Re: pkg/51122 (ffmpeg[23] compiled with gcc 5.x on x86 32-bit can lead to crashes)

[Joerg Sonnenberger <joerg%bec.de@localhost>]

The following reply was made to PR pkg/51122; it has been noted by GNATS.

From: Joerg Sonnenberger <joerg%bec.de@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: pkg/51122 (ffmpeg[23] compiled with gcc 5.x on x86 32-bit can
 lead to crashes)
Date: Sun, 26 Jun 2016 23:36:50 +0200

 On Sun, Jun 26, 2016 at 07:00:01PM +0000, David Holland wrote:
 > The following reply was made to PR pkg/51122; it has been noted by GNATS.
 > 
 > From: David Holland <dholland-pbugs%netbsd.org@localhost>
 > To: gnats-bugs%NetBSD.org@localhost
 > Cc: 
 > Subject: Re: pkg/51122 (ffmpeg[23] compiled with gcc 5.x on x86 32-bit can
 >  lead to crashes)
 > Date: Sun, 26 Jun 2016 18:56:08 +0000
 > 
 >  On Sun, Jun 26, 2016 at 06:35:00PM +0000, Joerg Sonnenberger wrote:
 >   >  > joerg wants to fix ffmpeg's assembler, everyone else is fine with
 >   >  > working around the problem...
 >   >  
 >   >  I've told you I don't touch this macro mess. If it was anywhere
 >   >  tractable I would have fixed it already. I said to fix the callers, but
 >   >  it seems you can't read. Since I got assigned the bug, I will just
 >   >  disable the assembler code again and remove the stackrealign hacks...
 >  
 >  If you don't want to fix it, set it back to pkg-manager.
 >  
 >  But then stop carping about what other people do to deal with it, ok?
 
 The only one carping about something here is you. The "fix" is
 introducing a performance penalty for every single codec. It doesn't fix
 the assembler. It doesn't apply the existing attributes to the caller of
 the broken assembler. In short, it is carpet bombing. But hey, I'm wrong
 to call it out as a crude hack and *I* should be fixing this?
 
 Joerg