alpha-1.ports.openbsd.org bulk build report

2016-11-05 Thread phessler 
bulk build on alpha-1.ports.openbsd.org
started on  Tue Oct 18 14:10:28 MDT 2016
finished at Thu Nov 3 22:23:04 MDT 2016
lasted 17D01h12m
done with kern.version=OpenBSD 6.0-current (GENERIC.MP) #0: Tue Oct 18 02:34:50 
MDT 2016

built packages:7528
Oct 18:421
Oct 19:327
Oct 20:392
Oct 21:175
Oct 22:531
Oct 23:301
Oct 24:257
Oct 25:172
Oct 26:268
Oct 27:209
Oct 28:231
Oct 29:257
Oct 30:304
Oct 31:308
Nov 1:377
Nov 2:437
Nov 3:2560

build failures: 19
http://build-failures.rhaalovely.net//alpha/2016-10-18/audio/moc.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/audio/ocp.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/cad/openscad.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/devel/ctftools.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/devel/elftoolchain.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/devel/reposurgeon.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/devel/ti-msp430gcc.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/games/brumbrumrally.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/graphics/freeimage.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/lang/gcc/4.6.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/lang/php/5.5.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/lang/php/5.6.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/lang/php/7.0.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/math/octave.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/multimedia/gstreamer1/plugins-bad.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/net/icinga/core2,-main.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/net/rtorrent.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/print/lyx.log
http://build-failures.rhaalovely.net//alpha/2016-10-18/x11/wmii.log
Base libs:
c.89.2 c.89.2.a crypto.38.1 curses.14.0 edit.5.2 event.4.1 expat.11.0
form.6.0 formw.6.0 fuse.1.1 iberty.12.0 m.10.0 menu.6.0 menuw.6.0
ncurses.14.0 ncursesw.14.0 objc.6.0 ossaudio.4.0 panel.6.0 panelw.6.0
pcap.8.1 perl.17.1 pthread.23.0 radius.1.0 readline.4.0 rpcsvc.2.0
skey.6.0 sndio.6.1 ssl.39.1 stdc++.57.0 termcap.14.0 .14.0 tls.11.4
usbhid.7.0 util.12.1 z.5.0 kvm.16.2

X libs:
FS.10.0 ICE.10.0 SM.9.0 X11-xcb.2.0 X11.16.1 XRes.5.0 Xau.10.0 Xaw.15.0
Xaw7.15.0 Xcomposite.4.0 Xcursor.5.0 Xdamage.4.0 Xdmcp.11.0 Xext.13.0
Xfixes.6.0 EGL.1.0 Xfontcache.5.0 GL.17.0 Xi.12.1 Xinerama.6.0 Xmu.11.0
Xmuu.6.0 Xpm.9.0 GLESv1_CM.1.0 Xrender.6.0 Xss.6.0 Xt.11.0 Xtst.11.0
Xv.6.0 XvMC.6.0 XvMCW.2.0 Xxf86dga.6.0 Xxf86misc.6.0 Xxf86vm.6.0 dmx.2.0
GLESv2.1.1 fontenc.4.0 GLU.9.0 pciaccess.2.0 pixman-1.32.6
pthread-stubs.2.0 xcb-composite.1.0 xcb-cursor.0.0 xcb-damage.1.0
xcb-dpms.1.0 xcb-dri2.1.1 xcb-dri3.0.0 xcb-ewmh.2.0 xcb-icccm.4.0
xcb-image.2.0 xcb-keysyms.3.0 xcb-render-util.2.0 xcb-res.1.1
xcb-shm.1.1 xcb-util.0.0 xcb-xevie.1.0 xcb-xf86dri.2.0 xcb-xinerama.1.0
xcb-xtest.1.0 xcb-xvmc.1.0 xkbfile.6.0 xkbui.5.0 GLw.6.0 OSMesa.10.0
drm_radeon.4.0 Xfont.13.0 Xft.10.0 Xrandr.7.1 epoxy.2.0 gbm.0.0
glapi.0.2 drm_amdgpu.1.1 drm_nouveau.3.0 txc_dxtn.0.0 fontconfig.11.0
freetype.26.0 xcb-glx.1.1 xcb-present.0.1 xcb-randr.2.2 xcb-record.1.1
xcb-render.1.1 xcb-screensaver.1.1 xcb-shape.1.1 xcb-sync.1.2
xcb-xfixes.1.2 xcb-xkb.0.1 xcb-xv.1.1 xcb-xprint.3.0 xcb.4.0 drm.7.2

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-19 Thread Sebastian Reitenbach 

On 02/19/16 08:18, Sebastian Reitenbach wrote:

Hi,

Am 2/18/2016 um 9:15 PM schrieb Landry Breuil:

On Wed, Feb 17, 2016 at 10:08:40PM +0100, Matthieu Herrb wrote:

On Wed, Feb 17, 2016 at 09:46:30PM +0100, Matthieu Herrb wrote:

On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:

On 2016-02-14, Matthieu Herrb  wrote:


I've taken the list of individual optimisation from gcc(1) for both -O1
and -O2 and replaced -O2 by this in CFLAGS.

gcc-local(1) is also worth a look...


+O2= ${O1} -fthread-jumps -fcrossjumping \
+-foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
+-fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
+-fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
+-fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
+-fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
+-freorder-functions -falign-functions -falign-jumps -falign-loops \
+-falign-labels -ftree-vrp -ftree-pre

Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
-ftree-vrp.


Here's a new patch, taking your remarks and jsg's advice on
USER_CFLAGS into account. I can't test it so please someone with an
alpha apply it and try a full xenocara build (see release(8)) with
it. Thanks.

Oops missed one important chunk. Use this one instead

Fwiw, mlarkin@ built xenocara with this patch on his alpha (thanks!),
and this successfully build libGL, libGLU, and even built games/abuse
and graphics/lsys without issues, so this is clearly an improvement.

I also finally got libGL and libGLU built with the patch. just going to build 
some ports,
now.


built some x11/gnustep/ stuff. that worked out well, and works.

Sebastian



Sebastian




Landry

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-18 Thread Sebastian Reitenbach 

Hi,

Am 2/18/2016 um 9:15 PM schrieb Landry Breuil:

On Wed, Feb 17, 2016 at 10:08:40PM +0100, Matthieu Herrb wrote:

On Wed, Feb 17, 2016 at 09:46:30PM +0100, Matthieu Herrb wrote:

On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:

On 2016-02-14, Matthieu Herrb  wrote:


I've taken the list of individual optimisation from gcc(1) for both -O1
and -O2 and replaced -O2 by this in CFLAGS.

gcc-local(1) is also worth a look...


+O2= ${O1} -fthread-jumps -fcrossjumping \
+   -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
+   -fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
+   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
+   -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
+   -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
+   -freorder-functions -falign-functions -falign-jumps -falign-loops \
+   -falign-labels -ftree-vrp -ftree-pre

Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
-ftree-vrp.


Here's a new patch, taking your remarks and jsg's advice on
USER_CFLAGS into account. I can't test it so please someone with an
alpha apply it and try a full xenocara build (see release(8)) with
it. Thanks.

Oops missed one important chunk. Use this one instead

Fwiw, mlarkin@ built xenocara with this patch on his alpha (thanks!),
and this successfully build libGL, libGLU, and even built games/abuse
and graphics/lsys without issues, so this is clearly an improvement.
I also finally got libGL and libGLU built with the patch. just going to 
build some ports,

now.

Sebastian




Landry

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-18 Thread Landry Breuil 
On Wed, Feb 17, 2016 at 10:08:40PM +0100, Matthieu Herrb wrote:
> On Wed, Feb 17, 2016 at 09:46:30PM +0100, Matthieu Herrb wrote:
> > On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:
> > > On 2016-02-14, Matthieu Herrb  wrote:
> > > 
> > > > I've taken the list of individual optimisation from gcc(1) for both -O1
> > > > and -O2 and replaced -O2 by this in CFLAGS.
> > > 
> > > gcc-local(1) is also worth a look...
> > > 
> > > > +O2= ${O1} -fthread-jumps -fcrossjumping \
> > > > +   -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
> > > > +   -fgcse  -fgcse-lm -fexpensive-optimizations 
> > > > -frerun-cse-after-loop \
> > > > +   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> > > > +   -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
> > > > +   -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks 
> > > > \
> > > > +   -freorder-functions -falign-functions -falign-jumps 
> > > > -falign-loops \
> > > > +   -falign-labels -ftree-vrp -ftree-pre
> > > 
> > > Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
> > > -ftree-vrp.
> > > 
> > 
> > Here's a new patch, taking your remarks and jsg's advice on
> > USER_CFLAGS into account. I can't test it so please someone with an
> > alpha apply it and try a full xenocara build (see release(8)) with
> > it. Thanks.
> 
> Oops missed one important chunk. Use this one instead

Fwiw, mlarkin@ built xenocara with this patch on his alpha (thanks!),
and this successfully build libGL, libGLU, and even built games/abuse
and graphics/lsys without issues, so this is clearly an improvement.

Landry


pgpsQwoi3pUsZ.pgp
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-17 Thread Matthieu Herrb 
On Thu, Feb 18, 2016 at 10:23:57AM +1100, Jonathan Gray wrote:
> On Wed, Feb 17, 2016 at 10:08:40PM +0100, Matthieu Herrb wrote:
> > On Wed, Feb 17, 2016 at 09:46:30PM +0100, Matthieu Herrb wrote:
> > > On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:
> > > > On 2016-02-14, Matthieu Herrb  wrote:
> > > > 
> > > > > I've taken the list of individual optimisation from gcc(1) for both 
> > > > > -O1
> > > > > and -O2 and replaced -O2 by this in CFLAGS.
> > > > 
> > > > gcc-local(1) is also worth a look...
> > > > 
> > > > > +O2= ${O1} -fthread-jumps -fcrossjumping \
> > > > > + -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
> > > > > + -fgcse  -fgcse-lm -fexpensive-optimizations 
> > > > > -frerun-cse-after-loop \
> > > > > + -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> > > > > + -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
> > > > > + -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks 
> > > > > \
> > > > > + -freorder-functions -falign-functions -falign-jumps 
> > > > > -falign-loops \
> > > > > + -falign-labels -ftree-vrp -ftree-pre
> > > > 
> > > > Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
> > > > -ftree-vrp.
> > > > 
> > > 
> > > Here's a new patch, taking your remarks and jsg's advice on
> > > USER_CFLAGS into account. I can't test it so please someone with an
> > > alpha apply it and try a full xenocara build (see release(8)) with
> > > it. Thanks.
> > 
> > Oops missed one important chunk. Use this one instead
> 
> Shouldn't this be 
> 
> CONFIGURE_ARGS+=USER_CFLAGS="-O0 ${O2}"
> 
> like the i386 portion of Makefile.bsd-wrapper dropping the
> bsd.xorg.mk part?

That would work too.
> 
> > 
> > Index: lib/mesa/Makefile.bsd-wrapper
> > ===
> > RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.bsd-wrapper,v
> > retrieving revision 1.4
> > diff -u -p -u -r1.4 Makefile.bsd-wrapper
> > --- lib/mesa/Makefile.bsd-wrapper   20 Dec 2015 12:31:01 -  1.4
> > +++ lib/mesa/Makefile.bsd-wrapper   17 Feb 2016 21:06:03 -
> > @@ -18,6 +18,25 @@ DRI_DRIVERS=swrast,radeon,r200
> >  GALLIUM_DRIVERS=swrast,r300,r600
> >  .endif
> >  
> > +.if ${MACHINE} == alpha
> > +O1= -fdefer-pop  -fguess-branch-probability -fcprop-registers \
> > +   -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce \
> > +   -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra \
> > +   -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time \
> > +   -fmerge-constants -fomit-frame-pointer
> > +
> > +O2= ${O1} -fthread-jumps -fcrossjumping -foptimize-sibling-calls \
> > +   -fcse-follow-jumps -fcse-skip-blocks -fgcse -fgcse-lm \
> > +   -fexpensive-optimizations -frerun-cse-after-loop \
> > +   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> > +   -fsched-interblock  -fsched-spec -fregmove \
> > +   -fdelete-null-pointer-checks -freorder-blocks \
> > +   -freorder-functions -falign-functions -falign-jumps -falign-loops \
> > +   -falign-labels -ftree-pre
> > +
> > +USER_CFLAGS= -O0 ${O2}
> > +.endif
> > +
> >  CONFIGURE_ARGS=--with-dri-drivers=${DRI_DRIVERS} \
> > --with-gallium-drivers=${GALLIUM_DRIVERS} \
> > --disable-silent-rules \
> > Index: lib/mesa/src/util/u_atomic.h
> > ===
> > RCS file: /cvs/OpenBSD/xenocara/lib/mesa/src/util/u_atomic.h,v
> > retrieving revision 1.2
> > diff -u -p -u -r1.2 u_atomic.h
> > --- lib/mesa/src/util/u_atomic.h22 Nov 2015 03:08:22 -  1.2
> > +++ lib/mesa/src/util/u_atomic.h17 Feb 2016 21:06:06 -
> > @@ -24,7 +24,8 @@
> >  #elif defined(_MSC_VER)
> >  #define PIPE_ATOMIC_MSVC_INTRINSIC
> >  #elif defined(__ARM_ARCH_4__) || defined(__ARM_ARCH_5__) || \
> > -  defined(__hppa__) || defined(__sparc__) || defined(__sh__)
> > +   defined(__hppa__) || defined(__sparc__) || defined(__sh__) \
> > +   || defined(__alpha__)
> >  #define PIPE_ATOMIC_UNLOCKED
> >  #elif defined(__GNUC__)
> >  #define PIPE_ATOMIC_GCC_INTRINSIC
> > Index: share/mk/bsd.xconf.mk
> > ===
> > RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xconf.mk,v
> > retrieving revision 1.31
> > diff -u -p -u -r1.31 bsd.xconf.mk
> > --- share/mk/bsd.xconf.mk   7 Feb 2016 10:06:50 -   1.31
> > +++ share/mk/bsd.xconf.mk   17 Feb 2016 21:06:06 -
> > @@ -8,7 +8,7 @@ XENOCARA_HAVE_SHARED_LIBS?=yes
> >  .endif
> >  
> >  # Build GL libs and apps?
> > -.if ${MACHINE_ARCH} == "m88k" || ${MACHINE_ARCH} == "alpha"
> > +.if ${MACHINE_ARCH} == "m88k"
> >  XENOCARA_BUILD_GL?=no
> >  XENOCARA_BUILD_DRI?=no
> >  .else
> > Index: share/mk/bsd.xorg.mk
> > ===
> > RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xorg.mk,v
> > retrieving

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-17 Thread Sebastian Reitenbach 

On 02/17/16 22:08, Matthieu Herrb wrote:

On Wed, Feb 17, 2016 at 09:46:30PM +0100, Matthieu Herrb wrote:

On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:

On 2016-02-14, Matthieu Herrb  wrote:


I've taken the list of individual optimisation from gcc(1) for both -O1
and -O2 and replaced -O2 by this in CFLAGS.


gcc-local(1) is also worth a look...


+O2= ${O1} -fthread-jumps -fcrossjumping \
+   -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
+   -fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
+   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
+   -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
+   -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
+   -freorder-functions -falign-functions -falign-jumps -falign-loops \
+   -falign-labels -ftree-vrp -ftree-pre


Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
-ftree-vrp.



Here's a new patch, taking your remarks and jsg's advice on
USER_CFLAGS into account. I can't test it so please someone with an
alpha apply it and try a full xenocara build (see release(8)) with
it. Thanks.


Oops missed one important chunk. Use this one instead


I'm going to throw that onto my Alpha. With a bit of luck, i.e. disk
not to small or other oddities, I'll report back tomorrow (:

Sebastian



Index: lib/mesa/Makefile.bsd-wrapper
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.bsd-wrapper,v
retrieving revision 1.4
diff -u -p -u -r1.4 Makefile.bsd-wrapper
--- lib/mesa/Makefile.bsd-wrapper   20 Dec 2015 12:31:01 -  1.4
+++ lib/mesa/Makefile.bsd-wrapper   17 Feb 2016 21:06:03 -
@@ -18,6 +18,25 @@ DRI_DRIVERS=swrast,radeon,r200
  GALLIUM_DRIVERS=swrast,r300,r600
  .endif

+.if ${MACHINE} == alpha
+O1= -fdefer-pop  -fguess-branch-probability -fcprop-registers \
+   -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce \
+   -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra \
+   -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time \
+   -fmerge-constants -fomit-frame-pointer
+
+O2= ${O1} -fthread-jumps -fcrossjumping -foptimize-sibling-calls \
+   -fcse-follow-jumps -fcse-skip-blocks -fgcse -fgcse-lm \
+   -fexpensive-optimizations -frerun-cse-after-loop \
+   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
+   -fsched-interblock  -fsched-spec -fregmove \
+   -fdelete-null-pointer-checks -freorder-blocks \
+   -freorder-functions -falign-functions -falign-jumps -falign-loops \
+   -falign-labels -ftree-pre
+
+USER_CFLAGS= -O0 ${O2}
+.endif
+
  CONFIGURE_ARGS=   --with-dri-drivers=${DRI_DRIVERS} \
--with-gallium-drivers=${GALLIUM_DRIVERS} \
--disable-silent-rules \
Index: lib/mesa/src/util/u_atomic.h
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/src/util/u_atomic.h,v
retrieving revision 1.2
diff -u -p -u -r1.2 u_atomic.h
--- lib/mesa/src/util/u_atomic.h22 Nov 2015 03:08:22 -  1.2
+++ lib/mesa/src/util/u_atomic.h17 Feb 2016 21:06:06 -
@@ -24,7 +24,8 @@
  #elif defined(_MSC_VER)
  #define PIPE_ATOMIC_MSVC_INTRINSIC
  #elif defined(__ARM_ARCH_4__) || defined(__ARM_ARCH_5__) || \
-  defined(__hppa__) || defined(__sparc__) || defined(__sh__)
+   defined(__hppa__) || defined(__sparc__) || defined(__sh__) \
+   || defined(__alpha__)
  #define PIPE_ATOMIC_UNLOCKED
  #elif defined(__GNUC__)
  #define PIPE_ATOMIC_GCC_INTRINSIC
Index: share/mk/bsd.xconf.mk
===
RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xconf.mk,v
retrieving revision 1.31
diff -u -p -u -r1.31 bsd.xconf.mk
--- share/mk/bsd.xconf.mk   7 Feb 2016 10:06:50 -   1.31
+++ share/mk/bsd.xconf.mk   17 Feb 2016 21:06:06 -
@@ -8,7 +8,7 @@ XENOCARA_HAVE_SHARED_LIBS?=yes
  .endif

  # Build GL libs and apps?
-.if ${MACHINE_ARCH} == "m88k" || ${MACHINE_ARCH} == "alpha"
+.if ${MACHINE_ARCH} == "m88k"
  XENOCARA_BUILD_GL?=no
  XENOCARA_BUILD_DRI?=no
  .else
Index: share/mk/bsd.xorg.mk
===
RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xorg.mk,v
retrieving revision 1.51
diff -u -p -u -r1.51 bsd.xorg.mk
--- share/mk/bsd.xorg.mk24 Aug 2014 15:53:08 -  1.51
+++ share/mk/bsd.xorg.mk17 Feb 2016 21:06:06 -
@@ -57,6 +57,7 @@ XENOCARA_PATH?=   /bin:/sbin:/usr/bin:/usr
  CONFIGURE_ENV=PKG_CONFIG_LIBDIR="$(PKG_CONFIG_LIBDIR)" \
CONFIG_SITE=$(CONFIG_SITE) \
CFLAGS="$(CFLAGS:C/ *$//)" \
+   USER_CFLAGS="$(USER_CFLAGS:C/ *$//)" \
MAKE="${MAKE}"

  CONFIGURE_ARGS+= --disable-silent-rules

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-17 Thread Jonathan Gray 
On Wed, Feb 17, 2016 at 10:08:40PM +0100, Matthieu Herrb wrote:
> On Wed, Feb 17, 2016 at 09:46:30PM +0100, Matthieu Herrb wrote:
> > On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:
> > > On 2016-02-14, Matthieu Herrb  wrote:
> > > 
> > > > I've taken the list of individual optimisation from gcc(1) for both -O1
> > > > and -O2 and replaced -O2 by this in CFLAGS.
> > > 
> > > gcc-local(1) is also worth a look...
> > > 
> > > > +O2= ${O1} -fthread-jumps -fcrossjumping \
> > > > +   -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
> > > > +   -fgcse  -fgcse-lm -fexpensive-optimizations 
> > > > -frerun-cse-after-loop \
> > > > +   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> > > > +   -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
> > > > +   -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks 
> > > > \
> > > > +   -freorder-functions -falign-functions -falign-jumps 
> > > > -falign-loops \
> > > > +   -falign-labels -ftree-vrp -ftree-pre
> > > 
> > > Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
> > > -ftree-vrp.
> > > 
> > 
> > Here's a new patch, taking your remarks and jsg's advice on
> > USER_CFLAGS into account. I can't test it so please someone with an
> > alpha apply it and try a full xenocara build (see release(8)) with
> > it. Thanks.
> 
> Oops missed one important chunk. Use this one instead

Shouldn't this be 

CONFIGURE_ARGS+=USER_CFLAGS="-O0 ${O2}"

like the i386 portion of Makefile.bsd-wrapper dropping the
bsd.xorg.mk part?

> 
> Index: lib/mesa/Makefile.bsd-wrapper
> ===
> RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.bsd-wrapper,v
> retrieving revision 1.4
> diff -u -p -u -r1.4 Makefile.bsd-wrapper
> --- lib/mesa/Makefile.bsd-wrapper 20 Dec 2015 12:31:01 -  1.4
> +++ lib/mesa/Makefile.bsd-wrapper 17 Feb 2016 21:06:03 -
> @@ -18,6 +18,25 @@ DRI_DRIVERS=swrast,radeon,r200
>  GALLIUM_DRIVERS=swrast,r300,r600
>  .endif
>  
> +.if ${MACHINE} == alpha
> +O1= -fdefer-pop  -fguess-branch-probability -fcprop-registers \
> + -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce \
> + -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra \
> + -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time \
> + -fmerge-constants -fomit-frame-pointer
> +
> +O2= ${O1} -fthread-jumps -fcrossjumping -foptimize-sibling-calls \
> + -fcse-follow-jumps -fcse-skip-blocks -fgcse -fgcse-lm \
> + -fexpensive-optimizations -frerun-cse-after-loop \
> + -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> + -fsched-interblock  -fsched-spec -fregmove \
> + -fdelete-null-pointer-checks -freorder-blocks \
> + -freorder-functions -falign-functions -falign-jumps -falign-loops \
> + -falign-labels -ftree-pre
> +
> +USER_CFLAGS= -O0 ${O2}
> +.endif
> +
>  CONFIGURE_ARGS=  --with-dri-drivers=${DRI_DRIVERS} \
>   --with-gallium-drivers=${GALLIUM_DRIVERS} \
>   --disable-silent-rules \
> Index: lib/mesa/src/util/u_atomic.h
> ===
> RCS file: /cvs/OpenBSD/xenocara/lib/mesa/src/util/u_atomic.h,v
> retrieving revision 1.2
> diff -u -p -u -r1.2 u_atomic.h
> --- lib/mesa/src/util/u_atomic.h  22 Nov 2015 03:08:22 -  1.2
> +++ lib/mesa/src/util/u_atomic.h  17 Feb 2016 21:06:06 -
> @@ -24,7 +24,8 @@
>  #elif defined(_MSC_VER)
>  #define PIPE_ATOMIC_MSVC_INTRINSIC
>  #elif defined(__ARM_ARCH_4__) || defined(__ARM_ARCH_5__) || \
> -  defined(__hppa__) || defined(__sparc__) || defined(__sh__)
> + defined(__hppa__) || defined(__sparc__) || defined(__sh__) \
> + || defined(__alpha__)
>  #define PIPE_ATOMIC_UNLOCKED
>  #elif defined(__GNUC__)
>  #define PIPE_ATOMIC_GCC_INTRINSIC
> Index: share/mk/bsd.xconf.mk
> ===
> RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xconf.mk,v
> retrieving revision 1.31
> diff -u -p -u -r1.31 bsd.xconf.mk
> --- share/mk/bsd.xconf.mk 7 Feb 2016 10:06:50 -   1.31
> +++ share/mk/bsd.xconf.mk 17 Feb 2016 21:06:06 -
> @@ -8,7 +8,7 @@ XENOCARA_HAVE_SHARED_LIBS?=yes
>  .endif
>  
>  # Build GL libs and apps?
> -.if ${MACHINE_ARCH} == "m88k" || ${MACHINE_ARCH} == "alpha"
> +.if ${MACHINE_ARCH} == "m88k"
>  XENOCARA_BUILD_GL?=no
>  XENOCARA_BUILD_DRI?=no
>  .else
> Index: share/mk/bsd.xorg.mk
> ===
> RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xorg.mk,v
> retrieving revision 1.51
> diff -u -p -u -r1.51 bsd.xorg.mk
> --- share/mk/bsd.xorg.mk  24 Aug 2014 15:53:08 -  1.51
> +++ share/mk/bsd.xorg.mk  17 Feb 2016 21:06:06 -
> @@ -57,6 +57,7 @@ XENOCARA_PATH?= /bin:/sbin:/usr/bin:/usr
>  CONFIGURE_ENV=

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-17 Thread Matthieu Herrb 
On Wed, Feb 17, 2016 at 09:46:30PM +0100, Matthieu Herrb wrote:
> On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:
> > On 2016-02-14, Matthieu Herrb  wrote:
> > 
> > > I've taken the list of individual optimisation from gcc(1) for both -O1
> > > and -O2 and replaced -O2 by this in CFLAGS.
> > 
> > gcc-local(1) is also worth a look...
> > 
> > > +O2= ${O1} -fthread-jumps -fcrossjumping \
> > > + -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
> > > + -fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
> > > + -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> > > + -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
> > > + -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
> > > + -freorder-functions -falign-functions -falign-jumps -falign-loops \
> > > + -falign-labels -ftree-vrp -ftree-pre
> > 
> > Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
> > -ftree-vrp.
> > 
> 
> Here's a new patch, taking your remarks and jsg's advice on
> USER_CFLAGS into account. I can't test it so please someone with an
> alpha apply it and try a full xenocara build (see release(8)) with
> it. Thanks.

Oops missed one important chunk. Use this one instead

Index: lib/mesa/Makefile.bsd-wrapper
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.bsd-wrapper,v
retrieving revision 1.4
diff -u -p -u -r1.4 Makefile.bsd-wrapper
--- lib/mesa/Makefile.bsd-wrapper   20 Dec 2015 12:31:01 -  1.4
+++ lib/mesa/Makefile.bsd-wrapper   17 Feb 2016 21:06:03 -
@@ -18,6 +18,25 @@ DRI_DRIVERS=swrast,radeon,r200
 GALLIUM_DRIVERS=swrast,r300,r600
 .endif
 
+.if ${MACHINE} == alpha
+O1= -fdefer-pop  -fguess-branch-probability -fcprop-registers \
+   -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce \
+   -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra \
+   -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time \
+   -fmerge-constants -fomit-frame-pointer
+
+O2= ${O1} -fthread-jumps -fcrossjumping -foptimize-sibling-calls \
+   -fcse-follow-jumps -fcse-skip-blocks -fgcse -fgcse-lm \
+   -fexpensive-optimizations -frerun-cse-after-loop \
+   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
+   -fsched-interblock  -fsched-spec -fregmove \
+   -fdelete-null-pointer-checks -freorder-blocks \
+   -freorder-functions -falign-functions -falign-jumps -falign-loops \
+   -falign-labels -ftree-pre
+
+USER_CFLAGS= -O0 ${O2}
+.endif
+
 CONFIGURE_ARGS=--with-dri-drivers=${DRI_DRIVERS} \
--with-gallium-drivers=${GALLIUM_DRIVERS} \
--disable-silent-rules \
Index: lib/mesa/src/util/u_atomic.h
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/src/util/u_atomic.h,v
retrieving revision 1.2
diff -u -p -u -r1.2 u_atomic.h
--- lib/mesa/src/util/u_atomic.h22 Nov 2015 03:08:22 -  1.2
+++ lib/mesa/src/util/u_atomic.h17 Feb 2016 21:06:06 -
@@ -24,7 +24,8 @@
 #elif defined(_MSC_VER)
 #define PIPE_ATOMIC_MSVC_INTRINSIC
 #elif defined(__ARM_ARCH_4__) || defined(__ARM_ARCH_5__) || \
-  defined(__hppa__) || defined(__sparc__) || defined(__sh__)
+   defined(__hppa__) || defined(__sparc__) || defined(__sh__) \
+   || defined(__alpha__)
 #define PIPE_ATOMIC_UNLOCKED
 #elif defined(__GNUC__)
 #define PIPE_ATOMIC_GCC_INTRINSIC
Index: share/mk/bsd.xconf.mk
===
RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xconf.mk,v
retrieving revision 1.31
diff -u -p -u -r1.31 bsd.xconf.mk
--- share/mk/bsd.xconf.mk   7 Feb 2016 10:06:50 -   1.31
+++ share/mk/bsd.xconf.mk   17 Feb 2016 21:06:06 -
@@ -8,7 +8,7 @@ XENOCARA_HAVE_SHARED_LIBS?=yes
 .endif
 
 # Build GL libs and apps?
-.if ${MACHINE_ARCH} == "m88k" || ${MACHINE_ARCH} == "alpha"
+.if ${MACHINE_ARCH} == "m88k"
 XENOCARA_BUILD_GL?=no
 XENOCARA_BUILD_DRI?=no
 .else
Index: share/mk/bsd.xorg.mk
===
RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xorg.mk,v
retrieving revision 1.51
diff -u -p -u -r1.51 bsd.xorg.mk
--- share/mk/bsd.xorg.mk24 Aug 2014 15:53:08 -  1.51
+++ share/mk/bsd.xorg.mk17 Feb 2016 21:06:06 -
@@ -57,6 +57,7 @@ XENOCARA_PATH?=   /bin:/sbin:/usr/bin:/usr
 CONFIGURE_ENV= PKG_CONFIG_LIBDIR="$(PKG_CONFIG_LIBDIR)" \
CONFIG_SITE=$(CONFIG_SITE) \
CFLAGS="$(CFLAGS:C/ *$//)" \
+   USER_CFLAGS="$(USER_CFLAGS:C/ *$//)" \
MAKE="${MAKE}"
 
 CONFIGURE_ARGS+= --disable-silent-rules

-- 
Matthieu Herrb


signature.asc
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-17 Thread Matthieu Herrb 
On Mon, Feb 15, 2016 at 03:44:02PM +, Christian Weisgerber wrote:
> On 2016-02-14, Matthieu Herrb  wrote:
> 
> > I've taken the list of individual optimisation from gcc(1) for both -O1
> > and -O2 and replaced -O2 by this in CFLAGS.
> 
> gcc-local(1) is also worth a look...
> 
> > +O2= ${O1} -fthread-jumps -fcrossjumping \
> > +   -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
> > +   -fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
> > +   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> > +   -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
> > +   -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
> > +   -freorder-functions -falign-functions -falign-jumps -falign-loops \
> > +   -falign-labels -ftree-vrp -ftree-pre
> 
> Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
> -ftree-vrp.
> 

Here's a new patch, taking your remarks and jsg's advice on
USER_CFLAGS into account. I can't test it so please someone with an
alpha apply it and try a full xenocara build (see release(8)) with
it. Thanks.

Index: share/mk/bsd.xorg.mk
===
RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xorg.mk,v
retrieving revision 1.51
diff -u -p -u -r1.51 bsd.xorg.mk
--- share/mk/bsd.xorg.mk24 Aug 2014 15:53:08 -  1.51
+++ share/mk/bsd.xorg.mk17 Feb 2016 20:44:59 -
@@ -57,6 +57,7 @@ XENOCARA_PATH?=   /bin:/sbin:/usr/bin:/usr
 CONFIGURE_ENV= PKG_CONFIG_LIBDIR="$(PKG_CONFIG_LIBDIR)" \
CONFIG_SITE=$(CONFIG_SITE) \
CFLAGS="$(CFLAGS:C/ *$//)" \
+   USER_CFLAGS="$(USER_CFLAGS:C/ *$//)" \
MAKE="${MAKE}"
 
 CONFIGURE_ARGS+= --disable-silent-rules
Index: lib/mesa/Makefile.bsd-wrapper
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.bsd-wrapper,v
retrieving revision 1.4
diff -u -p -u -r1.4 Makefile.bsd-wrapper
--- lib/mesa/Makefile.bsd-wrapper   20 Dec 2015 12:31:01 -  1.4
+++ lib/mesa/Makefile.bsd-wrapper   17 Feb 2016 20:44:59 -
@@ -18,6 +18,25 @@ DRI_DRIVERS=swrast,radeon,r200
 GALLIUM_DRIVERS=swrast,r300,r600
 .endif
 
+.if ${MACHINE} == alpha
+O1= -fdefer-pop  -fguess-branch-probability -fcprop-registers \
+   -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce \
+   -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra \
+   -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time \
+   -fmerge-constants -fomit-frame-pointer
+
+O2= ${O1} -fthread-jumps -fcrossjumping -foptimize-sibling-calls \
+   -fcse-follow-jumps -fcse-skip-blocks -fgcse -fgcse-lm \
+   -fexpensive-optimizations -frerun-cse-after-loop \
+   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
+   -fsched-interblock  -fsched-spec -fregmove \
+   -fdelete-null-pointer-checks -freorder-blocks \
+   -freorder-functions -falign-functions -falign-jumps -falign-loops \
+   -falign-labels -ftree-pre
+
+USER_CFLAGS= -O0 ${O2}
+.endif
+
 CONFIGURE_ARGS=--with-dri-drivers=${DRI_DRIVERS} \
--with-gallium-drivers=${GALLIUM_DRIVERS} \
--disable-silent-rules \
Index: lib/mesa/src/util/u_atomic.h
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/src/util/u_atomic.h,v
retrieving revision 1.2
diff -u -p -u -r1.2 u_atomic.h
--- lib/mesa/src/util/u_atomic.h22 Nov 2015 03:08:22 -  1.2
+++ lib/mesa/src/util/u_atomic.h17 Feb 2016 20:45:06 -
@@ -24,7 +24,8 @@
 #elif defined(_MSC_VER)
 #define PIPE_ATOMIC_MSVC_INTRINSIC
 #elif defined(__ARM_ARCH_4__) || defined(__ARM_ARCH_5__) || \
-  defined(__hppa__) || defined(__sparc__) || defined(__sh__)
+   defined(__hppa__) || defined(__sparc__) || defined(__sh__) \
+   || defined(__alpha__)
 #define PIPE_ATOMIC_UNLOCKED
 #elif defined(__GNUC__)
 #define PIPE_ATOMIC_GCC_INTRINSIC

-- 
Matthieu Herrb


signature.asc
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-15 Thread Christian Weisgerber 
On 2016-02-14, Matthieu Herrb  wrote:

> I've taken the list of individual optimisation from gcc(1) for both -O1
> and -O2 and replaced -O2 by this in CFLAGS.

gcc-local(1) is also worth a look...

> +O2= ${O1} -fthread-jumps -fcrossjumping \
> + -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
> + -fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
> + -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> + -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
> + -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
> + -freorder-functions -falign-functions -falign-jumps -falign-loops \
> + -falign-labels -ftree-vrp -ftree-pre

Our -O2 does not include -fstrict-aliasing, -fstrict-overflow,
-ftree-vrp.

-- 
Christian "naddy" Weisgerber  na...@mips.inka.de

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-14 Thread Jonathan Gray 
On Sun, Feb 14, 2016 at 08:57:10PM +0100, Matthieu Herrb wrote:
> On Sun, Feb 14, 2016 at 11:17:24PM +1100, Jonathan Gray wrote:
> > I would prefer if someone with time and an alpha can figure
> > out which optimisation flag breaks, then all of mesa
> > could be built with that disabled on alpha which would be
> > a much smaller diff.
> 
> Ok here is a new less intrusive diff. I don't understand why it works
> but it allows to build mesa on alpha.
> 
> I've taken the list of individual optimisation from gcc(1) for both -O1
> and -O2 and replaced -O2 by this in CFLAGS. My plan was to try to
> remove one after the other until it built, but both problematic files
> build ok with that combinaison.
> 
> I've only removed -fdelayed-branch  since gcc says that it's not supported
> on this architecture.

But this will break when building all of xenocara and not just
mesa if the i386 case is anything to go by, because of the cached
configure results.  That is why it uses USER_CFLAGS/USER_CXXFLAGS.

> 
> Index: share/mk/bsd.xconf.mk
> ===
> RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xconf.mk,v
> retrieving revision 1.30
> diff -u -p -u -r1.30 bsd.xconf.mk
> --- share/mk/bsd.xconf.mk 5 Feb 2016 13:59:11 -   1.30
> +++ share/mk/bsd.xconf.mk 14 Feb 2016 19:36:56 -
> @@ -8,7 +8,7 @@ XENOCARA_HAVE_SHARED_LIBS?=yes
>  .endif
>  
>  # Build GL libs and apps?
> -.if ${MACHINE_ARCH} == "m88k" || ${MACHINE_ARCH} == "alpha"
> +.if ${MACHINE_ARCH} == "m88k"
>  XENOCARA_BUILD_GL?=no
>  XENOCARA_BUILD_DRI?=no
>  .else
> Index: lib/mesa/Makefile.bsd-wrapper
> ===
> RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.bsd-wrapper,v
> retrieving revision 1.4
> diff -u -p -u -r1.4 Makefile.bsd-wrapper
> --- lib/mesa/Makefile.bsd-wrapper 20 Dec 2015 12:31:01 -  1.4
> +++ lib/mesa/Makefile.bsd-wrapper 14 Feb 2016 19:36:56 -
> @@ -18,6 +18,25 @@ DRI_DRIVERS=swrast,radeon,r200
>  GALLIUM_DRIVERS=swrast,r300,r600
>  .endif
>  
> +.if ${MACHINE} == alpha
> +O1= -fdefer-pop  -fguess-branch-probability -fcprop-registers \
> + -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce \
> + -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra \
> + -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time \
> + -fmerge-constants -fomit-frame-pointer
> +
> +O2= ${O1} -fthread-jumps -fcrossjumping \
> + -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
> + -fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
> + -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
> + -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
> + -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
> + -freorder-functions -falign-functions -falign-jumps -falign-loops \
> + -falign-labels -ftree-vrp -ftree-pre
> +
> +CFLAGS= ${O2} -pipe
> +.endif
> +
>  CONFIGURE_ARGS=  --with-dri-drivers=${DRI_DRIVERS} \
>   --with-gallium-drivers=${GALLIUM_DRIVERS} \
>   --disable-silent-rules \
> -- 
> Matthieu Herrb

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-14 Thread Matthieu Herrb 
On Sun, Feb 14, 2016 at 11:17:24PM +1100, Jonathan Gray wrote:
> I would prefer if someone with time and an alpha can figure
> out which optimisation flag breaks, then all of mesa
> could be built with that disabled on alpha which would be
> a much smaller diff.

Ok here is a new less intrusive diff. I don't understand why it works
but it allows to build mesa on alpha.

I've taken the list of individual optimisation from gcc(1) for both -O1
and -O2 and replaced -O2 by this in CFLAGS. My plan was to try to
remove one after the other until it built, but both problematic files
build ok with that combinaison.

I've only removed -fdelayed-branch  since gcc says that it's not supported
on this architecture.

Index: share/mk/bsd.xconf.mk
===
RCS file: /cvs/OpenBSD/xenocara/share/mk/bsd.xconf.mk,v
retrieving revision 1.30
diff -u -p -u -r1.30 bsd.xconf.mk
--- share/mk/bsd.xconf.mk   5 Feb 2016 13:59:11 -   1.30
+++ share/mk/bsd.xconf.mk   14 Feb 2016 19:36:56 -
@@ -8,7 +8,7 @@ XENOCARA_HAVE_SHARED_LIBS?=yes
 .endif
 
 # Build GL libs and apps?
-.if ${MACHINE_ARCH} == "m88k" || ${MACHINE_ARCH} == "alpha"
+.if ${MACHINE_ARCH} == "m88k"
 XENOCARA_BUILD_GL?=no
 XENOCARA_BUILD_DRI?=no
 .else
Index: lib/mesa/Makefile.bsd-wrapper
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.bsd-wrapper,v
retrieving revision 1.4
diff -u -p -u -r1.4 Makefile.bsd-wrapper
--- lib/mesa/Makefile.bsd-wrapper   20 Dec 2015 12:31:01 -  1.4
+++ lib/mesa/Makefile.bsd-wrapper   14 Feb 2016 19:36:56 -
@@ -18,6 +18,25 @@ DRI_DRIVERS=swrast,radeon,r200
 GALLIUM_DRIVERS=swrast,r300,r600
 .endif
 
+.if ${MACHINE} == alpha
+O1= -fdefer-pop  -fguess-branch-probability -fcprop-registers \
+   -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce \
+   -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra \
+   -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time \
+   -fmerge-constants -fomit-frame-pointer
+
+O2= ${O1} -fthread-jumps -fcrossjumping \
+   -foptimize-sibling-calls -fcse-follow-jumps -fcse-skip-blocks \
+   -fgcse  -fgcse-lm -fexpensive-optimizations -frerun-cse-after-loop \
+   -fcaller-saves -fpeephole2 -fschedule-insns -fschedule-insns2 \
+   -fsched-interblock  -fsched-spec -fregmove -fstrict-aliasing \
+   -fstrict-overflow -fdelete-null-pointer-checks -freorder-blocks \
+   -freorder-functions -falign-functions -falign-jumps -falign-loops \
+   -falign-labels -ftree-vrp -ftree-pre
+
+CFLAGS= ${O2} -pipe
+.endif
+
 CONFIGURE_ARGS=--with-dri-drivers=${DRI_DRIVERS} \
--with-gallium-drivers=${GALLIUM_DRIVERS} \
--disable-silent-rules \
-- 
Matthieu Herrb


signature.asc
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-14 Thread Jonathan Gray 
On Sun, Feb 14, 2016 at 10:43:30AM +0100, Matthieu Herrb wrote:
> On Sun, Feb 14, 2016 at 05:31:31PM +1100, Jonathan Gray wrote:
> > On Sat, Feb 13, 2016 at 07:17:35PM +0100, Matthieu Herrb wrote:
> > > So, here is a patch that should let libGL to build again on
> > > alpha. I've not done a full clean xenocara build with it yet though,
> > > given how slow my alpha is.
> > > 
> > > In addition to the non-existent sync operation, I had to force 2 files
> > > to build with -O0.
> > > 
> > > There is a lot of Makefile.in churn since I did some autotools
> > > hackery...
> > 
> > Any idea which of the -O1 options triggers it?
> > 
> > According to gcc(1):
> > 
> > -O turns on the following optimization flags: -fdefer-pop
> > -fdelayed-branch -fguess-branch-probability -fcprop-registers
> > -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce
> > -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra
> > -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time
> > -fmerge-constants
> 
> no idea. I just went directly to -O0. If someone else wants to narrow
> that down... I won't have time to try further build options in the
> coming week.

Ah, so it could be options enabled with -O2 as well then.

> 
> > 
> > So alpha is in the same situation mips64 was then?
> > 
> > Going by /sys/arch/alpha/include/atomic.h there are atomics,
> > which sync builtins does gcc not implement on alpha?
> 
> As I pointed out  in a previous mail it's p_atomic_cmpxchg() that is
> missing.
> 
> Again I went to the easy road to define PIPE_ATOMIC_UNLOCKED because I
> wanted something quick and easy and haven't seen a simple way to tell
> Mesa that it should fallback to unlocked only for specific operations.
> 
> The question is should I commit this now and let you or other people with
> alphas tweak it to get a libGL & friends available for 5.9 or do we
> postpone that after 5.9 ?

All my alphas are dead.  There are still things such as
https://gcc.gnu.org/ml/gcc-patches/2007-02/msg00722.html
that sound relevant I just can't build or test anything.

I would prefer if someone with time and an alpha can figure
out which optimisation flag breaks, then all of mesa
could be built with that disabled on alpha which would be
a much smaller diff.

Index: gnu/gcc/gcc/config/alpha/alpha.c
===
RCS file: /cvs/src/gnu/gcc/gcc/config/alpha/alpha.c,v
retrieving revision 1.4
diff -u -p -r1.4 alpha.c
--- gnu/gcc/gcc/config/alpha/alpha.c20 Dec 2012 13:58:06 -  1.4
+++ gnu/gcc/gcc/config/alpha/alpha.c14 Feb 2016 12:10:16 -
@@ -4551,6 +4551,8 @@ emit_insxl (enum machine_mode mode, rtx 
   else
fn = gen_inswl_le;
 }
+  /* The insbl and inswl patterns require a register operand.  */
+  op1 = force_reg (mode, op1);
   emit_insn (fn (ret, op1, op2));
 
   return ret;

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-14 Thread Landry Breuil 
On Sun, Feb 14, 2016 at 10:43:30AM +0100, Matthieu Herrb wrote:
> 
> The question is should I commit this now and let you or other people with
> alphas tweak it to get a libGL & friends available for 5.9 or do we
> postpone that after 5.9 ?

It all depends if we want to 'exercise the toolchain' more and build more
libGL-related packages/games/x11 things (which are by definition useless
without X server) - i doubt anyone runs gl games over ssh on alpha (or
runs anything on alpha).
Put it the other way, what could get more broken than it is now if you
commit it ?
Personally i stopped triaging/looking at alpha build failures, since 90%
relate to that. Maybe there are 'real' different failures due to other
gcc/toolchain issues. If they're worth fixing is a different topic :)

Landry


pgpOXVH7H9Ube.pgp
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-14 Thread Matthieu Herrb 
On Sun, Feb 14, 2016 at 05:31:31PM +1100, Jonathan Gray wrote:
> On Sat, Feb 13, 2016 at 07:17:35PM +0100, Matthieu Herrb wrote:
> > So, here is a patch that should let libGL to build again on
> > alpha. I've not done a full clean xenocara build with it yet though,
> > given how slow my alpha is.
> > 
> > In addition to the non-existent sync operation, I had to force 2 files
> > to build with -O0.
> > 
> > There is a lot of Makefile.in churn since I did some autotools
> > hackery...
> 
> Any idea which of the -O1 options triggers it?
> 
> According to gcc(1):
> 
> -O turns on the following optimization flags: -fdefer-pop
> -fdelayed-branch -fguess-branch-probability -fcprop-registers
> -fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce
> -ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra
> -ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time
> -fmerge-constants

no idea. I just went directly to -O0. If someone else wants to narrow
that down... I won't have time to try further build options in the
coming week.

> 
> So alpha is in the same situation mips64 was then?
> 
> Going by /sys/arch/alpha/include/atomic.h there are atomics,
> which sync builtins does gcc not implement on alpha?

As I pointed out  in a previous mail it's p_atomic_cmpxchg() that is
missing.

Again I went to the easy road to define PIPE_ATOMIC_UNLOCKED because I
wanted something quick and easy and haven't seen a simple way to tell
Mesa that it should fallback to unlocked only for specific operations.

The question is should I commit this now and let you or other people with
alphas tweak it to get a libGL & friends available for 5.9 or do we
postpone that after 5.9 ?
-- 
Matthieu Herrb


signature.asc
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-14 Thread Christian Weisgerber 
On 2016-02-14, Jonathan Gray  wrote:

>> In addition to the non-existent sync operation, I had to force 2 files
>> to build with -O0.
>
> Any idea which of the -O1 options triggers it?

-O1 is very broken on alpha and triggers ICEs a lot (always?).

-- 
Christian "naddy" Weisgerber  na...@mips.inka.de

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-13 Thread Matthieu Herrb 
So, here is a patch that should let libGL to build again on
alpha. I've not done a full clean xenocara build with it yet though,
given how slow my alpha is.

In addition to the non-existent sync operation, I had to force 2 files
to build with -O0.

There is a lot of Makefile.in churn since I did some autotools
hackery...

Index: lib/mesa/Makefile.in
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/Makefile.in,v
retrieving revision 1.3
diff -u -p -u -r1.3 Makefile.in
--- lib/mesa/Makefile.in23 Dec 2015 13:43:05 -  1.3
+++ lib/mesa/Makefile.in13 Feb 2016 15:41:54 -
@@ -322,6 +322,7 @@ OBJEXT = @OBJEXT@
 OMX_CFLAGS = @OMX_CFLAGS@
 OMX_LIBS = @OMX_LIBS@
 OMX_LIB_INSTALL_DIR = @OMX_LIB_INSTALL_DIR@
+OPENBSD_ALPHA = @OPENBSD_ALPHA@
 OPENCL_LIBNAME = @OPENCL_LIBNAME@
 OPENCL_VERSION = @OPENCL_VERSION@
 OSMESA_LIB = @OSMESA_LIB@
Index: lib/mesa/configure
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/configure,v
retrieving revision 1.4
diff -u -p -u -r1.4 configure
--- lib/mesa/configure  24 Jan 2016 01:21:23 -  1.4
+++ lib/mesa/configure  13 Feb 2016 15:41:54 -
@@ -903,6 +903,9 @@ HAVE_LD_DYNAMIC_LIST_TRUE
 HAVE_LD_VERSION_SCRIPT_FALSE
 HAVE_LD_VERSION_SCRIPT_TRUE
 LD_NO_UNDEFINED
+OPENBSD_ALPHA
+OPENBSD_ALPHA_FALSE
+OPENBSD_ALPHA_TRUE
 GC_SECTIONS
 BSYMBOLIC
 BUILD_SHARED_FALSE
@@ -19300,12 +19303,24 @@ openbsd*)
 case "$host_cpu" in
 powerpc*)
 GC_SECTIONS="" ;;
+alpha*)
+openbsd_alpha=1 ;;
 esac
 ;;
 esac
 
 
 
+ if test x$openbsd_alpha = x1; then
+  OPENBSD_ALPHA_TRUE=
+  OPENBSD_ALPHA_FALSE='#'
+else
+  OPENBSD_ALPHA_TRUE='#'
+  OPENBSD_ALPHA_FALSE=
+fi
+
+
+
 case "$host_os" in
 openbsd* | darwin* )
 LD_NO_UNDEFINED="" ;;
@@ -25658,6 +25673,10 @@ Usually this means the macro was only in
 fi
 if test -z "${BUILD_SHARED_TRUE}" && test -z "${BUILD_SHARED_FALSE}"; then
   as_fn_error $? "conditional \"BUILD_SHARED\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
+if test -z "${OPENBSD_ALPHA_TRUE}" && test -z "${OPENBSD_ALPHA_FALSE}"; then
+  as_fn_error $? "conditional \"OPENBSD_ALPHA\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${HAVE_LD_VERSION_SCRIPT_TRUE}" && test -z 
"${HAVE_LD_VERSION_SCRIPT_FALSE}"; then
Index: lib/mesa/configure.ac
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/configure.ac,v
retrieving revision 1.5
diff -u -p -u -r1.5 configure.ac
--- lib/mesa/configure.ac   23 Dec 2015 13:43:05 -  1.5
+++ lib/mesa/configure.ac   13 Feb 2016 15:41:54 -
@@ -475,11 +475,16 @@ openbsd*)
 case "$host_cpu" in
 powerpc*)
 GC_SECTIONS="" ;;
+alpha*)
+openbsd_alpha=1 ;;
 esac
 ;;
 esac
 
 AC_SUBST([GC_SECTIONS])
+
+AM_CONDITIONAL([OPENBSD_ALPHA], [test x$openbsd_alpha = x1])
+AC_SUBST([OPENBSD_ALPHA])
 
 dnl
 dnl OpenBSD does not have DT_NEEDED entries for libc by design
Index: lib/mesa/src/Makefile.in
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/src/Makefile.in,v
retrieving revision 1.3
diff -u -p -u -r1.3 Makefile.in
--- lib/mesa/src/Makefile.in23 Dec 2015 13:43:05 -  1.3
+++ lib/mesa/src/Makefile.in13 Feb 2016 15:41:55 -
@@ -332,6 +332,7 @@ OBJEXT = @OBJEXT@
 OMX_CFLAGS = @OMX_CFLAGS@
 OMX_LIBS = @OMX_LIBS@
 OMX_LIB_INSTALL_DIR = @OMX_LIB_INSTALL_DIR@
+OPENBSD_ALPHA = @OPENBSD_ALPHA@
 OPENCL_LIBNAME = @OPENCL_LIBNAME@
 OPENCL_VERSION = @OPENCL_VERSION@
 OSMESA_LIB = @OSMESA_LIB@
Index: lib/mesa/src/egl/Makefile.in
===
RCS file: /cvs/OpenBSD/xenocara/lib/mesa/src/egl/Makefile.in,v
retrieving revision 1.3
diff -u -p -u -r1.3 Makefile.in
--- lib/mesa/src/egl/Makefile.in23 Dec 2015 13:43:05 -  1.3
+++ lib/mesa/src/egl/Makefile.in13 Feb 2016 15:41:55 -
@@ -397,6 +397,7 @@ OBJEXT = @OBJEXT@
 OMX_CFLAGS = @OMX_CFLAGS@
 OMX_LIBS = @OMX_LIBS@
 OMX_LIB_INSTALL_DIR = @OMX_LIB_INSTALL_DIR@
+OPENBSD_ALPHA = @OPENBSD_ALPHA@
 OPENCL_LIBNAME = @OPENCL_LIBNAME@
 OPENCL_VERSION = @OPENCL_VERSION@
 OSMESA_LIB = @OSMESA_LIB@
Index: lib/mesa/src/egl/wayland/wayland-drm/Makefile.in
===
RCS file: 
/cvs/OpenBSD/xenocara/lib/mesa/src/egl/wayland/wayland-drm/Makefile.in,v
retrieving revision 1.3
diff -u -p -u -r1.3 Makefile.in
--- lib/mesa/src/egl/wayland/wayland-drm/Makefile.in23 Dec 2015 13:43:05 
-  1.3
+++ lib/mesa/src/egl/wayland/wayland-drm/Makefile.in13 Feb 2016 15:41:55 
-
@@ -264,6 +264,7 @@ OBJEXT = @OBJEXT@
 OMX_CFLAGS = @OMX_CFLAGS@
 OMX_LIBS = @OMX_LIBS@
 OMX_LIB_INSTALL_DIR = @OMX_LIB_INSTALL_DIR@
+OPENBSD_ALPHA = @OPENBSD_ALPHA@

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-13 Thread Jonathan Gray 
On Sat, Feb 13, 2016 at 07:17:35PM +0100, Matthieu Herrb wrote:
> So, here is a patch that should let libGL to build again on
> alpha. I've not done a full clean xenocara build with it yet though,
> given how slow my alpha is.
> 
> In addition to the non-existent sync operation, I had to force 2 files
> to build with -O0.
> 
> There is a lot of Makefile.in churn since I did some autotools
> hackery...

Any idea which of the -O1 options triggers it?

According to gcc(1):

-O turns on the following optimization flags: -fdefer-pop
-fdelayed-branch -fguess-branch-probability -fcprop-registers
-fif-conversion -fif-conversion2 -ftree-ccp -ftree-dce
-ftree-dominator-opts -ftree-dse -ftree-ter -ftree-lrs -ftree-sra
-ftree-copyrename -ftree-fre -ftree-ch -funit-at-a-time
-fmerge-constants

So alpha is in the same situation mips64 was then?

Going by /sys/arch/alpha/include/atomic.h there are atomics,
which sync builtins does gcc not implement on alpha?

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-10 Thread Matthieu Herrb 
On Wed, Feb 10, 2016 at 10:21:47AM +, Stuart Henderson wrote:
> On 2016/02/10 10:57, Karel Gardas wrote:
> > > It has been fixed upstream:
> > >
> > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66140
> > 
> > looks like in 5.1.1 which is GPL3 so probably not usable to be merge
> > into the OpenBSD's GPL2 4.2.1? But well, I'm not the lawyer here i
> > just see two different licenses so it raises red flag here...
> > 
> 
> We can't copy the patch directly. But it wouldn't be a big surprise
> if given the description of the bug someone else came up with a
> very similar diff though it would probably not have the whitespace
> change that is also in it.
> 
> Problem is that get_aligned_mem() and get_unaligned_address() need
> to look for replacements in memory addresses.

FWIW, even if we can't distribute the patched version, I could try it
on my alpha and the patch doesn't seem to fix libGL build.
.
-- 
Matthieu Herrb


signature.asc
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-10 Thread Christian Weisgerber 
On 2016-02-09, Jonathan Gray  wrote:

> /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: internal compiler 
> error: in extract_insn, at recog.c:2077

Our usual workaround for these types of errors is to compile the
affected source file (and only that file, and only on that arch)
without any optimization.  Did this still fail?

-- 
Christian "naddy" Weisgerber  na...@mips.inka.de

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-10 Thread Jonathan Gray 
On Thu, Feb 11, 2016 at 08:09:42AM +0100, Matthieu Herrb wrote:
> On Wed, Feb 10, 2016 at 08:32:51PM +, Christian Weisgerber wrote:
> > On 2016-02-09, Jonathan Gray  wrote:
> > 
> > > /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: internal 
> > > compiler error: in extract_insn, at recog.c:2077
> > 
> > Our usual workaround for these types of errors is to compile the
> > affected source file (and only that file, and only on that arch)
> > without any optimization.  Did this still fail?
> > 
> 
> Yes that was the 1st thing I tried of course.
> 
> Juest retried to be sure:
> 
> /share/OpenBSD/xenocara/lib/mesa/src/glsl/glsl_parser_extras.cpp:1697: error: 
> unrecognizable insn:
> (insn 57 56 58 3 
> /share/OpenBSD/xenocara/lib/mesa/src/glsl/glsl_parser_extras.cpp:1583 (set 
> (reg:DI 188)
> (ashift:DI (zero_extend:DI (const_int 1 [0x1]))
> (ashift:DI (reg/f:DI 183)
> (const_int 3 [0x3] -1 (nil)
> (nil))
> /share/OpenBSD/xenocara/lib/mesa/src/glsl/glsl_parser_extras.cpp:1697: 
> internal compiler error: in extract_insn, at recog.c:2077
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See  for instructions.
> 
> 
> Looking closer this is an atomic sync operation:
> 
> 1582:   if (ctx->Const.GenerateTemporaryNames)
> 1583: (void) 
> p_atomic_cmpxchg(_variable::temporaries_allocate_names,
> 1584:false, true);
> 
> Anyone has an idea to implement support for this ?

p_atomic_cmpxchg comes from /usr/xenocara/lib/mesa/src/util/u_atomic.h

which turns out to be __sync_val_compare_and_swap((v), (old), (_new))

There are kludged non atomic versions I added for other archs
that lack sync instructions under PIPE_ATOMIC_UNLOCKED

perhaps you could that version on alpha?

#define p_atomic_cmpxchg(_v, old, _new) ({  \
__typeof(*_v) _r = *(_v);   \
if (*(_v) == old)   \
*(_v) = (_new); \
_r; \
})

Here is the u_atomic_test.c test file out of the mesa source with
u_atomic.h included.  Should be easier to test against.

/**
 *
 * Copyright 2014 VMware, Inc.
 * All Rights Reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the
 * "Software"), to deal in the Software without restriction, including
 * without limitation the rights to use, copy, modify, merge, publish,
 * distribute, sub license, and/or sell copies of the Software, and to
 * permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 *
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
 * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 *
 **/


/* Force assertions, even on debug builds. */
#undef NDEBUG


#include 
#include 
#include 

/**
 * Many similar implementations exist. See for example libwsbm
 * or the linux kernel include/atomic.h
 *
 * No copyright claimed on this file.
 *
 */

#ifndef U_ATOMIC_H
#define U_ATOMIC_H

#include 

/* Favor OS-provided implementations.
 *
 * Where no OS-provided implementation is available, fall back to
 * locally coded assembly, compiler intrinsic or ultimately a
 * mutex-based implementation.
 */
#if defined(__sun)
#define PIPE_ATOMIC_OS_SOLARIS
#elif defined(_MSC_VER)
#define PIPE_ATOMIC_MSVC_INTRINSIC
#elif defined(__ARM_ARCH_4__) || defined(__ARM_ARCH_5__) || \
  defined(__hppa__) || defined(__sparc__) || defined(__sh__)
#define PIPE_ATOMIC_UNLOCKED
#elif defined(__GNUC__)
#define PIPE_ATOMIC_GCC_INTRINSIC
#else
#error "Unsupported platform"
#endif


/* Implementation using GCC-provided synchronization intrinsics
 */
#if defined(PIPE_ATOMIC_GCC_INTRINSIC)

#define PIPE_ATOMIC "GCC Sync Intrinsics"

#define p_atomic_set(_v, _i) (*(_v) = (_i))
#define p_atomic_read(_v) (*(_v))
#define p_atomic_dec_zero(v) (__sync_sub_and_fetch((v), 1) == 0)
#define p_atomic_inc(v) (void) __sync_add_and_fetch((v), 1)
#define p_atomic_dec(v) (void) __sync_sub_and_fetch((v), 1)
#define p_atomic_add(v, i) (void) __sync_add_and_fetch((v), (i))
#define p_atomic_inc_return(v) __sync_add_and_fetch((v), 1)
#define

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-10 Thread Matthieu Herrb 
On Wed, Feb 10, 2016 at 08:32:51PM +, Christian Weisgerber wrote:
> On 2016-02-09, Jonathan Gray  wrote:
> 
> > /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: internal compiler 
> > error: in extract_insn, at recog.c:2077
> 
> Our usual workaround for these types of errors is to compile the
> affected source file (and only that file, and only on that arch)
> without any optimization.  Did this still fail?
> 

Yes that was the 1st thing I tried of course.

Juest retried to be sure:

/share/OpenBSD/xenocara/lib/mesa/src/glsl/glsl_parser_extras.cpp:1697: error: 
unrecognizable insn:
(insn 57 56 58 3 
/share/OpenBSD/xenocara/lib/mesa/src/glsl/glsl_parser_extras.cpp:1583 (set 
(reg:DI 188)
(ashift:DI (zero_extend:DI (const_int 1 [0x1]))
(ashift:DI (reg/f:DI 183)
(const_int 3 [0x3] -1 (nil)
(nil))
/share/OpenBSD/xenocara/lib/mesa/src/glsl/glsl_parser_extras.cpp:1697: internal 
compiler error: in extract_insn, at recog.c:2077
Please submit a full bug report,
with preprocessed source if appropriate.
See  for instructions.


Looking closer this is an atomic sync operation:

1582:   if (ctx->Const.GenerateTemporaryNames)
1583: (void) p_atomic_cmpxchg(_variable::temporaries_allocate_names,
1584:  false, true);

Anyone has an idea to implement support for this ?

PS: I also had a look at trying to on;y build the indirect glx part of the
lib (since direct rendering part is useless without an X server, and
the error so far is in a file only used for direct rendering, but the
new Mesa build system doesn't provide support for that.


-- 
Matthieu Herrb


signature.asc
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-10 Thread Karel Gardas 
> It has been fixed upstream:
>
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66140

looks like in 5.1.1 which is GPL3 so probably not usable to be merge
into the OpenBSD's GPL2 4.2.1? But well, I'm not the lawyer here i
just see two different licenses so it raises red flag here...

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-10 Thread Stuart Henderson 
On 2016/02/10 10:57, Karel Gardas wrote:
> > It has been fixed upstream:
> >
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66140
> 
> looks like in 5.1.1 which is GPL3 so probably not usable to be merge
> into the OpenBSD's GPL2 4.2.1? But well, I'm not the lawyer here i
> just see two different licenses so it raises red flag here...
> 

We can't copy the patch directly. But it wouldn't be a big surprise
if given the description of the bug someone else came up with a
very similar diff though it would probably not have the whitespace
change that is also in it.

Problem is that get_aligned_mem() and get_unaligned_address() need
to look for replacements in memory addresses.

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-09 Thread Christian Weisgerber 
On 2016-02-08, Landry Breuil  wrote:

> for those who care about exotic archs, alpha has lots of failures
> because (among others) we dont build libGL/libGLU there anymore.

I don't see what can be done about this on the ports side, other
than marking affected ports as not for alpha or, possibly, disabling
some functionality depending on the arch (ewww, terrible).  None
of which actually fixes anything.

This needs to be addressed at the root of the problem: alpha needs
to be able to build libGL.

-- 
Christian "naddy" Weisgerber  na...@mips.inka.de

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-09 Thread Bryan Everly 
What is preventing libGL from being built on Alpha?  I have a machine
and am willing to help.

Thanks,
Bryan

> On Feb 9, 2016, at 10:42 AM, Christian Weisgerber  wrote:
>
>> On 2016-02-08, Landry Breuil  wrote:
>>
>> for those who care about exotic archs, alpha has lots of failures
>> because (among others) we dont build libGL/libGLU there anymore.
>
> I don't see what can be done about this on the ports side, other
> than marking affected ports as not for alpha or, possibly, disabling
> some functionality depending on the arch (ewww, terrible).  None
> of which actually fixes anything.
>
> This needs to be addressed at the root of the problem: alpha needs
> to be able to build libGL.
>
> --
> Christian "naddy" Weisgerber  na...@mips.inka.de
>

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-09 Thread Matthieu Herrb 
On Tue, Feb 09, 2016 at 12:21:16PM -0500, Bryan Everly wrote:
> What is preventing libGL from being built on Alpha?  I have a machine
> and am willing to help.

This is why it was disabled:
 |Author: Jonathan Gray 
 |Date:   Wed Nov 25 09:24:00 2015 +
 |
 |   Stop building Mesa on alpha for now as it hits a gcc ICE, and
 |   mips64 because
 |   it requires sync builtins not in gcc 4.2 and hits binutils issues.
 |   
 |   ok matthieu@ deraadt@

To fix this

get /usr/xenocara,
edit /usr/xenocara/share/mk/bsd.xconf.mk to re-enable GL on alpha  
and follow release(8) to build xenocara.
Alternatively you can just run

cd /usr/xenocara/lib/mesa
make -f Makefile.bsd-wrapper obj
make -f Makefile.bsd-wrapper build

to try to build only mesa.

Look at errors and send patches.

> 
> Thanks,
> Bryan
> 
> > On Feb 9, 2016, at 10:42 AM, Christian Weisgerber  
> > wrote:
> >
> >> On 2016-02-08, Landry Breuil  wrote:
> >>
> >> for those who care about exotic archs, alpha has lots of failures
> >> because (among others) we dont build libGL/libGLU there anymore.
> >
> > I don't see what can be done about this on the ports side, other
> > than marking affected ports as not for alpha or, possibly, disabling
> > some functionality depending on the arch (ewww, terrible).  None
> > of which actually fixes anything.
> >
> > This needs to be addressed at the root of the problem: alpha needs
> > to be able to build libGL.
> >
> > --
> > Christian "naddy" Weisgerber  na...@mips.inka.de
> >

-- 
Matthieu Herrb


pgpPnKYpmu3LE.pgp
Description: PGP signature

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-09 Thread Jonathan Gray 
Toolchain bugs.

Something along the lines of

lib/mesa/src/mesa/main/format_pack.c  -fPIC -DPIC -o main/.libs/format_pack.o
/usr/xenocara/lib/mesa/src/mesa/main/format_pack.c: In function 
'pack_float_bgr_srgb8':
/usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: error: unrecognizable 
insn:
(insn 259 11 13 2 (set (reg:SF 1 $1)
(const_double:SF 1.220703125e-4 [0x0.8p-12])) -1 (nil)
(nil))
/usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: internal compiler 
error: in extract_insn, at recog.c:2077

was what I was told previously.  No working alpha here.

On Tue, Feb 09, 2016 at 12:21:16PM -0500, Bryan Everly wrote:
> What is preventing libGL from being built on Alpha?  I have a machine
> and am willing to help.
>
> Thanks,
> Bryan
>
> > On Feb 9, 2016, at 10:42 AM, Christian Weisgerber  
> > wrote:
> >
> >> On 2016-02-08, Landry Breuil  wrote:
> >>
> >> for those who care about exotic archs, alpha has lots of failures
> >> because (among others) we dont build libGL/libGLU there anymore.
> >
> > I don't see what can be done about this on the ports side, other
> > than marking affected ports as not for alpha or, possibly, disabling
> > some functionality depending on the arch (ewww, terrible).  None
> > of which actually fixes anything.
> >
> > This needs to be addressed at the root of the problem: alpha needs
> > to be able to build libGL.
> >
> > --
> > Christian "naddy" Weisgerber  na...@mips.inka.de
> >
>

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-09 Thread David Coppa 
Il 10/feb/2016 08:24, "Landry Breuil"  ha scritto:
>
> On Wed, Feb 10, 2016 at 09:52:22AM +1100, Jonathan Gray wrote:
> > Toolchain bugs.
> >
> > Something along the lines of
> >
> > lib/mesa/src/mesa/main/format_pack.c  -fPIC -DPIC -o
main/.libs/format_pack.o
> > /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c: In function
'pack_float_bgr_srgb8':
> > /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: error:
unrecognizable insn:
> > (insn 259 11 13 2 (set (reg:SF 1 $1)
> > (const_double:SF 1.220703125e-4 [0x0.8p-12])) -1 (nil)
> > (nil))
> > /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: internal
compiler error: in extract_insn, at recog.c:2077
>
> And if you look at all the ports build failure, 'lots' of them in the
> past have the same 'internal compiler error: in extract_insn, at
> recog.c' issue.

It has been fixed upstream:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66140

Ciao!
David

Re: alpha-1.ports.openbsd.org bulk build report

2016-02-09 Thread Landry Breuil 
On Wed, Feb 10, 2016 at 09:52:22AM +1100, Jonathan Gray wrote:
> Toolchain bugs.
> 
> Something along the lines of
> 
> lib/mesa/src/mesa/main/format_pack.c  -fPIC -DPIC -o main/.libs/format_pack.o
> /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c: In function 
> 'pack_float_bgr_srgb8':
> /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: error: 
> unrecognizable insn:
> (insn 259 11 13 2 (set (reg:SF 1 $1)
> (const_double:SF 1.220703125e-4 [0x0.8p-12])) -1 (nil)
> (nil))
> /usr/xenocara/lib/mesa/src/mesa/main/format_pack.c:6890: internal compiler 
> error: in extract_insn, at recog.c:2077

And if you look at all the ports build failure, 'lots' of them in the
past have the same 'internal compiler error: in extract_insn, at
recog.c' issue.

Landry

alpha-1.ports.openbsd.org bulk build report

2016-02-08 Thread Landry Breuil 
Fwiw,

for those who care about exotic archs, alpha has lots of failures
because (among others) we dont build libGL/libGLU there anymore.
other failures are probably toolchain segfaults.

Landry

- Forwarded message from Landry Breuil  
-

bulk build on alpha-1.ports.openbsd.org
started on  Sat Jan 30 03:06:07 MST 2016
finished at Mon Feb 8 14:58:45 MST 2016
lasted 10D04h52m
done with kern.version=OpenBSD 5.9-beta (GENERIC.MP) #348: Fri Jan 29 00:21:55 
MST 2016

built packages:6692
Jan 30:578
Jan 31:273
Feb 1:766
Feb 2:380
Feb 3:359
Feb 4:362
Feb 5:492
Feb 6:508
Feb 7:487
Feb 8:2486

build failures: 85
http://build-failures.rhaalovely.net//alpha/2016-01-30/audio/xmms.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/databases/tdb.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/cppcheck.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/fox.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/libmemcached.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/libsigc++-2.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/libtalloc.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/liburcu.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/ocaml-menhir.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/ode.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/p5-SDL.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/plib.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/py-subvertpy.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/sdl2.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/devel/ti-msp430gcc.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/emulators/dosbox.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/emulators/mednafen.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/emulators/vice.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/abuse.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/alephone/alephone.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/armagetronad.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/astromenace.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/blobby.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/boswars.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/brumbrumrally.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/dangerdeep.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/egoboo.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/f1spirit.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/foobillard.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/freedroidrpg.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/lincity-ng.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/manaplus.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/meandmyshadow.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/naev/base.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/oolite.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/pacman-arena.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/pingus.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/prboom.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/prboom-plus.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/residualvm.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/stone-soup.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/taisei.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/valyriatear.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/games/wizznic.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/cfdg.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/clutter/cogl.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/freeglut.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/freeimage.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/ftgl.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/glew.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/glpng.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/lsys.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/py-cairo.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/py3-cairo.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/quesoglc.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/tesseract/tesseract.log
http://build-failures.rhaalovely.net//alpha/2016-01-30/graphics/waffle.log