public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Carlos O'Donell <carlos@redhat.com>
To: Yang Yanchao <yangyanchao6@huawei.com>, libc-alpha@sourceware.org
Cc: linfeilong@huawei.com, fweimer@redhat.com, schwab@linux-m68k.org
Subject: Re: [PATCH v3] elf: fixes compile error when both enable -Werror and -DNDEBUG
Date: Mon, 11 Apr 2022 10:29:44 -0400	[thread overview]
Message-ID: <6484ccbc-0297-de8a-3d2d-13257eab8fa4@redhat.com> (raw)
In-Reply-To: <20220402095152.3063-1-yangyanchao6@huawei.com>

On 4/2/22 05:51, Yang Yanchao via Libc-alpha wrote:
> Use -Werror and -DNDEBUG at the same time will
> causes the following compilation errors:

Thank you for following up with a v3. A little more work and I think
this will be ready. Thank you for testing with -Werror and -DNDEBUG.
This is a configuration we used to test in Fedora, but have since
stopped using -DNDEBUG to keep the asserts in place. I need to consider
adding such an option to build-many-glibcs testing.

> cache.c: In function 'save_cache':
> cache.c:758:15: error: unused variable 'old_offset' [-Werror=unused-variable]
>   758 |       off64_t old_offset = lseek64 (fd, extension_offset, SEEK_SET);
>       |               ^~~~~~~~~~
> 
> -DNDEBUG will disables the assertion.
> Therefore, only the variables used by assertions do not take effect.
> Fooling the Compiler with Type Conversions.
> ---
>  elf/cache.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/elf/cache.c b/elf/cache.c
> index dbf4c83a7a..68cd4d0828 100644
> --- a/elf/cache.c
> +++ b/elf/cache.c
> @@ -754,6 +754,7 @@ save_cache (const char *cache_name)
>      {
>        /* Align file position to 4.  */
>        off64_t old_offset = lseek64 (fd, extension_offset, SEEK_SET);
> +      (void) old_offset;

The compiler unused-variable warnings are very aggressive, and this may not fix all issues.

In cases like these we try to improve the situation by using specific pragmas.

e.g.

/* With GCC X.Y when compiling with -Werror, -DNDEBUG, and optimizations, the
   compiler may be able to eliminate the use of extension_offset and so we
   get an unused variable warning for extension_offset.  We want to continue to
   support the assert, so we disable this warning for this code.  */
DIAG_PUSH_NEEDS_COMMENT;
DIAG_IGNORE_NEEDS_COMMENT (X, "-Werror=unused-variable")
assert ((unsigned long long int) (extension_offset - old_offset) < 4);
DIAG_POP_NEEDS_COMMENT;

This is just an example, with X being the major version of gcc, and y being the minor.

You'll need to test this out, and double check CI/CD passes:
https://patchwork.sourceware.org/project/glibc/list/
You can view your posts there with success/warn/fail results.

>        assert ((unsigned long long int) (extension_offset - old_offset) < 4);
>        write_extensions (fd, str_offset, extension_offset);
>      }

Looking forward to a v4.

-- 
Cheers,
Carlos.


      reply	other threads:[~2022-04-11 14:29 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-04-02  9:51 Yang Yanchao
2022-04-11 14:29 ` Carlos O'Donell [this message]

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=6484ccbc-0297-de8a-3d2d-13257eab8fa4@redhat.com \
    --to=carlos@redhat.com \
    --cc=fweimer@redhat.com \
    --cc=libc-alpha@sourceware.org \
    --cc=linfeilong@huawei.com \
    --cc=schwab@linux-m68k.org \
    --cc=yangyanchao6@huawei.com \
    /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).