Re: not forwarded elfutils mips patches

2020-07-29 Thread YunQiang Su
Matthias Klose  于2020年7月28日周二 下午6:15写道:
>
> Hi,
>
> with elfutils 0.180, I disabled some mips patches which didn't apply anymore,
> and then forgot about them for the upload.
>
> mips_readelf_w.patch
> 0001-Ignore-differences-between-mips-machine-identifiers.patch
> 0002-Add-support-for-mips64-abis-in-mips_retval.c.patch
> 0003-Add-mips-n64-relocation-format-hack.patch
> mips_cfi.patch
>

It seems all of them are needed. I will try to refresh them and send
them upstream.

> Please could yu have alook which ones are still necessary, and probably 
> forward
> those upstream?
>
> Thanks, Matthias
>


-- 
YunQiang Su



Re: C++ symbols files (Re: GCC 10 transition)

2020-07-29 Thread Adrian Bunk
On Tue, Jul 28, 2020 at 12:41:42PM +0200, Matthias Klose wrote:
>...
> Suggested ways to solve these issues:
> 
>  - Use plain shlibs files, without using symbols files, although
>debhelper wrongly warns about missing symbols files.

Usually this is the better option for C++ libraries,
for more discussion see the bottom half of
  https://wiki.debian.org/UsingSymbolsFiles

>  - Use a better tool like abi-compliance-checker or abigail, although
>both are missing dpkg integration like symbols files.
>...

Symbols files have two purposes:
1. detect ABI breakages
2. don't create unnecessarily strict dependencies on the
   latest version of a library

You are talking about 1., and I agree that any proper ABI checking tools 
would be more useful than people just blindly updating symbols files 
from the buildds without properly reviewing every single symbol change.

There is also the point 2. which was AFAIR the main original motivation 
for symbols files, but IMHO the C++ symbols files mess is not worth it.

The wiki contains some information from Simon McVittie regarding 
upstream of a C++ library solving this with proper symbol visibility, 
but this is something for upstream to implement.

> Matthias

cu
Adrian



[Bug target/35488] A incorrect result in a simple division, only in 32-bit gcc.

2020-07-29 Thread vries at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=35488

Tom de Vries  changed:

   What|Removed |Added

 CC||vries at gcc dot gnu.org

--- Comment #11 from Tom de Vries  ---
FTR, this PR is linked to from here (
https://lemire.me/blog/2020/06/26/gcc-not-nearest/ ).

-- 
You are receiving this mail because:
You are on the CC list for the bug.