wget-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: wget2 | * configure.ac: Remove redundant sanitizer warnings (!506)


From: @rockdaboot
Subject: Re: wget2 | * configure.ac: Remove redundant sanitizer warnings (!506)
Date: Sun, 24 Apr 2022 18:49:53 +0000



Tim Rühsen commented on a discussion on configure.ac: 
https://gitlab.com/gnuwget/wget2/-/merge_requests/506#note_922053609

> +  if test "$CCNAME" = "gcc"; then
> +    gl_WARN_ADD([-fsanitize=float-cast-overflow])
> +  fi
>  fi
>  
>  if test "$gl_cc_sanitize_asan" = yes; then
>    gl_WARN_ADD([-fsanitize=address])
>    gl_WARN_ADD([-fsanitize=leak])
> -  gl_WARN_ADD([-fsanitize=nonnull-attribute])
>    gl_WARN_ADD([-fno-omit-frame-pointer])
> +  gl_WARN_ADD([-fno-optimize-sibling-calls])
>  
> -  # additional clang option
> -  gl_WARN_ADD([-fsanitize-address-use-after-scope])
> +  if test "$CCNAME" = "gcc"; then
> +    gl_WARN_ADD([-fsanitize=pointer-compare])

LLVM docs don't tell us since which version this is true. While it is true for 
the current version, what about e.g. llvm/clang 3.x ? I mentioned before that 
we possibly need some comments about which version we tested with. Just becasue 
we don't want to go back and forth (or argue about it) if someone complains 
that an older version of llvm/clang x.y has any issues with our assumptions. 
Once we add a lose comment with the gcc and clang versions, we should be fine.

Now sure if that was understandable, but my fear is that some earlier versions 
did not include `-fsanitize-address-use-after-scope` in `-fsanitize=address`. 
My memory is not reliable on this, but it deems to me that there where some 
options (maybe not exactly this one) in the past that were not included, while 
they are with recent compiler versions. And I remember that there were always 
slight differences between clang and gcc.

-- 
Reply to this email directly or view it on GitLab: 
https://gitlab.com/gnuwget/wget2/-/merge_requests/506#note_922053609
You're receiving this email because of your account on gitlab.com.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]