public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Noah Goldstein <goldstein.w.n@gmail.com>
To: "H.J. Lu" <hjl.tools@gmail.com>
Cc: libc-alpha@sourceware.org, carlos@systemhalted.org
Subject: Re: [PATCH v1] x86: Fix strncat-avx2.S when `src` has no null-term [BZ #30065]
Date: Tue, 31 Jan 2023 16:37:05 -0600	[thread overview]
Message-ID: <CAFUsyfLezDNp=CMWd7pP=oP9CCutgvyHS4c76rqYtQbX_cD2wQ@mail.gmail.com> (raw)
In-Reply-To: <CAMe9rOoGFx6FwTkh5113=tJX+2S5EJyz+chsAu0tX4jv4h7B1g@mail.gmail.com>

On Tue, Jan 31, 2023 at 4:29 PM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Tue, Jan 31, 2023 at 1:37 PM Noah Goldstein <goldstein.w.n@gmail.com> wrote:
> >
> > Two issue:
> >
> > 1) Zero-length check is doing:
> > ```
> >     test    %rdx, %rdx
> >     jl      L(zero_len)
> > ```
> > which doesn't actually check zero (was at some point `decq` and the
> > flag never got updated).
> >
> > The fix is just make the flag `jle` i.e:
> > ```
> >     test    %rdx, %rdx
> >     jle     L(zero_len)
> > ```
> >
> > 2) Length check in page-cross case checking if we should continue is
> > doing:
> > ```
> >     cmpq    %r8, %rdx
> >     jb      L(page_cross_small)
> > ```
> > which means we will continue searching for null-term if length ends at
> > the end of a page and there was no null-term in `src`.
>
> It is not purely about null-term.   strncat shouldn't read beyond the limit.
> In this case, src may point to PROT_NONE memory.
>

If there was a null-term the later check:
```
shll $CHAR_SIZE, %ecx
jz L(page_cross_continue)
```
would catch it, this can only occur if there is no null-term.
> > The fix is to make the flag:
> > ```
> >     cmpq    %r8, %rdx
> >     jbe     L(page_cross_small)
> > ```
> > ---
> >  string/test-strncat.c                   | 25 ++++++++++++++++++++++++-
> >  sysdeps/x86_64/multiarch/strncat-avx2.S |  4 ++--
> >  2 files changed, 26 insertions(+), 3 deletions(-)
> >
> > diff --git a/string/test-strncat.c b/string/test-strncat.c
> > index e03d329e1c..c0cde206ee 100644
> > --- a/string/test-strncat.c
> > +++ b/string/test-strncat.c
> > @@ -28,6 +28,7 @@
> >  # define CHAR char
> >  # define UCHAR unsigned char
> >  # define SIMPLE_STRNCAT simple_strncat
> > +# define STRNLEN strnlen
> >  # define STRLEN strlen
> >  # define MEMSET memset
> >  # define MEMCPY memcpy
> > @@ -40,6 +41,7 @@
> >  # define CHAR wchar_t
> >  # define UCHAR wchar_t
> >  # define SIMPLE_STRNCAT simple_wcsncat
> > +# define STRNLEN wcsnlen
> >  # define STRLEN wcslen
> >  # define MEMSET wmemset
> >  # define MEMCPY wmemcpy
> > @@ -78,7 +80,7 @@ do_one_test (impl_t *impl, CHAR *dst, const CHAR *src, size_t n)
> >        return;
> >      }
> >
> > -  size_t len = STRLEN (src);
> > +  size_t len = STRNLEN (src, n);
> >    if (MEMCMP (dst + k, src, len + 1 > n ? n : len + 1) != 0)
>
> >      {
> >        error (0, 0, "Incorrect concatenation in function %s",
> > @@ -95,6 +97,26 @@ do_one_test (impl_t *impl, CHAR *dst, const CHAR *src, size_t n)
> >      }
> >  }
> >
> > +static void
> > +do_test_src_no_nullterm_bz30065 (void)
> > +{
> > +  /* NB: "src does not need to be null-terminated if it contains n or more
> > +   * bytes." */
> > +  CHAR *s1, *s2;
> > +  size_t bound = page_size / sizeof (CHAR);
> > +  s1 = (CHAR *) (buf1 + BUF1PAGES * page_size);
> > +  s2 = (CHAR *) buf2;
> > +  MEMSET (s1 - bound, -1, bound);
> > +  for (size_t n = 0; n < bound; ++n)
> > +    {
> > +      FOR_EACH_IMPL (impl, 0)
> > +       {
> > +         s2[0] = '\0';
> > +         do_one_test (impl, s2, s1 - n, n);
> > +       }
> > +    }
> > +}
> > +
> >  static void
> >  do_test (size_t align1, size_t align2, size_t len1, size_t len2,
> >          size_t n, int max_char)
> > @@ -372,6 +394,7 @@ test_main (void)
> >
> >    do_random_tests ();
> >    do_overflow_tests ();
> > +  do_test_src_no_nullterm_bz30065 ();
> >    return ret;
> >  }
> >
> > diff --git a/sysdeps/x86_64/multiarch/strncat-avx2.S b/sysdeps/x86_64/multiarch/strncat-avx2.S
> > index b380e8e11c..c2ff202238 100644
> > --- a/sysdeps/x86_64/multiarch/strncat-avx2.S
> > +++ b/sysdeps/x86_64/multiarch/strncat-avx2.S
> > @@ -66,7 +66,7 @@ ENTRY(STRNCAT)
> >         salq    $2, %rdx
> >  # else
> >         test    %rdx, %rdx
> > -       jl      L(zero_len)
> > +       jle     L(zero_len)
> >  # endif
> >         vpxor   %VZERO_128, %VZERO_128, %VZERO_128
> >
> > @@ -387,7 +387,7 @@ L(page_cross):
> >         subl    %esi, %r8d
> >         andl    $(VEC_SIZE - 1), %r8d
> >         cmpq    %r8, %rdx
> > -       jb      L(page_cross_small)
> > +       jbe     L(page_cross_small)
> >
> >         /* Optimizing more aggressively for space as this is very cold
> >            code. This saves 2x cache lines.  */
> > --
> > 2.34.1
> >
>
>
> --
> H.J.

  reply	other threads:[~2023-01-31 22:37 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-31 21:36 Noah Goldstein
2023-01-31 22:28 ` H.J. Lu
2023-01-31 22:37   ` Noah Goldstein [this message]
2023-01-31 22:42     ` H.J. Lu
2023-01-31 22:52       ` Noah Goldstein
2023-01-31 23:01         ` H.J. Lu
2023-01-31 23:09           ` Noah Goldstein
2023-01-31 23:40             ` H.J. Lu
2023-01-31 23:47               ` Noah Goldstein
2023-01-31 23:46 ` [PATCH v2] x86: Fix strncat-avx2.S reading past length " Noah Goldstein
2023-02-01  0:23   ` H.J. Lu
2023-02-01  3:10   ` Carlos O'Donell
2023-02-12  1:00     ` Sunil Pandey
2023-02-12  1:36       ` H.J. Lu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAFUsyfLezDNp=CMWd7pP=oP9CCutgvyHS4c76rqYtQbX_cD2wQ@mail.gmail.com' \
    --to=goldstein.w.n@gmail.com \
    --cc=carlos@systemhalted.org \
    --cc=hjl.tools@gmail.com \
    --cc=libc-alpha@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).