From: Martin Sebor <msebor@gmail.com>
To: Paul Eggert <eggert@cs.ucla.edu>
Cc: GNU C Library <libc-alpha@sourceware.org>,
Carlos O'Donell <carlos@redhat.com>
Subject: [PATCH v2] remove attribute access from regexec
Date: Wed, 18 Aug 2021 09:53:50 -0600 [thread overview]
Message-ID: <ba40f5e9-0a2f-cc6d-e96a-8959d45fbafa@gmail.com> (raw)
In-Reply-To: <bebb6028-ddae-67f7-e14c-7f95b733c7cc@gmail.com>
[-- Attachment #1: Type: text/plain, Size: 3113 bytes --]
Let me repost an updated patch as v2 to let the patch tester know
about the revision, and to also include the corresponding change
to regexec.c needed to avoid the -Wvla-parameter I mentioned.
Apparently the first patch failed to apply and the second one
wasn't picked up because the subject line hasn't changed. Thanks
for letting me know, Carlos!
Here's a link to the Patchwork check:
http://patchwork.sourceware.org/project/glibc/patch/15a32181-8060-4135-cb72-9e79831697d5@gmail.com/
On 8/14/21 2:08 PM, Martin Sebor wrote:
> On 8/13/21 4:34 PM, Paul Eggert wrote:
>> On 8/13/21 2:30 PM, Martin Sebor wrote:
>>> Attached is a revised patch with this approach.
>>
>> The revised patch is to include/regex.h but the original patch was to
>> posix/regex.h. Is that intentional?
>
> Yes, they need to be consistent, otherwise GCC issues -Wvla-parameter.
> (That's to help detect inadvertent array/VLA mismatches as well as
> mismatches in the VLA parameter bounds.)
>
>>
>> We need to check whether __STDC_VERSION__ is defined. Also, no need
>> for parens around arg of 'defined'. Something like this perhaps:
>>
>> #if (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__ \
>> && !defined __STDC_NO_VLA__)
>>
>> Also, the duplication of the declarations make the headers harder to
>> read and encourage typos (I noticed one typo: "_Restrict_arr" without
>> the trailing "_"). Instead, I suggest something like this:
>>
>> #if (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__ \
>> && !defined __STDC_NO_VLA__)
>> # define _REGEX_VLA(arg) arg
>> #else
>> # define _REGEX_VLA(arg)
>> #endif
>>
>> That way, we can simply change "regmatch_t __pmatch[_Restrict_arr_]"
>> to "regmatch_t __pmatch[_Restrict_arr_ _REGEX_VLA (__nmatch)]" without
>> having to duplicate the entire function declaration.
>
> Sounds good. I've defined the macro in cdefs.h and mamed it _VLA_ARG
> to make it usable in other contexts. Please see the attached revision.
>
>>
>>> PS POSIX says regexec() ignores pnmatch when REG_NOSUB is set, so
>>> strictly speaking, warning for such calls to it in that case is
>>> also a false positive.
>>
>> Ouch, this casts doubt on the entire exercise. It's not simply about
>> warnings: it's about the code being generated for the matcher. For
>> example, for:
>>
>> int
>> f (_Bool flag, unsigned long n, int a[n])
>> {
>> return n == 0 ? 0 : flag ? a[n - 1] : a[0];
>> }
>>
>> a compiler is allowed to generate code that loads a[n - 1] even when
>> FLAG is false. Similarly, if we add this VLA business to regexec, the
>> generated machine code could dereference pmatch unconditionally even
>> if our source code makes the dereferencing conditional on REG_NOSUB,
>> and the resulting behavior would fail to conform to POSIX.
>
> The VLA bound by itself doesn't affect codegen. I suspect you're
> thinking of a[static n]? With just a[n], without static, there
> is no requirement that a point to an array with n elements. It
> simply declares an ordinary pointer, same as [] or *.
>
> Martin
[-- Attachment #2: glibc-28170.diff --]
[-- Type: text/x-patch, Size: 2305 bytes --]
diff --git a/include/regex.h b/include/regex.h
index 24eca2c297..5cf3ef7636 100644
--- a/include/regex.h
+++ b/include/regex.h
@@ -37,7 +37,8 @@ extern int __regcomp (regex_t *__preg, const char *__pattern, int __cflags);
libc_hidden_proto (__regcomp)
extern int __regexec (const regex_t *__preg, const char *__string,
- size_t __nmatch, regmatch_t __pmatch[], int __eflags);
+ size_t __nmatch, regmatch_t __pmatch[_VLA_ARG (__nmatch)],
+ int __eflags);
libc_hidden_proto (__regexec)
extern size_t __regerror (int __errcode, const regex_t *__preg,
diff --git a/misc/sys/cdefs.h b/misc/sys/cdefs.h
index e490fc1aeb..6b6e4c233c 100644
--- a/misc/sys/cdefs.h
+++ b/misc/sys/cdefs.h
@@ -632,4 +632,14 @@ _Static_assert (0, "IEEE 128-bits long double requires redirection on this platf
# define __attribute_returns_twice__ /* Ignore. */
#endif
+
+#if (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__ \
+ && !defined __STDC_NO_VLA__)
+/* Used to specify a variable bound in a declaration of a function
+ VLA-like parameter, as in 'int f (int n, int[_VLA_ARG (n)n]);' */
+# define _VLA_ARG(arg) arg
+#else
+# define _VLA_ARG(arg)
+#endif
+
#endif /* sys/cdefs.h */
diff --git a/posix/regex.h b/posix/regex.h
index 14fb1d8364..52ccc4b577 100644
--- a/posix/regex.h
+++ b/posix/regex.h
@@ -654,9 +654,8 @@ extern int regcomp (regex_t *_Restrict_ __preg,
extern int regexec (const regex_t *_Restrict_ __preg,
const char *_Restrict_ __String, size_t __nmatch,
- regmatch_t __pmatch[_Restrict_arr_],
- int __eflags)
- __attr_access ((__write_only__, 4, 3));
+ regmatch_t __pmatch[_Restrict_arr_ _VLA_ARG (__nmatch)],
+ int __eflags);
extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg,
char *_Restrict_ __errbuf, size_t __errbuf_size)
diff --git a/posix/regexec.c b/posix/regexec.c
index f7b4f9cfc3..7b9d0ee65f 100644
--- a/posix/regexec.c
+++ b/posix/regexec.c
@@ -190,7 +190,7 @@ static reg_errcode_t extend_buffers (re_match_context_t *mctx, int min_len);
int
regexec (const regex_t *__restrict preg, const char *__restrict string,
- size_t nmatch, regmatch_t pmatch[], int eflags)
+ size_t nmatch, regmatch_t pmatch[_VLA_ARG (nmatch)], int eflags)
{
reg_errcode_t err;
Idx start, length;
next prev parent reply other threads:[~2021-08-18 15:53 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-13 18:26 [PATCH] " Martin Sebor
2021-08-13 20:11 ` Paul Eggert
2021-08-13 21:30 ` Martin Sebor
2021-08-13 22:34 ` Paul Eggert
2021-08-14 20:08 ` Martin Sebor
2021-08-18 15:53 ` Martin Sebor [this message]
2021-08-18 19:52 ` Paul Eggert
2021-08-19 23:50 ` [PATCH v3] " Martin Sebor
2021-08-22 5:06 ` Paul Eggert
2021-08-26 15:06 ` Martin Sebor
2021-08-26 16:24 ` Paul Eggert
2021-08-26 16:47 ` Martin Sebor
2021-08-27 8:52 ` Paul Eggert
2021-08-27 16:34 ` Martin Sebor
2021-08-27 17:50 ` Allow #pragma GCC in headers in conformtest [committed] (was: Re: [PATCH v3] remove attribute access from regexec) Joseph Myers
2021-08-27 18:57 ` [PATCH v3] remove attribute access from regexec Paul Eggert
2021-09-20 20:46 ` Joseph Myers
2021-09-21 6:52 ` Paul Eggert
2021-09-21 13:48 ` Joseph Myers
2021-09-21 15:00 ` Paul Eggert
2021-10-19 16:39 ` Carlos O'Donell
2021-10-19 17:06 ` Martin Sebor
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=ba40f5e9-0a2f-cc6d-e96a-8959d45fbafa@gmail.com \
--to=msebor@gmail.com \
--cc=carlos@redhat.com \
--cc=eggert@cs.ucla.edu \
--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).