Commit Graph

5467 Commits

Author SHA1 Message Date
deraadt
b2afdad445 sync 2015-12-05 19:37:40 +00:00
kettenis
d9a335dd99 Revert previous commit; the "blt" codepath is obviously not tested very well
and makes X crash.
2015-12-03 22:36:46 +00:00
kettenis
b1132884a7 Enable glamor where we support it.
Requested by jsg@
2015-12-01 21:22:56 +00:00
kettenis
c7a6143bc4 Enable glamor on architectures where we have OpenGL.
ok jsg@, matthieu@
2015-12-01 21:20:21 +00:00
kettenis
2e2a8984c6 Remove a check for DRI3 in glamor_name_from_pixmap(). This makes DRI2 work
on OpenBSD where we don't have support for DRI3 yet.

ok jsg@, matthieu@
2015-12-01 21:18:47 +00:00
kettenis
bfcc545caa Use an ARGB8888 based texture even when DRI3 isn't enabled. This achieves the
same effect as a change that was recently made upstream.  The code changed
significantly in the meantime.

ok jsg@, matthieu@
2015-12-01 21:17:49 +00:00
kettenis
e6977614b0 Call miSyncSetup() if we don't call miSyncShmScreenInit() because
HAVE_XSHMFENCE isn't defined.  This makes sure the relevant "private key"
gets installed such that we don't hit a fatal assert later.

ok jsg@, matthieu@
2015-12-01 21:13:48 +00:00
deraadt
d6749ba35a sync 2015-11-28 19:19:33 +00:00
deraadt
98abda57b7 sync 2015-11-28 18:06:03 +00:00
deraadt
a0f789e26b sync 2015-11-27 21:44:33 +00:00
kettenis
b31a77d567 On i386, compile with -march=i586. The i965 driver requires 64-bit atomic
operations and older CPUs lack the needed instructions.  The hardware
supported by that driver will never be used together those older CPUs.
This might mean that even the software rasterizer doesn't work anymore on
those.  But they're so slow that you probably wouldn't want to anyway.

ok jsg@
2015-11-26 19:05:49 +00:00
jsg
0d062359fc 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@
2015-11-25 09:23:59 +00:00
semarie
d0c47f20a0 add pledge promise "exec" for few cases in xterm
in particular if the user use `exec-formatted' or `exec-selectable' actions in key binding.

problem reported by jirib at devio dot us.
2015-11-25 08:56:50 +00:00
jsg
90e9a7268a update 2015-11-22 10:05:14 +00:00
jsg
d9d2d52f84 sync 2015-11-22 07:59:26 +00:00
jsg
45966989b6 Switch to Mesa 11.0.6 built with autoconf instead of a set of a custom
makefiles.  Thanks to Emil Velikov for removing requirements on gnu make
upstream and sorting out a bunch of other things to make this possible.
2015-11-22 07:57:10 +00:00
jsg
987bf76a95 regen 2015-11-22 07:14:29 +00:00
jsg
a8524181dd Disable the bison/lex/python rules to regenerate sources that were
included in the tarball.

cvs import seems to have changed timestamps of files which caused
these rules to trigger and the build to fail.
2015-11-22 06:57:53 +00:00
jsg
88d4ee1fc0 disable --gc-sections on ppc to avoid an ld 2.17 segfault 2015-11-22 03:13:27 +00:00
jsg
3204a661ed add a makefile for xenocara 2015-11-22 03:11:30 +00:00
jsg
61bd9e6bfc provide an unlocked atomic fallback path
sparc < v9, hppa, sh, arm < v6 don't provide these builtins
and we don't have kernel assistance to fake them currently.
2015-11-22 03:08:22 +00:00
jsg
4f35c98dd8 don't install /etc/drirc it's just a bunch of workarounds for closed
linux programs
2015-11-22 03:05:02 +00:00
jsg
0ecb55906d don't try to symlink files that don't exist 2015-11-22 03:01:23 +00:00
jsg
0cd8e2ac98 dlopen libGL.so not libGL.so.1 2015-11-22 02:58:43 +00:00
jsg
7966c15bd5 remove the hard requirement of llvm for gallium on i386/amd64 2015-11-22 02:55:55 +00:00
jsg
1fe54f218f import Mesa 11.0.6 2015-11-22 02:36:47 +00:00
okan
eaf23378dd pledge fvwm and FvwmPager (since it's started by default); committing to
solicit further feedback and testing.

ok matthieu and encouragement from deraadt
2015-11-18 17:41:42 +00:00
okan
402ba9437e Fix a typo from r1.201; fixes window_grouptoggle binding. 2015-11-17 15:19:19 +00:00
okan
448a04ca44 If a client does not set increment values, use 'moveamount' as a way to
scale keyboard based resizes; extend kbfunc_amount().

Behaviour noted by, tested by, and ok sthen@
2015-11-17 14:32:38 +00:00
okan
5dd7f2f304 Inline the only use of mousefunc_sweep_calc. 2015-11-17 14:31:28 +00:00
aoyama
184a98336b Pass --disable-libdrm option to configure when the machine does not
support DRI.  Found while building Xorg server on luna88k.

"fine for now" matthieu@
2015-11-17 10:49:57 +00:00
kettenis
0929ee3ac4 Partly disable acceleration on Broadwell. The inteldrm(4) driver fails to
restore something properly after resume which makes the render ring get stuck.
By defaulting to the "blt" acceleration method we avoid the use of this ring
and the machine remains usable after resume.

People who really want full 2D acceleration can still override the acceleration
method in their xorg.conf.

ok jsg@
2015-11-15 15:33:57 +00:00
matthieu
1e036e407b Remove the -configure option. This has been broken for a loong time
and no one is going to fix it.
2015-11-15 08:28:24 +00:00
deraadt
106ed06de4 sync 2015-11-14 20:27:41 +00:00
matthieu
78005e8a07 pledge xconsole(1). 2015-11-14 17:13:21 +00:00
matthieu
5092a69bf9 Don't try to revoque privileges if priv_init is called without
privileges. The parent process will be used for operations
that can't be pledged.
2015-11-14 17:12:14 +00:00
shadchin
d98e5b49b4 sync (xkeyboard-config-2.16) 2015-11-14 15:32:30 +00:00
shadchin
aecbad4093 Update to xkeyboard-config 2.16.
ok matthieu@
2015-11-14 15:30:00 +00:00
okan
403542f5b2 more client vs screen context differences 2015-11-12 21:28:03 +00:00
okan
6e55f9d89b If a client sets hints, honor them for kb resize requests, just like we
do for mouse based resize requests.

Based on a patch from Vadim Vygonets.
2015-11-12 18:33:30 +00:00
okan
db0907a53d Move kb pointer movement out of the kbfunc_client_moveresize since it's
got nothing to do with clients, thus doing flags work causes lots of
waste and almost useless jumpy pointer movements; while here, split out
move and resize since they share almost no code, just like mouse client
move/resize; factor out amount and factor. Still wonder why this is
here, but it works now.
2015-11-12 18:26:41 +00:00
okan
4db32a8463 pledge "stdio rpath proc exec" cwm before main event loop, after
init/setup - mostly for menu building.

ok semarie@ (another cwm user)
2015-11-12 17:56:54 +00:00
matthieu
27549fce93 Reduce gcc warnings. No functional changes. 2015-11-11 21:20:12 +00:00
matthieu
300ba5c057 use HAVE_ instead of HAS_ macros for consistency. 2015-11-11 21:18:12 +00:00
matthieu
f08e325b48 regen 2015-11-11 21:12:43 +00:00
matthieu
eedb4e77fa pledge for xclock. okan@ got to the same patch. I added autoconf goo
to make it possible to send it upstreams. ok deraadt@
2015-11-11 21:12:19 +00:00
matthieu
f7d98a310c pledge(2) for the X server privileged process. ok deraadt@ 2015-11-11 21:07:49 +00:00
okan
a7bbf5197e Partial revert of replacing screen_area() with region_find(); until a
fix for a regression is found; this bug has been around for a long time
it seems, but this change exposed it. Likely need to track clients in to
and out of regions.
2015-11-11 14:22:01 +00:00
deraadt
3ab7d7d094 avoid using err(), use xtermWarning() instead; ok semarie 2015-11-11 07:09:50 +00:00
okan
72507096a1 Use position on root to figure out region. 2015-11-10 22:06:57 +00:00