From: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
To: libc-alpha@sourceware.org, Joe Simmons-Talbott <josimmon@redhat.com>
Subject: Re: [PATCH v5] vfprintf-internal: Get rid of alloca.
Date: Wed, 19 Jul 2023 15:04:09 -0300 [thread overview]
Message-ID: <f6512ede-2311-5793-d0bf-ad6f77e3a7d8@linaro.org> (raw)
In-Reply-To: <20230620132644.2984023-1-josimmon@redhat.com>
On 20/06/23 10:26, Joe Simmons-Talbott via Libc-alpha wrote:
> Avoid potential stack overflow from unbounded alloca. Use the existing
> scratch_buffer instead.
>
> Add a testcase to exercise the code as suggested by Adhemerval Zanella Netto.
> ---
> Changes to v4:
> * Rebased to latest commits.
>
> stdio-common/tst-vfprintf-user-type.c | 43 +++++++++++++++++++++++++++
> stdio-common/vfprintf-internal.c | 23 +++++++++++++-
> 2 files changed, 65 insertions(+), 1 deletion(-)
>
> diff --git a/stdio-common/tst-vfprintf-user-type.c b/stdio-common/tst-vfprintf-user-type.c
> index 7cc69dc716..bd2ac04351 100644
> --- a/stdio-common/tst-vfprintf-user-type.c
> +++ b/stdio-common/tst-vfprintf-user-type.c
> @@ -194,6 +194,49 @@ do_test (void)
> TEST_COMPARE_STRING (str, "[[(123, 457.000000) (123, 457.000000)]]");
> free (str);
>
> + str = NULL;
> + TEST_VERIFY (asprintf_alias (&str, "%1$P %2$P %3$P %4$P %5$P %6$P"
> + "%7$P %8$P %9$P %10$P %11$P %12$P"
> + "%13$P %14$P %15$P %16$P %17$P %18$P"
> + "%19$P %20$P %21$P %22$P %23$P %24$P"
> + "%25$P %26$P %27$P %28$P", /*%*29$P %30$P",*/
> + 1L, 1.0,
> + 2L, 2.0,
> + 3L, 3.0,
> + 4L, 4.0,
> + 5L, 6.0,
> + 6L, 6.0,
> + 7L, 7.0,
> + 8L, 8.0,
> + 9L, 9.0,
> + 10L, 10.0,
> + 11L, 11.0,
> + 12L, 12.0,
> + 13L, 13.0,
> + 14L, 14.0,
> + 15L, 15.0,
> + 16L, 16.0,
> + 17L, 17.0,
> + 18L, 18.0,
> + 19L, 19.0,
> + 20L, 20.0,
> + 21L, 21.0,
> + 22L, 22.0,
> + 23L, 23.0,
> + 24L, 34.0,
> + 25L, 25.0,
> + 26L, 26.0,
> + 27L, 27.0,
> + 28L, 28.0)
> + >= 0);
> + printf ("str=%s\n", str);
Remove the printf, I added while debugging.
> + free (str);
> +
> + str = NULL;
> + TEST_VERIFY (asprintf_alias (&str, "[[%1$P %1$P]]", 123L, 457.0) >= 0);
> + TEST_COMPARE_STRING (str, "[[(123, 457.000000) (123, 457.000000)]]");
> + free (str);
> +
> str = NULL;
> TEST_VERIFY (asprintf_alias (&str, "[[%.1P]]", 1L, 2.0) >= 0);
> TEST_COMPARE_STRING (str, "[[{(1, 2.000000)}]]");
Could you tune the test to check for multiple different argument numbers? I added
28 to be assured that the printf_positional path were triggered, but it wold be
good to test for different argument numbers as well.
> diff --git a/stdio-common/vfprintf-internal.c b/stdio-common/vfprintf-internal.c
> index f30a9e9f3a..6d6d96e800 100644
> --- a/stdio-common/vfprintf-internal.c
> +++ b/stdio-common/vfprintf-internal.c
> @@ -1130,6 +1130,8 @@ printf_positional (struct Xprintf_buffer * buf, const CHAR_T *format,
> union printf_arg *args_value;
> int *args_size;
> int *args_type;
> + void *args_pa_user;
> + size_t args_pa_user_offset;
> {
> /* Calculate total size needed to represent a single argument
> across all three argument-related arrays. */
> @@ -1146,6 +1148,7 @@ printf_positional (struct Xprintf_buffer * buf, const CHAR_T *format,
> now. */
> args_size = &args_value[nargs].pa_int;
> args_type = &args_size[nargs];
> + args_pa_user = &args_type[nargs];
> memset (args_type, (mode_flags & PRINTF_FORTIFY) != 0 ? '\xff' : '\0',
> nargs * sizeof (*args_type));
> }
> @@ -1235,7 +1238,25 @@ printf_positional (struct Xprintf_buffer * buf, const CHAR_T *format,
> else if (__glibc_unlikely (__printf_va_arg_table != NULL)
> && __printf_va_arg_table[args_type[cnt] - PA_LAST] != NULL)
> {
> - args_value[cnt].pa_user = alloca (args_size[cnt]);
> + while (args_pa_user + args_size[cnt] >
> + argsbuf.data + argsbuf.length)
> + {
> + args_pa_user_offset = args_pa_user - (void *) &args_type[nargs];
> + if (!scratch_buffer_grow_preserve (&argsbuf))
> + {
> + Xprintf_buffer_mark_failed (buf);
> + goto all_done;
> + }
> + args_value = argsbuf.data;
> + /* Set up the remaining two arrays to each point past the end of
> + the prior array, since space for all three has been allocated
> + now. */
> + args_size = &args_value[nargs].pa_int;
> + args_type = &args_size[nargs];
> + args_pa_user = (void *) &args_type[nargs] + args_pa_user_offset;
> + }
> + args_value[cnt].pa_user = args_pa_user;
> + args_pa_user += args_size[cnt];
> (*__printf_va_arg_table[args_type[cnt] - PA_LAST])
> (args_value[cnt].pa_user, ap_savep);
> }
next prev parent reply other threads:[~2023-07-19 18:04 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-20 13:26 Joe Simmons-Talbott
2023-07-07 17:21 ` Joe Simmons-Talbott
2023-07-19 18:04 ` Adhemerval Zanella Netto [this message]
2023-07-20 13:47 ` Joe Simmons-Talbott
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=f6512ede-2311-5793-d0bf-ad6f77e3a7d8@linaro.org \
--to=adhemerval.zanella@linaro.org \
--cc=josimmon@redhat.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).