Message ID | 201610030151.25089.cehoyos@ag.or.at |
---|---|
State | Superseded |
Headers | show |
Hi, On Sun, Oct 2, 2016 at 7:51 PM, Carl Eugen Hoyos <cehoyos@ag.or.at> wrote: > Hi! > > Attached patch adds a musl section to doc/platform. This is counter-productive... Let's work with the musl-devs and fix their libc-alternative? Ronald
2016-10-03 2:36 GMT+02:00 Ronald S. Bultje <rsbultje@gmail.com>: > Hi, > > On Sun, Oct 2, 2016 at 7:51 PM, Carl Eugen Hoyos <cehoyos@ag.or.at> wrote: > >> Attached patch adds a musl section to doc/platform. > > This is counter-productive... Sorry, I just wanted to be productive wrt a bug (actually bugs) that we will not be able to fix quickly. Carl Eugen
Hi, On Sun, Oct 2, 2016 at 8:42 PM, Carl Eugen Hoyos <ceffmpeg@gmail.com> wrote: > 2016-10-03 2:36 GMT+02:00 Ronald S. Bultje <rsbultje@gmail.com>: > > Hi, > > > > On Sun, Oct 2, 2016 at 7:51 PM, Carl Eugen Hoyos <cehoyos@ag.or.at> > wrote: > > > >> Attached patch adds a musl section to doc/platform. > > > > This is counter-productive... > > Sorry, I just wanted to be productive wrt a bug (actually > bugs) that we will not be able to fix quickly. Have we reported it upstream? That seems to be the least we can do. Ronald
On 03/10/2016 01:44, Ronald S. Bultje wrote: > Hi, > > On Sun, Oct 2, 2016 at 8:42 PM, Carl Eugen Hoyos <ceffmpeg@gmail.com> wrote: > >> 2016-10-03 2:36 GMT+02:00 Ronald S. Bultje <rsbultje@gmail.com>: >>> Hi, >>> >>> On Sun, Oct 2, 2016 at 7:51 PM, Carl Eugen Hoyos <cehoyos@ag.or.at> >> wrote: >>> >>>> Attached patch adds a musl section to doc/platform. >>> >>> This is counter-productive... >> >> Sorry, I just wanted to be productive wrt a bug (actually >> bugs) that we will not be able to fix quickly. > > > Have we reported it upstream? That seems to be the least we can do. > Not sure it's worth reporting upstream, we're at fault here. The SysV calling convention ABI explicitly states: 'If the function does not return a floating-point value, then this register must be empty. This register must be empty before entry to a function.' (in regards to the floating point registers). See: http://sco.com/developers/devspecs/abi386-4.pdf pages 37-38
On Mon, 3 Oct 2016 01:51:25 +0200 Carl Eugen Hoyos <cehoyos@ag.or.at> wrote: > From 35bf44b909658864c9e8cca8dbc855dbde1e1ef5 Mon Sep 17 00:00:00 2001 > From: Carl Eugen Hoyos <cehoyos@ag.or.at> > Date: Mon, 3 Oct 2016 01:41:37 +0200 > Subject: [PATCH] doc/platform: Mention musl where x86_32 is not supported. > > --- > doc/platform.texi | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/doc/platform.texi b/doc/platform.texi > index 21b135f..3404407 100644 > --- a/doc/platform.texi > +++ b/doc/platform.texi > @@ -77,6 +77,12 @@ optimized assembly functions. @uref{http://www.finkproject.org/, Fink}, > @uref{https://mxcl.github.com/homebrew/, Homebrew} > or @uref{http://www.macports.org, MacPorts} can easily provide it. > > +@section musl > + > +Compilation with @uref{http://www.musl-libc.org/, musl} on x86-64 is supposed > +to work out-of-the-box. > +Do not compile FFmpeg with musl on x86_32, random crashes have to be expected. > + > > @chapter DOS > You could mention that this is due to FFmpeg violating the ABI (apparently).
Le duodi 12 vendémiaire, an CCXXV, wm4 a écrit : > You could mention that this is due to FFmpeg violating the ABI > (apparently). Still, using floats in malloc() looks like a very bad idea, borderline recipe for disaster. Especially on x86_32. Regards,
On Mon, Oct 3, 2016 at 2:36 AM, Ronald S. Bultje <rsbultje@gmail.com> wrote: > Hi, > > On Sun, Oct 2, 2016 at 7:51 PM, Carl Eugen Hoyos <cehoyos@ag.or.at> wrote: > >> Hi! >> >> Attached patch adds a musl section to doc/platform. > > > This is counter-productive... Let's work with the musl-devs and fix their > libc-alternative? > I agree that adding such a note is not really going to help anything, if anything its going to cause more sillyness. Lets just fix the bugs, and not call something simply unsupported. - Hendrik
2016-10-03 12:30 GMT+02:00 Hendrik Leppkes <h.leppkes@gmail.com>:
> Lets just fix the bugs
Sorry: Which bugs exactly should we just fix?
Carl Eugen
On Mon, Oct 03, 2016 at 12:37:29PM +0200, Carl Eugen Hoyos wrote: > 2016-10-03 12:30 GMT+02:00 Hendrik Leppkes <h.leppkes@gmail.com>: > > > Lets just fix the bugs > > Sorry: Which bugs exactly should we just fix? > > Carl Eugen Please fix the UB in the MMX-code. Rune
diff --git a/doc/platform.texi b/doc/platform.texi index 21b135f..3404407 100644 --- a/doc/platform.texi +++ b/doc/platform.texi @@ -77,6 +77,12 @@ optimized assembly functions. @uref{http://www.finkproject.org/, Fink}, @uref{https://mxcl.github.com/homebrew/, Homebrew} or @uref{http://www.macports.org, MacPorts} can easily provide it. +@section musl + +Compilation with @uref{http://www.musl-libc.org/, musl} on x86-64 is supposed +to work out-of-the-box. +Do not compile FFmpeg with musl on x86_32, random crashes have to be expected. + @chapter DOS
Hi! Attached patch adds a musl section to doc/platform. Please comment, Carl Eugen From 35bf44b909658864c9e8cca8dbc855dbde1e1ef5 Mon Sep 17 00:00:00 2001 From: Carl Eugen Hoyos <cehoyos@ag.or.at> Date: Mon, 3 Oct 2016 01:41:37 +0200 Subject: [PATCH] doc/platform: Mention musl where x86_32 is not supported. --- doc/platform.texi | 6 ++++++ 1 file changed, 6 insertions(+)