compiler.h, kasan: Avoid duplicating __read_once_size_nocheck()
Subsystems (4)
PATCH 1/5 compiler.h, kasan: Avoid duplicating __read_once_size_nocheck()
Instead of having two identical __read_once_size_nocheck() functions with different attributes, consolidate all the difference in new macro __no_kasan_or_inline and use it. No functional changes.
--- include/linux/compiler.h | 14 ++++++-------- 1 file changed, 6 insertions(+), 8 deletions(-)
include/linux/compiler.h +6 -8
@@ -190,23 +190,21 @@ void __read_once_size(const volatile void *p, void *res, int size)#ifdef CONFIG_KASAN /* -* This function is not 'inline' because __no_sanitize_address confilcts +* We can't declare function 'inline' because __no_sanitize_address confilcts * with inlining. Attempt to inline it may cause a build failure. * https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67368 * '__maybe_unused' allows us to avoid defined-but-not-used warnings. */ -static __no_sanitize_address __maybe_unused -void __read_once_size_nocheck(const volatile void *p, void *res, int size) -{ - __READ_ONCE_SIZE; -} +# define __no_kasan_or_inline __no_sanitize_address __maybe_unused#else -static __always_inline +# define __no_kasan_or_inline __always_inline +#endif + +static __no_kasan_or_inline void __read_once_size_nocheck(const volatile void *p, void *res, int size) { __READ_ONCE_SIZE; } -#endif static __always_inline void __write_once_size(volatile void *p, void *res, int size) {
Signature · 1 line
2.13.6
PATCH 2/5 compiler.h: Add read_word_at_a_time() function.
Sometimes we know that it's safe to do potentially out-of-bounds access because we know it won't cross a page boundary. Still, KASAN will report this as a bug.
Add read_word_at_a_time() function which is supposed to be used in such cases. In read_word_at_a_time() KASAN performs relaxed check - only the first byte of access is validated.
--- include/linux/compiler.h | 8 ++++++++ 1 file changed, 8 insertions(+)
include/linux/compiler.h +8 -0
@@ -243,6 +243,7 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s* required ordering. */ #include <asm/barrier.h> +#include <linux/kasan-checks.h> #define __READ_ONCE (x, check) \ ({ \
@@ -262,6 +263,13 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s*/ #define READ_ONCE_NOCHECK (x) __READ_ONCE(x, 0) +static __no_kasan_or_inline +unsigned long read_word_at_a_time (const void *addr) +{ +kasan_check_read (addr,1 ); +return *(unsigned long *)addr; +} +#define WRITE_ONCE (x, val) \ ({ \ union { typeof(x) __val; char __c[1]; } __u = \
Signature · 1 line
2.13.6
PATCH 3/5 lib/strscpy: Shut up KASAN false-positives in strscpy()
In reply to Andrey Ryabininstrscpy() performs the word-at-a-time optimistic reads. So it may may access the memory past the end of the object, which is perfectly fine since strscpy() doesn't use that (past-the-end) data and makes sure the optimistic read won't cross a page boundary.
Use new read_word_at_a_time() to shut up the KASAN.
Note that this potentially could hide some bugs. In example bellow, stscpy() will copy more than we should (1-3 extra uninitialized bytes):
char dst[8];
char *src;
src = kmalloc(5, GFP_KERNEL);
memset(src, 0xff, 5);
strscpy(dst, src, 8);--- lib/string.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
lib/string.c +1 -1
@@ -203,7 +203,7 @@ ssize_t strscpy(char *dest, const char *src, size_t count)while (max >=sizeof (unsigned long )) {unsigned long c, data; - c = *(unsigned long *)(src+res); + c =read_word_at_a_time (src+res);if (has_zero (c, &data, &constants)) { data =prep_zero_mask (c, data, &constants); data =create_zero_mask (data);
Signature · 1 line
2.13.6
PATCH 4/5 fs/dcache: Use read_word_at_a_time() in dentry_string_cmp()
In reply to Andrey Ryabinindentry_string_cmp() performs the word-at-a-time reads from 'cs' and may read slightly more than it was requested in kmallac(). Normally this would make KASAN to report out-of-bounds access, but this was workarounded by commit df4c0e36f1b1 ("fs: dcache: manually unpoison dname after allocation to shut up kasan's reports").
This workaround is not perfect, since it allows out-of-bounds access to dentry's name for all the code, not just in dentry_string_cmp().
So it would be better to use read_word_at_a_time() instead and revert commit df4c0e36f1b1.
--- fs/dcache.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
fs/dcache.c +1 -1
@@ -193,7 +193,7 @@ static inline int dentry_string_cmp(const unsigned char *cs, const unsigned charunsigned long a,b,mask;for (;;) { - a = *(unsigned long *)cs; + a =read_word_at_a_time (cs); b =load_unaligned_zeropad (ct);if (tcount <sizeof (unsigned long ))break ;
Signature · 1 line
2.13.6
PATCH 5/5 fs: dcache: Revert "manually unpoison dname after allocation to shut up kasan's reports"
In reply to Andrey RyabininThis reverts commit df4c0e36f1b1782b0611a77c52cc240e5c4752dd. It's no longer needed since dentry_string_cmp() now uses read_word_at_a_time() to avoid kasan's reports.
--- fs/dcache.c | 5 ----- 1 file changed, 5 deletions(-)
fs/dcache.c +0 -5
@@ -37,8 +37,6 @@#include <linux/prefetch.h> #include <linux/ratelimit.h> #include <linux/list_lru.h> -#include <linux/kasan.h> -#include "internal.h" #include "mount.h"
@@ -1629,9 +1627,6 @@ struct dentry *__d_alloc(struct super_block *sb, const struct qstr *name) }atomic_set (&p->u.count,1 ); dname = p->name; -if (IS_ENABLED (CONFIG_DCACHE_WORD_ACCESS )) -kasan_unpoison_shadow (dname, -round_up (name->len +1 ,sizeof (unsigned long ))); }else { dname = dentry->d_iname; }
Signature · 1 line
2.13.6
Ack for the whole series.
Will this go through Andrew's mm tree or what? I can pick it up directly too.
Linus
Quoted 5 lines · Andrey Ryabinin
Instead of having two identical
__read_once_size_nocheck()functions with different attributes, consolidate all the difference in new macro__no_kasan_or_inlineand use it. No functional changes.Signed-off-by Andrey Ryabinin <[email protected]>
Quoted 4 lines · Linus Torvalds
Ack for the whole series.
Will this go through Andrew's mm tree or what? I can pick it up directly too.
Whatever you'd prefer.
Quoted 5 lines · Andrey Ryabinin
Quoted 2 lines · Linus Torvalds
Will this go through Andrew's mm tree or what? I can pick it up directly too.
Whatever you'd prefer.
Ok, I just took it directly as-is,
Linus