From 0f8f308925ebe0480bd9831d32963ee0b885e24b Mon Sep 17 00:00:00 2001 From: Hiroshi Shimamoto Date: Thu, 26 Mar 2009 10:03:08 -0700 Subject: [PATCH 1/9] x86: signal: check sas_ss_size instead of sas_ss_flags() Impact: fix redundant and incorrect check Oleg Nesterov noticed wrt commit: 14fc9fb: x86: signal: check signal stack overflow properly >> No need to check SA_ONSTACK if we're already using alternate signal stack. > > Yes, but this also mean that we don't need sas_ss_flags() under > "if (!onsigstack)", Checking on_sig_stack() in sas_ss_flags() at get_sigframe() is redundant and not correct on 64 bit. To check sas_ss_size is enough. Reported-by: Oleg Nesterov Signed-off-by: Hiroshi Shimamoto Cc: roland@redhat.com LKML-Reference: <49CBB54C.5080201@ct.jp.nec.com> Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index dfcc74ab0ab6..14425166b8e3 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -221,7 +221,7 @@ get_sigframe(struct k_sigaction *ka, struct pt_regs *regs, size_t frame_size, if (!onsigstack) { /* This is the X/Open sanctioned signal stack switching. */ if (ka->sa.sa_flags & SA_ONSTACK) { - if (sas_ss_flags(sp) == 0) + if (current->sas_ss_size) sp = current->sas_ss_sp + current->sas_ss_size; } else { #ifdef CONFIG_X86_32 From eb12ce60c81826a99eadbc56401e08ceb37a0cc2 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 1 Apr 2009 15:06:33 +0900 Subject: [PATCH 2/9] x86,percpu: fix inverted NUMA test in setup_pcpu_remap() setup_percpu_remap() is for NUMA machines yet it bailed out with -EINVAL if pcpu_need_numa(). Fix the inverted condition. This problem was reported by David Miller and verified by Yinhai Lu. Reported-by: David Miller Reported-by: Yinghai Lu Signed-off-by: Tejun Heo LKML-Reference: <49D30469.8020006@kernel.org> Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup_percpu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 400331b50a53..876b1271c1dc 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -162,7 +162,7 @@ static ssize_t __init setup_pcpu_remap(size_t static_size) * If large page isn't supported, there's no benefit in doing * this. Also, on non-NUMA, embedding is better. */ - if (!cpu_has_pse || pcpu_need_numa()) + if (!cpu_has_pse || !pcpu_need_numa()) return -EINVAL; last = NULL; From 3de46fda4c104deef17ec70f85361f5c6b84ce0e Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 1 Apr 2009 00:27:44 -0700 Subject: [PATCH 3/9] x86: remove duplicated code with pcpu_need_numa() Impact: clean up those code pcpu_need_numa(), should be removed. Signed-off-by: Yinghai Lu Acked-by: Tejun Heo Acked-by: David Miller LKML-Reference: <49D31770.9090502@kernel.org> Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup_percpu.c | 14 -------------- 1 file changed, 14 deletions(-) diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 876b1271c1dc..3a97a4cf1872 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -153,7 +153,6 @@ static struct page * __init pcpur_get_page(unsigned int cpu, int pageno) static ssize_t __init setup_pcpu_remap(size_t static_size) { static struct vm_struct vm; - pg_data_t *last; size_t ptrs_size, dyn_size; unsigned int cpu; ssize_t ret; @@ -165,19 +164,6 @@ static ssize_t __init setup_pcpu_remap(size_t static_size) if (!cpu_has_pse || !pcpu_need_numa()) return -EINVAL; - last = NULL; - for_each_possible_cpu(cpu) { - int node = early_cpu_to_node(cpu); - - if (node_online(node) && NODE_DATA(node) && - last && last != NODE_DATA(node)) - goto proceed; - - last = NODE_DATA(node); - } - return -EINVAL; - -proceed: /* * Currently supports only single page. Supporting multiple * pages won't be too difficult if it ever becomes necessary. From a7f8c50d90a4e983c456ae75e534b5cd6c03674b Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Thu, 2 Apr 2009 16:01:26 +0900 Subject: [PATCH 4/9] x86, mm: fix misuse of debug_kmap_atomic Impact: fix CONFIG_DEBUG_HIGHMEM=y breakage Commit 7ca43e756 ("mm: use debug_kmap_atomic") introduced some debug_kmap_atomic() calls in the wrong places. Signed-off-by: Akinobu Mita Cc: Andrew Morton Cc: Linus Torvalds LKML-Reference: <20090402070126.GA3951@localhost.localdomain> Signed-off-by: Ingo Molnar --- arch/x86/mm/highmem_32.c | 1 - arch/x86/mm/iomap_32.c | 2 +- 2 files changed, 1 insertion(+), 2 deletions(-) diff --git a/arch/x86/mm/highmem_32.c b/arch/x86/mm/highmem_32.c index 5bc5d1688c1c..8126e8d1a2a4 100644 --- a/arch/x86/mm/highmem_32.c +++ b/arch/x86/mm/highmem_32.c @@ -40,7 +40,6 @@ void *kmap_atomic_prot(struct page *page, enum km_type type, pgprot_t prot) debug_kmap_atomic(type); - debug_kmap_atomic(type); idx = type + KM_TYPE_NR*smp_processor_id(); vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx); BUG_ON(!pte_none(*(kmap_pte-idx))); diff --git a/arch/x86/mm/iomap_32.c b/arch/x86/mm/iomap_32.c index bff0c9032f8c..e331f77348a7 100644 --- a/arch/x86/mm/iomap_32.c +++ b/arch/x86/mm/iomap_32.c @@ -39,6 +39,7 @@ void *kmap_atomic_prot_pfn(unsigned long pfn, enum km_type type, pgprot_t prot) pagefault_disable(); + debug_kmap_atomic(type); idx = type + KM_TYPE_NR * smp_processor_id(); vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx); set_pte(kmap_pte - idx, pfn_pte(pfn, prot)); @@ -72,7 +73,6 @@ iounmap_atomic(void *kvaddr, enum km_type type) unsigned long vaddr = (unsigned long) kvaddr & PAGE_MASK; enum fixed_addresses idx = type + KM_TYPE_NR*smp_processor_id(); - debug_kmap_atomic(type); /* * Force other mappings to Oops if they'll try to access this pte * without first remap it. Keeping stale mappings around is a bad idea From 67796bf7dc54c035fd97f2681a72e5d2bf2a234a Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 2 Apr 2009 15:55:55 +0200 Subject: [PATCH 5/9] x86/dma: unify definition of pci_unmap_addr* and pci_unmap_len macros Impact: unification of pci-dma macros and pci_32.h removal This patch unifies the definition of the pci_unmap_addr*, pci_unmap_len* and DECLARE_PCI_UNMAP* macros. This makes sense because the pci_unmap functions are no longer no-ops anymore when the kernel runs with CONFIG_DMA_API_DEBUG. Without an iommu or DMA_API_DEBUG it is a no-op on 32 bit because the dma mapping path returns a physical address and therefore the dma-api implementation has no internal state which needs to be destroyed with an unmap call. This unification also simplifies the port of x86_64 iommu drivers to 32 bit x86 and let us get rid of pci_32.h. Signed-off-by: Joerg Roedel Acked-by: Stephen Hemminger --- arch/x86/include/asm/pci.h | 36 +++++++++++++++++++++++++++++++---- arch/x86/include/asm/pci_32.h | 34 --------------------------------- arch/x86/include/asm/pci_64.h | 22 --------------------- 3 files changed, 32 insertions(+), 60 deletions(-) delete mode 100644 arch/x86/include/asm/pci_32.h diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h index a0301bfeb954..e545ea01abcf 100644 --- a/arch/x86/include/asm/pci.h +++ b/arch/x86/include/asm/pci.h @@ -89,12 +89,40 @@ extern void pci_iommu_alloc(void); /* MSI arch hook */ #define arch_setup_msi_irqs arch_setup_msi_irqs +#define PCI_DMA_BUS_IS_PHYS (dma_ops->is_phys) + +#if defined(CONFIG_X86_64) || defined(CONFIG_DMA_API_DEBUG) + +#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) \ + dma_addr_t ADDR_NAME; +#define DECLARE_PCI_UNMAP_LEN(LEN_NAME) \ + __u32 LEN_NAME; +#define pci_unmap_addr(PTR, ADDR_NAME) \ + ((PTR)->ADDR_NAME) +#define pci_unmap_addr_set(PTR, ADDR_NAME, VAL) \ + (((PTR)->ADDR_NAME) = (VAL)) +#define pci_unmap_len(PTR, LEN_NAME) \ + ((PTR)->LEN_NAME) +#define pci_unmap_len_set(PTR, LEN_NAME, VAL) \ + (((PTR)->LEN_NAME) = (VAL)) + +#else + +#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) dma_addr_t ADDR_NAME[0]; +#define DECLARE_PCI_UNMAP_LEN(LEN_NAME) unsigned LEN_NAME[0]; +#define pci_unmap_addr(PTR, ADDR_NAME) sizeof((PTR)->ADDR_NAME) +#define pci_unmap_addr_set(PTR, ADDR_NAME, VAL) \ + do { break; } while (pci_unmap_addr(PTR, ADDR_NAME)) +#define pci_unmap_len(PTR, LEN_NAME) sizeof((PTR)->LEN_NAME) +#define pci_unmap_len_set(PTR, LEN_NAME, VAL) \ + do { break; } while (pci_unmap_len(PTR, LEN_NAME)) + +#endif + #endif /* __KERNEL__ */ -#ifdef CONFIG_X86_32 -# include "pci_32.h" -#else -# include "pci_64.h" +#ifdef CONFIG_X86_64 +#include "pci_64.h" #endif /* implement the pci_ DMA API in terms of the generic device dma_ one */ diff --git a/arch/x86/include/asm/pci_32.h b/arch/x86/include/asm/pci_32.h deleted file mode 100644 index 6f1213a6ef4f..000000000000 --- a/arch/x86/include/asm/pci_32.h +++ /dev/null @@ -1,34 +0,0 @@ -#ifndef _ASM_X86_PCI_32_H -#define _ASM_X86_PCI_32_H - - -#ifdef __KERNEL__ - - -/* Dynamic DMA mapping stuff. - * i386 has everything mapped statically. - */ - -struct pci_dev; - -/* The PCI address space does equal the physical memory - * address space. The networking and block device layers use - * this boolean for bounce buffer decisions. - */ -#define PCI_DMA_BUS_IS_PHYS (1) - -/* pci_unmap_{page,single} is a nop so... */ -#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) dma_addr_t ADDR_NAME[0]; -#define DECLARE_PCI_UNMAP_LEN(LEN_NAME) unsigned LEN_NAME[0]; -#define pci_unmap_addr(PTR, ADDR_NAME) sizeof((PTR)->ADDR_NAME) -#define pci_unmap_addr_set(PTR, ADDR_NAME, VAL) \ - do { break; } while (pci_unmap_addr(PTR, ADDR_NAME)) -#define pci_unmap_len(PTR, LEN_NAME) sizeof((PTR)->LEN_NAME) -#define pci_unmap_len_set(PTR, LEN_NAME, VAL) \ - do { break; } while (pci_unmap_len(PTR, LEN_NAME)) - - -#endif /* __KERNEL__ */ - - -#endif /* _ASM_X86_PCI_32_H */ diff --git a/arch/x86/include/asm/pci_64.h b/arch/x86/include/asm/pci_64.h index 4da207982777..ae5e40f67daf 100644 --- a/arch/x86/include/asm/pci_64.h +++ b/arch/x86/include/asm/pci_64.h @@ -24,28 +24,6 @@ extern int (*pci_config_write)(int seg, int bus, int dev, int fn, extern void dma32_reserve_bootmem(void); -/* The PCI address space does equal the physical memory - * address space. The networking and block device layers use - * this boolean for bounce buffer decisions - * - * On AMD64 it mostly equals, but we set it to zero if a hardware - * IOMMU (gart) of sotware IOMMU (swiotlb) is available. - */ -#define PCI_DMA_BUS_IS_PHYS (dma_ops->is_phys) - -#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME) \ - dma_addr_t ADDR_NAME; -#define DECLARE_PCI_UNMAP_LEN(LEN_NAME) \ - __u32 LEN_NAME; -#define pci_unmap_addr(PTR, ADDR_NAME) \ - ((PTR)->ADDR_NAME) -#define pci_unmap_addr_set(PTR, ADDR_NAME, VAL) \ - (((PTR)->ADDR_NAME) = (VAL)) -#define pci_unmap_len(PTR, LEN_NAME) \ - ((PTR)->LEN_NAME) -#define pci_unmap_len_set(PTR, LEN_NAME, VAL) \ - (((PTR)->LEN_NAME) = (VAL)) - #endif /* __KERNEL__ */ #endif /* _ASM_X86_PCI_64_H */ From 95a38f34635bdf06089de763b4becbc957694977 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Wed, 1 Apr 2009 17:35:00 -0700 Subject: [PATCH 6/9] x86, setup: compile with -DDISABLE_BRANCH_PROFILING Impact: code size reduction (possibly critical) The x86 boot and decompression code has no use of the branch profiling constructs, so disable them. This would bloat the setup code by as much as 14K, eating up a fairly large chunk of the 32K area we are guaranteed to have. Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/boot/Makefile | 3 ++- arch/x86/boot/compressed/Makefile | 1 + 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/arch/x86/boot/Makefile b/arch/x86/boot/Makefile index fb737ce5888d..6633b6e7505a 100644 --- a/arch/x86/boot/Makefile +++ b/arch/x86/boot/Makefile @@ -57,6 +57,7 @@ $(obj)/cpustr.h: $(obj)/mkcpustr FORCE # How to compile the 16-bit code. Note we always compile for -march=i386, # that way we can complain to the user if the CPU is insufficient. KBUILD_CFLAGS := $(LINUXINCLUDE) -g -Os -D_SETUP -D__KERNEL__ \ + -DDISABLE_BRANCH_PROFILING \ -Wall -Wstrict-prototypes \ -march=i386 -mregparm=3 \ -include $(srctree)/$(src)/code16gcc.h \ @@ -66,7 +67,7 @@ KBUILD_CFLAGS := $(LINUXINCLUDE) -g -Os -D_SETUP -D__KERNEL__ \ $(call cc-option, -fno-unit-at-a-time)) \ $(call cc-option, -fno-stack-protector) \ $(call cc-option, -mpreferred-stack-boundary=2) -KBUILD_CFLAGS += $(call cc-option,-m32) +KBUILD_CFLAGS += $(call cc-option, -m32) KBUILD_AFLAGS := $(KBUILD_CFLAGS) -D__ASSEMBLY__ $(obj)/bzImage: asflags-y := $(SVGA_MODE) diff --git a/arch/x86/boot/compressed/Makefile b/arch/x86/boot/compressed/Makefile index 3ca4c194b8e5..65551c9f8571 100644 --- a/arch/x86/boot/compressed/Makefile +++ b/arch/x86/boot/compressed/Makefile @@ -8,6 +8,7 @@ targets := vmlinux vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2 vmlinux.bin.lzma h KBUILD_CFLAGS := -m$(BITS) -D__KERNEL__ $(LINUX_INCLUDE) -O2 KBUILD_CFLAGS += -fno-strict-aliasing -fPIC +KBUILD_CFLAGS += -DDISABLE_BRANCH_PROFILING cflags-$(CONFIG_X86_64) := -mcmodel=small KBUILD_CFLAGS += $(cflags-y) KBUILD_CFLAGS += $(call cc-option,-ffreestanding) From 6a491e2e3e52a64c6d88a192c56499d931842ac5 Mon Sep 17 00:00:00 2001 From: Andrew Morton Date: Thu, 2 Apr 2009 16:44:38 -0700 Subject: [PATCH 7/9] x86: fix is_io_mapping_possible() build warning on i386 allnoconfig i386 allnoconfig: arch/x86/mm/iomap_32.c: In function 'is_io_mapping_possible': arch/x86/mm/iomap_32.c:27: warning: comparison is always false due to limited range of data type Signed-off-by: Andrew Morton Signed-off-by: Ingo Molnar --- arch/x86/mm/iomap_32.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/x86/mm/iomap_32.c b/arch/x86/mm/iomap_32.c index e331f77348a7..8056545e2d39 100644 --- a/arch/x86/mm/iomap_32.c +++ b/arch/x86/mm/iomap_32.c @@ -23,7 +23,7 @@ int is_io_mapping_possible(resource_size_t base, unsigned long size) { -#ifndef CONFIG_X86_PAE +#if !defined(CONFIG_X86_PAE) && defined(CONFIG_PHYS_ADDR_T_64BIT) /* There is no way to map greater than 1 << 32 address without PAE */ if (base + size > 0x100000000ULL) return 0; From 9b7b89efa3bdaceaa2efb93e2d635391835da209 Mon Sep 17 00:00:00 2001 From: Joseph Cihula Date: Mon, 30 Mar 2009 14:03:01 -0700 Subject: [PATCH 8/9] x86: disable stack-protector for __restore_processor_state() The __restore_processor_state() fn restores %gs on resume from S3. As such, it cannot be protected by the stack-protector guard since %gs will not be correct on function entry. There are only a few other fns in this file and it should not negatively impact kernel security that they will also have the stack-protector guard removed (and so it's not worth moving them to another file). Without this change, S3 resume on a kernel built with CONFIG_CC_STACKPROTECTOR_ALL=y will fail. Signed-off-by: Joseph Cihula Tested-by: Chris Wright Cc: Arjan van de Ven Cc: Tejun Heo LKML-Reference: <49D13385.5060900@intel.com> Signed-off-by: Ingo Molnar --- arch/x86/power/Makefile | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/arch/x86/power/Makefile b/arch/x86/power/Makefile index 9ff4d5b55ad1..58b32db33125 100644 --- a/arch/x86/power/Makefile +++ b/arch/x86/power/Makefile @@ -1,2 +1,7 @@ +# __restore_processor_state() restores %gs after S3 resume and so should not +# itself be stack-protected +nostackp := $(call cc-option, -fno-stack-protector) +CFLAGS_cpu_$(BITS).o := $(nostackp) + obj-$(CONFIG_PM_SLEEP) += cpu_$(BITS).o obj-$(CONFIG_HIBERNATION) += hibernate_$(BITS).o hibernate_asm_$(BITS).o From c5c67c7cba6a652d1c62dce45b0c130e5cb2a802 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sat, 4 Apr 2009 00:31:02 +0200 Subject: [PATCH 9/9] x86, mtrr: remove debug message The MTRR code grew a new debug message which triggers commonly: [ 40.142276] get_mtrr: cpu0 reg00 base=0000000000 size=0000080000 write-back [ 40.142280] get_mtrr: cpu0 reg01 base=0000080000 size=0000040000 write-back [ 40.142284] get_mtrr: cpu0 reg02 base=0000100000 size=0000040000 write-back [ 40.142311] get_mtrr: cpu0 reg00 base=0000000000 size=0000080000 write-back [ 40.142314] get_mtrr: cpu0 reg01 base=0000080000 size=0000040000 write-back [ 40.142317] get_mtrr: cpu0 reg02 base=0000100000 size=0000040000 write-back Remove this annoyance. Reported-by: Linus Torvalds Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mtrr/generic.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c index 37f28fc7cf95..0b776c09aff3 100644 --- a/arch/x86/kernel/cpu/mtrr/generic.c +++ b/arch/x86/kernel/cpu/mtrr/generic.c @@ -462,9 +462,6 @@ static void generic_get_mtrr(unsigned int reg, unsigned long *base, *base = base_hi << (32 - PAGE_SHIFT) | base_lo >> PAGE_SHIFT; *type = base_lo & 0xff; - printk(KERN_DEBUG " get_mtrr: cpu%d reg%02d base=%010lx size=%010lx %s\n", - cpu, reg, *base, *size, - mtrr_attrib_to_str(*type & 0xff)); out_put_cpu: put_cpu(); }