From 2ca052a3710fac208eee690faefdeb8bbd4586a1 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Mon, 2 Apr 2012 16:15:33 -0700 Subject: x86: Use correct byte-sized register constraint in __xchg_op() x86-64 can access the low half of any register, but i386 can only do it with a subset of registers. 'r' causes compilation failures on i386, but 'q' expresses the constraint properly. Signed-off-by: Jeremy Fitzhardinge Link: http://lkml.kernel.org/r/4F7A3315.501@goop.org Reported-by: Leigh Scott Tested-by: Thomas Reitmayr Signed-off-by: H. Peter Anvin Cc: v3.3 --- arch/x86/include/asm/cmpxchg.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/include/asm/cmpxchg.h b/arch/x86/include/asm/cmpxchg.h index b3b733262909..bc18d0ed459a 100644 --- a/arch/x86/include/asm/cmpxchg.h +++ b/arch/x86/include/asm/cmpxchg.h @@ -43,7 +43,7 @@ extern void __add_wrong_size(void) switch (sizeof(*(ptr))) { \ case __X86_CASE_B: \ asm volatile (lock #op "b %b0, %1\n" \ - : "+r" (__ret), "+m" (*(ptr)) \ + : "+q" (__ret), "+m" (*(ptr)) \ : : "memory", "cc"); \ break; \ case __X86_CASE_W: \ -- cgit v1.2.3 From 8c91c5325e107ec17e40a59a47c6517387d64eb7 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Fri, 6 Apr 2012 09:30:57 -0700 Subject: x86: Use correct byte-sized register constraint in __add() Similar to: 2ca052a x86: Use correct byte-sized register constraint in __xchg_op() ... the __add() macro also needs to use a "q" constraint in the byte-sized case, lest we try to generate an illegal register. Link: http://lkml.kernel.org/r/4F7A3315.501@goop.org Signed-off-by: H. Peter Anvin Cc: Jeremy Fitzhardinge Cc: Leigh Scott Cc: Thomas Reitmayr Cc: v3.3 --- arch/x86/include/asm/cmpxchg.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/include/asm/cmpxchg.h b/arch/x86/include/asm/cmpxchg.h index bc18d0ed459a..99480e55973d 100644 --- a/arch/x86/include/asm/cmpxchg.h +++ b/arch/x86/include/asm/cmpxchg.h @@ -173,7 +173,7 @@ extern void __add_wrong_size(void) switch (sizeof(*(ptr))) { \ case __X86_CASE_B: \ asm volatile (lock "addb %b1, %0\n" \ - : "+m" (*(ptr)) : "ri" (inc) \ + : "+m" (*(ptr)) : "qi" (inc) \ : "memory", "cc"); \ break; \ case __X86_CASE_W: \ -- cgit v1.2.3 From 92ae03f2ef99fbc23bfa9080d6b58f25227bd7ef Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Fri, 6 Apr 2012 14:32:32 -0700 Subject: x86: merge 32/64-bit versions of 'strncpy_from_user()' and speed it up This merges the 32- and 64-bit versions of the x86 strncpy_from_user() by just rewriting it in C rather than the ancient inline asm versions that used lodsb/stosb and had been duplicated for (trivial) differences between the 32-bit and 64-bit versions. While doing that, it also speeds them up by doing the accesses a word at a time. Finally, the new routines also properly handle the case of hitting the end of the address space, which we have never done correctly before (fs/namei.c has a hack around it for that reason). Despite all these improvements, it actually removes more lines than it adds, due to the de-duplication. Also, we no longer export (or define) the legacy __strncpy_from_user() function (that was defined to not do the user permission checks), since it's not actually used anywhere, and the user address space checks are built in to the new code. Other architecture maintainers have been notified that the old hack in fs/namei.c will be going away in the 3.5 merge window, in case they copied the x86 approach of being a bit cavalier about the end of the address space. Cc: linux-arch@vger.kernel.org Cc: Ingo Molnar Cc: Peter Anvin" Signed-off-by: Linus Torvalds --- arch/x86/include/asm/uaccess.h | 2 + arch/x86/include/asm/uaccess_32.h | 5 -- arch/x86/include/asm/uaccess_64.h | 4 -- arch/x86/lib/usercopy.c | 103 ++++++++++++++++++++++++++++++++++++++ arch/x86/lib/usercopy_32.c | 87 -------------------------------- arch/x86/lib/usercopy_64.c | 49 ------------------ 6 files changed, 105 insertions(+), 145 deletions(-) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/include/asm/uaccess.h b/arch/x86/include/asm/uaccess.h index 8be5f54d9360..e0544597cfe7 100644 --- a/arch/x86/include/asm/uaccess.h +++ b/arch/x86/include/asm/uaccess.h @@ -557,6 +557,8 @@ struct __large_struct { unsigned long buf[100]; }; extern unsigned long copy_from_user_nmi(void *to, const void __user *from, unsigned long n); +extern __must_check long +strncpy_from_user(char *dst, const char __user *src, long count); /* * movsl can be slow when source and dest are not both 8-byte aligned diff --git a/arch/x86/include/asm/uaccess_32.h b/arch/x86/include/asm/uaccess_32.h index 566e803cc602..8084bc73b18c 100644 --- a/arch/x86/include/asm/uaccess_32.h +++ b/arch/x86/include/asm/uaccess_32.h @@ -213,11 +213,6 @@ static inline unsigned long __must_check copy_from_user(void *to, return n; } -long __must_check strncpy_from_user(char *dst, const char __user *src, - long count); -long __must_check __strncpy_from_user(char *dst, - const char __user *src, long count); - /** * strlen_user: - Get the size of a string in user space. * @str: The string to measure. diff --git a/arch/x86/include/asm/uaccess_64.h b/arch/x86/include/asm/uaccess_64.h index 1c66d30971ad..fcd4b6f3ef02 100644 --- a/arch/x86/include/asm/uaccess_64.h +++ b/arch/x86/include/asm/uaccess_64.h @@ -208,10 +208,6 @@ int __copy_in_user(void __user *dst, const void __user *src, unsigned size) } } -__must_check long -strncpy_from_user(char *dst, const char __user *src, long count); -__must_check long -__strncpy_from_user(char *dst, const char __user *src, long count); __must_check long strnlen_user(const char __user *str, long n); __must_check long __strnlen_user(const char __user *str, long n); __must_check long strlen_user(const char __user *str); diff --git a/arch/x86/lib/usercopy.c b/arch/x86/lib/usercopy.c index 97be9cb54483..57252c928f56 100644 --- a/arch/x86/lib/usercopy.c +++ b/arch/x86/lib/usercopy.c @@ -7,6 +7,8 @@ #include #include +#include + /* * best effort, GUP based copy_from_user() that is NMI-safe */ @@ -41,3 +43,104 @@ copy_from_user_nmi(void *to, const void __user *from, unsigned long n) return len; } EXPORT_SYMBOL_GPL(copy_from_user_nmi); + +static inline unsigned long count_bytes(unsigned long mask) +{ + mask = (mask - 1) & ~mask; + mask >>= 7; + return count_masked_bytes(mask); +} + +/* + * Do a strncpy, return length of string without final '\0'. + * 'count' is the user-supplied count (return 'count' if we + * hit it), 'max' is the address space maximum (and we return + * -EFAULT if we hit it). + */ +static inline long do_strncpy_from_user(char *dst, const char __user *src, long count, long max) +{ + long res = 0; + + /* + * Truncate 'max' to the user-specified limit, so that + * we only have one limit we need to check in the loop + */ + if (max > count) + max = count; + + while (max >= sizeof(unsigned long)) { + unsigned long c; + + /* Fall back to byte-at-a-time if we get a page fault */ + if (unlikely(__get_user(c,(unsigned long __user *)(src+res)))) + break; + /* This can write a few bytes past the NUL character, but that's ok */ + *(unsigned long *)(dst+res) = c; + c = has_zero(c); + if (c) + return res + count_bytes(c); + res += sizeof(unsigned long); + max -= sizeof(unsigned long); + } + + while (max) { + char c; + + if (unlikely(__get_user(c,src+res))) + return -EFAULT; + dst[res] = c; + if (!c) + return res; + res++; + max--; + } + + /* + * Uhhuh. We hit 'max'. But was that the user-specified maximum + * too? If so, that's ok - we got as much as the user asked for. + */ + if (res >= count) + return count; + + /* + * Nope: we hit the address space limit, and we still had more + * characters the caller would have wanted. That's an EFAULT. + */ + return -EFAULT; +} + +/** + * strncpy_from_user: - Copy a NUL terminated string from userspace. + * @dst: Destination address, in kernel space. This buffer must be at + * least @count bytes long. + * @src: Source address, in user space. + * @count: Maximum number of bytes to copy, including the trailing NUL. + * + * Copies a NUL-terminated string from userspace to kernel space. + * + * On success, returns the length of the string (not including the trailing + * NUL). + * + * If access to userspace fails, returns -EFAULT (some data may have been + * copied). + * + * If @count is smaller than the length of the string, copies @count bytes + * and returns @count. + */ +long +strncpy_from_user(char *dst, const char __user *src, long count) +{ + unsigned long max_addr, src_addr; + + if (unlikely(count <= 0)) + return 0; + + max_addr = current_thread_info()->addr_limit.seg; + src_addr = (unsigned long)src; + if (likely(src_addr < max_addr)) { + unsigned long max = max_addr - src_addr; + return do_strncpy_from_user(dst, src, count, max); + } + return -EFAULT; +} +EXPORT_SYMBOL(strncpy_from_user); diff --git a/arch/x86/lib/usercopy_32.c b/arch/x86/lib/usercopy_32.c index d9b094ca7aaa..ef2a6a5d78e3 100644 --- a/arch/x86/lib/usercopy_32.c +++ b/arch/x86/lib/usercopy_32.c @@ -32,93 +32,6 @@ static inline int __movsl_is_ok(unsigned long a1, unsigned long a2, unsigned lon #define movsl_is_ok(a1, a2, n) \ __movsl_is_ok((unsigned long)(a1), (unsigned long)(a2), (n)) -/* - * Copy a null terminated string from userspace. - */ - -#define __do_strncpy_from_user(dst, src, count, res) \ -do { \ - int __d0, __d1, __d2; \ - might_fault(); \ - __asm__ __volatile__( \ - " testl %1,%1\n" \ - " jz 2f\n" \ - "0: lodsb\n" \ - " stosb\n" \ - " testb %%al,%%al\n" \ - " jz 1f\n" \ - " decl %1\n" \ - " jnz 0b\n" \ - "1: subl %1,%0\n" \ - "2:\n" \ - ".section .fixup,\"ax\"\n" \ - "3: movl %5,%0\n" \ - " jmp 2b\n" \ - ".previous\n" \ - _ASM_EXTABLE(0b,3b) \ - : "=&d"(res), "=&c"(count), "=&a" (__d0), "=&S" (__d1), \ - "=&D" (__d2) \ - : "i"(-EFAULT), "0"(count), "1"(count), "3"(src), "4"(dst) \ - : "memory"); \ -} while (0) - -/** - * __strncpy_from_user: - Copy a NUL terminated string from userspace, with less checking. - * @dst: Destination address, in kernel space. This buffer must be at - * least @count bytes long. - * @src: Source address, in user space. - * @count: Maximum number of bytes to copy, including the trailing NUL. - * - * Copies a NUL-terminated string from userspace to kernel space. - * Caller must check the specified block with access_ok() before calling - * this function. - * - * On success, returns the length of the string (not including the trailing - * NUL). - * - * If access to userspace fails, returns -EFAULT (some data may have been - * copied). - * - * If @count is smaller than the length of the string, copies @count bytes - * and returns @count. - */ -long -__strncpy_from_user(char *dst, const char __user *src, long count) -{ - long res; - __do_strncpy_from_user(dst, src, count, res); - return res; -} -EXPORT_SYMBOL(__strncpy_from_user); - -/** - * strncpy_from_user: - Copy a NUL terminated string from userspace. - * @dst: Destination address, in kernel space. This buffer must be at - * least @count bytes long. - * @src: Source address, in user space. - * @count: Maximum number of bytes to copy, including the trailing NUL. - * - * Copies a NUL-terminated string from userspace to kernel space. - * - * On success, returns the length of the string (not including the trailing - * NUL). - * - * If access to userspace fails, returns -EFAULT (some data may have been - * copied). - * - * If @count is smaller than the length of the string, copies @count bytes - * and returns @count. - */ -long -strncpy_from_user(char *dst, const char __user *src, long count) -{ - long res = -EFAULT; - if (access_ok(VERIFY_READ, src, 1)) - __do_strncpy_from_user(dst, src, count, res); - return res; -} -EXPORT_SYMBOL(strncpy_from_user); - /* * Zero Userspace */ diff --git a/arch/x86/lib/usercopy_64.c b/arch/x86/lib/usercopy_64.c index b7c2849ffb66..0d0326f388c0 100644 --- a/arch/x86/lib/usercopy_64.c +++ b/arch/x86/lib/usercopy_64.c @@ -8,55 +8,6 @@ #include #include -/* - * Copy a null terminated string from userspace. - */ - -#define __do_strncpy_from_user(dst,src,count,res) \ -do { \ - long __d0, __d1, __d2; \ - might_fault(); \ - __asm__ __volatile__( \ - " testq %1,%1\n" \ - " jz 2f\n" \ - "0: lodsb\n" \ - " stosb\n" \ - " testb %%al,%%al\n" \ - " jz 1f\n" \ - " decq %1\n" \ - " jnz 0b\n" \ - "1: subq %1,%0\n" \ - "2:\n" \ - ".section .fixup,\"ax\"\n" \ - "3: movq %5,%0\n" \ - " jmp 2b\n" \ - ".previous\n" \ - _ASM_EXTABLE(0b,3b) \ - : "=&r"(res), "=&c"(count), "=&a" (__d0), "=&S" (__d1), \ - "=&D" (__d2) \ - : "i"(-EFAULT), "0"(count), "1"(count), "3"(src), "4"(dst) \ - : "memory"); \ -} while (0) - -long -__strncpy_from_user(char *dst, const char __user *src, long count) -{ - long res; - __do_strncpy_from_user(dst, src, count, res); - return res; -} -EXPORT_SYMBOL(__strncpy_from_user); - -long -strncpy_from_user(char *dst, const char __user *src, long count) -{ - long res = -EFAULT; - if (access_ok(VERIFY_READ, src, 1)) - return __strncpy_from_user(dst, src, count); - return res; -} -EXPORT_SYMBOL(strncpy_from_user); - /* * Zero Userspace */ -- cgit v1.2.3 From 68894632afb2729a1d8785c877840953894c7283 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Mon, 2 Apr 2012 18:06:48 +0200 Subject: x86/platform: Remove incorrect error message in x86_default_fixup_cpu_id() It's only called from amd.c:srat_detect_node(). The introduced condition for calling the fixup code is true for all AMD multi-node processors, e.g. Magny-Cours and Interlagos. There we have 2 NUMA nodes on one socket. Thus there are cores having different numa-node-id but with equal phys_proc_id. There is no point to print error messages in such a situation. The confusing/misleading error message was introduced with commit 64be4c1c2428e148de6081af235e2418e6a66dda ("x86: Add x86_init platform override to fix up NUMA core numbering"). Remove the default fixup function (especially the error message) and replace it by a NULL pointer check, move the Numascale-specific condition for calling the fixup into the fixup-function itself and slightly adapt the comment. Signed-off-by: Andreas Herrmann Acked-by: Borislav Petkov Cc: Cc: Cc: Cc: Link: http://lkml.kernel.org/r/20120402160648.GR27684@alberich.amd.com Signed-off-by: Ingo Molnar --- arch/x86/include/asm/x86_init.h | 1 - arch/x86/kernel/apic/apic_numachip.c | 7 +++++-- arch/x86/kernel/cpu/amd.c | 7 ++++--- arch/x86/kernel/cpu/common.c | 9 --------- arch/x86/kernel/x86_init.c | 1 - 5 files changed, 9 insertions(+), 16 deletions(-) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/include/asm/x86_init.h b/arch/x86/include/asm/x86_init.h index baaca8defec8..764b66a4cf89 100644 --- a/arch/x86/include/asm/x86_init.h +++ b/arch/x86/include/asm/x86_init.h @@ -195,6 +195,5 @@ extern struct x86_msi_ops x86_msi; extern void x86_init_noop(void); extern void x86_init_uint_noop(unsigned int unused); -extern void x86_default_fixup_cpu_id(struct cpuinfo_x86 *c, int node); #endif diff --git a/arch/x86/kernel/apic/apic_numachip.c b/arch/x86/kernel/apic/apic_numachip.c index 899803e03214..23e75422e013 100644 --- a/arch/x86/kernel/apic/apic_numachip.c +++ b/arch/x86/kernel/apic/apic_numachip.c @@ -207,8 +207,11 @@ static void __init map_csrs(void) static void fixup_cpu_id(struct cpuinfo_x86 *c, int node) { - c->phys_proc_id = node; - per_cpu(cpu_llc_id, smp_processor_id()) = node; + + if (c->phys_proc_id != node) { + c->phys_proc_id = node; + per_cpu(cpu_llc_id, smp_processor_id()) = node; + } } static int __init numachip_system_init(void) diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 1248f9ceabc1..1c67ca100e4c 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -353,10 +353,11 @@ static void __cpuinit srat_detect_node(struct cpuinfo_x86 *c) node = per_cpu(cpu_llc_id, cpu); /* - * If core numbers are inconsistent, it's likely a multi-fabric platform, - * so invoke platform-specific handler + * On multi-fabric platform (e.g. Numascale NumaChip) a + * platform-specific handler needs to be called to fixup some + * IDs of the CPU. */ - if (c->phys_proc_id != node) + if (x86_cpuinit.fixup_cpu_id) x86_cpuinit.fixup_cpu_id(c, node); if (!node_online(node)) { diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 67e258362a3d..cf79302198a6 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -1162,15 +1162,6 @@ static void dbg_restore_debug_regs(void) #define dbg_restore_debug_regs() #endif /* ! CONFIG_KGDB */ -/* - * Prints an error where the NUMA and configured core-number mismatch and the - * platform didn't override this to fix it up - */ -void __cpuinit x86_default_fixup_cpu_id(struct cpuinfo_x86 *c, int node) -{ - pr_err("NUMA core number %d differs from configured core number %d\n", node, c->phys_proc_id); -} - /* * cpu_init() initializes state that is per-CPU. Some data is already * initialized (naturally) in the bootstrap process, such as the GDT diff --git a/arch/x86/kernel/x86_init.c b/arch/x86/kernel/x86_init.c index e9f265fd79ae..9cf71d0b2d37 100644 --- a/arch/x86/kernel/x86_init.c +++ b/arch/x86/kernel/x86_init.c @@ -93,7 +93,6 @@ struct x86_init_ops x86_init __initdata = { struct x86_cpuinit_ops x86_cpuinit __cpuinitdata = { .early_percpu_clock_init = x86_init_noop, .setup_percpu_clockev = setup_secondary_APIC_clock, - .fixup_cpu_id = x86_default_fixup_cpu_id, }; static void default_nmi_init(void) { }; -- cgit v1.2.3 From 98e5272fe70d62e193f70acf9951667beab27aba Mon Sep 17 00:00:00 2001 From: "H.J. Lu" Date: Mon, 23 Apr 2012 14:51:14 -0700 Subject: x32: Check __ILP32__ instead of __LP64__ for x32 Check __LP64__ isn't a reliable way to tell if we are compiling for x32 since __LP64__ isnn't specified by x86-64 psABI. Not all x86-64 compilers define __LP64__, which was added to GCC 3.3. The updated x32 psABI: https://sites.google.com/site/x32abi/documents definse _ILP32 and __ILP32__ for x32. GCC trunk and 4.7 branch have been updated to define _ILP32 and __ILP32__ for x32. This patch replaces __LP64__ check with __ILP32__. Signed-off-by: H.J. Lu Signed-off-by: H. Peter Anvin --- arch/x86/include/asm/posix_types.h | 6 +++--- arch/x86/include/asm/sigcontext.h | 2 +- arch/x86/include/asm/unistd.h | 6 +++--- 3 files changed, 7 insertions(+), 7 deletions(-) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/include/asm/posix_types.h b/arch/x86/include/asm/posix_types.h index 3427b7798dbc..7ef7c3020e5c 100644 --- a/arch/x86/include/asm/posix_types.h +++ b/arch/x86/include/asm/posix_types.h @@ -7,9 +7,9 @@ #else # ifdef __i386__ # include "posix_types_32.h" -# elif defined(__LP64__) -# include "posix_types_64.h" -# else +# elif defined(__ILP32__) # include "posix_types_x32.h" +# else +# include "posix_types_64.h" # endif #endif diff --git a/arch/x86/include/asm/sigcontext.h b/arch/x86/include/asm/sigcontext.h index 4a085383af27..5ca71c065eef 100644 --- a/arch/x86/include/asm/sigcontext.h +++ b/arch/x86/include/asm/sigcontext.h @@ -257,7 +257,7 @@ struct sigcontext { __u64 oldmask; __u64 cr2; struct _fpstate __user *fpstate; /* zero when no FPU context */ -#ifndef __LP64__ +#ifdef __ILP32__ __u32 __fpstate_pad; #endif __u64 reserved1[8]; diff --git a/arch/x86/include/asm/unistd.h b/arch/x86/include/asm/unistd.h index 37cdc9d99bb1..4437001d8e3d 100644 --- a/arch/x86/include/asm/unistd.h +++ b/arch/x86/include/asm/unistd.h @@ -63,10 +63,10 @@ #else # ifdef __i386__ # include -# elif defined(__LP64__) -# include -# else +# elif defined(__ILP32__) # include +# else +# include # endif #endif -- cgit v1.2.3 From 89b8835ec865dddd6673a8dd7003581bf2377176 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 23 Apr 2012 16:34:12 -0700 Subject: x32, siginfo: Provide proper overrides for x32 siginfo_t Provide the proper override macros for x32 siginfo_t. The combination of a special type here and an overall alignment constraint actually ends up with all the types being properly aligned, but the hack is needed to keep the substructures inside siginfo_t from adding padding. Note: use __attribute__((aligned())) since __aligned() is not exported to user space. [ v2: fix stray semicolon ] Reported-by: H.J. Lu Cc: Bruce J. Beare Cc: Arnd Bergmann Link: http://lkml.kernel.org/r/CAMe9rOqF6Kh6-NK7oP0Fpzkd4SBAWU%2BG53hwBbSD4iA2UzyxuA@mail.gmail.com Signed-off-by: H. Peter Anvin --- arch/x86/include/asm/siginfo.h | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/include/asm/siginfo.h b/arch/x86/include/asm/siginfo.h index fc1aa5535646..34c47b3341c0 100644 --- a/arch/x86/include/asm/siginfo.h +++ b/arch/x86/include/asm/siginfo.h @@ -2,7 +2,13 @@ #define _ASM_X86_SIGINFO_H #ifdef __x86_64__ -# define __ARCH_SI_PREAMBLE_SIZE (4 * sizeof(int)) +# ifdef __ILP32__ /* x32 */ +typedef long long __kernel_si_clock_t __attribute__((aligned(4))); +# define __ARCH_SI_CLOCK_T __kernel_si_clock_t +# define __ARCH_SI_ATTRIBUTES __attribute__((aligned(8))) +# else /* x86-64 */ +# define __ARCH_SI_PREAMBLE_SIZE (4 * sizeof(int)) +# endif #endif #include -- cgit v1.2.3 From e419b4cc585680940bc42f8ca8a071d6023fb1bb Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Thu, 3 May 2012 10:16:43 -0700 Subject: vfs: make word-at-a-time accesses handle a non-existing page It turns out that there are more cases than CONFIG_DEBUG_PAGEALLOC that can have holes in the kernel address space: it seems to happen easily with Xen, and it looks like the AMD gart64 code will also punch holes dynamically. Actually hitting that case is still very unlikely, so just do the access, and take an exception and fix it up for the very unlikely case of it being a page-crosser with no next page. And hey, this abstraction might even help other architectures that have other issues with unaligned word accesses than the possible missing next page. IOW, this could do the byte order magic too. Peter Anvin fixed a thinko in the shifting for the exception case. Reported-and-tested-by: Jana Saout Cc: Peter Anvin Signed-off-by: Linus Torvalds --- arch/x86/Kconfig | 2 +- arch/x86/include/asm/word-at-a-time.h | 33 +++++++++++++++++++++++++++++++++ fs/dcache.c | 26 ++++++++++++++++++++++---- fs/namei.c | 4 ++-- 4 files changed, 58 insertions(+), 7 deletions(-) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 1d14cc6b79ad..c9866b0b77d8 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -81,7 +81,7 @@ config X86 select CLKEVT_I8253 select ARCH_HAVE_NMI_SAFE_CMPXCHG select GENERIC_IOMAP - select DCACHE_WORD_ACCESS if !DEBUG_PAGEALLOC + select DCACHE_WORD_ACCESS config INSTRUCTION_DECODER def_bool (KPROBES || PERF_EVENTS) diff --git a/arch/x86/include/asm/word-at-a-time.h b/arch/x86/include/asm/word-at-a-time.h index 6fe6767b7124..e58f03b206c3 100644 --- a/arch/x86/include/asm/word-at-a-time.h +++ b/arch/x86/include/asm/word-at-a-time.h @@ -43,4 +43,37 @@ static inline unsigned long has_zero(unsigned long a) return ((a - REPEAT_BYTE(0x01)) & ~a) & REPEAT_BYTE(0x80); } +/* + * Load an unaligned word from kernel space. + * + * In the (very unlikely) case of the word being a page-crosser + * and the next page not being mapped, take the exception and + * return zeroes in the non-existing part. + */ +static inline unsigned long load_unaligned_zeropad(const void *addr) +{ + unsigned long ret, dummy; + + asm( + "1:\tmov %2,%0\n" + "2:\n" + ".section .fixup,\"ax\"\n" + "3:\t" + "lea %2,%1\n\t" + "and %3,%1\n\t" + "mov (%1),%0\n\t" + "leal %2,%%ecx\n\t" + "andl %4,%%ecx\n\t" + "shll $3,%%ecx\n\t" + "shr %%cl,%0\n\t" + "jmp 2b\n" + ".previous\n" + _ASM_EXTABLE(1b, 3b) + :"=&r" (ret),"=&c" (dummy) + :"m" (*(unsigned long *)addr), + "i" (-sizeof(unsigned long)), + "i" (sizeof(unsigned long)-1)); + return ret; +} + #endif /* _ASM_WORD_AT_A_TIME_H */ diff --git a/fs/dcache.c b/fs/dcache.c index b60ddc41d783..b80531c91779 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -141,18 +141,29 @@ int proc_nr_dentry(ctl_table *table, int write, void __user *buffer, * Compare 2 name strings, return 0 if they match, otherwise non-zero. * The strings are both count bytes long, and count is non-zero. */ +#ifdef CONFIG_DCACHE_WORD_ACCESS + +#include +/* + * NOTE! 'cs' and 'scount' come from a dentry, so it has a + * aligned allocation for this particular component. We don't + * strictly need the load_unaligned_zeropad() safety, but it + * doesn't hurt either. + * + * In contrast, 'ct' and 'tcount' can be from a pathname, and do + * need the careful unaligned handling. + */ static inline int dentry_cmp(const unsigned char *cs, size_t scount, const unsigned char *ct, size_t tcount) { -#ifdef CONFIG_DCACHE_WORD_ACCESS unsigned long a,b,mask; if (unlikely(scount != tcount)) return 1; for (;;) { - a = *(unsigned long *)cs; - b = *(unsigned long *)ct; + a = load_unaligned_zeropad(cs); + b = load_unaligned_zeropad(ct); if (tcount < sizeof(unsigned long)) break; if (unlikely(a != b)) @@ -165,7 +176,13 @@ static inline int dentry_cmp(const unsigned char *cs, size_t scount, } mask = ~(~0ul << tcount*8); return unlikely(!!((a ^ b) & mask)); +} + #else + +static inline int dentry_cmp(const unsigned char *cs, size_t scount, + const unsigned char *ct, size_t tcount) +{ if (scount != tcount) return 1; @@ -177,9 +194,10 @@ static inline int dentry_cmp(const unsigned char *cs, size_t scount, tcount--; } while (tcount); return 0; -#endif } +#endif + static void __d_free(struct rcu_head *head) { struct dentry *dentry = container_of(head, struct dentry, d_u.d_rcu); diff --git a/fs/namei.c b/fs/namei.c index 0062dd17eb55..c42791914f82 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -1429,7 +1429,7 @@ unsigned int full_name_hash(const unsigned char *name, unsigned int len) unsigned long hash = 0; for (;;) { - a = *(unsigned long *)name; + a = load_unaligned_zeropad(name); if (len < sizeof(unsigned long)) break; hash += a; @@ -1459,7 +1459,7 @@ static inline unsigned long hash_name(const char *name, unsigned int *hashp) do { hash = (hash + a) * 9; len += sizeof(unsigned long); - a = *(unsigned long *)(name+len); + a = load_unaligned_zeropad(name+len); /* Do we have any NUL or '/' bytes in this word? */ mask = has_zero(a) | has_zero(a ^ REPEAT_BYTE('/')); } while (!mask); -- cgit v1.2.3 From c3709e6734daa4d9b37fe31592ebb0eb57bae1bb Mon Sep 17 00:00:00 2001 From: Alan Cox Date: Mon, 14 May 2012 18:41:09 +0100 Subject: x86, kvm: KVM paravirt kernels don't check for CPUID being unavailable We set cpuid_level to -1 if there is no CPUID instruction (only possible on i386). Signed-off-by: Alan Cox Link: http://lkml.kernel.org/r/20120514174059.30236.1064.stgit@bluebook Resolves-bug: https://bugzilla.kernel.org/show_bug.cgi?id=12122 Cc: Avi Kivity Cc: Marcelo Tosatti Signed-off-by: H. Peter Anvin --- arch/x86/include/asm/kvm_para.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'arch/x86/include/asm') diff --git a/arch/x86/include/asm/kvm_para.h b/arch/x86/include/asm/kvm_para.h index 734c3767cfac..183922e13de1 100644 --- a/arch/x86/include/asm/kvm_para.h +++ b/arch/x86/include/asm/kvm_para.h @@ -170,6 +170,9 @@ static inline int kvm_para_available(void) unsigned int eax, ebx, ecx, edx; char signature[13]; + if (boot_cpu_data.cpuid_level < 0) + return 0; /* So we don't blow up on old processors */ + cpuid(KVM_CPUID_SIGNATURE, &eax, &ebx, &ecx, &edx); memcpy(signature + 0, &ebx, 4); memcpy(signature + 4, &ecx, 4); -- cgit v1.2.3