
-{
- pfm_context_t *ctx;
- u64 psr;
-
- ctx = PFM_GET_CTX(task);
- if (ctx == NULL) return;
-
- /*
- * save current PSR: needed because we modify it
- */
- psr = pfm_get_psr();
-
- BUG_ON(psr & (IA64_PSR_I));
-
- /*
- * stop monitoring:
- * This is the last instruction which may generate an overflow
- *
- * We do not need to set psr.sp because, it is irrelevant in kernel.
- * It will be restored from ipsr when going back to user level
- */
- pfm_clear_psr_up();
-
- /*
- * keep a copy of psr.up (for reload)
- */
- ctx->ctx_saved_psr_up = psr & IA64_PSR_UP;
-}
-
-static void
-pfm_lazy_save_regs (struct task_struct *task)
-{
- pfm_context_t *ctx;
- unsigned long flags;
-
- { u64 psr = pfm_get_psr();
- BUG_ON(psr & IA64_PSR_UP);
- }
-
- ctx = PFM_GET_CTX(task);
-
- /*
- * we need to mask PMU overflow here to
- * make sure that we maintain pmc0 until
- * we save it. overflow interrupts are
- * treated as spurious if there is no
- * owner.
- *
- * XXX: I don't think this is necessary
- */
- PROTECT_CTX(ctx,flags);
-
- /*
- * release ownership of this PMU.
- * must be done before we save the registers.
- *
- * after this call any PMU interrupt is treated
- * as spurious.
- */
- SET_PMU_OWNER(NULL, NULL);
-
- /*
- * save all the pmds we use
- */
- pfm_save_pmds(ctx->th_pmds, ctx->ctx_used_pmds[0]);
-
- /*
- * save pmc0 ia64_srlz_d() done in pfm_save_pmds()
- * it is needed to check for pended overflow
- * on the restore path
- */
- ctx->th_pmcs[0] = ia64_get_pmc(0);
-
- /*
- * unfreeze PMU if had pending overflows
- */
- if (ctx->th_pmcs[0] & ~0x1UL) pfm_unfreeze_pmu();
-
- /*
- * now get can unmask PMU interrupts, they will
- * be treated as purely spurious and we will not
- * lose any information
- */
- UNPROTECT_CTX(ctx,flags);
-}
-#endif /* CONFIG_SMP */
-
-#ifdef CONFIG_SMP
-/*
- * in 2.6, interrupts are masked when we come here and the runqueue lock is held
- */
-void
-pfm_load_regs (struct task_struct *task)
-{
- pfm_context_t *ctx;
- unsigned long pmc_mask = 0UL, pmd_mask = 0UL;
- unsigned long flags;
- u64 psr, psr_up;
- int need_irq_resend;
-
- ctx = PFM_GET_CTX(task);
- if (unlikely(ctx == NULL)) return;
-
- BUG_ON(GET_PMU_OWNER());
-
- /*
- * possible on unload
- */
- if (unlikely((task->thread.flags & IA64_THREAD_PM_VALID) == 0)) return;
-
- /*
- * we always come here with interrupts ALREADY disabled by
- * the scheduler. So we simply need to protect against concurrent
- * access, not CPU concurrency.
- */
- flags = pfm_protect_ctx_ctxsw(ctx);
- psr = pfm_get_psr();
-
- need_irq_resend = pmu_conf->flags & PFM_PMU_IRQ_RESEND;
-
- BUG_ON(psr & (IA64_PSR_UP|IA64_PSR_PP));
- BUG_ON(psr & IA64_PSR_I);
-
- if (unlikely(ctx->ctx_state == PFM_CTX_ZOMBIE)) {
- struct pt_regs *regs = task_pt_regs(task);
-
- BUG_ON(ctx->ctx_smpl_hdr);
-
- pfm_force_cleanup(ctx, regs);
-
- pfm_unprotect_ctx_ctxsw(ctx, flags);
-
- /*
- * this one (kmalloc'ed) is fine with interrupts disabled
- */
- pfm_context_free(ctx);
-
- return;
- }
-
- /*
- * we restore ALL the debug registers to avoid picking up
- * stale state.
- */
- if (ctx->ctx_fl_using_dbreg) {
- pfm_restore_ibrs(ctx->ctx_ibrs, pmu_conf->num_ibrs);
- pfm_restore_dbrs(ctx->ctx_dbrs, pmu_conf->num_dbrs);
- }
- /*
- * retrieve saved psr.up
- */
- psr_up = ctx->ctx_saved_psr_up;
-
- /*
- * if we were the last user of the PMU on that CPU,
- * then nothing to do except restore psr
- */
- if (GET_LAST_CPU(ctx) == smp_processor_id() && ctx->ctx_last_activation == GET_ACTIVATION()) {
-
- /*
- * retrieve partial reload masks (due to user modifications)
- */
- pmc_mask = ctx->ctx_reload_pmcs[0];
- pmd_mask = ctx->ctx_reload_pmds[0];
-
- } else {
- /*
- * To avoid leaking information to the user level when psr.sp=0,
- * we must reload ALL implemented pmds (even the ones we don't use).
- * In the kernel we only allow PFM_READ_PMDS on registers which
- * we initialized or requested (sampling) so there is no risk there.
- */
- pmd_mask = pfm_sysctl.fastctxsw ? ctx->ctx_used_pmds[0] : ctx->ctx_all_pmds[0];
-
- /*
- * ALL accessible PMCs are systematically reloaded, unused registers
- * get their default (from pfm_reset_pmu_state()) values to avoid picking
- * up stale configuration.
- *
- * PMC0 is never in the mask. It is always restored separately.
- */
- pmc_mask = ctx->ctx_all_pmcs[0];
- }
- /*
- * when context is MASKED, we will restore PMC with plm=0
- * and PMD with stale information, but that's ok, nothing
- * will be captured.
- *
- * XXX: optimize here
- */
- if (pmd_mask) pfm_restore_pmds(ctx->th_pmds, pmd_mask);
- if (pmc_mask) pfm_restore_pmcs(ctx->th_pmcs, pmc_mask);
-
- /*
- * check for pending overflow at the time the state
- * was saved.
- */
- if (unlikely(PMC0_HAS_OVFL(ctx->th_pmcs[0]))) {
- /*
- * reload pmc0 with the overflow information
- * On McKinley PMU, this will trigger a PMU interrupt
- */
- ia64_set_pmc(0, ctx->th_pmcs[0]);
- ia64_srlz_d();
- ctx->th_pmcs[0] = 0UL;
-
- /*
- * will replay the PMU interrupt
- */
- if (need_irq_resend) ia64_resend_irq(IA64_PERFMON_VECTOR);
-
- pfm_stats[smp_processor_id()].pfm_replay_ovfl_intr_count++;
- }
-
- /*
- * we just did a reload, so we reset the partial reload fields
- */
- ctx->ctx_reload_pmcs[0] = 0UL;
- ctx->ctx_reload_pmds[0] = 0UL;
-
- SET_LAST_CPU(ctx, smp_processor_id());
-
- /*
- * dump activation value for this PMU
- */
- INC_ACTIVATION();
- /*
- * record current activation for this context
- */
- SET_ACTIVATION(ctx);
-
- /*
- * establish new ownership.
- */
- SET_PMU_OWNER(task, ctx);
-
- /*
- * restore the psr.up bit. measurement
- * is active again.
- * no PMU interrupt can happen at this point
- * because we still have interrupts disabled.
- */
- if (likely(psr_up)) pfm_set_psr_up();
-
- /*
- * allow concurrent access to context
- */
- pfm_unprotect_ctx_ctxsw(ctx, flags);
-}
-#else /* !CONFIG_SMP */
-/*
- * reload PMU state for UP kernels
- * in 2.5 we come here with interrupts disabled
- */
-void
-pfm_load_regs (struct task_struct *task)
-{
- pfm_context_t *ctx;
- struct task_struct *owner;
- unsigned long pmd_mask, pmc_mask;
- u64 psr, psr_up;
- int need_irq_resend;
-
- owner = GET_PMU_OWNER();
- ctx = PFM_GET_CTX(task);
- psr = pfm_get_psr();
-
- BUG_ON(psr & (IA64_PSR_UP|IA64_PSR_PP));
- BUG_ON(psr & IA64_PSR_I);
-
- /*
- * we restore ALL the debug registers to avoid picking up
- * stale state.
- *
- * This must be done even when the task is still the owner
- * as the registers may have been modified via ptrace()
- * (not perfmon) by the previous task.
- */
- if (ctx->ctx_fl_using_dbreg) {
- pfm_restore_ibrs(ctx->ctx_ibrs, pmu_conf->num_ibrs);
- pfm_restore_dbrs(ctx->ctx_dbrs, pmu_conf->num_dbrs);
- }
-
- /*
- * retrieved saved psr.up
- */
- psr_up = ctx->ctx_saved_psr_up;
- need_irq_resend = pmu_conf->flags & PFM_PMU_IRQ_RESEND;
-
- /*
- * short path, our state is still there, just
- * need to restore psr and we go
- *
- * we do not touch either PMC nor PMD. the psr is not touched
- * by the overflow_handler. So we are safe w.r.t. to interrupt
- * concurrency even without interrupt masking.
- */
- if (likely(owner == task)) {
- if (likely(psr_up)) pfm_set_psr_up();
- return;
- }
-
- /*
- * someone else is still using the PMU, first push it out and
- * then we'll be able to install our stuff !
- *
- * Upon return, there will be no owner for the current PMU
- */
- if (owner) pfm_lazy_save_regs(owner);
-
- /*
- * To avoid leaking information to the user level when psr.sp=0,
- * we must reload ALL implemented pmds (even the ones we don't use).
- * In the kernel we only allow PFM_READ_PMDS on registers which
- * we initialized or requested (sampling) so there is no risk there.
- */
- pmd_mask = pfm_sysctl.fastctxsw ? ctx->ctx_used_pmds[0] : ctx->ctx_all_pmds[0];
-
- /*
- * ALL accessible PMCs are systematically reloaded, unused registers
- * get their default (from pfm_reset_pmu_state()) values to avoid picking
- * up stale configuration.
- *
- * PMC0 is never in the mask. It is always restored separately
- */
- pmc_mask = ctx->ctx_all_pmcs[0];
-
- pfm_restore_pmds(ctx->th_pmds, pmd_mask);
- pfm_restore_pmcs(ctx->th_pmcs, pmc_mask);
-
- /*
- * check for pending overflow at the time the state
- * was saved.
- */
- if (unlikely(PMC0_HAS_OVFL(ctx->th_pmcs[0]))) {
- /*
- * reload pmc0 with the overflow information
- * On McKinley PMU, this will trigger a PMU interrupt
- */
- ia64_set_pmc(0, ctx->th_pmcs[0]);
- ia64_srlz_d();
-
- ctx->th_pmcs[0] = 0UL;
-
- /*
- * will replay the PMU interrupt
- */
- if (need_irq_resend) ia64_resend_irq(IA64_PERFMON_VECTOR);
-
- pfm_stats[smp_processor_id()].pfm_replay_ovfl_intr_count++;
- }
-
- /*
- * establish new ownership.
- */
- SET_PMU_OWNER(task, ctx);
-
- /*
- * restore the psr.up bit. measurement
- * is active again.
- * no PMU interrupt can happen at this point
- * because we still have interrupts disabled.
- */
- if (likely(psr_up)) pfm_set_psr_up();
-}
-#endif /* CONFIG_SMP */
-
-/*
- * this function assumes monitoring is stopped
- */
-static void
-pfm_flush_pmds(struct task_struct *task, pfm_context_t *ctx)
-{
- u64 pmc0;
- unsigned long mask2, val, pmd_val, ovfl_val;
- int i, can_access_pmu = 0;
- int is_self;
-
- /*
- * is the caller the task being monitored (or which initiated the
- * session for system wide measurements)
- */
- is_self = ctx->ctx_task == task ? 1 : 0;
-
- /*
- * can access PMU is task is the owner of the PMU state on the current CPU
- * or if we are running on the CPU bound to the context in system-wide mode
- * (that is not necessarily the task the context is attached to in this mode).
- * In system-wide we always have can_access_pmu true because a task running on an
- * invalid processor is flagged earlier in the call stack (see pfm_stop).
- */
- can_access_pmu = (GET_PMU_OWNER() == task) || (ctx->ctx_fl_system && ctx->ctx_cpu == smp_processor_id());
- if (can_access_pmu) {
- /*
- * Mark the PMU as not owned
- * This will cause the interrupt handler to do nothing in case an overflow
- * interrupt was in-flight
- * This also guarantees that pmc0 will contain the final state
- * It virtually gives us full control on overflow processing from that point
- * on.
- */
- SET_PMU_OWNER(NULL, NULL);
- DPRINT(("releasing ownership\n"));
-
- /*
- * read current overflow status:
- *
- * we are guaranteed to read the final stable state
- */
- ia64_srlz_d();
- pmc0 = ia64_get_pmc(0); /* slow */
-
- /*
- * reset freeze bit, overflow status information destroyed
- */
- pfm_unfreeze_pmu();
- } else {
- pmc0 = ctx->th_pmcs[0];
- /*
- * clear whatever overflow status bits there were
- */
- ctx->th_pmcs[0] = 0;
- }
- ovfl_val = pmu_conf->ovfl_val;
- /*
- * we save all the used pmds
- * we take care of overflows for counting PMDs
- *
- * XXX: sampling situation is not taken into account here
- */
- mask2 = ctx->ctx_used_pmds[0];
-
- DPRINT(("is_self=%d ovfl_val=0x%lx mask2=0x%lx\n", is_self, ovfl_val, mask2));
-
- for (i = 0; mask2; i++, mask2>>=1) {
-
- /* skip non used pmds */
- if ((mask2 & 0x1) == 0) continue;
-
- /*
- * can access PMU always true in system wide mode
- */
- val = pmd_val = can_access_pmu ? ia64_get_pmd(i) : ctx->th_pmds[i];
-
- if (PMD_IS_COUNTING(i)) {
- DPRINT(("[%d] pmd[%d] ctx_pmd=0x%lx hw_pmd=0x%lx\n",
- task_pid_nr(task),
- i,
- ctx->ctx_pmds[i].val,
- val & ovfl_val));
-
- /*
- * we rebuild the full 64 bit value of the counter
- */
- val = ctx->ctx_pmds[i].val + (val & ovfl_val);
-
- /*
- * now everything is in ctx_pmds[] and we need
- * to clear the saved context from save_regs() such that
- * pfm_read_pmds() gets the correct value
- */
- pmd_val = 0UL;
-
- /*
- * take care of overflow inline
- */
- if (pmc0 & (1UL << i)) {
- val += 1 + ovfl_val;
- DPRINT(("[%d] pmd[%d] overflowed\n", task_pid_nr(task), i));
- }
- }
-
- DPRINT(("[%d] ctx_pmd[%d]=0x%lx pmd_val=0x%lx\n", task_pid_nr(task), i, val, pmd_val));
-
- if (is_self) ctx->th_pmds[i] = pmd_val;
-
- ctx->ctx_pmds[i].val = val;
- }
-}
-
-static struct irqaction perfmon_irqaction = {
- .handler = pfm_interrupt_handler,
- .name = "perfmon"
-};
-
-static void
-pfm_alt_save_pmu_state(void *data)
-{
- struct pt_regs *regs;
-
- regs = task_pt_regs(current);
-
- DPRINT(("called\n"));
-
- /*
- * should not be necessary but
- * let's take not risk
- */
- pfm_clear_psr_up();
- pfm_clear_psr_pp();
- ia64_psr(regs)->pp = 0;
-
- /*
- * This call is required
- * May cause a spurious interrupt on some processors
- */
- pfm_freeze_pmu();
-
- ia64_srlz_d();
-}
-
-void
-pfm_alt_restore_pmu_state(void *data)
-{
- struct pt_regs *regs;
-
- regs = task_pt_regs(current);
-
- DPRINT(("called\n"));
-
- /*
- * put PMU back in state expected
- * by perfmon
- */
- pfm_clear_psr_up();
- pfm_clear_psr_pp();
- ia64_psr(regs)->pp = 0;
-
- /*
- * perfmon runs with PMU unfrozen at all times
- */
- pfm_unfreeze_pmu();
-
- ia64_srlz_d();
-}
-
-int
-pfm_install_alt_pmu_interrupt(pfm_intr_handler_desc_t *hdl)
-{
- int ret, i;
- int reserve_cpu;
-
- /* some sanity checks */
- if (hdl == NULL || hdl->handler == NULL) return -EINVAL;
-
- /* do the easy test first */
- if (pfm_alt_intr_handler) return -EBUSY;
-
- /* one at a time in the install or remove, just fail the others */
- if (!spin_trylock(&pfm_alt_install_check)) {
- return -EBUSY;
- }
-
- /* reserve our session */
- for_each_online_cpu(reserve_cpu) {
- ret = pfm_reserve_session(NULL, 1, reserve_cpu);
- if (ret) goto cleanup_reserve;
- }
-
- /* save the current system wide pmu states */
- on_each_cpu(pfm_alt_save_pmu_state, NULL, 1);
-
- /* officially change to the alternate interrupt handler */
- pfm_alt_intr_handler = hdl;
-
- spin_unlock(&pfm_alt_install_check);
-
- return 0;
-
-cleanup_reserve:
- for_each_online_cpu(i) {
- /* don't unreserve more than we reserved */
- if (i >= reserve_cpu) break;
-
- pfm_unreserve_session(NULL, 1, i);
- }
-
- spin_unlock(&pfm_alt_install_check);
-
- return ret;
-}
-EXPORT_SYMBOL_GPL(pfm_install_alt_pmu_interrupt);
-
-int
-pfm_remove_alt_pmu_interrupt(pfm_intr_handler_desc_t *hdl)
-{
- int i;
-
- if (hdl == NULL) return -EINVAL;
-
- /* cannot remove someone else's handler! */
- if (pfm_alt_intr_handler != hdl) return -EINVAL;
-
- /* one at a time in the install or remove, just fail the others */
- if (!spin_trylock(&pfm_alt_install_check)) {
- return -EBUSY;
- }
-
- pfm_alt_intr_handler = NULL;
-
- on_each_cpu(pfm_alt_restore_pmu_state, NULL, 1);
-
- for_each_online_cpu(i) {
- pfm_unreserve_session(NULL, 1, i);
- }
-
- spin_unlock(&pfm_alt_install_check);
-
- return 0;
-}
-EXPORT_SYMBOL_GPL(pfm_remove_alt_pmu_interrupt);
-
-/*
- * perfmon initialization routine, called from the initcall() table
- */
-static int init_pfm_fs(void);
-
-static int __init
-pfm_probe_pmu(void)
-{
- pmu_config_t **p;
- int family;
-
- family = local_cpu_data->family;
- p = pmu_confs;
-
- while(*p) {
- if ((*p)->probe) {
- if ((*p)->probe() == 0) goto found;
- } else if ((*p)->pmu_family == family || (*p)->pmu_family == 0xff) {
- goto found;
- }
- p++;
- }
- return -1;
-found:
- pmu_conf = *p;
- return 0;
-}
-
-int __init
-pfm_init(void)
-{
- unsigned int n, n_counters, i;
-
- printk("perfmon: version %u.%u IRQ %u\n",
- PFM_VERSION_MAJ,
- PFM_VERSION_MIN,
- IA64_PERFMON_VECTOR);
-
- if (pfm_probe_pmu()) {
- printk(KERN_INFO "perfmon: disabled, there is no support for processor family %d\n",
- local_cpu_data->family);
- return -ENODEV;
- }
-
- /*
- * compute the number of implemented PMD/PMC from the
- * description tables
- */
- n = 0;
- for (i=0; PMC_IS_LAST(i) == 0; i++) {
- if (PMC_IS_IMPL(i) == 0) continue;
- pmu_conf->impl_pmcs[i>>6] |= 1UL << (i&63);
- n++;
- }
- pmu_conf->num_pmcs = n;
-
- n = 0; n_counters = 0;
- for (i=0; PMD_IS_LAST(i) == 0; i++) {
- if (PMD_IS_IMPL(i) == 0) continue;
- pmu_conf->impl_pmds[i>>6] |= 1UL << (i&63);
- n++;
- if (PMD_IS_COUNTING(i)) n_counters++;
- }
- pmu_conf->num_pmds = n;
- pmu_conf->num_counters = n_counters;
-
- /*
- * sanity checks on the number of debug registers
- */
- if (pmu_conf->use_rr_dbregs) {
- if (pmu_conf->num_ibrs > IA64_NUM_DBG_REGS) {
- printk(KERN_INFO "perfmon: unsupported number of code debug registers (%u)\n", pmu_conf->num_ibrs);
- pmu_conf = NULL;
- return -1;
- }
- if (pmu_conf->num_dbrs > IA64_NUM_DBG_REGS) {
- printk(KERN_INFO "perfmon: unsupported number of data debug registers (%u)\n", pmu_conf->num_ibrs);
- pmu_conf = NULL;
- return -1;
- }
- }
-
- printk("perfmon: %s PMU detected, %u PMCs, %u PMDs, %u counters (%lu bits)\n",
- pmu_conf->pmu_name,
- pmu_conf->num_pmcs,
- pmu_conf->num_pmds,
- pmu_conf->num_counters,
- ffz(pmu_conf->ovfl_val));
-
- /* sanity check */
- if (pmu_conf->num_pmds >= PFM_NUM_PMD_REGS || pmu_conf->num_pmcs >= PFM_NUM_PMC_REGS) {
- printk(KERN_ERR "perfmon: not enough pmc/pmd, perfmon disabled\n");
- pmu_conf = NULL;
- return -1;
- }
-
- /*
- * create /proc/perfmon (mostly for debugging purposes)
- */
- perfmon_dir = proc_create_seq("perfmon", S_IRUGO, NULL, &pfm_seq_ops);
- if (perfmon_dir == NULL) {
- printk(KERN_ERR "perfmon: cannot create /proc entry, perfmon disabled\n");
- pmu_conf = NULL;
- return -1;
- }
-
- /*
- * create /proc/sys/kernel/perfmon (for debugging purposes)
- */
- pfm_sysctl_header = register_sysctl_table(pfm_sysctl_root);
-
- /*
- * initialize all our spinlocks
- */
- spin_lock_init(&pfm_sessions.pfs_lock);
- spin_lock_init(&pfm_buffer_fmt_lock);
-
- init_pfm_fs();
-
- for(i=0; i < NR_CPUS; i++) pfm_stats[i].pfm_ovfl_intr_cycles_min = ~0UL;
-
- return 0;
-}
-
-__initcall(pfm_init);
-
-/*
- * this function is called before pfm_init()
- */
-void
-pfm_init_percpu (void)
-{
- static int first_time=1;
- /*
- * make sure no measurement is active
- * (may inherit programmed PMCs from EFI).
- */
- pfm_clear_psr_pp();
- pfm_clear_psr_up();
-
- /*
- * we run with the PMU not frozen at all times
- */
- pfm_unfreeze_pmu();
-
- if (first_time) {
- register_percpu_irq(IA64_PERFMON_VECTOR, &perfmon_irqaction);
- first_time=0;
- }
-
- ia64_setreg(_IA64_REG_CR_PMV, IA64_PERFMON_VECTOR);
- ia64_srlz_d();
-}
-
-/*
- * used for debug purposes only
- */
-void
-dump_pmu_state(const char *from)
-{
- struct task_struct *task;
- struct pt_regs *regs;
- pfm_context_t *ctx;
- unsigned long psr, dcr, info, flags;
- int i, this_cpu;
-
- local_irq_save(flags);
-
- this_cpu = smp_processor_id();
- regs = task_pt_regs(current);
- info = PFM_CPUINFO_GET();
- dcr = ia64_getreg(_IA64_REG_CR_DCR);
-
- if (info == 0 && ia64_psr(regs)->pp == 0 && (dcr & IA64_DCR_PP) == 0) {
- local_irq_restore(flags);
- return;
- }
-
- printk("CPU%d from %s() current [%d] iip=0x%lx %s\n",
- this_cpu,
- from,
- task_pid_nr(current),
- regs->cr_iip,
- current->comm);
-
- task = GET_PMU_OWNER();
- ctx = GET_PMU_CTX();
-
- printk("->CPU%d owner [%d] ctx=%p\n", this_cpu, task ? task_pid_nr(task) : -1, ctx);
-
- psr = pfm_get_psr();
-
- printk("->CPU%d pmc0=0x%lx psr.pp=%d psr.up=%d dcr.pp=%d syst_info=0x%lx user_psr.up=%d user_psr.pp=%d\n",
- this_cpu,
- ia64_get_pmc(0),
- psr & IA64_PSR_PP ? 1 : 0,
- psr & IA64_PSR_UP ? 1 : 0,
- dcr & IA64_DCR_PP ? 1 : 0,
- info,
- ia64_psr(regs)->up,
- ia64_psr(regs)->pp);
-
- ia64_psr(regs)->up = 0;
- ia64_psr(regs)->pp = 0;
-
- for (i=1; PMC_IS_LAST(i) == 0; i++) {
- if (PMC_IS_IMPL(i) == 0) continue;
- printk("->CPU%d pmc[%d]=0x%lx thread_pmc[%d]=0x%lx\n", this_cpu, i, ia64_get_pmc(i), i, ctx->th_pmcs[i]);
- }
-
- for (i=1; PMD_IS_LAST(i) == 0; i++) {
- if (PMD_IS_IMPL(i) == 0) continue;
- printk("->CPU%d pmd[%d]=0x%lx thread_pmd[%d]=0x%lx\n", this_cpu, i, ia64_get_pmd(i), i, ctx->th_pmds[i]);
- }
-
- if (ctx) {
- printk("->CPU%d ctx_state=%d vaddr=%p addr=%p fd=%d ctx_task=[%d] saved_psr_up=0x%lx\n",
- this_cpu,
- ctx->ctx_state,
- ctx->ctx_smpl_vaddr,
- ctx->ctx_smpl_hdr,
- ctx->ctx_msgq_head,
- ctx->ctx_msgq_tail,
- ctx->ctx_saved_psr_up);
- }
- local_irq_restore(flags);
-}
-
-/*
- * called from process.c:copy_thread(). task is new child.
- */
-void
-pfm_inherit(struct task_struct *task, struct pt_regs *regs)
-{
- struct thread_struct *thread;
-
- DPRINT(("perfmon: pfm_inherit clearing state for [%d]\n", task_pid_nr(task)));
-
- thread = &task->thread;
-
- /*
- * cut links inherited from parent (current)
- */
- thread->pfm_context = NULL;
-
- PFM_SET_WORK_PENDING(task, 0);
-
- /*
- * the psr bits are already set properly in copy_threads()
- */
-}
-#else /* !CONFIG_PERFMON */
-asmlinkage long
-sys_perfmonctl (int fd, int cmd, void *arg, int count)
-{
- return -ENOSYS;
-}
-#endif /* CONFIG_PERFMON */
diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
index 968b5f3..c9ff879 100644
--- a/arch/ia64/kernel/process.c
+++ b/arch/ia64/kernel/process.c
@@ -40,7 +40,6 @@
#include <asm/elf.h>
#include <asm/irq.h>
#include <asm/kexec.h>
-#include <asm/pgalloc.h>
#include <asm/processor.h>
#include <asm/sal.h>
#include <asm/switch_to.h>
@@ -48,13 +47,10 @@
#include <linux/uaccess.h>
#include <asm/unwind.h>
#include <asm/user.h>
+#include <asm/xtp.h>
#include "entry.h"
-#ifdef CONFIG_PERFMON
-# include <asm/perfmon.h>
-#endif
-
#include "sigframe.h"
void (*ia64_mark_idle)(int);
@@ -64,12 +60,13 @@
void (*pm_power_off) (void);
EXPORT_SYMBOL(pm_power_off);
-void
+static void
ia64_do_show_stack (struct unw_frame_info *info, void *arg)
{
unsigned long ip, sp, bsp;
+ const char *loglvl = arg;
- printk("\nCall Trace:\n");
+ printk("%s\nCall Trace:\n", loglvl);
do {
unw_get_ip(info, &ip);
if (ip == 0)
@@ -77,22 +74,22 @@
unw_get_sp(info, &sp);
unw_get_bsp(info, &bsp);
- printk(" [<%016lx>] %pS\n"
+ printk("%s [<%016lx>] %pS\n"
" sp=%016lx bsp=%016lx\n",
- ip, (void *)ip, sp, bsp);
+ loglvl, ip, (void *)ip, sp, bsp);
} while (unw_unwind(info) >= 0);
}
void
-show_stack (struct task_struct *task, unsigned long *sp)
+show_stack (struct task_struct *task, unsigned long *sp, const char *loglvl)
{
if (!task)
- unw_init_running(ia64_do_show_stack, NULL);
+ unw_init_running(ia64_do_show_stack, (void *)loglvl);
else {
struct unw_frame_info info;
unw_init_from_blocked_task(&info, task);
- ia64_do_show_stack(&info, NULL);
+ ia64_do_show_stack(&info, (void *)loglvl);
}
}
@@ -150,7 +147,7 @@
((i == sof - 1) || (i % 3) == 2) ? "\n" : " ");
}
} else
- show_stack(NULL, NULL);
+ show_stack(NULL, NULL, KERN_DEFAULT);
}
/* local support for deprecated console_print */
@@ -173,22 +170,13 @@
return;
}
-#ifdef CONFIG_PERFMON
- if (current->thread.pfm_needs_checking)
- /*
- * Note: pfm_handle_work() allow us to call it with interrupts
- * disabled, and may enable interrupts within the function.
- */
- pfm_handle_work();
-#endif
-
/* deal with pending signal delivery */
if (test_thread_flag(TIF_SIGPENDING)) {
local_irq_enable(); /* force interrupt enable */
ia64_do_signal(scr, in_syscall);
}
- if (test_and_clear_thread_flag(TIF_NOTIFY_RESUME)) {
+ if (test_thread_flag(TIF_NOTIFY_RESUME)) {
local_irq_enable(); /* force interrupt enable */
tracehook_notify_resume(&scr->pt);
}
@@ -251,7 +239,7 @@
if (mark_idle)
(*mark_idle)(1);
- safe_halt();
+ raw_safe_halt();
if (mark_idle)
(*mark_idle)(0);
@@ -263,41 +251,15 @@
void
ia64_save_extra (struct task_struct *task)
{
-#ifdef CONFIG_PERFMON
- unsigned long info;
-#endif
-
if ((task->thread.flags & IA64_THREAD_DBG_VALID) != 0)
ia64_save_debug_regs(&task->thread.dbr[0]);
-
-#ifdef CONFIG_PERFMON
- if ((task->thread.flags & IA64_THREAD_PM_VALID) != 0)
- pfm_save_regs(task);
-
- info = __this_cpu_read(pfm_syst_info);
- if (info & PFM_CPUINFO_SYST_WIDE)
- pfm_syst_wide_update_task(task, info, 0);
-#endif
}
void
ia64_load_extra (struct task_struct *task)
{
-#ifdef CONFIG_PERFMON
- unsigned long info;
-#endif
-
if ((task->thread.flags & IA64_THREAD_DBG_VALID) != 0)
ia64_load_debug_regs(&task->thread.dbr[0]);
-
-#ifdef CONFIG_PERFMON
- if ((task->thread.flags & IA64_THREAD_PM_VALID) != 0)
- pfm_load_regs(task);
-
- info = __this_cpu_read(pfm_syst_info);
- if (info & PFM_CPUINFO_SYST_WIDE)
- pfm_syst_wide_update_task(task, info, 1);
-#endif
}
/*
@@ -309,7 +271,7 @@
*
* <clone syscall> <some kernel call frames>
* sys_clone :
- * do_fork do_fork
+ * kernel_clone kernel_clone
* copy_thread copy_thread
*
* This means that the stack layout is as follows:
@@ -332,9 +294,8 @@
* so there is nothing to worry about.
*/
int
-copy_thread(unsigned long clone_flags,
- unsigned long user_stack_base, unsigned long user_stack_size,
- struct task_struct *p)
+copy_thread(unsigned long clone_flags, unsigned long user_stack_base,
+ unsigned long user_stack_size, struct task_struct *p, unsigned long tls)
{
extern char ia64_ret_from_clone;
struct switch_stack *child_stack, *stack;
@@ -415,7 +376,7 @@
rbs_size = stack->ar_bspstore - rbs;
memcpy((void *) child_rbs, (void *) rbs, rbs_size);
if (clone_flags & CLONE_SETTLS)
- child_ptregs->r13 = regs->r16; /* see sys_clone2() in entry.S */
+ child_ptregs->r13 = tls;
if (user_stack_base) {
child_ptregs->r12 = user_stack_base + user_stack_size - 16;
child_ptregs->ar_bspstore = user_stack_base;
@@ -432,19 +393,32 @@
*/
child_ptregs->cr_ipsr = ((child_ptregs->cr_ipsr | IA64_PSR_BITS_TO_SET)
& ~(IA64_PSR_BITS_TO_CLEAR | IA64_PSR_PP | IA64_PSR_UP));
-
-#ifdef CONFIG_PERFMON
- if (current->thread.pfm_context)
- pfm_inherit(p, child_ptregs);
-#endif
return retval;
}
+asmlinkage long ia64_clone(unsigned long clone_flags, unsigned long stack_start,
+ unsigned long stack_size, unsigned long parent_tidptr,
+ unsigned long child_tidptr, unsigned long tls)
+{
+ struct kernel_clone_args args = {
+ .flags = (lower_32_bits(clone_flags) & ~CSIGNAL),
+ .pidfd = (int __user *)parent_tidptr,
+ .child_tid = (int __user *)child_tidptr,
+ .parent_tid = (int __user *)parent_tidptr,
+ .exit_signal = (lower_32_bits(clone_flags) & CSIGNAL),
+ .stack = stack_start,
+ .stack_size = stack_size,
+ .tls = tls,
+ };
+
+ return kernel_clone(&args);
+}
+
static void
do_copy_task_regs (struct task_struct *task, struct unw_frame_info *info, void *arg)
{
unsigned long mask, sp, nat_bits = 0, ar_rnat, urbs_end, cfm;
- unsigned long uninitialized_var(ip); /* GCC be quiet */
+ unsigned long ip;
elf_greg_t *dst = arg;
struct pt_regs *pt;
char nat;
@@ -514,51 +488,17 @@
}
void
-do_dump_task_fpu (struct task_struct *task, struct unw_frame_info *info, void *arg)
-{
- elf_fpreg_t *dst = arg;
- int i;
-
- memset(dst, 0, sizeof(elf_fpregset_t)); /* don't leak any "random" bits */
-
- if (unw_unwind_to_user(info) < 0)
- return;
-
- /* f0 is 0.0, f1 is 1.0 */
-
- for (i = 2; i < 32; ++i)
- unw_get_fr(info, i, dst + i);
-
- ia64_flush_fph(task);
- if ((task->thread.flags & IA64_THREAD_FPH_VALID) != 0)
- memcpy(dst + 32, task->thread.fph, 96*16);
-}
-
-void
do_copy_regs (struct unw_frame_info *info, void *arg)
{
do_copy_task_regs(current, info, arg);
}
void
-do_dump_fpu (struct unw_frame_info *info, void *arg)
-{
- do_dump_task_fpu(current, info, arg);
-}
-
-void
ia64_elf_core_copy_regs (struct pt_regs *pt, elf_gregset_t dst)
{
unw_init_running(do_copy_regs, dst);
}
-int
-dump_fpu (struct pt_regs *pt, elf_fpregset_t dst)
-{
- unw_init_running(do_dump_fpu, dst);
- return 1; /* f0-f31 are always valid so we always return 1 */
-}
-
/*
* Flush thread state. This is called when a thread does an execve().
*/
@@ -579,15 +519,6 @@
{
ia64_drop_fpu(tsk);
-#ifdef CONFIG_PERFMON
- /* if needed, stop monitoring and flush state to perfmon context */
- if (tsk->thread.pfm_context)
- pfm_exit_thread(tsk);
-
- /* free debug register resources */
- if (tsk->thread.flags & IA64_THREAD_DBG_VALID)
- pfm_release_debug_registers(tsk);
-#endif
}
unsigned long
@@ -646,14 +577,8 @@
void machine_shutdown(void)
{
-#ifdef CONFIG_HOTPLUG_CPU
- int cpu;
+ smp_shutdown_nonboot_cpus(reboot_cpu);
- for_each_online_cpu(cpu) {
- if (cpu != smp_processor_id())
- cpu_down(cpu);
- }
-#endif
#ifdef CONFIG_KEXEC
kexec_disable_iosapic();
#endif
@@ -681,3 +606,4 @@
machine_halt();
}
+EXPORT_SYMBOL(ia64_delay_loop);
diff --git a/arch/ia64/kernel/ptrace.c b/arch/ia64/kernel/ptrace.c
index 54e12b0..dad3a60 100644
--- a/arch/ia64/kernel/ptrace.c
+++ b/arch/ia64/kernel/ptrace.c
@@ -25,15 +25,11 @@
#include <linux/elf.h>
#include <linux/tracehook.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/ptrace_offsets.h>
#include <asm/rse.h>
#include <linux/uaccess.h>
#include <asm/unwind.h>
-#ifdef CONFIG_PERFMON
-#include <asm/perfmon.h>
-#endif
#include "entry.h"
@@ -1274,52 +1270,43 @@
int ret;
};
+static const ptrdiff_t pt_offsets[32] =
+{
+#define R(n) offsetof(struct pt_regs, r##n)
+ [0] = -1, R(1), R(2), R(3),
+ [4] = -1, [5] = -1, [6] = -1, [7] = -1,
+ R(8), R(9), R(10), R(11), R(12), R(13), R(14), R(15),
+ R(16), R(17), R(18), R(19), R(20), R(21), R(22), R(23),
+ R(24), R(25), R(26), R(27), R(28), R(29), R(30), R(31),
+#undef R
+};
+
static int
access_elf_gpreg(struct task_struct *target, struct unw_frame_info *info,
unsigned long addr, unsigned long *data, int write_access)
{
- struct pt_regs *pt;
- unsigned long *ptr = NULL;
- int ret;
- char nat = 0;
+ struct pt_regs *pt = task_pt_regs(target);
+ unsigned reg = addr / sizeof(unsigned long);
+ ptrdiff_t d = pt_offsets[reg];
- pt = task_pt_regs(target);
- switch (addr) {
- case ELF_GR_OFFSET(1):
- ptr = &pt->r1;
- break;
- case ELF_GR_OFFSET(2):
- case ELF_GR_OFFSET(3):
- ptr = (void *)&pt->r2 + (addr - ELF_GR_OFFSET(2));
- break;
- case ELF_GR_OFFSET(4) ... ELF_GR_OFFSET(7):
+ if (d >= 0) {
+ unsigned long *ptr = (void *)pt + d;
+ if (write_access)
+ *ptr = *data;
+ else
+ *data = *ptr;
+ return 0;
+ } else {
+ char nat = 0;
if (write_access) {
/* read NaT bit first: */
unsigned long dummy;
-
- ret = unw_get_gr(info, addr/8, &dummy, &nat);
+ int ret = unw_get_gr(info, reg, &dummy, &nat);
if (ret < 0)
return ret;
}
- return unw_access_gr(info, addr/8, data, &nat, write_access);
- case ELF_GR_OFFSET(8) ... ELF_GR_OFFSET(11):
- ptr = (void *)&pt->r8 + addr - ELF_GR_OFFSET(8);
- break;
- case ELF_GR_OFFSET(12):
- case ELF_GR_OFFSET(13):
- ptr = (void *)&pt->r12 + addr - ELF_GR_OFFSET(12);
- break;
- case ELF_GR_OFFSET(14):
- ptr = &pt->r14;
- break;
- case ELF_GR_OFFSET(15):
- ptr = &pt->r15;
+ return unw_access_gr(info, reg, data, &nat, write_access);
}
- if (write_access)
- *ptr = *data;
- else
- *data = *ptr;
- return 0;
}
static int
@@ -1491,7 +1478,7 @@
access_elf_reg(struct task_struct *target, struct unw_frame_info *info,
unsigned long addr, unsigned long *data, int write_access)
{
- if (addr >= ELF_GR_OFFSET(1) && addr <= ELF_GR_OFFSET(15))
+ if (addr >= ELF_GR_OFFSET(1) && addr <= ELF_GR_OFFSET(31))
return access_elf_gpreg(target, info, addr, data, write_access);
else if (addr >= ELF_BR_OFFSET(0) && addr <= ELF_BR_OFFSET(7))
return access_elf_breg(target, info, addr, data, write_access);
@@ -1499,12 +1486,17 @@
return access_elf_areg(target, info, addr, data, write_access);
}
+struct regset_membuf {
+ struct membuf to;
+ int ret;
+};
+
void do_gpregs_get(struct unw_frame_info *info, void *arg)
{
- struct pt_regs *pt;
- struct regset_getset *dst = arg;
- elf_greg_t tmp[16];
- unsigned int i, index, min_copy;
+ struct regset_membuf *dst = arg;
+ struct membuf to = dst->to;
+ unsigned int n;
+ elf_greg_t reg;
if (unw_unwind_to_user(info) < 0)
return;
@@ -1522,165 +1514,53 @@
/* Skip r0 */
- if (dst->count > 0 && dst->pos < ELF_GR_OFFSET(1)) {
- dst->ret = user_regset_copyout_zero(&dst->pos, &dst->count,
- &dst->u.get.kbuf,
- &dst->u.get.ubuf,
- 0, ELF_GR_OFFSET(1));
- if (dst->ret || dst->count == 0)
+ membuf_zero(&to, 8);
+ for (n = 8; to.left && n < ELF_AR_END_OFFSET; n += 8) {
+ if (access_elf_reg(info->task, info, n, ®, 0) < 0) {
+ dst->ret = -EIO;
return;
- }
-
- /* gr1 - gr15 */
- if (dst->count > 0 && dst->pos < ELF_GR_OFFSET(16)) {
- index = (dst->pos - ELF_GR_OFFSET(1)) / sizeof(elf_greg_t);
- min_copy = ELF_GR_OFFSET(16) > (dst->pos + dst->count) ?
- (dst->pos + dst->count) : ELF_GR_OFFSET(16);
- for (i = dst->pos; i < min_copy; i += sizeof(elf_greg_t),
- index++)
- if (access_elf_reg(dst->target, info, i,
- &tmp[index], 0) < 0) {
- dst->ret = -EIO;
- return;
- }
- dst->ret = user_regset_copyout(&dst->pos, &dst->count,
- &dst->u.get.kbuf, &dst->u.get.ubuf, tmp,
- ELF_GR_OFFSET(1), ELF_GR_OFFSET(16));
- if (dst->ret || dst->count == 0)
- return;
- }
-
- /* r16-r31 */
- if (dst->count > 0 && dst->pos < ELF_NAT_OFFSET) {
- pt = task_pt_regs(dst->target);
- dst->ret = user_regset_copyout(&dst->pos, &dst->count,
- &dst->u.get.kbuf, &dst->u.get.ubuf, &pt->r16,
- ELF_GR_OFFSET(16), ELF_NAT_OFFSET);
- if (dst->ret || dst->count == 0)
- return;
- }
-
- /* nat, pr, b0 - b7 */
- if (dst->count > 0 && dst->pos < ELF_CR_IIP_OFFSET) {
- index = (dst->pos - ELF_NAT_OFFSET) / sizeof(elf_greg_t);
- min_copy = ELF_CR_IIP_OFFSET > (dst->pos + dst->count) ?
- (dst->pos + dst->count) : ELF_CR_IIP_OFFSET;
- for (i = dst->pos; i < min_copy; i += sizeof(elf_greg_t),
- index++)
- if (access_elf_reg(dst->target, info, i,
- &tmp[index], 0) < 0) {
- dst->ret = -EIO;
- return;
- }
- dst->ret = user_regset_copyout(&dst->pos, &dst->count,
- &dst->u.get.kbuf, &dst->u.get.ubuf, tmp,
- ELF_NAT_OFFSET, ELF_CR_IIP_OFFSET);
- if (dst->ret || dst->count == 0)
- return;
- }
-
- /* ip cfm psr ar.rsc ar.bsp ar.bspstore ar.rnat
- * ar.ccv ar.unat ar.fpsr ar.pfs ar.lc ar.ec ar.csd ar.ssd
- */
- if (dst->count > 0 && dst->pos < (ELF_AR_END_OFFSET)) {
- index = (dst->pos - ELF_CR_IIP_OFFSET) / sizeof(elf_greg_t);
- min_copy = ELF_AR_END_OFFSET > (dst->pos + dst->count) ?
- (dst->pos + dst->count) : ELF_AR_END_OFFSET;
- for (i = dst->pos; i < min_copy; i += sizeof(elf_greg_t),
- index++)
- if (access_elf_reg(dst->target, info, i,
- &tmp[index], 0) < 0) {
- dst->ret = -EIO;
- return;
- }
- dst->ret = user_regset_copyout(&dst->pos, &dst->count,
- &dst->u.get.kbuf, &dst->u.get.ubuf, tmp,
- ELF_CR_IIP_OFFSET, ELF_AR_END_OFFSET);
+ }
+ membuf_store(&to, reg);
}
}
void do_gpregs_set(struct unw_frame_info *info, void *arg)
{
- struct pt_regs *pt;
struct regset_getset *dst = arg;
- elf_greg_t tmp[16];
- unsigned int i, index;
if (unw_unwind_to_user(info) < 0)
return;
+ if (!dst->count)
+ return;
/* Skip r0 */
- if (dst->count > 0 && dst->pos < ELF_GR_OFFSET(1)) {
+ if (dst->pos < ELF_GR_OFFSET(1)) {
dst->ret = user_regset_copyin_ignore(&dst->pos, &dst->count,
&dst->u.set.kbuf,
&dst->u.set.ubuf,
0, ELF_GR_OFFSET(1));
- if (dst->ret || dst->count == 0)
- return;
- }
-
- /* gr1-gr15 */
- if (dst->count > 0 && dst->pos < ELF_GR_OFFSET(16)) {
- i = dst->pos;
- index = (dst->pos - ELF_GR_OFFSET(1)) / sizeof(elf_greg_t);
- dst->ret = user_regset_copyin(&dst->pos, &dst->count,
- &dst->u.set.kbuf, &dst->u.set.ubuf, tmp,
- ELF_GR_OFFSET(1), ELF_GR_OFFSET(16));
if (dst->ret)
return;
- for ( ; i < dst->pos; i += sizeof(elf_greg_t), index++)
- if (access_elf_reg(dst->target, info, i,
- &tmp[index], 1) < 0) {
- dst->ret = -EIO;
- return;
- }
- if (dst->count == 0)
- return;
}
- /* gr16-gr31 */
- if (dst->count > 0 && dst->pos < ELF_NAT_OFFSET) {
- pt = task_pt_regs(dst->target);
- dst->ret = user_regset_copyin(&dst->pos, &dst->count,
- &dst->u.set.kbuf, &dst->u.set.ubuf, &pt->r16,
- ELF_GR_OFFSET(16), ELF_NAT_OFFSET);
- if (dst->ret || dst->count == 0)
- return;
- }
+ while (dst->count && dst->pos < ELF_AR_END_OFFSET) {
+ unsigned int n, from, to;
+ elf_greg_t tmp[16];
- /* nat, pr, b0 - b7 */
- if (dst->count > 0 && dst->pos < ELF_CR_IIP_OFFSET) {
- i = dst->pos;
- index = (dst->pos - ELF_NAT_OFFSET) / sizeof(elf_greg_t);
+ from = dst->pos;
+ to = from + sizeof(tmp);
+ if (to > ELF_AR_END_OFFSET)
+ to = ELF_AR_END_OFFSET;
+ /* get up to 16 values */
dst->ret = user_regset_copyin(&dst->pos, &dst->count,
&dst->u.set.kbuf, &dst->u.set.ubuf, tmp,
- ELF_NAT_OFFSET, ELF_CR_IIP_OFFSET);
+ from, to);
if (dst->ret)
return;
- for (; i < dst->pos; i += sizeof(elf_greg_t), index++)
- if (access_elf_reg(dst->target, info, i,
- &tmp[index], 1) < 0) {
- dst->ret = -EIO;
- return;
- }
- if (dst->count == 0)
- return;
- }
-
- /* ip cfm psr ar.rsc ar.bsp ar.bspstore ar.rnat
- * ar.ccv ar.unat ar.fpsr ar.pfs ar.lc ar.ec ar.csd ar.ssd
- */
- if (dst->count > 0 && dst->pos < (ELF_AR_END_OFFSET)) {
- i = dst->pos;
- index = (dst->pos - ELF_CR_IIP_OFFSET) / sizeof(elf_greg_t);
- dst->ret = user_regset_copyin(&dst->pos, &dst->count,
- &dst->u.set.kbuf, &dst->u.set.ubuf, tmp,
- ELF_CR_IIP_OFFSET, ELF_AR_END_OFFSET);
- if (dst->ret)
- return;
- for ( ; i < dst->pos; i += sizeof(elf_greg_t), index++)
- if (access_elf_reg(dst->target, info, i,
- &tmp[index], 1) < 0) {
+ /* now copy them into registers */
+ for (n = 0; from < dst->pos; from += sizeof(elf_greg_t), n++)
+ if (access_elf_reg(dst->target, info, from,
+ &tmp[n], 1) < 0) {
dst->ret = -EIO;
return;
}
@@ -1691,60 +1571,36 @@
void do_fpregs_get(struct unw_frame_info *info, void *arg)
{
- struct regset_getset *dst = arg;
- struct task_struct *task = dst->target;
- elf_fpreg_t tmp[30];
- int index, min_copy, i;
+ struct task_struct *task = info->task;
+ struct regset_membuf *dst = arg;
+ struct membuf to = dst->to;
+ elf_fpreg_t reg;
+ unsigned int n;
if (unw_unwind_to_user(info) < 0)
return;
/* Skip pos 0 and 1 */
- if (dst->count > 0 && dst->pos < ELF_FP_OFFSET(2)) {
- dst->ret = user_regset_copyout_zero(&dst->pos, &dst->count,
- &dst->u.get.kbuf,
- &dst->u.get.ubuf,
- 0, ELF_FP_OFFSET(2));
- if (dst->count == 0 || dst->ret)
- return;
- }
+ membuf_zero(&to, 2 * sizeof(elf_fpreg_t));
/* fr2-fr31 */
- if (dst->count > 0 && dst->pos < ELF_FP_OFFSET(32)) {
- index = (dst->pos - ELF_FP_OFFSET(2)) / sizeof(elf_fpreg_t);
-
- min_copy = min(((unsigned int)ELF_FP_OFFSET(32)),
- dst->pos + dst->count);
- for (i = dst->pos; i < min_copy; i += sizeof(elf_fpreg_t),
- index++)
- if (unw_get_fr(info, i / sizeof(elf_fpreg_t),
- &tmp[index])) {
- dst->ret = -EIO;
- return;
- }
- dst->ret = user_regset_copyout(&dst->pos, &dst->count,
- &dst->u.get.kbuf, &dst->u.get.ubuf, tmp,
- ELF_FP_OFFSET(2), ELF_FP_OFFSET(32));
- if (dst->count == 0 || dst->ret)
+ for (n = 2; to.left && n < 32; n++) {
+ if (unw_get_fr(info, n, ®)) {
+ dst->ret = -EIO;
return;
+ }
+ membuf_write(&to, ®, sizeof(reg));
}
/* fph */
- if (dst->count > 0) {
- ia64_flush_fph(dst->target);
- if (task->thread.flags & IA64_THREAD_FPH_VALID)
- dst->ret = user_regset_copyout(
- &dst->pos, &dst->count,
- &dst->u.get.kbuf, &dst->u.get.ubuf,
- &dst->target->thread.fph,
- ELF_FP_OFFSET(32), -1);
- else
- /* Zero fill instead. */
- dst->ret = user_regset_copyout_zero(
- &dst->pos, &dst->count,
- &dst->u.get.kbuf, &dst->u.get.ubuf,
- ELF_FP_OFFSET(32), -1);
- }
+ if (!to.left)
+ return;
+
+ ia64_flush_fph(task);
+ if (task->thread.flags & IA64_THREAD_FPH_VALID)
+ membuf_write(&to, &task->thread.fph, 96 * sizeof(reg));
+ else
+ membuf_zero(&to, 96 * sizeof(reg));
}
void do_fpregs_set(struct unw_frame_info *info, void *arg)
@@ -1820,6 +1676,20 @@
}
}
+static void
+unwind_and_call(void (*call)(struct unw_frame_info *, void *),
+ struct task_struct *target, void *data)
+{
+ if (target == current)
+ unw_init_running(call, data);
+ else {
+ struct unw_frame_info info;
+ memset(&info, 0, sizeof(info));
+ unw_init_from_blocked_task(&info, target);
+ (*call)(&info, data);
+ }
+}
+
static int
do_regset_call(void (*call)(struct unw_frame_info *, void *),
struct task_struct *target,
@@ -1831,27 +1701,18 @@
.pos = pos, .count = count,
.u.set = { .kbuf = kbuf, .ubuf = ubuf },
.ret = 0 };
-
- if (target == current)
- unw_init_running(call, &info);
- else {
- struct unw_frame_info ufi;
- memset(&ufi, 0, sizeof(ufi));
- unw_init_from_blocked_task(&ufi, target);
- (*call)(&ufi, &info);
- }
-
+ unwind_and_call(call, target, &info);
return info.ret;
}
static int
gpregs_get(struct task_struct *target,
const struct user_regset *regset,
- unsigned int pos, unsigned int count,
- void *kbuf, void __user *ubuf)
+ struct membuf to)
{
- return do_regset_call(do_gpregs_get, target, regset, pos, count,
- kbuf, ubuf);
+ struct regset_membuf info = {.to = to};
+ unwind_and_call(do_gpregs_get, target, &info);
+ return info.ret;
}
static int gpregs_set(struct task_struct *target,
@@ -1893,11 +1754,11 @@
static int fpregs_get(struct task_struct *target,
const struct user_regset *regset,
- unsigned int pos, unsigned int count,
- void *kbuf, void __user *ubuf)
+ struct membuf to)
{
- return do_regset_call(do_fpregs_get, target, regset, pos, count,
- kbuf, ubuf);
+ struct regset_membuf info = {.to = to};
+ unwind_and_call(do_fpregs_get, target, &info);
+ return info.ret;
}
static int fpregs_set(struct task_struct *target,
@@ -1914,7 +1775,6 @@
unsigned long *data, int write_access)
{
unsigned int pos = -1; /* an invalid value */
- int ret;
unsigned long *ptr, regnum;
if ((addr & 0x7) != 0) {
@@ -1946,14 +1806,39 @@
}
if (pos != -1) {
- if (write_access)
- ret = fpregs_set(child, NULL, pos,
- sizeof(unsigned long), data, NULL);
- else
- ret = fpregs_get(child, NULL, pos,
- sizeof(unsigned long), data, NULL);
- if (ret != 0)
- return -1;
+ unsigned reg = pos / sizeof(elf_fpreg_t);
+ int which_half = (pos / sizeof(unsigned long)) & 1;
+
+ if (reg < 32) { /* fr2-fr31 */
+ struct unw_frame_info info;
+ elf_fpreg_t fpreg;
+
+ memset(&info, 0, sizeof(info));
+ unw_init_from_blocked_task(&info, child);
+ if (unw_unwind_to_user(&info) < 0)
+ return 0;
+
+ if (unw_get_fr(&info, reg, &fpreg))
+ return -1;
+ if (write_access) {
+ fpreg.u.bits[which_half] = *data;
+ if (unw_set_fr(&info, reg, fpreg))
+ return -1;
+ } else {
+ *data = fpreg.u.bits[which_half];
+ }
+ } else { /* fph */
+ elf_fpreg_t *p = &child->thread.fph[reg - 32];
+ unsigned long *bits = &p->u.bits[which_half];
+
+ ia64_sync_fph(child);
+ if (write_access)
+ *bits = *data;
+ else if (child->thread.flags & IA64_THREAD_FPH_VALID)
+ *data = *bits;
+ else
+ *data = 0;
+ }
return 0;
}
@@ -2039,15 +1924,14 @@
}
if (pos != -1) {
- if (write_access)
- ret = gpregs_set(child, NULL, pos,
- sizeof(unsigned long), data, NULL);
- else
- ret = gpregs_get(child, NULL, pos,
- sizeof(unsigned long), data, NULL);
- if (ret != 0)
- return -1;
- return 0;
+ struct unw_frame_info info;
+
+ memset(&info, 0, sizeof(info));
+ unw_init_from_blocked_task(&info, child);
+ if (unw_unwind_to_user(&info) < 0)
+ return 0;
+
+ return access_elf_reg(child, &info, pos, data, write_access);
}
/* access debug registers */
@@ -2064,27 +1948,6 @@
"address 0x%lx\n", addr);
return -1;
}
-#ifdef CONFIG_PERFMON
- /*
- * Check if debug registers are used by perfmon. This
- * test must be done once we know that we can do the
- * operation, i.e. the arguments are all valid, but
- * before we start modifying the state.
- *
- * Perfmon needs to keep a count of how many processes
- * are trying to modify the debug registers for system
- * wide monitoring sessions.
- *
- * We also include read access here, because they may
- * cause the PMU-installed debug register state
- * (dbr[], ibr[]) to be reset. The two arrays are also
- * used by perfmon, but we do not use
- * IA64_THREAD_DBG_VALID. The registers are restored
- * by the PMU context switch code.
- */
- if (pfm_use_debug_registers(child))
- return -1;
-#endif
if (!(child->thread.flags & IA64_THREAD_DBG_VALID)) {
child->thread.flags |= IA64_THREAD_DBG_VALID;
@@ -2113,14 +1976,14 @@
.core_note_type = NT_PRSTATUS,
.n = ELF_NGREG,
.size = sizeof(elf_greg_t), .align = sizeof(elf_greg_t),
- .get = gpregs_get, .set = gpregs_set,
+ .regset_get = gpregs_get, .set = gpregs_set,
.writeback = gpregs_writeback
},
{
.core_note_type = NT_PRFPREG,
.n = ELF_NFPREG,
.size = sizeof(elf_fpreg_t), .align = sizeof(elf_fpreg_t),
- .get = fpregs_get, .set = fpregs_set, .active = fpregs_active
+ .regset_get = fpregs_get, .set = fpregs_set, .active = fpregs_active
},
};
diff --git a/arch/ia64/kernel/relocate_kernel.S b/arch/ia64/kernel/relocate_kernel.S
index 7124fe7..527a7b8 100644
--- a/arch/ia64/kernel/relocate_kernel.S
+++ b/arch/ia64/kernel/relocate_kernel.S
@@ -8,10 +8,10 @@
* Copyright (C) 2005 Khalid Aziz <khalid.aziz@hp.com>
* Copyright (C) 2005 Intel Corp, Zou Nan hai <nanhai.zou@intel.com>
*/
+#include <linux/pgtable.h>
#include <asm/asmmacro.h>
#include <asm/kregs.h>
#include <asm/page.h>
-#include <asm/pgtable.h>
#include <asm/mca_asm.h>
/* Must be relocatable PIC code callable as a C function
@@ -319,5 +319,3 @@
;;
br.ret.sptk.many rp
END(ia64_dump_cpu_regs)
-
-
diff --git a/arch/ia64/kernel/sal.c b/arch/ia64/kernel/sal.c
index c455ece..e4f0705 100644
--- a/arch/ia64/kernel/sal.c
+++ b/arch/ia64/kernel/sal.c
@@ -18,6 +18,7 @@
#include <asm/page.h>
#include <asm/sal.h>
#include <asm/pal.h>
+#include <asm/xtp.h>
__cacheline_aligned DEFINE_SPINLOCK(sal_lock);
unsigned long sal_platform_features;
diff --git a/arch/ia64/kernel/salinfo.c b/arch/ia64/kernel/salinfo.c
index b392c0a..a25ab9b 100644
--- a/arch/ia64/kernel/salinfo.c
+++ b/arch/ia64/kernel/salinfo.c
@@ -331,10 +331,10 @@
return size;
}
-static const struct file_operations salinfo_event_fops = {
- .open = salinfo_event_open,
- .read = salinfo_event_read,
- .llseek = noop_llseek,
+static const struct proc_ops salinfo_event_proc_ops = {
+ .proc_open = salinfo_event_open,
+ .proc_read = salinfo_event_read,
+ .proc_lseek = noop_llseek,
};
static int
@@ -534,12 +534,12 @@
return count;
}
-static const struct file_operations salinfo_data_fops = {
- .open = salinfo_log_open,
- .release = salinfo_log_release,
- .read = salinfo_log_read,
- .write = salinfo_log_write,
- .llseek = default_llseek,
+static const struct proc_ops salinfo_data_proc_ops = {
+ .proc_open = salinfo_log_open,
+ .proc_release = salinfo_log_release,
+ .proc_read = salinfo_log_read,
+ .proc_write = salinfo_log_write,
+ .proc_lseek = default_llseek,
};
static int salinfo_cpu_online(unsigned int cpu)
@@ -617,13 +617,13 @@
continue;
entry = proc_create_data("event", S_IRUSR, dir,
- &salinfo_event_fops, data);
+ &salinfo_event_proc_ops, data);
if (!entry)
continue;
*sdir++ = entry;
entry = proc_create_data("data", S_IRUSR | S_IWUSR, dir,
- &salinfo_data_fops, data);
+ &salinfo_data_proc_ops, data);
if (!entry)
continue;
*sdir++ = entry;
diff --git a/arch/ia64/kernel/setup.c b/arch/ia64/kernel/setup.c
index bb320c6..dd595fb 100644
--- a/arch/ia64/kernel/setup.c
+++ b/arch/ia64/kernel/setup.c
@@ -25,6 +25,7 @@
*/
#include <linux/module.h>
#include <linux/init.h>
+#include <linux/pgtable.h>
#include <linux/acpi.h>
#include <linux/console.h>
@@ -56,7 +57,6 @@
#include <asm/meminit.h>
#include <asm/page.h>
#include <asm/patch.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/sal.h>
#include <asm/sections.h>
@@ -65,6 +65,7 @@
#include <asm/tlbflush.h>
#include <asm/unistd.h>
#include <asm/uv/uv.h>
+#include <asm/xtp.h>
#if defined(CONFIG_SMP) && (IA64_CPU_SIZE > PAGE_SIZE)
# error "struct cpuinfo_ia64 too big!"
@@ -289,7 +290,7 @@
}
if (!check_crashkernel_memory(base, size)) {
- pr_warning("crashkernel: There would be kdump memory "
+ pr_warn("crashkernel: There would be kdump memory "
"at %ld GB but this is unusable because it "
"must\nbe below 4 GB. Change the memory "
"configuration of the machine.\n",
@@ -608,9 +609,6 @@
#ifdef CONFIG_VT
if (!conswitchp) {
-# if defined(CONFIG_DUMMY_CONSOLE)
- conswitchp = &dummy_con;
-# endif
# if defined(CONFIG_VGA_CONSOLE)
/*
* Non-legacy systems may route legacy VGA MMIO range to system
diff --git a/arch/ia64/kernel/signal.c b/arch/ia64/kernel/signal.c
index d07ed65..e67b22f 100644
--- a/arch/ia64/kernel/signal.c
+++ b/arch/ia64/kernel/signal.c
@@ -374,7 +374,7 @@
/* note: scr->pt.r10 is already -1 */
break;
}
- /*FALLTHRU*/
+ fallthrough;
case ERESTARTNOINTR:
ia64_decrement_ip(&scr->pt);
restart = 0; /* don't restart twice if handle_signal() fails... */
diff --git a/arch/ia64/kernel/smp.c b/arch/ia64/kernel/smp.c
index de35c54..7b7b64e 100644
--- a/arch/ia64/kernel/smp.c
+++ b/arch/ia64/kernel/smp.c
@@ -39,14 +39,13 @@
#include <asm/io.h>
#include <asm/irq.h>
#include <asm/page.h>
-#include <asm/pgalloc.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/sal.h>
#include <asm/tlbflush.h>
#include <asm/unistd.h>
#include <asm/mca.h>
+#include <asm/xtp.h>
/*
* Note: alignment of 4 entries/cacheline was empirically determined
diff --git a/arch/ia64/kernel/smpboot.c b/arch/ia64/kernel/smpboot.c
index 6501d9a..0cad990 100644
--- a/arch/ia64/kernel/smpboot.c
+++ b/arch/ia64/kernel/smpboot.c
@@ -49,8 +49,6 @@
#include <asm/irq.h>
#include <asm/mca.h>
#include <asm/page.h>
-#include <asm/pgalloc.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/sal.h>
@@ -357,10 +355,6 @@
extern void ia64_init_itm(void);
extern volatile int time_keeper_id;
-#ifdef CONFIG_PERFMON
- extern void pfm_init_percpu(void);
-#endif
-
cpuid = smp_processor_id();
phys_id = hard_smp_processor_id();
itc_master = time_keeper_id;
@@ -391,10 +385,6 @@
ia64_mca_cmc_vector_setup(); /* Setup vector on AP */
-#ifdef CONFIG_PERFMON
- pfm_init_percpu();
-#endif
-
local_irq_enable();
if (!(sal_platform_features & IA64_SAL_PLATFORM_FEATURE_ITC_DRIFT)) {
@@ -450,7 +440,6 @@
#endif
efi_map_pal_code();
cpu_init();
- preempt_disable();
smp_callin();
cpu_startup_entry(CPUHP_AP_ONLINE_IDLE);
diff --git a/arch/ia64/kernel/syscalls/syscall.tbl b/arch/ia64/kernel/syscalls/syscall.tbl
index 36d5faf..b96ed8b 100644
--- a/arch/ia64/kernel/syscalls/syscall.tbl
+++ b/arch/ia64/kernel/syscalls/syscall.tbl
@@ -135,7 +135,7 @@
123 common writev sys_writev
124 common pread64 sys_pread64
125 common pwrite64 sys_pwrite64
-126 common _sysctl sys_sysctl
+126 common _sysctl sys_ni_syscall
127 common mmap sys_mmap
128 common munmap sys_munmap
129 common mlock sys_mlock
@@ -160,7 +160,7 @@
148 common mmap2 sys_mmap2
149 common pciconfig_read sys_pciconfig_read
150 common pciconfig_write sys_pciconfig_write
-151 common perfmonctl sys_perfmonctl
+151 common perfmonctl sys_ni_syscall
152 common sigaltstack sys_sigaltstack
153 common rt_sigaction sys_rt_sigaction
154 common rt_sigpending sys_rt_sigpending
@@ -356,3 +356,8 @@
433 common fspick sys_fspick
434 common pidfd_open sys_pidfd_open
# 435 reserved for clone3
+436 common close_range sys_close_range
+437 common openat2 sys_openat2
+438 common pidfd_getfd sys_pidfd_getfd
+439 common faccessat2 sys_faccessat2
+440 common process_madvise sys_process_madvise
diff --git a/arch/ia64/kernel/syscalls/syscallhdr.sh b/arch/ia64/kernel/syscalls/syscallhdr.sh
index 0c2d2c7..f407b6e 100644
--- a/arch/ia64/kernel/syscalls/syscallhdr.sh
+++ b/arch/ia64/kernel/syscalls/syscallhdr.sh
@@ -32,5 +32,5 @@
printf "#define __NR_syscalls\t%s\n" "${nxt}"
printf "#endif\n"
printf "\n"
- printf "#endif /* %s */" "${fileguard}"
+ printf "#endif /* %s */\n" "${fileguard}"
) > "$out"
diff --git a/arch/ia64/kernel/time.c b/arch/ia64/kernel/time.c
index 1e95d32..7abc5f3 100644
--- a/arch/ia64/kernel/time.c
+++ b/arch/ia64/kernel/time.c
@@ -32,6 +32,7 @@
#include <asm/sections.h>
#include "fsyscall_gtod_data.h"
+#include "irq.h"
static u64 itc_get_cycles(struct clocksource *cs);
@@ -132,7 +133,7 @@
return delta_stime;
}
-void vtime_account_system(struct task_struct *tsk)
+void vtime_account_kernel(struct task_struct *tsk)
{
struct thread_info *ti = task_thread_info(tsk);
__u64 stime = vtime_delta(tsk);
@@ -146,7 +147,7 @@
else
ti->stime += stime;
}
-EXPORT_SYMBOL_GPL(vtime_account_system);
+EXPORT_SYMBOL_GPL(vtime_account_kernel);
void vtime_account_idle(struct task_struct *tsk)
{
@@ -380,13 +381,6 @@
return now;
}
-
-static struct irqaction timer_irqaction = {
- .handler = timer_interrupt,
- .flags = IRQF_IRQPOLL,
- .name = "timer"
-};
-
void read_persistent_clock64(struct timespec64 *ts)
{
efi_gettimeofday(ts);
@@ -395,7 +389,8 @@
void __init
time_init (void)
{
- register_percpu_irq(IA64_TIMER_VECTOR, &timer_irqaction);
+ register_percpu_irq(IA64_TIMER_VECTOR, timer_interrupt, IRQF_IRQPOLL,
+ "timer");
ia64_init_itm();
}
diff --git a/arch/ia64/kernel/unaligned.c b/arch/ia64/kernel/unaligned.c
index 2d4e65b..6c1a895 100644
--- a/arch/ia64/kernel/unaligned.c
+++ b/arch/ia64/kernel/unaligned.c
@@ -1431,7 +1431,7 @@
if (u.insn.x)
/* oops, really a semaphore op (cmpxchg, etc) */
goto failure;
- /*FALLTHRU*/
+ fallthrough;
case LDS_IMM_OP:
case LDSA_IMM_OP:
case LDFS_OP:
@@ -1459,7 +1459,7 @@
if (u.insn.x)
/* oops, really a semaphore op (cmpxchg, etc) */
goto failure;
- /*FALLTHRU*/
+ fallthrough;
case LD_IMM_OP:
case LDA_IMM_OP:
case LDBIAS_IMM_OP:
@@ -1475,7 +1475,7 @@
if (u.insn.x)
/* oops, really a semaphore op (cmpxchg, etc) */
goto failure;
- /*FALLTHRU*/
+ fallthrough;
case ST_IMM_OP:
case STREL_IMM_OP:
ret = emulate_store_int(ifa, u.insn, regs);
diff --git a/arch/ia64/kernel/uncached.c b/arch/ia64/kernel/uncached.c
index 3776ef2..0750f36 100644
--- a/arch/ia64/kernel/uncached.c
+++ b/arch/ia64/kernel/uncached.c
@@ -19,9 +19,9 @@
#include <linux/nmi.h>
#include <linux/genalloc.h>
#include <linux/gfp.h>
+#include <linux/pgtable.h>
#include <asm/page.h>
#include <asm/pal.h>
-#include <asm/pgtable.h>
#include <linux/atomic.h>
#include <asm/tlbflush.h>
diff --git a/arch/ia64/kernel/unwind.c b/arch/ia64/kernel/unwind.c
index 7601fe0..6bd64c3 100644
--- a/arch/ia64/kernel/unwind.c
+++ b/arch/ia64/kernel/unwind.c
@@ -324,7 +324,7 @@
return 0;
}
}
- /* fall through */
+ fallthrough;
case UNW_NAT_NONE:
dummy_nat = 0;
nat_addr = &dummy_nat;
diff --git a/arch/ia64/kernel/unwind_i.h b/arch/ia64/kernel/unwind_i.h
index 67994a7..1dd57ba 100644
--- a/arch/ia64/kernel/unwind_i.h
+++ b/arch/ia64/kernel/unwind_i.h
@@ -42,7 +42,7 @@
struct unw_info_block {
u64 header;
- u64 desc[0]; /* unwind descriptors */
+ u64 desc[]; /* unwind descriptors */
/* personality routine and language-specific data follow behind descriptors */
};
diff --git a/arch/ia64/kernel/vmlinux.lds.S b/arch/ia64/kernel/vmlinux.lds.S
index d9d4e21..9b26578 100644
--- a/arch/ia64/kernel/vmlinux.lds.S
+++ b/arch/ia64/kernel/vmlinux.lds.S
@@ -1,10 +1,13 @@
/* SPDX-License-Identifier: GPL-2.0 */
+#include <linux/pgtable.h>
#include <asm/cache.h>
#include <asm/ptrace.h>
-#include <asm/pgtable.h>
#include <asm/thread_info.h>
+#define EMITS_PT_NOTE
+#define RO_EXCEPTION_TABLE_ALIGN 16
+
#include <asm-generic/vmlinux.lds.h>
OUTPUT_FORMAT("elf64-ia64-little")
@@ -13,7 +16,7 @@
jiffies = jiffies_64;
PHDRS {
- code PT_LOAD;
+ text PT_LOAD;
percpu PT_LOAD;
data PT_LOAD;
note PT_NOTE;
@@ -36,7 +39,7 @@
phys_start = _start - LOAD_OFFSET;
code : {
- } :code
+ } :text
. = KERNEL_START;
_text = .;
@@ -51,6 +54,8 @@
CPUIDLE_TEXT
LOCK_TEXT
KPROBES_TEXT
+ IRQENTRY_TEXT
+ SOFTIRQENTRY_TEXT
*(.gnu.linkonce.t*)
}
@@ -68,11 +73,6 @@
/*
* Read-only data
*/
- NOTES :code :note /* put .notes in text and mark in PT_NOTE */
- code_continues : {
- } : code /* switch back to regular program... */
-
- EXCEPTION_TABLE(16)
/* MCA table */
. = ALIGN(16);
@@ -102,11 +102,11 @@
__start_unwind = .;
*(.IA_64.unwind*)
__end_unwind = .;
- } :code :unwind
+ } :text :unwind
code_continues2 : {
- } : code
+ } :text
- RODATA
+ RO_DATA(4096)
.opd : AT(ADDR(.opd) - LOAD_OFFSET) {
__start_opd = .;
@@ -214,10 +214,11 @@
_end = .;
code : {
- } :code
+ } :text
STABS_DEBUG
DWARF_DEBUG
+ ELF_DETAILS
/* Default discards */
DISCARDS
diff --git a/arch/ia64/lib/Makefile b/arch/ia64/lib/Makefile
index 82118b3..081fcba 100644
--- a/arch/ia64/lib/Makefile
+++ b/arch/ia64/lib/Makefile
@@ -12,7 +12,6 @@
lib-$(CONFIG_ITANIUM) += copy_page.o copy_user.o memcpy.o
lib-$(CONFIG_MCKINLEY) += copy_page_mck.o memcpy_mck.o
-lib-$(CONFIG_PERFMON) += carta_random.o
AFLAGS___divdi3.o =
AFLAGS___udivdi3.o = -DUNSIGNED
diff --git a/arch/ia64/lib/carta_random.S b/arch/ia64/lib/carta_random.S
deleted file mode 100644
index 1a4a639..0000000
--- a/arch/ia64/lib/carta_random.S
+++ /dev/null
@@ -1,55 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-/*
- * Fast, simple, yet decent quality random number generator based on
- * a paper by David G. Carta ("Two Fast Implementations of the
- * `Minimal Standard' Random Number Generator," Communications of the
- * ACM, January, 1990).
- *
- * Copyright (C) 2002 Hewlett-Packard Co
- * David Mosberger-Tang <davidm@hpl.hp.com>
- */
-
-#include <asm/asmmacro.h>
-
-#define a r2
-#define m r3
-#define lo r8
-#define hi r9
-#define t0 r16
-#define t1 r17
-#define seed r32
-
-GLOBAL_ENTRY(carta_random32)
- movl a = (16807 << 16) | 16807
- ;;
- pmpyshr2.u t0 = a, seed, 0
- pmpyshr2.u t1 = a, seed, 16
- ;;
- unpack2.l t0 = t1, t0
- dep m = -1, r0, 0, 31
- ;;
- zxt4 lo = t0
- shr.u hi = t0, 32
- ;;
- dep t0 = 0, hi, 15, 49 // t0 = (hi & 0x7fff)
- ;;
- shl t0 = t0, 16 // t0 = (hi & 0x7fff) << 16
- shr t1 = hi, 15 // t1 = (hi >> 15)
- ;;
- add lo = lo, t0
- ;;
- cmp.gtu p6, p0 = lo, m
- ;;
-(p6) and lo = lo, m
- ;;
-(p6) add lo = 1, lo
- ;;
- add lo = lo, t1
- ;;
- cmp.gtu p6, p0 = lo, m
- ;;
-(p6) and lo = lo, m
- ;;
-(p6) add lo = 1, lo
- br.ret.sptk.many rp
-END(carta_random32)
diff --git a/arch/ia64/lib/csum_partial_copy.c b/arch/ia64/lib/csum_partial_copy.c
index bf9396b..917e313 100644
--- a/arch/ia64/lib/csum_partial_copy.c
+++ b/arch/ia64/lib/csum_partial_copy.c
@@ -12,7 +12,7 @@
#include <linux/types.h>
#include <linux/string.h>
-#include <linux/uaccess.h>
+#include <net/checksum.h>
/*
* XXX Fixme: those 2 inlines are meant for debugging and will go away
@@ -96,46 +96,3 @@
out:
return result;
}
-
-/*
- * XXX Fixme
- *
- * This is very ugly but temporary. THIS NEEDS SERIOUS ENHANCEMENTS.
- * But it's very tricky to get right even in C.
- */
-extern unsigned long do_csum(const unsigned char *, long);
-
-__wsum
-csum_partial_copy_from_user(const void __user *src, void *dst,
- int len, __wsum psum, int *errp)
-{
- unsigned long result;
-
- /* XXX Fixme
- * for now we separate the copy from checksum for obvious
- * alignment difficulties. Look at the Alpha code and you'll be
- * scared.
- */
-
- if (__copy_from_user(dst, src, len) != 0 && errp)
- *errp = -EFAULT;
-
- result = do_csum(dst, len);
-
- /* add in old sum, and carry.. */
- result += (__force u32)psum;
- /* 32+c bits -> 32 bits */
- result = (result & 0xffffffff) + (result >> 32);
- return (__force __wsum)result;
-}
-
-EXPORT_SYMBOL(csum_partial_copy_from_user);
-
-__wsum
-csum_partial_copy_nocheck(const void *src, void *dst, int len, __wsum sum)
-{
- return csum_partial_copy_from_user((__force const void __user *)src,
- dst, len, sum, NULL);
-}
-
-EXPORT_SYMBOL(csum_partial_copy_nocheck);
diff --git a/arch/ia64/mm/contig.c b/arch/ia64/mm/contig.c
index 5b00dc3..e30e360 100644
--- a/arch/ia64/mm/contig.c
+++ b/arch/ia64/mm/contig.c
@@ -21,8 +21,6 @@
#include <linux/swap.h>
#include <asm/meminit.h>
-#include <asm/pgalloc.h>
-#include <asm/pgtable.h>
#include <asm/sections.h>
#include <asm/mca.h>
@@ -210,6 +208,6 @@
printk("Virtual mem_map starts at 0x%p\n", mem_map);
}
#endif /* !CONFIG_VIRTUAL_MEM_MAP */
- free_area_init_nodes(max_zone_pfns);
+ free_area_init(max_zone_pfns);
zero_page_memmap_ptr = virt_to_page(ia64_imva(empty_zero_page));
}
diff --git a/arch/ia64/mm/discontig.c b/arch/ia64/mm/discontig.c
index 41d243c..4d08134 100644
--- a/arch/ia64/mm/discontig.c
+++ b/arch/ia64/mm/discontig.c
@@ -24,7 +24,6 @@
#include <linux/efi.h>
#include <linux/nodemask.h>
#include <linux/slab.h>
-#include <asm/pgalloc.h>
#include <asm/tlb.h>
#include <asm/meminit.h>
#include <asm/numa.h>
@@ -180,7 +179,7 @@
void __init setup_per_cpu_areas(void)
{
struct pcpu_alloc_info *ai;
- struct pcpu_group_info *uninitialized_var(gi);
+ struct pcpu_group_info *gi;
unsigned int *cpu_map;
void *base;
unsigned long base_offset;
@@ -601,7 +600,6 @@
max_dma = virt_to_phys((void *) MAX_DMA_ADDRESS) >> PAGE_SHIFT;
- sparse_memory_present_with_active_regions(MAX_NUMNODES);
sparse_init();
#ifdef CONFIG_VIRTUAL_MEM_MAP
@@ -627,7 +625,7 @@
max_zone_pfns[ZONE_DMA32] = max_dma;
#endif
max_zone_pfns[ZONE_NORMAL] = max_pfn;
- free_area_init_nodes(max_zone_pfns);
+ free_area_init(max_zone_pfns);
zero_page_memmap_ptr = virt_to_page(ia64_imva(empty_zero_page));
}
@@ -656,7 +654,7 @@
int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
struct vmem_altmap *altmap)
{
- return vmemmap_populate_basepages(start, end, node);
+ return vmemmap_populate_basepages(start, end, node, NULL);
}
void vmemmap_free(unsigned long start, unsigned long end,
diff --git a/arch/ia64/mm/fault.c b/arch/ia64/mm/fault.c
index c2f299f..cd9766d 100644
--- a/arch/ia64/mm/fault.c
+++ b/arch/ia64/mm/fault.c
@@ -14,8 +14,8 @@
#include <linux/kdebug.h>
#include <linux/prefetch.h>
#include <linux/uaccess.h>
+#include <linux/perf_event.h>
-#include <asm/pgtable.h>
#include <asm/processor.h>
#include <asm/exception.h>
@@ -29,6 +29,7 @@
mapped_kernel_page_is_present (unsigned long address)
{
pgd_t *pgd;
+ p4d_t *p4d;
pud_t *pud;
pmd_t *pmd;
pte_t *ptep, pte;
@@ -37,7 +38,11 @@
if (pgd_none(*pgd) || pgd_bad(*pgd))
return 0;
- pud = pud_offset(pgd, address);
+ p4d = p4d_offset(pgd, address);
+ if (p4d_none(*p4d) || p4d_bad(*p4d))
+ return 0;
+
+ pud = pud_offset(p4d, address);
if (pud_none(*pud) || pud_bad(*pud))
return 0;
@@ -65,13 +70,13 @@
struct mm_struct *mm = current->mm;
unsigned long mask;
vm_fault_t fault;
- unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
+ unsigned int flags = FAULT_FLAG_DEFAULT;
mask = ((((isr >> IA64_ISR_X_BIT) & 1UL) << VM_EXEC_BIT)
| (((isr >> IA64_ISR_W_BIT) & 1UL) << VM_WRITE_BIT));
- /* mmap_sem is performance critical.... */
- prefetchw(&mm->mmap_sem);
+ /* mmap_lock is performance critical.... */
+ prefetchw(&mm->mmap_lock);
/*
* If we're in an interrupt or have no user context, we must not take the fault..
@@ -82,7 +87,7 @@
#ifdef CONFIG_VIRTUAL_MEM_MAP
/*
* If fault is in region 5 and we are in the kernel, we may already
- * have the mmap_sem (pfn_valid macro is called during mmap). There
+ * have the mmap_lock (pfn_valid macro is called during mmap). There
* is no vma for region 5 addr's anyway, so skip getting the semaphore
* and go directly to the exception handling code.
*/
@@ -101,8 +106,10 @@
flags |= FAULT_FLAG_USER;
if (mask & VM_WRITE)
flags |= FAULT_FLAG_WRITE;
+
+ perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
retry:
- down_read(&mm->mmap_sem);
+ mmap_read_lock(mm);
vma = find_vma_prev(mm, address, &prev_vma);
if (!vma && !prev_vma )
@@ -139,9 +146,9 @@
* sure we exit gracefully rather than endlessly redo the
* fault.
*/
- fault = handle_mm_fault(vma, address, flags);
+ fault = handle_mm_fault(vma, address, flags, regs);
- if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
+ if (fault_signal_pending(fault, regs))
return;
if (unlikely(fault & VM_FAULT_ERROR)) {
@@ -162,15 +169,10 @@
}
if (flags & FAULT_FLAG_ALLOW_RETRY) {
- if (fault & VM_FAULT_MAJOR)
- current->maj_flt++;
- else
- current->min_flt++;
if (fault & VM_FAULT_RETRY) {
- flags &= ~FAULT_FLAG_ALLOW_RETRY;
flags |= FAULT_FLAG_TRIED;
- /* No need to up_read(&mm->mmap_sem) as we would
+ /* No need to mmap_read_unlock(mm) as we would
* have already released it in __lock_page_or_retry
* in mm/filemap.c.
*/
@@ -179,7 +181,7 @@
}
}
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
return;
check_expansion:
@@ -210,7 +212,7 @@
goto good_area;
bad_area:
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
#ifdef CONFIG_VIRTUAL_MEM_MAP
bad_area_no_up:
#endif
@@ -276,7 +278,7 @@
return;
out_of_memory:
- up_read(&mm->mmap_sem);
+ mmap_read_unlock(mm);
if (!user_mode(regs))
goto no_context;
pagefault_out_of_memory();
diff --git a/arch/ia64/mm/hugetlbpage.c b/arch/ia64/mm/hugetlbpage.c
index d16e419..b331f94 100644
--- a/arch/ia64/mm/hugetlbpage.c
+++ b/arch/ia64/mm/hugetlbpage.c
@@ -18,7 +18,6 @@
#include <linux/sysctl.h>
#include <linux/log2.h>
#include <asm/mman.h>
-#include <asm/pgalloc.h>
#include <asm/tlb.h>
#include <asm/tlbflush.h>
@@ -30,12 +29,14 @@
{
unsigned long taddr = htlbpage_to_page(addr);
pgd_t *pgd;
+ p4d_t *p4d;
pud_t *pud;
pmd_t *pmd;
pte_t *pte = NULL;
pgd = pgd_offset(mm, taddr);
- pud = pud_alloc(mm, pgd, taddr);
+ p4d = p4d_offset(pgd, taddr);
+ pud = pud_alloc(mm, p4d, taddr);
if (pud) {
pmd = pmd_alloc(mm, pud, taddr);
if (pmd)
@@ -49,17 +50,21 @@
{
unsigned long taddr = htlbpage_to_page(addr);
pgd_t *pgd;
+ p4d_t *p4d;
pud_t *pud;
pmd_t *pmd;
pte_t *pte = NULL;
pgd = pgd_offset(mm, taddr);
if (pgd_present(*pgd)) {
- pud = pud_offset(pgd, taddr);
- if (pud_present(*pud)) {
- pmd = pmd_offset(pud, taddr);
- if (pmd_present(*pmd))
- pte = pte_offset_map(pmd, taddr);
+ p4d = p4d_offset(pgd, addr);
+ if (p4d_present(*p4d)) {
+ pud = pud_offset(p4d, taddr);
+ if (pud_present(*pud)) {
+ pmd = pmd_offset(pud, taddr);
+ if (pmd_present(*pmd))
+ pte = pte_offset_map(pmd, taddr);
+ }
}
}
diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c
index ee50506..f316a83 100644
--- a/arch/ia64/mm/init.c
+++ b/arch/ia64/mm/init.c
@@ -8,7 +8,7 @@
#include <linux/kernel.h>
#include <linux/init.h>
-#include <linux/dma-noncoherent.h>
+#include <linux/dma-map-ops.h>
#include <linux/dmar.h>
#include <linux/efi.h>
#include <linux/elf.h>
@@ -73,8 +73,7 @@
* DMA can be marked as "clean" so that lazy_mmu_prot_update() doesn't have to
* flush them when they get mapped into an executable vm-area.
*/
-void arch_sync_dma_for_cpu(struct device *dev, phys_addr_t paddr,
- size_t size, enum dma_data_direction dir)
+void arch_dma_mark_clean(phys_addr_t paddr, size_t size)
{
unsigned long pfn = PHYS_PFN(paddr);
@@ -118,13 +117,13 @@
vma->vm_end = vma->vm_start + PAGE_SIZE;
vma->vm_flags = VM_DATA_DEFAULT_FLAGS|VM_GROWSUP|VM_ACCOUNT;
vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
- down_write(¤t->mm->mmap_sem);
+ mmap_write_lock(current->mm);
if (insert_vm_struct(current->mm, vma)) {
- up_write(¤t->mm->mmap_sem);
+ mmap_write_unlock(current->mm);
vm_area_free(vma);
return;
}
- up_write(¤t->mm->mmap_sem);
+ mmap_write_unlock(current->mm);
}
/* map NaT-page at address zero to speed up speculative dereferencing of NULL: */
@@ -136,13 +135,13 @@
vma->vm_page_prot = __pgprot(pgprot_val(PAGE_READONLY) | _PAGE_MA_NAT);
vma->vm_flags = VM_READ | VM_MAYREAD | VM_IO |
VM_DONTEXPAND | VM_DONTDUMP;
- down_write(¤t->mm->mmap_sem);
+ mmap_write_lock(current->mm);
if (insert_vm_struct(current->mm, vma)) {
- up_write(¤t->mm->mmap_sem);
+ mmap_write_unlock(current->mm);
vm_area_free(vma);
return;
}
- up_write(¤t->mm->mmap_sem);
+ mmap_write_unlock(current->mm);
}
}
}
@@ -208,6 +207,7 @@
put_kernel_page (struct page *page, unsigned long address, pgprot_t pgprot)
{
pgd_t *pgd;
+ p4d_t *p4d;
pud_t *pud;
pmd_t *pmd;
pte_t *pte;
@@ -215,7 +215,10 @@
pgd = pgd_offset_k(address); /* note: this is NOT pgd_offset()! */
{
- pud = pud_alloc(&init_mm, pgd, address);
+ p4d = p4d_alloc(&init_mm, pgd, address);
+ if (!p4d)
+ goto out;
+ pud = pud_alloc(&init_mm, p4d, address);
if (!pud)
goto out;
pmd = pmd_alloc(&init_mm, pud, address);
@@ -382,6 +385,7 @@
do {
pgd_t *pgd;
+ p4d_t *p4d;
pud_t *pud;
pmd_t *pmd;
pte_t *pte;
@@ -392,7 +396,13 @@
continue;
}
- pud = pud_offset(pgd, end_address);
+ p4d = p4d_offset(pgd, end_address);
+ if (p4d_none(*p4d)) {
+ end_address += P4D_SIZE;
+ continue;
+ }
+
+ pud = pud_offset(p4d, end_address);
if (pud_none(*pud)) {
end_address += PUD_SIZE;
continue;
@@ -430,6 +440,7 @@
struct page *map_start, *map_end;
int node;
pgd_t *pgd;
+ p4d_t *p4d;
pud_t *pud;
pmd_t *pmd;
pte_t *pte;
@@ -444,12 +455,20 @@
for (address = start_page; address < end_page; address += PAGE_SIZE) {
pgd = pgd_offset_k(address);
if (pgd_none(*pgd)) {
+ p4d = memblock_alloc_node(PAGE_SIZE, PAGE_SIZE, node);
+ if (!p4d)
+ goto err_alloc;
+ pgd_populate(&init_mm, pgd, p4d);
+ }
+ p4d = p4d_offset(pgd, address);
+
+ if (p4d_none(*p4d)) {
pud = memblock_alloc_node(PAGE_SIZE, PAGE_SIZE, node);
if (!pud)
goto err_alloc;
- pgd_populate(&init_mm, pgd, pud);
+ p4d_populate(&init_mm, p4d, pud);
}
- pud = pud_offset(pgd, address);
+ pud = pud_offset(p4d, address);
if (pud_none(*pud)) {
pmd = memblock_alloc_node(PAGE_SIZE, PAGE_SIZE, node);
@@ -517,18 +536,18 @@
if (map_start < map_end)
memmap_init_zone((unsigned long)(map_end - map_start),
- args->nid, args->zone, page_to_pfn(map_start),
- MEMINIT_EARLY, NULL);
+ args->nid, args->zone, page_to_pfn(map_start), page_to_pfn(map_end),
+ MEMINIT_EARLY, NULL, MIGRATE_MOVABLE);
return 0;
}
void __meminit
-memmap_init (unsigned long size, int nid, unsigned long zone,
+arch_memmap_init (unsigned long size, int nid, unsigned long zone,
unsigned long start_pfn)
{
if (!vmem_map) {
- memmap_init_zone(size, nid, zone, start_pfn,
- MEMINIT_EARLY, NULL);
+ memmap_init_zone(size, nid, zone, start_pfn, start_pfn + size,
+ MEMINIT_EARLY, NULL, MIGRATE_MOVABLE);
} else {
struct page *start;
struct memmap_init_callback_data args;
@@ -543,6 +562,10 @@
}
}
+void __init memmap_init(void)
+{
+}
+
int
ia64_pfn_valid (unsigned long pfn)
{
@@ -670,13 +693,16 @@
#ifdef CONFIG_MEMORY_HOTPLUG
int arch_add_memory(int nid, u64 start, u64 size,
- struct mhp_restrictions *restrictions)
+ struct mhp_params *params)
{
unsigned long start_pfn = start >> PAGE_SHIFT;
unsigned long nr_pages = size >> PAGE_SHIFT;
int ret;
- ret = __add_pages(nid, start_pfn, nr_pages, restrictions);
+ if (WARN_ON_ONCE(params->pgprot.pgprot != PAGE_KERNEL.pgprot))
+ return -EINVAL;
+
+ ret = __add_pages(nid, start_pfn, nr_pages, params);
if (ret)
printk("%s: Problem encountered in __add_pages() as ret=%d\n",
__func__, ret);
diff --git a/arch/ia64/mm/ioremap.c b/arch/ia64/mm/ioremap.c
index 0c0de2c..55fd3eb 100644
--- a/arch/ia64/mm/ioremap.c
+++ b/arch/ia64/mm/ioremap.c
@@ -8,6 +8,7 @@
#include <linux/module.h>
#include <linux/efi.h>
#include <linux/io.h>
+#include <linux/mm.h>
#include <linux/vmalloc.h>
#include <asm/io.h>
#include <asm/meminit.h>
@@ -99,14 +100,14 @@
EXPORT_SYMBOL(ioremap);
void __iomem *
-ioremap_nocache (unsigned long phys_addr, unsigned long size)
+ioremap_uc(unsigned long phys_addr, unsigned long size)
{
if (kern_mem_attribute(phys_addr, size) & EFI_MEMORY_WB)
return NULL;
return __ioremap_uc(phys_addr);
}
-EXPORT_SYMBOL(ioremap_nocache);
+EXPORT_SYMBOL(ioremap_uc);
void
early_iounmap (volatile void __iomem *addr, unsigned long size)
diff --git a/arch/ia64/mm/numa.c b/arch/ia64/mm/numa.c
index 5e1015e..f349642 100644
--- a/arch/ia64/mm/numa.c
+++ b/arch/ia64/mm/numa.c
@@ -106,7 +106,5 @@
return 0;
return nid;
}
-
-EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid);
#endif
#endif
diff --git a/arch/ia64/mm/tlb.c b/arch/ia64/mm/tlb.c
index 72cc568..135b513 100644
--- a/arch/ia64/mm/tlb.c
+++ b/arch/ia64/mm/tlb.c
@@ -27,7 +27,6 @@
#include <asm/delay.h>
#include <asm/mmu_context.h>
-#include <asm/pgalloc.h>
#include <asm/pal.h>
#include <asm/tlbflush.h>
#include <asm/dma.h>
@@ -369,7 +368,7 @@
void ia64_tlb_init(void)
{
- ia64_ptce_info_t uninitialized_var(ptce_info); /* GCC be quiet */
+ ia64_ptce_info_t ptce_info;
u64 tr_pgbits;
long status;
pal_vm_info_1_u_t vm_info_1;
diff --git a/arch/ia64/oprofile/Makefile b/arch/ia64/oprofile/Makefile
index cd134d6..fc7944d 100644
--- a/arch/ia64/oprofile/Makefile
+++ b/arch/ia64/oprofile/Makefile
@@ -8,4 +8,3 @@
timer_int.o )
oprofile-y := $(DRIVER_OBJS) init.o backtrace.o
-oprofile-$(CONFIG_PERFMON) += perfmon.o
diff --git a/arch/ia64/oprofile/init.c b/arch/ia64/oprofile/init.c
index 31b545c..a692ba1 100644
--- a/arch/ia64/oprofile/init.c
+++ b/arch/ia64/oprofile/init.c
@@ -18,21 +18,11 @@
int __init oprofile_arch_init(struct oprofile_operations *ops)
{
- int ret = -ENODEV;
-
-#ifdef CONFIG_PERFMON
- /* perfmon_init() can fail, but we have no way to report it */
- ret = perfmon_init(ops);
-#endif
ops->backtrace = ia64_backtrace;
-
- return ret;
+ return -ENODEV;
}
void oprofile_arch_exit(void)
{
-#ifdef CONFIG_PERFMON
- perfmon_exit();
-#endif
}
diff --git a/arch/ia64/oprofile/perfmon.c b/arch/ia64/oprofile/perfmon.c
deleted file mode 100644
index 192d3e8..0000000
--- a/arch/ia64/oprofile/perfmon.c
+++ /dev/null
@@ -1,99 +0,0 @@
-/**
- * @file perfmon.c
- *
- * @remark Copyright 2003 OProfile authors
- * @remark Read the file COPYING
- *
- * @author John Levon <levon@movementarian.org>
- */
-
-#include <linux/kernel.h>
-#include <linux/oprofile.h>
-#include <linux/sched.h>
-#include <asm/perfmon.h>
-#include <asm/ptrace.h>
-#include <asm/errno.h>
-
-static int allow_ints;
-
-static int
-perfmon_handler(struct task_struct *task, void *buf, pfm_ovfl_arg_t *arg,
- struct pt_regs *regs, unsigned long stamp)
-{
- int event = arg->pmd_eventid;
-
- arg->ovfl_ctrl.bits.reset_ovfl_pmds = 1;
-
- /* the owner of the oprofile event buffer may have exited
- * without perfmon being shutdown (e.g. SIGSEGV)
- */
- if (allow_ints)
- oprofile_add_sample(regs, event);
- return 0;
-}
-
-
-static int perfmon_start(void)
-{
- allow_ints = 1;
- return 0;
-}
-
-
-static void perfmon_stop(void)
-{
- allow_ints = 0;
-}
-
-
-#define OPROFILE_FMT_UUID { \
- 0x77, 0x7a, 0x6e, 0x61, 0x20, 0x65, 0x73, 0x69, 0x74, 0x6e, 0x72, 0x20, 0x61, 0x65, 0x0a, 0x6c }
-
-static pfm_buffer_fmt_t oprofile_fmt = {
- .fmt_name = "oprofile_format",
- .fmt_uuid = OPROFILE_FMT_UUID,
- .fmt_handler = perfmon_handler,
-};
-
-
-static char *get_cpu_type(void)
-{
- __u8 family = local_cpu_data->family;
-
- switch (family) {
- case 0x07:
- return "ia64/itanium";
- case 0x1f:
- return "ia64/itanium2";
- default:
- return "ia64/ia64";
- }
-}
-
-
-/* all the ops are handled via userspace for IA64 perfmon */
-
-static int using_perfmon;
-
-int perfmon_init(struct oprofile_operations *ops)
-{
- int ret = pfm_register_buffer_fmt(&oprofile_fmt);
- if (ret)
- return -ENODEV;
-
- ops->cpu_type = get_cpu_type();
- ops->start = perfmon_start;
- ops->stop = perfmon_stop;
- using_perfmon = 1;
- printk(KERN_INFO "oprofile: using perfmon.\n");
- return 0;
-}
-
-
-void perfmon_exit(void)
-{
- if (!using_perfmon)
- return;
-
- pfm_unregister_buffer_fmt(oprofile_fmt.fmt_uuid);
-}