* [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591]
@ 2022-09-21 0:58 Noah Goldstein via Libc-alpha
2022-09-21 22:01 ` H.J. Lu via Libc-alpha
0 siblings, 1 reply; 6+ messages in thread
From: Noah Goldstein via Libc-alpha @ 2022-09-21 0:58 UTC (permalink / raw)
To: libc-alpha
Previous implementation was adjusting length (rsi) to match
bytes (eax), but since there is no bound to length this can cause
overflow.
Fix is to just convert the byte-count (eax) to length by dividing by
sizeof (wchar_t) before the comparison.
Full check passes on x86-64 and build succeeds w/ and w/o multiarch.
---
string/test-strnlen.c | 70 +++++++++++++++-----------
sysdeps/x86_64/multiarch/strlen-avx2.S | 7 +--
2 files changed, 43 insertions(+), 34 deletions(-)
diff --git a/string/test-strnlen.c b/string/test-strnlen.c
index 4a9375112a..5cbaf4b734 100644
--- a/string/test-strnlen.c
+++ b/string/test-strnlen.c
@@ -73,7 +73,7 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
{
size_t i;
- align &= 63;
+ align &= (getpagesize () / sizeof (CHAR) - 1);
if ((align + len) * sizeof (CHAR) >= page_size)
return;
@@ -90,38 +90,50 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
static void
do_overflow_tests (void)
{
- size_t i, j, len;
+ size_t i, j, al_idx, repeats, len;
const size_t one = 1;
uintptr_t buf_addr = (uintptr_t) buf1;
+ const size_t alignments[] = { 0, 1, 7, 9, 31, 33, 63, 65, 95, 97, 127, 129 };
- for (i = 0; i < 750; ++i)
+ for (al_idx = 0; al_idx < sizeof (alignments) / sizeof (alignments[0]);
+ al_idx++)
{
- do_test (1, i, SIZE_MAX, BIG_CHAR);
-
- do_test (0, i, SIZE_MAX - i, BIG_CHAR);
- do_test (0, i, i - buf_addr, BIG_CHAR);
- do_test (0, i, -buf_addr - i, BIG_CHAR);
- do_test (0, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
- do_test (0, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
-
- len = 0;
- for (j = 8 * sizeof(size_t) - 1; j ; --j)
- {
- len |= one << j;
- do_test (0, i, len - i, BIG_CHAR);
- do_test (0, i, len + i, BIG_CHAR);
- do_test (0, i, len - buf_addr - i, BIG_CHAR);
- do_test (0, i, len - buf_addr + i, BIG_CHAR);
-
- do_test (0, i, ~len - i, BIG_CHAR);
- do_test (0, i, ~len + i, BIG_CHAR);
- do_test (0, i, ~len - buf_addr - i, BIG_CHAR);
- do_test (0, i, ~len - buf_addr + i, BIG_CHAR);
-
- do_test (0, i, -buf_addr, BIG_CHAR);
- do_test (0, i, j - buf_addr, BIG_CHAR);
- do_test (0, i, -buf_addr - j, BIG_CHAR);
- }
+ for (repeats = 0; repeats < 2; ++repeats)
+ {
+ size_t align = repeats ? (getpagesize () - alignments[al_idx])
+ : alignments[al_idx];
+ align /= sizeof (CHAR);
+ for (i = 0; i < 750; ++i)
+ {
+ do_test (align, i, SIZE_MAX, BIG_CHAR);
+
+ do_test (align, i, SIZE_MAX - i, BIG_CHAR);
+ do_test (align, i, i - buf_addr, BIG_CHAR);
+ do_test (align, i, -buf_addr - i, BIG_CHAR);
+ do_test (align, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
+ do_test (align, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
+
+ len = 0;
+ for (j = 8 * sizeof (size_t) - 1; j; --j)
+ {
+ len |= one << j;
+ do_test (align, i, len, BIG_CHAR);
+ do_test (align, i, len - i, BIG_CHAR);
+ do_test (align, i, len + i, BIG_CHAR);
+ do_test (align, i, len - buf_addr - i, BIG_CHAR);
+ do_test (align, i, len - buf_addr + i, BIG_CHAR);
+
+ do_test (align, i, ~len - i, BIG_CHAR);
+ do_test (align, i, ~len + i, BIG_CHAR);
+ do_test (align, i, ~len - buf_addr - i, BIG_CHAR);
+ do_test (align, i, ~len - buf_addr + i, BIG_CHAR);
+
+ do_test (align, i, -buf_addr, BIG_CHAR);
+ do_test (align, i, j - buf_addr, BIG_CHAR);
+ do_test (align, i, -buf_addr - j, BIG_CHAR);
+ }
+ }
+ }
}
}
diff --git a/sysdeps/x86_64/multiarch/strlen-avx2.S b/sysdeps/x86_64/multiarch/strlen-avx2.S
index 0593fb303b..b9b58ef599 100644
--- a/sysdeps/x86_64/multiarch/strlen-avx2.S
+++ b/sysdeps/x86_64/multiarch/strlen-avx2.S
@@ -544,14 +544,11 @@ L(return_vzeroupper):
L(cross_page_less_vec):
tzcntl %eax, %eax
# ifdef USE_AS_WCSLEN
- /* NB: Multiply length by 4 to get byte count. */
- sall $2, %esi
+ /* NB: Divide by 4 to convert from byte-count to length. */
+ shrl $2, %eax
# endif
cmpq %rax, %rsi
cmovb %esi, %eax
-# ifdef USE_AS_WCSLEN
- shrl $2, %eax
-# endif
VZEROUPPER_RETURN
# endif
--
2.34.1
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591]
2022-09-21 0:58 [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591] Noah Goldstein via Libc-alpha
@ 2022-09-21 22:01 ` H.J. Lu via Libc-alpha
2022-11-23 22:20 ` Sunil Pandey via Libc-alpha
0 siblings, 1 reply; 6+ messages in thread
From: H.J. Lu via Libc-alpha @ 2022-09-21 22:01 UTC (permalink / raw)
To: Noah Goldstein; +Cc: libc-alpha
On Tue, Sep 20, 2022 at 5:58 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
>
> Previous implementation was adjusting length (rsi) to match
> bytes (eax), but since there is no bound to length this can cause
> overflow.
>
> Fix is to just convert the byte-count (eax) to length by dividing by
> sizeof (wchar_t) before the comparison.
>
> Full check passes on x86-64 and build succeeds w/ and w/o multiarch.
> ---
> string/test-strnlen.c | 70 +++++++++++++++-----------
> sysdeps/x86_64/multiarch/strlen-avx2.S | 7 +--
> 2 files changed, 43 insertions(+), 34 deletions(-)
>
> diff --git a/string/test-strnlen.c b/string/test-strnlen.c
> index 4a9375112a..5cbaf4b734 100644
> --- a/string/test-strnlen.c
> +++ b/string/test-strnlen.c
> @@ -73,7 +73,7 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> {
> size_t i;
>
> - align &= 63;
> + align &= (getpagesize () / sizeof (CHAR) - 1);
> if ((align + len) * sizeof (CHAR) >= page_size)
> return;
>
> @@ -90,38 +90,50 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> static void
> do_overflow_tests (void)
> {
> - size_t i, j, len;
> + size_t i, j, al_idx, repeats, len;
> const size_t one = 1;
> uintptr_t buf_addr = (uintptr_t) buf1;
> + const size_t alignments[] = { 0, 1, 7, 9, 31, 33, 63, 65, 95, 97, 127, 129 };
>
> - for (i = 0; i < 750; ++i)
> + for (al_idx = 0; al_idx < sizeof (alignments) / sizeof (alignments[0]);
> + al_idx++)
> {
> - do_test (1, i, SIZE_MAX, BIG_CHAR);
> -
> - do_test (0, i, SIZE_MAX - i, BIG_CHAR);
> - do_test (0, i, i - buf_addr, BIG_CHAR);
> - do_test (0, i, -buf_addr - i, BIG_CHAR);
> - do_test (0, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> - do_test (0, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> -
> - len = 0;
> - for (j = 8 * sizeof(size_t) - 1; j ; --j)
> - {
> - len |= one << j;
> - do_test (0, i, len - i, BIG_CHAR);
> - do_test (0, i, len + i, BIG_CHAR);
> - do_test (0, i, len - buf_addr - i, BIG_CHAR);
> - do_test (0, i, len - buf_addr + i, BIG_CHAR);
> -
> - do_test (0, i, ~len - i, BIG_CHAR);
> - do_test (0, i, ~len + i, BIG_CHAR);
> - do_test (0, i, ~len - buf_addr - i, BIG_CHAR);
> - do_test (0, i, ~len - buf_addr + i, BIG_CHAR);
> -
> - do_test (0, i, -buf_addr, BIG_CHAR);
> - do_test (0, i, j - buf_addr, BIG_CHAR);
> - do_test (0, i, -buf_addr - j, BIG_CHAR);
> - }
> + for (repeats = 0; repeats < 2; ++repeats)
> + {
> + size_t align = repeats ? (getpagesize () - alignments[al_idx])
> + : alignments[al_idx];
> + align /= sizeof (CHAR);
> + for (i = 0; i < 750; ++i)
> + {
> + do_test (align, i, SIZE_MAX, BIG_CHAR);
> +
> + do_test (align, i, SIZE_MAX - i, BIG_CHAR);
> + do_test (align, i, i - buf_addr, BIG_CHAR);
> + do_test (align, i, -buf_addr - i, BIG_CHAR);
> + do_test (align, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> + do_test (align, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> +
> + len = 0;
> + for (j = 8 * sizeof (size_t) - 1; j; --j)
> + {
> + len |= one << j;
> + do_test (align, i, len, BIG_CHAR);
> + do_test (align, i, len - i, BIG_CHAR);
> + do_test (align, i, len + i, BIG_CHAR);
> + do_test (align, i, len - buf_addr - i, BIG_CHAR);
> + do_test (align, i, len - buf_addr + i, BIG_CHAR);
> +
> + do_test (align, i, ~len - i, BIG_CHAR);
> + do_test (align, i, ~len + i, BIG_CHAR);
> + do_test (align, i, ~len - buf_addr - i, BIG_CHAR);
> + do_test (align, i, ~len - buf_addr + i, BIG_CHAR);
> +
> + do_test (align, i, -buf_addr, BIG_CHAR);
> + do_test (align, i, j - buf_addr, BIG_CHAR);
> + do_test (align, i, -buf_addr - j, BIG_CHAR);
> + }
> + }
> + }
> }
> }
>
> diff --git a/sysdeps/x86_64/multiarch/strlen-avx2.S b/sysdeps/x86_64/multiarch/strlen-avx2.S
> index 0593fb303b..b9b58ef599 100644
> --- a/sysdeps/x86_64/multiarch/strlen-avx2.S
> +++ b/sysdeps/x86_64/multiarch/strlen-avx2.S
> @@ -544,14 +544,11 @@ L(return_vzeroupper):
> L(cross_page_less_vec):
> tzcntl %eax, %eax
> # ifdef USE_AS_WCSLEN
> - /* NB: Multiply length by 4 to get byte count. */
> - sall $2, %esi
> + /* NB: Divide by 4 to convert from byte-count to length. */
> + shrl $2, %eax
> # endif
> cmpq %rax, %rsi
> cmovb %esi, %eax
> -# ifdef USE_AS_WCSLEN
> - shrl $2, %eax
> -# endif
> VZEROUPPER_RETURN
> # endif
>
> --
> 2.34.1
>
LGTM.
Thanks.
--
H.J.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591]
2022-09-21 22:01 ` H.J. Lu via Libc-alpha
@ 2022-11-23 22:20 ` Sunil Pandey via Libc-alpha
2022-11-24 0:22 ` H.J. Lu via Libc-alpha
0 siblings, 1 reply; 6+ messages in thread
From: Sunil Pandey via Libc-alpha @ 2022-11-23 22:20 UTC (permalink / raw)
To: H.J. Lu, Libc-stable Mailing List; +Cc: Noah Goldstein, libc-alpha
On Wed, Sep 21, 2022 at 3:02 PM H.J. Lu via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> On Tue, Sep 20, 2022 at 5:58 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
> >
> > Previous implementation was adjusting length (rsi) to match
> > bytes (eax), but since there is no bound to length this can cause
> > overflow.
> >
> > Fix is to just convert the byte-count (eax) to length by dividing by
> > sizeof (wchar_t) before the comparison.
> >
> > Full check passes on x86-64 and build succeeds w/ and w/o multiarch.
> > ---
> > string/test-strnlen.c | 70 +++++++++++++++-----------
> > sysdeps/x86_64/multiarch/strlen-avx2.S | 7 +--
> > 2 files changed, 43 insertions(+), 34 deletions(-)
> >
> > diff --git a/string/test-strnlen.c b/string/test-strnlen.c
> > index 4a9375112a..5cbaf4b734 100644
> > --- a/string/test-strnlen.c
> > +++ b/string/test-strnlen.c
> > @@ -73,7 +73,7 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > {
> > size_t i;
> >
> > - align &= 63;
> > + align &= (getpagesize () / sizeof (CHAR) - 1);
> > if ((align + len) * sizeof (CHAR) >= page_size)
> > return;
> >
> > @@ -90,38 +90,50 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > static void
> > do_overflow_tests (void)
> > {
> > - size_t i, j, len;
> > + size_t i, j, al_idx, repeats, len;
> > const size_t one = 1;
> > uintptr_t buf_addr = (uintptr_t) buf1;
> > + const size_t alignments[] = { 0, 1, 7, 9, 31, 33, 63, 65, 95, 97, 127, 129 };
> >
> > - for (i = 0; i < 750; ++i)
> > + for (al_idx = 0; al_idx < sizeof (alignments) / sizeof (alignments[0]);
> > + al_idx++)
> > {
> > - do_test (1, i, SIZE_MAX, BIG_CHAR);
> > -
> > - do_test (0, i, SIZE_MAX - i, BIG_CHAR);
> > - do_test (0, i, i - buf_addr, BIG_CHAR);
> > - do_test (0, i, -buf_addr - i, BIG_CHAR);
> > - do_test (0, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > - do_test (0, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > -
> > - len = 0;
> > - for (j = 8 * sizeof(size_t) - 1; j ; --j)
> > - {
> > - len |= one << j;
> > - do_test (0, i, len - i, BIG_CHAR);
> > - do_test (0, i, len + i, BIG_CHAR);
> > - do_test (0, i, len - buf_addr - i, BIG_CHAR);
> > - do_test (0, i, len - buf_addr + i, BIG_CHAR);
> > -
> > - do_test (0, i, ~len - i, BIG_CHAR);
> > - do_test (0, i, ~len + i, BIG_CHAR);
> > - do_test (0, i, ~len - buf_addr - i, BIG_CHAR);
> > - do_test (0, i, ~len - buf_addr + i, BIG_CHAR);
> > -
> > - do_test (0, i, -buf_addr, BIG_CHAR);
> > - do_test (0, i, j - buf_addr, BIG_CHAR);
> > - do_test (0, i, -buf_addr - j, BIG_CHAR);
> > - }
> > + for (repeats = 0; repeats < 2; ++repeats)
> > + {
> > + size_t align = repeats ? (getpagesize () - alignments[al_idx])
> > + : alignments[al_idx];
> > + align /= sizeof (CHAR);
> > + for (i = 0; i < 750; ++i)
> > + {
> > + do_test (align, i, SIZE_MAX, BIG_CHAR);
> > +
> > + do_test (align, i, SIZE_MAX - i, BIG_CHAR);
> > + do_test (align, i, i - buf_addr, BIG_CHAR);
> > + do_test (align, i, -buf_addr - i, BIG_CHAR);
> > + do_test (align, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > + do_test (align, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > +
> > + len = 0;
> > + for (j = 8 * sizeof (size_t) - 1; j; --j)
> > + {
> > + len |= one << j;
> > + do_test (align, i, len, BIG_CHAR);
> > + do_test (align, i, len - i, BIG_CHAR);
> > + do_test (align, i, len + i, BIG_CHAR);
> > + do_test (align, i, len - buf_addr - i, BIG_CHAR);
> > + do_test (align, i, len - buf_addr + i, BIG_CHAR);
> > +
> > + do_test (align, i, ~len - i, BIG_CHAR);
> > + do_test (align, i, ~len + i, BIG_CHAR);
> > + do_test (align, i, ~len - buf_addr - i, BIG_CHAR);
> > + do_test (align, i, ~len - buf_addr + i, BIG_CHAR);
> > +
> > + do_test (align, i, -buf_addr, BIG_CHAR);
> > + do_test (align, i, j - buf_addr, BIG_CHAR);
> > + do_test (align, i, -buf_addr - j, BIG_CHAR);
> > + }
> > + }
> > + }
> > }
> > }
> >
> > diff --git a/sysdeps/x86_64/multiarch/strlen-avx2.S b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > index 0593fb303b..b9b58ef599 100644
> > --- a/sysdeps/x86_64/multiarch/strlen-avx2.S
> > +++ b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > @@ -544,14 +544,11 @@ L(return_vzeroupper):
> > L(cross_page_less_vec):
> > tzcntl %eax, %eax
> > # ifdef USE_AS_WCSLEN
> > - /* NB: Multiply length by 4 to get byte count. */
> > - sall $2, %esi
> > + /* NB: Divide by 4 to convert from byte-count to length. */
> > + shrl $2, %eax
> > # endif
> > cmpq %rax, %rsi
> > cmovb %esi, %eax
> > -# ifdef USE_AS_WCSLEN
> > - shrl $2, %eax
> > -# endif
> > VZEROUPPER_RETURN
> > # endif
> >
> > --
> > 2.34.1
> >
>
> LGTM.
>
> Thanks.
>
> --
> H.J.
I would like to backport this patch to affected release branches from
2.36 to 2.33.
Any comments/suggestions or objections on this.
--Sunil
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591]
2022-11-23 22:20 ` Sunil Pandey via Libc-alpha
@ 2022-11-24 0:22 ` H.J. Lu via Libc-alpha
2022-11-24 3:04 ` Sunil Pandey via Libc-alpha
0 siblings, 1 reply; 6+ messages in thread
From: H.J. Lu via Libc-alpha @ 2022-11-24 0:22 UTC (permalink / raw)
To: Sunil Pandey; +Cc: Libc-stable Mailing List, Noah Goldstein, libc-alpha
On Wed, Nov 23, 2022 at 2:21 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
>
> On Wed, Sep 21, 2022 at 3:02 PM H.J. Lu via Libc-alpha
> <libc-alpha@sourceware.org> wrote:
> >
> > On Tue, Sep 20, 2022 at 5:58 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
> > >
> > > Previous implementation was adjusting length (rsi) to match
> > > bytes (eax), but since there is no bound to length this can cause
> > > overflow.
> > >
> > > Fix is to just convert the byte-count (eax) to length by dividing by
> > > sizeof (wchar_t) before the comparison.
> > >
> > > Full check passes on x86-64 and build succeeds w/ and w/o multiarch.
> > > ---
> > > string/test-strnlen.c | 70 +++++++++++++++-----------
> > > sysdeps/x86_64/multiarch/strlen-avx2.S | 7 +--
> > > 2 files changed, 43 insertions(+), 34 deletions(-)
> > >
> > > diff --git a/string/test-strnlen.c b/string/test-strnlen.c
> > > index 4a9375112a..5cbaf4b734 100644
> > > --- a/string/test-strnlen.c
> > > +++ b/string/test-strnlen.c
> > > @@ -73,7 +73,7 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > > {
> > > size_t i;
> > >
> > > - align &= 63;
> > > + align &= (getpagesize () / sizeof (CHAR) - 1);
> > > if ((align + len) * sizeof (CHAR) >= page_size)
> > > return;
> > >
> > > @@ -90,38 +90,50 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > > static void
> > > do_overflow_tests (void)
> > > {
> > > - size_t i, j, len;
> > > + size_t i, j, al_idx, repeats, len;
> > > const size_t one = 1;
> > > uintptr_t buf_addr = (uintptr_t) buf1;
> > > + const size_t alignments[] = { 0, 1, 7, 9, 31, 33, 63, 65, 95, 97, 127, 129 };
> > >
> > > - for (i = 0; i < 750; ++i)
> > > + for (al_idx = 0; al_idx < sizeof (alignments) / sizeof (alignments[0]);
> > > + al_idx++)
> > > {
> > > - do_test (1, i, SIZE_MAX, BIG_CHAR);
> > > -
> > > - do_test (0, i, SIZE_MAX - i, BIG_CHAR);
> > > - do_test (0, i, i - buf_addr, BIG_CHAR);
> > > - do_test (0, i, -buf_addr - i, BIG_CHAR);
> > > - do_test (0, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > > - do_test (0, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > > -
> > > - len = 0;
> > > - for (j = 8 * sizeof(size_t) - 1; j ; --j)
> > > - {
> > > - len |= one << j;
> > > - do_test (0, i, len - i, BIG_CHAR);
> > > - do_test (0, i, len + i, BIG_CHAR);
> > > - do_test (0, i, len - buf_addr - i, BIG_CHAR);
> > > - do_test (0, i, len - buf_addr + i, BIG_CHAR);
> > > -
> > > - do_test (0, i, ~len - i, BIG_CHAR);
> > > - do_test (0, i, ~len + i, BIG_CHAR);
> > > - do_test (0, i, ~len - buf_addr - i, BIG_CHAR);
> > > - do_test (0, i, ~len - buf_addr + i, BIG_CHAR);
> > > -
> > > - do_test (0, i, -buf_addr, BIG_CHAR);
> > > - do_test (0, i, j - buf_addr, BIG_CHAR);
> > > - do_test (0, i, -buf_addr - j, BIG_CHAR);
> > > - }
> > > + for (repeats = 0; repeats < 2; ++repeats)
> > > + {
> > > + size_t align = repeats ? (getpagesize () - alignments[al_idx])
> > > + : alignments[al_idx];
> > > + align /= sizeof (CHAR);
> > > + for (i = 0; i < 750; ++i)
> > > + {
> > > + do_test (align, i, SIZE_MAX, BIG_CHAR);
> > > +
> > > + do_test (align, i, SIZE_MAX - i, BIG_CHAR);
> > > + do_test (align, i, i - buf_addr, BIG_CHAR);
> > > + do_test (align, i, -buf_addr - i, BIG_CHAR);
> > > + do_test (align, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > > + do_test (align, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > > +
> > > + len = 0;
> > > + for (j = 8 * sizeof (size_t) - 1; j; --j)
> > > + {
> > > + len |= one << j;
> > > + do_test (align, i, len, BIG_CHAR);
> > > + do_test (align, i, len - i, BIG_CHAR);
> > > + do_test (align, i, len + i, BIG_CHAR);
> > > + do_test (align, i, len - buf_addr - i, BIG_CHAR);
> > > + do_test (align, i, len - buf_addr + i, BIG_CHAR);
> > > +
> > > + do_test (align, i, ~len - i, BIG_CHAR);
> > > + do_test (align, i, ~len + i, BIG_CHAR);
> > > + do_test (align, i, ~len - buf_addr - i, BIG_CHAR);
> > > + do_test (align, i, ~len - buf_addr + i, BIG_CHAR);
> > > +
> > > + do_test (align, i, -buf_addr, BIG_CHAR);
> > > + do_test (align, i, j - buf_addr, BIG_CHAR);
> > > + do_test (align, i, -buf_addr - j, BIG_CHAR);
> > > + }
> > > + }
> > > + }
> > > }
> > > }
> > >
> > > diff --git a/sysdeps/x86_64/multiarch/strlen-avx2.S b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > index 0593fb303b..b9b58ef599 100644
> > > --- a/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > +++ b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > @@ -544,14 +544,11 @@ L(return_vzeroupper):
> > > L(cross_page_less_vec):
> > > tzcntl %eax, %eax
> > > # ifdef USE_AS_WCSLEN
> > > - /* NB: Multiply length by 4 to get byte count. */
> > > - sall $2, %esi
> > > + /* NB: Divide by 4 to convert from byte-count to length. */
> > > + shrl $2, %eax
> > > # endif
> > > cmpq %rax, %rsi
> > > cmovb %esi, %eax
> > > -# ifdef USE_AS_WCSLEN
> > > - shrl $2, %eax
> > > -# endif
> > > VZEROUPPER_RETURN
> > > # endif
> > >
> > > --
> > > 2.34.1
> > >
> >
> > LGTM.
> >
> > Thanks.
> >
> > --
> > H.J.
>
> I would like to backport this patch to affected release branches from
> 2.36 to 2.33.
>
> Any comments/suggestions or objections on this.
>
OK.
Thanks.
--
H.J.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591]
2022-11-24 0:22 ` H.J. Lu via Libc-alpha
@ 2022-11-24 3:04 ` Sunil Pandey via Libc-alpha
2022-11-24 3:12 ` H.J. Lu via Libc-alpha
0 siblings, 1 reply; 6+ messages in thread
From: Sunil Pandey via Libc-alpha @ 2022-11-24 3:04 UTC (permalink / raw)
To: H.J. Lu; +Cc: Libc-stable Mailing List, Noah Goldstein, libc-alpha
On Wed, Nov 23, 2022 at 4:23 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Wed, Nov 23, 2022 at 2:21 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
> >
> > On Wed, Sep 21, 2022 at 3:02 PM H.J. Lu via Libc-alpha
> > <libc-alpha@sourceware.org> wrote:
> > >
> > > On Tue, Sep 20, 2022 at 5:58 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
> > > >
> > > > Previous implementation was adjusting length (rsi) to match
> > > > bytes (eax), but since there is no bound to length this can cause
> > > > overflow.
> > > >
> > > > Fix is to just convert the byte-count (eax) to length by dividing by
> > > > sizeof (wchar_t) before the comparison.
> > > >
> > > > Full check passes on x86-64 and build succeeds w/ and w/o multiarch.
> > > > ---
> > > > string/test-strnlen.c | 70 +++++++++++++++-----------
> > > > sysdeps/x86_64/multiarch/strlen-avx2.S | 7 +--
> > > > 2 files changed, 43 insertions(+), 34 deletions(-)
> > > >
> > > > diff --git a/string/test-strnlen.c b/string/test-strnlen.c
> > > > index 4a9375112a..5cbaf4b734 100644
> > > > --- a/string/test-strnlen.c
> > > > +++ b/string/test-strnlen.c
> > > > @@ -73,7 +73,7 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > > > {
> > > > size_t i;
> > > >
> > > > - align &= 63;
> > > > + align &= (getpagesize () / sizeof (CHAR) - 1);
> > > > if ((align + len) * sizeof (CHAR) >= page_size)
> > > > return;
> > > >
> > > > @@ -90,38 +90,50 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > > > static void
> > > > do_overflow_tests (void)
> > > > {
> > > > - size_t i, j, len;
> > > > + size_t i, j, al_idx, repeats, len;
> > > > const size_t one = 1;
> > > > uintptr_t buf_addr = (uintptr_t) buf1;
> > > > + const size_t alignments[] = { 0, 1, 7, 9, 31, 33, 63, 65, 95, 97, 127, 129 };
> > > >
> > > > - for (i = 0; i < 750; ++i)
> > > > + for (al_idx = 0; al_idx < sizeof (alignments) / sizeof (alignments[0]);
> > > > + al_idx++)
> > > > {
> > > > - do_test (1, i, SIZE_MAX, BIG_CHAR);
> > > > -
> > > > - do_test (0, i, SIZE_MAX - i, BIG_CHAR);
> > > > - do_test (0, i, i - buf_addr, BIG_CHAR);
> > > > - do_test (0, i, -buf_addr - i, BIG_CHAR);
> > > > - do_test (0, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > > > - do_test (0, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > > > -
> > > > - len = 0;
> > > > - for (j = 8 * sizeof(size_t) - 1; j ; --j)
> > > > - {
> > > > - len |= one << j;
> > > > - do_test (0, i, len - i, BIG_CHAR);
> > > > - do_test (0, i, len + i, BIG_CHAR);
> > > > - do_test (0, i, len - buf_addr - i, BIG_CHAR);
> > > > - do_test (0, i, len - buf_addr + i, BIG_CHAR);
> > > > -
> > > > - do_test (0, i, ~len - i, BIG_CHAR);
> > > > - do_test (0, i, ~len + i, BIG_CHAR);
> > > > - do_test (0, i, ~len - buf_addr - i, BIG_CHAR);
> > > > - do_test (0, i, ~len - buf_addr + i, BIG_CHAR);
> > > > -
> > > > - do_test (0, i, -buf_addr, BIG_CHAR);
> > > > - do_test (0, i, j - buf_addr, BIG_CHAR);
> > > > - do_test (0, i, -buf_addr - j, BIG_CHAR);
> > > > - }
> > > > + for (repeats = 0; repeats < 2; ++repeats)
> > > > + {
> > > > + size_t align = repeats ? (getpagesize () - alignments[al_idx])
> > > > + : alignments[al_idx];
> > > > + align /= sizeof (CHAR);
> > > > + for (i = 0; i < 750; ++i)
> > > > + {
> > > > + do_test (align, i, SIZE_MAX, BIG_CHAR);
> > > > +
> > > > + do_test (align, i, SIZE_MAX - i, BIG_CHAR);
> > > > + do_test (align, i, i - buf_addr, BIG_CHAR);
> > > > + do_test (align, i, -buf_addr - i, BIG_CHAR);
> > > > + do_test (align, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > > > + do_test (align, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > > > +
> > > > + len = 0;
> > > > + for (j = 8 * sizeof (size_t) - 1; j; --j)
> > > > + {
> > > > + len |= one << j;
> > > > + do_test (align, i, len, BIG_CHAR);
> > > > + do_test (align, i, len - i, BIG_CHAR);
> > > > + do_test (align, i, len + i, BIG_CHAR);
> > > > + do_test (align, i, len - buf_addr - i, BIG_CHAR);
> > > > + do_test (align, i, len - buf_addr + i, BIG_CHAR);
> > > > +
> > > > + do_test (align, i, ~len - i, BIG_CHAR);
> > > > + do_test (align, i, ~len + i, BIG_CHAR);
> > > > + do_test (align, i, ~len - buf_addr - i, BIG_CHAR);
> > > > + do_test (align, i, ~len - buf_addr + i, BIG_CHAR);
> > > > +
> > > > + do_test (align, i, -buf_addr, BIG_CHAR);
> > > > + do_test (align, i, j - buf_addr, BIG_CHAR);
> > > > + do_test (align, i, -buf_addr - j, BIG_CHAR);
> > > > + }
> > > > + }
> > > > + }
> > > > }
> > > > }
> > > >
> > > > diff --git a/sysdeps/x86_64/multiarch/strlen-avx2.S b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > > index 0593fb303b..b9b58ef599 100644
> > > > --- a/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > > +++ b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > > @@ -544,14 +544,11 @@ L(return_vzeroupper):
> > > > L(cross_page_less_vec):
> > > > tzcntl %eax, %eax
> > > > # ifdef USE_AS_WCSLEN
> > > > - /* NB: Multiply length by 4 to get byte count. */
> > > > - sall $2, %esi
> > > > + /* NB: Divide by 4 to convert from byte-count to length. */
> > > > + shrl $2, %eax
> > > > # endif
> > > > cmpq %rax, %rsi
> > > > cmovb %esi, %eax
> > > > -# ifdef USE_AS_WCSLEN
> > > > - shrl $2, %eax
> > > > -# endif
> > > > VZEROUPPER_RETURN
> > > > # endif
> > > >
> > > > --
> > > > 2.34.1
> > > >
> > >
> > > LGTM.
> > >
> > > Thanks.
> > >
> > > --
> > > H.J.
> >
> > I would like to backport this patch to affected release branches from
> > 2.36 to 2.33.
> >
> > Any comments/suggestions or objections on this.
> >
>
> OK.
>
> Thanks.
>
>
> --
> H.J.
Just ran testing from 2.32 to 2.26. All of them have this issue.
Ok for 2.32 to 2.26 branches?
--Sunil
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591]
2022-11-24 3:04 ` Sunil Pandey via Libc-alpha
@ 2022-11-24 3:12 ` H.J. Lu via Libc-alpha
0 siblings, 0 replies; 6+ messages in thread
From: H.J. Lu via Libc-alpha @ 2022-11-24 3:12 UTC (permalink / raw)
To: Sunil Pandey; +Cc: Libc-stable Mailing List, Noah Goldstein, libc-alpha
On Wed, Nov 23, 2022 at 7:04 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
>
> On Wed, Nov 23, 2022 at 4:23 PM H.J. Lu <hjl.tools@gmail.com> wrote:
> >
> > On Wed, Nov 23, 2022 at 2:21 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
> > >
> > > On Wed, Sep 21, 2022 at 3:02 PM H.J. Lu via Libc-alpha
> > > <libc-alpha@sourceware.org> wrote:
> > > >
> > > > On Tue, Sep 20, 2022 at 5:58 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
> > > > >
> > > > > Previous implementation was adjusting length (rsi) to match
> > > > > bytes (eax), but since there is no bound to length this can cause
> > > > > overflow.
> > > > >
> > > > > Fix is to just convert the byte-count (eax) to length by dividing by
> > > > > sizeof (wchar_t) before the comparison.
> > > > >
> > > > > Full check passes on x86-64 and build succeeds w/ and w/o multiarch.
> > > > > ---
> > > > > string/test-strnlen.c | 70 +++++++++++++++-----------
> > > > > sysdeps/x86_64/multiarch/strlen-avx2.S | 7 +--
> > > > > 2 files changed, 43 insertions(+), 34 deletions(-)
> > > > >
> > > > > diff --git a/string/test-strnlen.c b/string/test-strnlen.c
> > > > > index 4a9375112a..5cbaf4b734 100644
> > > > > --- a/string/test-strnlen.c
> > > > > +++ b/string/test-strnlen.c
> > > > > @@ -73,7 +73,7 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > > > > {
> > > > > size_t i;
> > > > >
> > > > > - align &= 63;
> > > > > + align &= (getpagesize () / sizeof (CHAR) - 1);
> > > > > if ((align + len) * sizeof (CHAR) >= page_size)
> > > > > return;
> > > > >
> > > > > @@ -90,38 +90,50 @@ do_test (size_t align, size_t len, size_t maxlen, int max_char)
> > > > > static void
> > > > > do_overflow_tests (void)
> > > > > {
> > > > > - size_t i, j, len;
> > > > > + size_t i, j, al_idx, repeats, len;
> > > > > const size_t one = 1;
> > > > > uintptr_t buf_addr = (uintptr_t) buf1;
> > > > > + const size_t alignments[] = { 0, 1, 7, 9, 31, 33, 63, 65, 95, 97, 127, 129 };
> > > > >
> > > > > - for (i = 0; i < 750; ++i)
> > > > > + for (al_idx = 0; al_idx < sizeof (alignments) / sizeof (alignments[0]);
> > > > > + al_idx++)
> > > > > {
> > > > > - do_test (1, i, SIZE_MAX, BIG_CHAR);
> > > > > -
> > > > > - do_test (0, i, SIZE_MAX - i, BIG_CHAR);
> > > > > - do_test (0, i, i - buf_addr, BIG_CHAR);
> > > > > - do_test (0, i, -buf_addr - i, BIG_CHAR);
> > > > > - do_test (0, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > > > > - do_test (0, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > > > > -
> > > > > - len = 0;
> > > > > - for (j = 8 * sizeof(size_t) - 1; j ; --j)
> > > > > - {
> > > > > - len |= one << j;
> > > > > - do_test (0, i, len - i, BIG_CHAR);
> > > > > - do_test (0, i, len + i, BIG_CHAR);
> > > > > - do_test (0, i, len - buf_addr - i, BIG_CHAR);
> > > > > - do_test (0, i, len - buf_addr + i, BIG_CHAR);
> > > > > -
> > > > > - do_test (0, i, ~len - i, BIG_CHAR);
> > > > > - do_test (0, i, ~len + i, BIG_CHAR);
> > > > > - do_test (0, i, ~len - buf_addr - i, BIG_CHAR);
> > > > > - do_test (0, i, ~len - buf_addr + i, BIG_CHAR);
> > > > > -
> > > > > - do_test (0, i, -buf_addr, BIG_CHAR);
> > > > > - do_test (0, i, j - buf_addr, BIG_CHAR);
> > > > > - do_test (0, i, -buf_addr - j, BIG_CHAR);
> > > > > - }
> > > > > + for (repeats = 0; repeats < 2; ++repeats)
> > > > > + {
> > > > > + size_t align = repeats ? (getpagesize () - alignments[al_idx])
> > > > > + : alignments[al_idx];
> > > > > + align /= sizeof (CHAR);
> > > > > + for (i = 0; i < 750; ++i)
> > > > > + {
> > > > > + do_test (align, i, SIZE_MAX, BIG_CHAR);
> > > > > +
> > > > > + do_test (align, i, SIZE_MAX - i, BIG_CHAR);
> > > > > + do_test (align, i, i - buf_addr, BIG_CHAR);
> > > > > + do_test (align, i, -buf_addr - i, BIG_CHAR);
> > > > > + do_test (align, i, SIZE_MAX - buf_addr - i, BIG_CHAR);
> > > > > + do_test (align, i, SIZE_MAX - buf_addr + i, BIG_CHAR);
> > > > > +
> > > > > + len = 0;
> > > > > + for (j = 8 * sizeof (size_t) - 1; j; --j)
> > > > > + {
> > > > > + len |= one << j;
> > > > > + do_test (align, i, len, BIG_CHAR);
> > > > > + do_test (align, i, len - i, BIG_CHAR);
> > > > > + do_test (align, i, len + i, BIG_CHAR);
> > > > > + do_test (align, i, len - buf_addr - i, BIG_CHAR);
> > > > > + do_test (align, i, len - buf_addr + i, BIG_CHAR);
> > > > > +
> > > > > + do_test (align, i, ~len - i, BIG_CHAR);
> > > > > + do_test (align, i, ~len + i, BIG_CHAR);
> > > > > + do_test (align, i, ~len - buf_addr - i, BIG_CHAR);
> > > > > + do_test (align, i, ~len - buf_addr + i, BIG_CHAR);
> > > > > +
> > > > > + do_test (align, i, -buf_addr, BIG_CHAR);
> > > > > + do_test (align, i, j - buf_addr, BIG_CHAR);
> > > > > + do_test (align, i, -buf_addr - j, BIG_CHAR);
> > > > > + }
> > > > > + }
> > > > > + }
> > > > > }
> > > > > }
> > > > >
> > > > > diff --git a/sysdeps/x86_64/multiarch/strlen-avx2.S b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > > > index 0593fb303b..b9b58ef599 100644
> > > > > --- a/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > > > +++ b/sysdeps/x86_64/multiarch/strlen-avx2.S
> > > > > @@ -544,14 +544,11 @@ L(return_vzeroupper):
> > > > > L(cross_page_less_vec):
> > > > > tzcntl %eax, %eax
> > > > > # ifdef USE_AS_WCSLEN
> > > > > - /* NB: Multiply length by 4 to get byte count. */
> > > > > - sall $2, %esi
> > > > > + /* NB: Divide by 4 to convert from byte-count to length. */
> > > > > + shrl $2, %eax
> > > > > # endif
> > > > > cmpq %rax, %rsi
> > > > > cmovb %esi, %eax
> > > > > -# ifdef USE_AS_WCSLEN
> > > > > - shrl $2, %eax
> > > > > -# endif
> > > > > VZEROUPPER_RETURN
> > > > > # endif
> > > > >
> > > > > --
> > > > > 2.34.1
> > > > >
> > > >
> > > > LGTM.
> > > >
> > > > Thanks.
> > > >
> > > > --
> > > > H.J.
> > >
> > > I would like to backport this patch to affected release branches from
> > > 2.36 to 2.33.
> > >
> > > Any comments/suggestions or objections on this.
> > >
> >
> > OK.
> >
> > Thanks.
> >
> >
> > --
> > H.J.
>
> Just ran testing from 2.32 to 2.26. All of them have this issue.
>
> Ok for 2.32 to 2.26 branches?
>
> --Sunil
Let's stop at 2.28.
--
H.J.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2022-11-24 3:13 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-09-21 0:58 [PATCH v1] x86: Fix wcsnlen-avx2 page cross length comparison [BZ #29591] Noah Goldstein via Libc-alpha
2022-09-21 22:01 ` H.J. Lu via Libc-alpha
2022-11-23 22:20 ` Sunil Pandey via Libc-alpha
2022-11-24 0:22 ` H.J. Lu via Libc-alpha
2022-11-24 3:04 ` Sunil Pandey via Libc-alpha
2022-11-24 3:12 ` H.J. Lu via Libc-alpha
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).