| Message ID | 20181002191136.8792-1-raj.khem@gmail.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | mesa: Disable asm on musl | expand |
Isn't this a bit overkill? Is the assembler actually broken? Ross On Tue, 2 Oct 2018 at 20:21, Khem Raj <raj.khem@gmail.com> wrote: > > some unknown part of mesa's x86 assembly > code is broken by readonly text segments > > [ YOCTO #12918 ] > > Signed-off-by: Khem Raj <raj.khem@gmail.com> > --- > meta/recipes-graphics/mesa/mesa.inc | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/meta/recipes-graphics/mesa/mesa.inc b/meta/recipes-graphics/mesa/mesa.inc > index 8d0e2cb67c..e6f3712706 100644 > --- a/meta/recipes-graphics/mesa/mesa.inc > +++ b/meta/recipes-graphics/mesa/mesa.inc > @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = "--enable-${OSMESA},--disable-${OSMESA}" > PACKAGECONFIG[unwind] = "--enable-libunwind,--disable-libunwind,libunwind" > > EXTRA_OECONF_remove_libc-musl = "--enable-glx-tls" > -EXTRA_OECONF_append_libc-musl = " --disable-glx-tls" > +EXTRA_OECONF_append_libc-musl = " --disable-asm --disable-glx-tls" > # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2) > FULL_OPTIMIZATION_append = " -fno-omit-frame-pointer" > > -- > 2.19.0 > > -- > _______________________________________________ > Openembedded-core mailing list > Openembedded-core@lists.openembedded.org > http://lists.openembedded.org/mailman/listinfo/openembedded-core -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
I think it is. The way th calls are set for performance reasons seems to be causing this Ideally we should ensure to use right dialect for tls calls I think musl is exposing a bug in Mesa Which we covered by disabling tls in Glx once we disable that then this problem pops up so we disable ASM So this is a workaround until then On Tue, Oct 2, 2018 at 12:28 PM Burton, Ross <ross.burton@intel.com> wrote: > Isn't this a bit overkill? Is the assembler actually broken? > > Ross > On Tue, 2 Oct 2018 at 20:21, Khem Raj <raj.khem@gmail.com> wrote: > > > > some unknown part of mesa's x86 assembly > > code is broken by readonly text segments > > > > [ YOCTO #12918 ] > > > > Signed-off-by: Khem Raj <raj.khem@gmail.com> > > --- > > meta/recipes-graphics/mesa/mesa.inc | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/meta/recipes-graphics/mesa/mesa.inc > b/meta/recipes-graphics/mesa/mesa.inc > > index 8d0e2cb67c..e6f3712706 100644 > > --- a/meta/recipes-graphics/mesa/mesa.inc > > +++ b/meta/recipes-graphics/mesa/mesa.inc > > @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = > "--enable-${OSMESA},--disable-${OSMESA}" > > PACKAGECONFIG[unwind] = > "--enable-libunwind,--disable-libunwind,libunwind" > > > > EXTRA_OECONF_remove_libc-musl = "--enable-glx-tls" > > -EXTRA_OECONF_append_libc-musl = " --disable-glx-tls" > > +EXTRA_OECONF_append_libc-musl = " --disable-asm --disable-glx-tls" > > # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2) > > FULL_OPTIMIZATION_append = " -fno-omit-frame-pointer" > > > > -- > > 2.19.0 > > > > -- > > _______________________________________________ > > Openembedded-core mailing list > > Openembedded-core@lists.openembedded.org > > http://lists.openembedded.org/mailman/listinfo/openembedded-core > <div><div dir="auto">I think it is. The way th calls are set for performance reasons seems to be causing this</div></div><div dir="auto">Ideally we should ensure to use right dialect for tls calls I think musl is exposing a bug in Mesa </div><div dir="auto">Which we covered by disabling tls in Glx once we disable that then this problem pops up so we disable ASM </div><div dir="auto"><br></div><div dir="auto">So this is a workaround until then </div><div><br><div class="gmail_quote"><div dir="ltr">On Tue, Oct 2, 2018 at 12:28 PM Burton, Ross <<a href="mailto:ross.burton@intel.com">ross.burton@intel.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Isn't this a bit overkill? Is the assembler actually broken?<br> <br> Ross<br> On Tue, 2 Oct 2018 at 20:21, Khem Raj <<a href="mailto:raj.khem@gmail.com" target="_blank">raj.khem@gmail.com</a>> wrote:<br> ><br> > some unknown part of mesa's x86 assembly<br> > code is broken by readonly text segments<br> ><br> > [ YOCTO #12918 ]<br> ><br> > Signed-off-by: Khem Raj <<a href="mailto:raj.khem@gmail.com" target="_blank">raj.khem@gmail.com</a>><br> > ---<br> > meta/recipes-graphics/mesa/mesa.inc | 2 +-<br> > 1 file changed, 1 insertion(+), 1 deletion(-)<br> ><br> > diff --git a/meta/recipes-graphics/mesa/mesa.inc b/meta/recipes-graphics/mesa/mesa.inc<br> > index 8d0e2cb67c..e6f3712706 100644<br> > --- a/meta/recipes-graphics/mesa/mesa.inc<br> > +++ b/meta/recipes-graphics/mesa/mesa.inc<br> > @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = "--enable-${OSMESA},--disable-${OSMESA}"<br> > PACKAGECONFIG[unwind] = "--enable-libunwind,--disable-libunwind,libunwind"<br> ><br> > EXTRA_OECONF_remove_libc-musl = "--enable-glx-tls"<br> > -EXTRA_OECONF_append_libc-musl = " --disable-glx-tls"<br> > +EXTRA_OECONF_append_libc-musl = " --disable-asm --disable-glx-tls"<br> > # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2)<br> > FULL_OPTIMIZATION_append = " -fno-omit-frame-pointer"<br> ><br> > --<br> > 2.19.0<br> ><br> > --<br> > _______________________________________________<br> > Openembedded-core mailing list<br> > <a href="mailto:Openembedded-core@lists.openembedded.org" target="_blank">Openembedded-core@lists.openembedded.org</a><br> > <a href="http://lists.openembedded.org/mailman/listinfo/openembedded-core" rel="noreferrer" target="_blank">http://lists.openembedded.org/mailman/listinfo/openembedded-core</a><br> </blockquote></div></div> -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
Can you file a bug with mesa so we don't forget and Mesa knows? Ross On Tue, 2 Oct 2018 at 20:37, Khem Raj <raj.khem@gmail.com> wrote: > > I think it is. The way th calls are set for performance reasons seems to be causing this > Ideally we should ensure to use right dialect for tls calls I think musl is exposing a bug in Mesa > Which we covered by disabling tls in Glx once we disable that then this problem pops up so we disable ASM > > So this is a workaround until then > > On Tue, Oct 2, 2018 at 12:28 PM Burton, Ross <ross.burton@intel.com> wrote: >> >> Isn't this a bit overkill? Is the assembler actually broken? >> >> Ross >> On Tue, 2 Oct 2018 at 20:21, Khem Raj <raj.khem@gmail.com> wrote: >> > >> > some unknown part of mesa's x86 assembly >> > code is broken by readonly text segments >> > >> > [ YOCTO #12918 ] >> > >> > Signed-off-by: Khem Raj <raj.khem@gmail.com> >> > --- >> > meta/recipes-graphics/mesa/mesa.inc | 2 +- >> > 1 file changed, 1 insertion(+), 1 deletion(-) >> > >> > diff --git a/meta/recipes-graphics/mesa/mesa.inc b/meta/recipes-graphics/mesa/mesa.inc >> > index 8d0e2cb67c..e6f3712706 100644 >> > --- a/meta/recipes-graphics/mesa/mesa.inc >> > +++ b/meta/recipes-graphics/mesa/mesa.inc >> > @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = "--enable-${OSMESA},--disable-${OSMESA}" >> > PACKAGECONFIG[unwind] = "--enable-libunwind,--disable-libunwind,libunwind" >> > >> > EXTRA_OECONF_remove_libc-musl = "--enable-glx-tls" >> > -EXTRA_OECONF_append_libc-musl = " --disable-glx-tls" >> > +EXTRA_OECONF_append_libc-musl = " --disable-asm --disable-glx-tls" >> > # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2) >> > FULL_OPTIMIZATION_append = " -fno-omit-frame-pointer" >> > >> > -- >> > 2.19.0 >> > >> > -- >> > _______________________________________________ >> > Openembedded-core mailing list >> > Openembedded-core@lists.openembedded.org >> > http://lists.openembedded.org/mailman/listinfo/openembedded-core -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
Yes, will do. I think it will need some time since this needs to be well explained so mesa devs can see the reasoning behind it On Tue, Oct 2, 2018 at 1:42 PM Burton, Ross <ross.burton@intel.com> wrote: > > Can you file a bug with mesa so we don't forget and Mesa knows? > > Ross > On Tue, 2 Oct 2018 at 20:37, Khem Raj <raj.khem@gmail.com> wrote: > > > > I think it is. The way th calls are set for performance reasons seems to be causing this > > Ideally we should ensure to use right dialect for tls calls I think musl is exposing a bug in Mesa > > Which we covered by disabling tls in Glx once we disable that then this problem pops up so we disable ASM > > > > So this is a workaround until then > > > > On Tue, Oct 2, 2018 at 12:28 PM Burton, Ross <ross.burton@intel.com> wrote: > >> > >> Isn't this a bit overkill? Is the assembler actually broken? > >> > >> Ross > >> On Tue, 2 Oct 2018 at 20:21, Khem Raj <raj.khem@gmail.com> wrote: > >> > > >> > some unknown part of mesa's x86 assembly > >> > code is broken by readonly text segments > >> > > >> > [ YOCTO #12918 ] > >> > > >> > Signed-off-by: Khem Raj <raj.khem@gmail.com> > >> > --- > >> > meta/recipes-graphics/mesa/mesa.inc | 2 +- > >> > 1 file changed, 1 insertion(+), 1 deletion(-) > >> > > >> > diff --git a/meta/recipes-graphics/mesa/mesa.inc b/meta/recipes-graphics/mesa/mesa.inc > >> > index 8d0e2cb67c..e6f3712706 100644 > >> > --- a/meta/recipes-graphics/mesa/mesa.inc > >> > +++ b/meta/recipes-graphics/mesa/mesa.inc > >> > @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = "--enable-${OSMESA},--disable-${OSMESA}" > >> > PACKAGECONFIG[unwind] = "--enable-libunwind,--disable-libunwind,libunwind" > >> > > >> > EXTRA_OECONF_remove_libc-musl = "--enable-glx-tls" > >> > -EXTRA_OECONF_append_libc-musl = " --disable-glx-tls" > >> > +EXTRA_OECONF_append_libc-musl = " --disable-asm --disable-glx-tls" > >> > # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2) > >> > FULL_OPTIMIZATION_append = " -fno-omit-frame-pointer" > >> > > >> > -- > >> > 2.19.0 > >> > > >> > -- > >> > _______________________________________________ > >> > Openembedded-core mailing list > >> > Openembedded-core@lists.openembedded.org > >> > http://lists.openembedded.org/mailman/listinfo/openembedded-core -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
Can we have bit more detail in the commit message about what the issues are? We're not supposed to rely solely on a bug report reference to do that; also if it's considered a workaround and we're working with upstream to fix it, the commit message should state that as well. Thanks Paul On Wednesday, 3 October 2018 8:37:10 AM NZDT Khem Raj wrote: > I think it is. The way th calls are set for performance reasons seems to be > causing this > Ideally we should ensure to use right dialect for tls calls I think musl > is exposing a bug in Mesa > Which we covered by disabling tls in Glx once we disable that then this > problem pops up so we disable ASM > > So this is a workaround until then > > On Tue, Oct 2, 2018 at 12:28 PM Burton, Ross <ross.burton@intel.com> wrote: > > > Isn't this a bit overkill? Is the assembler actually broken? > > > > Ross > > On Tue, 2 Oct 2018 at 20:21, Khem Raj <raj.khem@gmail.com> wrote: > > > > > > some unknown part of mesa's x86 assembly > > > code is broken by readonly text segments > > > > > > [ YOCTO #12918 ] > > > > > > Signed-off-by: Khem Raj <raj.khem@gmail.com> > > > --- > > > meta/recipes-graphics/mesa/mesa.inc | 2 +- > > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > > > diff --git a/meta/recipes-graphics/mesa/mesa.inc > > b/meta/recipes-graphics/mesa/mesa.inc > > > index 8d0e2cb67c..e6f3712706 100644 > > > --- a/meta/recipes-graphics/mesa/mesa.inc > > > +++ b/meta/recipes-graphics/mesa/mesa.inc > > > @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = > > "--enable-${OSMESA},--disable-${OSMESA}" > > > PACKAGECONFIG[unwind] = > > "--enable-libunwind,--disable-libunwind,libunwind" > > > > > > EXTRA_OECONF_remove_libc-musl = "--enable-glx-tls" > > > -EXTRA_OECONF_append_libc-musl = " --disable-glx-tls" > > > +EXTRA_OECONF_append_libc-musl = " --disable-asm --disable-glx-tls" > > > # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2) > > > FULL_OPTIMIZATION_append = " -fno-omit-frame-pointer" > > > > > > -- > > > 2.19.0 > > > > > > -- > > > _______________________________________________ > > > Openembedded-core mailing list > > > Openembedded-core@lists.openembedded.org > > > http://lists.openembedded.org/mailman/listinfo/openembedded-core > > > -- Paul Eggleton Intel Open Source Technology Centre -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
diff --git a/meta/recipes-graphics/mesa/mesa.inc b/meta/recipes-graphics/mesa/mesa.inc index 8d0e2cb67c..e6f3712706 100644 --- a/meta/recipes-graphics/mesa/mesa.inc +++ b/meta/recipes-graphics/mesa/mesa.inc @@ -107,7 +107,7 @@ PACKAGECONFIG[osmesa] = "--enable-${OSMESA},--disable-${OSMESA}" PACKAGECONFIG[unwind] = "--enable-libunwind,--disable-libunwind,libunwind" EXTRA_OECONF_remove_libc-musl = "--enable-glx-tls" -EXTRA_OECONF_append_libc-musl = " --disable-glx-tls" +EXTRA_OECONF_append_libc-musl = " --disable-asm --disable-glx-tls" # llvmpipe is slow if compiled with -fomit-frame-pointer (e.g. -O2) FULL_OPTIMIZATION_append = " -fno-omit-frame-pointer"
some unknown part of mesa's x86 assembly code is broken by readonly text segments [ YOCTO #12918 ] Signed-off-by: Khem Raj <raj.khem@gmail.com> --- meta/recipes-graphics/mesa/mesa.inc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) -- 2.19.0 -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core