Discussion:
[Xcb] _XReply stuck in xcb_wait_for_reply64
Marius Cirsta
2016-11-30 22:50:51 UTC
Permalink
Hello,

I have a problem that came once I upgraded my Linux OS ( less known
one called Fruglware OS ).
For some reason with libxcb 1.12 libx11 1.6.4 I'm having this weird
freeze in plasmashell:

#0 0x00007fc2cf9cd3e1 in poll () from /lib/libc.so.6
#1 0x00007fc2d39831e0 in _xcb_conn_wait (c=***@entry=0x117aaf0,
cond=***@entry=0x7fff869b5cd0, vector=***@entry=0x0,
count=***@entry=0x0) at xcb_conn.c:479
#2 0x00007fc2d3984bdf in wait_for_reply (c=***@entry=0x117aaf0,
request=***@entry=10682, e=***@entry=0x7fff869b5d98) at xcb_in.c:516
#3 0x00007fc2d3984d31 in xcb_wait_for_reply64 (c=0x117aaf0,
request=10682, e=0x7fff869b5d98) at xcb_in.c:560
#4 0x00007fc2d3bea2e8 in _XReply (dpy=0x11798a0, rep=0x7fff869b5e40,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc2cc781cd2 in ?? () from /usr/lib/libGL.so.1
#6 0x00007fc2cc781ff7 in ?? () from /usr/lib/libGL.so.1
#7 0x00007fc2216f9239 in ?? () from /usr/lib/dri/r600_dri.so
#8 0x00007fc2216f3ab4 in ?? () from /usr/lib/dri/r600_dri.so
#9 0x00007fc2215dd92c in ?? () from /usr/lib/dri/r600_dri.so
#10 0x00007fc2215de84c in ?? () from /usr/lib/dri/r600_dri.so
#11 0x00007fc22158a67b in ?? () from /usr/lib/dri/r600_dri.so
#12 0x00007fc221592eb4 in ?? () from /usr/lib/dri/r600_dri.so
#13 0x00007fc2d3201216 in QSGBatchRenderer::Renderer::renderBatches()
() from /usr/lib/libQt5Quick.so.5
#14 0x00007fc2d3206835 in QSGBatchRenderer::Renderer::render() () from
/usr/lib/libQt5Quick.so.5
#15 0x00007fc2d3211daf in QSGRenderer::renderScene(QSGBindable const&)
() from /usr/lib/libQt5Quick.so.5
#16 0x00007fc2d321246b in QSGRenderer::renderScene(unsigned int) ()
from /usr/lib/libQt5Quick.so.5
#17 0x00007fc2d322186e in
QSGRenderContext::renderNextFrame(QSGRenderer*, unsigned int) () from
/usr/lib/libQt5Quick.so.5
#18 0x00007fc2d32682ee in QQuickWindowPrivate::renderSceneGraph(QSize
const&) () from /usr/lib/libQt5Quick.so.5
#19 0x00007fc2d3237986 in ?? () from /usr/lib/libQt5Quick.so.5
#20 0x00007fc2d3272595 in QQuickWindow::event(QEvent*) () from
/usr/lib/libQt5Quick.so.5
#21 0x00007fc2d4b8e8a3 in PlasmaQuick::Dialog::event(QEvent*) () from
/usr/lib/libKF5PlasmaQuick.so.5
#22 0x00007fc2d0a9ab2c in QApplicationPrivate::notify_helper(QObject*,
QEvent*) () from /usr/lib/libQt5Widgets.so.5
#23 0x00007fc2d0aa1cb1 in QApplication::notify(QObject*, QEvent*) ()
from /usr/lib/libQt5Widgets.so.5
#24 0x00007fc2d0249e3a in QCoreApplication::notifyInternal2(QObject*,
QEvent*) () from /usr/lib/libQt5Core.so.5
#25 0x00007fc2d059f18e in QWindowPrivate::deliverUpdateRequest() ()
from /usr/lib/libQt5Gui.so.5
#26 0x00007fc2d059f669 in QWindow::event(QEvent*) () from
/usr/lib/libQt5Gui.so.5
#27 0x00007fc2d3272403 in QQuickWindow::event(QEvent*) () from
/usr/lib/libQt5Quick.so.5
#28 0x00007fc2d4b8e8a3 in PlasmaQuick::Dialog::event(QEvent*) () from
/usr/lib/libKF5PlasmaQuick.so.5
#29 0x00007fc2d0a9ab2c in QApplicationPrivate::notify_helper(QObject*,
QEvent*) () from /usr/lib/libQt5Widgets.so.5
#30 0x00007fc2d0aa1cb1 in QApplication::notify(QObject*, QEvent*) ()
from /usr/lib/libQt5Widgets.so.5
#31 0x00007fc2d0249e3a in QCoreApplication::notifyInternal2(QObject*,
QEvent*) () from /usr/lib/libQt5Core.so.5
#32 0x00007fc2d0293fa3 in QTimerInfoList::activateTimers() () from
/usr/lib/libQt5Core.so.5
#33 0x00007fc2d02944f1 in ?? () from /usr/lib/libQt5Core.so.5
#34 0x00007fc2cb7a5ea7 in g_main_context_dispatch () from
/usr/lib/libglib-2.0.so.0
#35 0x00007fc2cb7a60d8 in ?? () from /usr/lib/libglib-2.0.so.0
#36 0x00007fc2cb7a617c in g_main_context_iteration () from
/usr/lib/libglib-2.0.so.0
#37 0x00007fc2d02948ff in
QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>)
() from /usr/lib/libQt5Core.so.5
#38 0x00007fc2d0248eea in
QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () from
/usr/lib/libQt5Core.so.5
#39 0x00007fc2d02505dd in QCoreApplication::exec() () from
/usr/lib/libQt5Core.so.5
#40 0x000000000041ea8a in ?? ()
#41 0x00007fc2cf90e271 in __libc_start_main () from /lib/libc.so.6
#42 0x000000000041ecba in _start ()

I tried to look at it because I'm a C/C++ developer but I'm not really
familiar with this.

It seems that for some reason even though a request was made there's
not response.
I assume this is probably not caused libxcb or libx11 but any help in
debugging this would be highly appreciated.

If any additional info would help I can provide it.

Thank you.
Michel Dänzer
2016-12-01 01:42:32 UTC
Permalink
Post by Marius Cirsta
Hello,
I have a problem that came once I upgraded my Linux OS ( less known
one called Fruglware OS ).
For some reason with libxcb 1.12 libx11 1.6.4 I'm having this weird
#0 0x00007fc2cf9cd3e1 in poll () from /lib/libc.so.6
#3 0x00007fc2d3984d31 in xcb_wait_for_reply64 (c=0x117aaf0,
request=10682, e=0x7fff869b5d98) at xcb_in.c:560
#4 0x00007fc2d3bea2e8 in _XReply (dpy=0x11798a0, rep=0x7fff869b5e40,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc2cc781cd2 in ?? () from /usr/lib/libGL.so.1
#6 0x00007fc2cc781ff7 in ?? () from /usr/lib/libGL.so.1
#7 0x00007fc2216f9239 in ?? () from /usr/lib/dri/r600_dri.so
#8 0x00007fc2216f3ab4 in ?? () from /usr/lib/dri/r600_dri.so
#9 0x00007fc2215dd92c in ?? () from /usr/lib/dri/r600_dri.so
#10 0x00007fc2215de84c in ?? () from /usr/lib/dri/r600_dri.so
#11 0x00007fc22158a67b in ?? () from /usr/lib/dri/r600_dri.so
#12 0x00007fc221592eb4 in ?? () from /usr/lib/dri/r600_dri.so
Please get a backtrace with debugging symbols available for
/usr/lib/libGL.so.1 and /usr/lib/dri/r600_dri.so as well.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Marius Cirsta
2016-12-01 23:27:55 UTC
Permalink
#0 0x00007fc37f1673dd in poll () from /lib/libc.so.6
#1 0x00007fc38311d1e0 in _xcb_conn_wait (c=***@entry=0x2584af0,
cond=***@entry=0x7ffe6ac997f0, vector=***@entry=0x0,
count=***@entry=0x0) at xcb_conn.c:479
#2 0x00007fc38311ebdf in wait_for_reply (c=***@entry=0x2584af0,
request=***@entry=65199, e=***@entry=0x7ffe6ac998b8) at xcb_in.c:516
#3 0x00007fc38311ed31 in xcb_wait_for_reply64 (c=0x2584af0,
request=65199, e=0x7ffe6ac998b8) at xcb_in.c:560
#4 0x00007fc3833842e8 in _XReply (dpy=0x25838a0, rep=0x7ffe6ac99960,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc37bf1b772 in DRI2GetBuffersWithFormat (dpy=0x25838a0,
drawable=37748763, width=***@entry=0x4f54578,
height=***@entry=0x4f5457c, attachments=0x7ffe6ac99af0, count=1,
outCount=0x7ffe6ac99ad0) at dri2.c:491
#6 0x00007fc37bf1ba97 in dri2GetBuffersWithFormat
(driDrawable=<optimized out>, width=0x4f54578, height=0x4f5457c,
attachments=<optimized out>, count=<optimized out>,
out_count=0x7ffe6ac99ad0, loaderPrivate=0x5260100) at dri2_glx.c:894
#7 0x00007fc358d6ac20 in dri2_drawable_get_buffers (count=<synthetic
pointer>, atts=0x41f3338, drawable=0x438fbe0) at dri2.c:285
#8 dri2_allocate_textures (ctx=0x25a7270, drawable=0x438fbe0,
statts=0x41f3338, statts_count=2) at dri2.c:480
#9 0x00007fc358d63d94 in dri_st_framebuffer_validate
(stctx=<optimized out>, stfbi=<optimized out>, statts=0x41f3338,
count=2, out=0x7ffe6ac99c60) at dri_drawable.c:83
#10 0x00007fc358c3948c in st_framebuffer_validate (stfb=0x41f2ed0,
st=***@entry=0x2d1f250) at state_tracker/st_manager.c:202
#11 0x00007fc358c3a7de in st_api_make_current (stapi=<optimized out>,
stctxi=0x2d1f250, stdrawi=0x438fbe0, streadi=0x438fbe0) at
state_tracker/st_manager.c:781
#12 0x00007fc358d636a1 in dri_make_current (cPriv=<optimized out>,
driDrawPriv=0x4f54550, driReadPriv=0x4f54550) at dri_context.c:245
#13 0x00007fc358d62606 in driBindContext (pcp=<optimized out>,
pdp=<optimized out>, prp=<optimized out>) at dri_util.c:553
#14 0x00007fc37bf1d41b in dri2_bind_context (context=0x25a4350,
old=<optimized out>, draw=37748763, read=37748763) at dri2_glx.c:154
#15 0x00007fc37bef7b67 in MakeContextCurrent (dpy=0x25838a0,
draw=37748763, read=37748763, gc_user=0x25a4350) at glxcurrent.c:228
#16 0x00007fc3849763d8 in ?? () from
/usr/lib/qt5/plugins/xcbglintegrations/libqxcb-glx-integration.so
#17 0x00007fc37fd63870 in QOpenGLContext::makeCurrent(QSurface*) ()
from /usr/lib/libQt5Gui.so.5
#18 0x00007fc3829d1860 in ?? () from /usr/lib/libQt5Quick.so.5

Also mesa is at 13.0.2 and kernel is 4.8.10. Video card is an AMD
HD5670 which as far as I know is not faulty ( doesn't show any signs
of screen corruption or problems in Windows ).
I have to do some stuff to recreate this but I'm able to consistently
recreate it.
Post by Michel Dänzer
Post by Marius Cirsta
Hello,
I have a problem that came once I upgraded my Linux OS ( less known
one called Fruglware OS ).
For some reason with libxcb 1.12 libx11 1.6.4 I'm having this weird
#0 0x00007fc2cf9cd3e1 in poll () from /lib/libc.so.6
#3 0x00007fc2d3984d31 in xcb_wait_for_reply64 (c=0x117aaf0,
request=10682, e=0x7fff869b5d98) at xcb_in.c:560
#4 0x00007fc2d3bea2e8 in _XReply (dpy=0x11798a0, rep=0x7fff869b5e40,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc2cc781cd2 in ?? () from /usr/lib/libGL.so.1
#6 0x00007fc2cc781ff7 in ?? () from /usr/lib/libGL.so.1
#7 0x00007fc2216f9239 in ?? () from /usr/lib/dri/r600_dri.so
#8 0x00007fc2216f3ab4 in ?? () from /usr/lib/dri/r600_dri.so
#9 0x00007fc2215dd92c in ?? () from /usr/lib/dri/r600_dri.so
#10 0x00007fc2215de84c in ?? () from /usr/lib/dri/r600_dri.so
#11 0x00007fc22158a67b in ?? () from /usr/lib/dri/r600_dri.so
#12 0x00007fc221592eb4 in ?? () from /usr/lib/dri/r600_dri.so
Please get a backtrace with debugging symbols available for
/usr/lib/libGL.so.1 and /usr/lib/dri/r600_dri.so as well.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Marius Cirsta
2016-12-02 00:08:13 UTC
Permalink
Googling a bit ( like a good engineer does, right ? ) I found this
very similar problem:

https://lkml.org/lkml/2016/1/18/261

https://lists.freedesktop.org/archives/dri-devel/2016-January/099151.html

And then then some explanations here:

http://code.metager.de/source/history/linux/stable/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
Post by Marius Cirsta
#0 0x00007fc37f1673dd in poll () from /lib/libc.so.6
#3 0x00007fc38311ed31 in xcb_wait_for_reply64 (c=0x2584af0,
request=65199, e=0x7ffe6ac998b8) at xcb_in.c:560
#4 0x00007fc3833842e8 in _XReply (dpy=0x25838a0, rep=0x7ffe6ac99960,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc37bf1b772 in DRI2GetBuffersWithFormat (dpy=0x25838a0,
outCount=0x7ffe6ac99ad0) at dri2.c:491
#6 0x00007fc37bf1ba97 in dri2GetBuffersWithFormat
(driDrawable=<optimized out>, width=0x4f54578, height=0x4f5457c,
attachments=<optimized out>, count=<optimized out>,
out_count=0x7ffe6ac99ad0, loaderPrivate=0x5260100) at dri2_glx.c:894
#7 0x00007fc358d6ac20 in dri2_drawable_get_buffers (count=<synthetic
pointer>, atts=0x41f3338, drawable=0x438fbe0) at dri2.c:285
#8 dri2_allocate_textures (ctx=0x25a7270, drawable=0x438fbe0,
statts=0x41f3338, statts_count=2) at dri2.c:480
#9 0x00007fc358d63d94 in dri_st_framebuffer_validate
(stctx=<optimized out>, stfbi=<optimized out>, statts=0x41f3338,
count=2, out=0x7ffe6ac99c60) at dri_drawable.c:83
#10 0x00007fc358c3948c in st_framebuffer_validate (stfb=0x41f2ed0,
#11 0x00007fc358c3a7de in st_api_make_current (stapi=<optimized out>,
stctxi=0x2d1f250, stdrawi=0x438fbe0, streadi=0x438fbe0) at
state_tracker/st_manager.c:781
#12 0x00007fc358d636a1 in dri_make_current (cPriv=<optimized out>,
driDrawPriv=0x4f54550, driReadPriv=0x4f54550) at dri_context.c:245
#13 0x00007fc358d62606 in driBindContext (pcp=<optimized out>,
pdp=<optimized out>, prp=<optimized out>) at dri_util.c:553
#14 0x00007fc37bf1d41b in dri2_bind_context (context=0x25a4350,
old=<optimized out>, draw=37748763, read=37748763) at dri2_glx.c:154
#15 0x00007fc37bef7b67 in MakeContextCurrent (dpy=0x25838a0,
draw=37748763, read=37748763, gc_user=0x25a4350) at glxcurrent.c:228
#16 0x00007fc3849763d8 in ?? () from
/usr/lib/qt5/plugins/xcbglintegrations/libqxcb-glx-integration.so
#17 0x00007fc37fd63870 in QOpenGLContext::makeCurrent(QSurface*) ()
from /usr/lib/libQt5Gui.so.5
#18 0x00007fc3829d1860 in ?? () from /usr/lib/libQt5Quick.so.5
Also mesa is at 13.0.2 and kernel is 4.8.10. Video card is an AMD
HD5670 which as far as I know is not faulty ( doesn't show any signs
of screen corruption or problems in Windows ).
I have to do some stuff to recreate this but I'm able to consistently
recreate it.
Post by Michel Dänzer
Post by Marius Cirsta
Hello,
I have a problem that came once I upgraded my Linux OS ( less known
one called Fruglware OS ).
For some reason with libxcb 1.12 libx11 1.6.4 I'm having this weird
#0 0x00007fc2cf9cd3e1 in poll () from /lib/libc.so.6
#3 0x00007fc2d3984d31 in xcb_wait_for_reply64 (c=0x117aaf0,
request=10682, e=0x7fff869b5d98) at xcb_in.c:560
#4 0x00007fc2d3bea2e8 in _XReply (dpy=0x11798a0, rep=0x7fff869b5e40,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc2cc781cd2 in ?? () from /usr/lib/libGL.so.1
#6 0x00007fc2cc781ff7 in ?? () from /usr/lib/libGL.so.1
#7 0x00007fc2216f9239 in ?? () from /usr/lib/dri/r600_dri.so
#8 0x00007fc2216f3ab4 in ?? () from /usr/lib/dri/r600_dri.so
#9 0x00007fc2215dd92c in ?? () from /usr/lib/dri/r600_dri.so
#10 0x00007fc2215de84c in ?? () from /usr/lib/dri/r600_dri.so
#11 0x00007fc22158a67b in ?? () from /usr/lib/dri/r600_dri.so
#12 0x00007fc221592eb4 in ?? () from /usr/lib/dri/r600_dri.so
Please get a backtrace with debugging symbols available for
/usr/lib/libGL.so.1 and /usr/lib/dri/r600_dri.so as well.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Marius Cirsta
2016-12-02 00:41:30 UTC
Permalink
Sorry for spamming here but I think I found my bug indeed:

https://bugs.freedesktop.org/show_bug.cgi?id=98645

Funny thing is that I have an X99 chipset too, I actually have this
motherboard: http://www.asrock.com/mb/Intel/X99%20Extreme4/?cat=Specifications

I only have a single screen though and it's all it takes to reproduce
this apparently.

I think maybe we should continue the discussion on that bug, right ?
I'll also try with radeon.msi=0 but need to do it tomorrow.

These problems only started appearing recently though so I suspect
maybe a kernel update or something. I'll could try with an older
version of the kernel I suppose and maybe even bisect this.

Thanks.
Post by Marius Cirsta
Googling a bit ( like a good engineer does, right ? ) I found this
https://lkml.org/lkml/2016/1/18/261
https://lists.freedesktop.org/archives/dri-devel/2016-January/099151.html
http://code.metager.de/source/history/linux/stable/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
Post by Marius Cirsta
#0 0x00007fc37f1673dd in poll () from /lib/libc.so.6
#3 0x00007fc38311ed31 in xcb_wait_for_reply64 (c=0x2584af0,
request=65199, e=0x7ffe6ac998b8) at xcb_in.c:560
#4 0x00007fc3833842e8 in _XReply (dpy=0x25838a0, rep=0x7ffe6ac99960,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc37bf1b772 in DRI2GetBuffersWithFormat (dpy=0x25838a0,
outCount=0x7ffe6ac99ad0) at dri2.c:491
#6 0x00007fc37bf1ba97 in dri2GetBuffersWithFormat
(driDrawable=<optimized out>, width=0x4f54578, height=0x4f5457c,
attachments=<optimized out>, count=<optimized out>,
out_count=0x7ffe6ac99ad0, loaderPrivate=0x5260100) at dri2_glx.c:894
#7 0x00007fc358d6ac20 in dri2_drawable_get_buffers (count=<synthetic
pointer>, atts=0x41f3338, drawable=0x438fbe0) at dri2.c:285
#8 dri2_allocate_textures (ctx=0x25a7270, drawable=0x438fbe0,
statts=0x41f3338, statts_count=2) at dri2.c:480
#9 0x00007fc358d63d94 in dri_st_framebuffer_validate
(stctx=<optimized out>, stfbi=<optimized out>, statts=0x41f3338,
count=2, out=0x7ffe6ac99c60) at dri_drawable.c:83
#10 0x00007fc358c3948c in st_framebuffer_validate (stfb=0x41f2ed0,
#11 0x00007fc358c3a7de in st_api_make_current (stapi=<optimized out>,
stctxi=0x2d1f250, stdrawi=0x438fbe0, streadi=0x438fbe0) at
state_tracker/st_manager.c:781
#12 0x00007fc358d636a1 in dri_make_current (cPriv=<optimized out>,
driDrawPriv=0x4f54550, driReadPriv=0x4f54550) at dri_context.c:245
#13 0x00007fc358d62606 in driBindContext (pcp=<optimized out>,
pdp=<optimized out>, prp=<optimized out>) at dri_util.c:553
#14 0x00007fc37bf1d41b in dri2_bind_context (context=0x25a4350,
old=<optimized out>, draw=37748763, read=37748763) at dri2_glx.c:154
#15 0x00007fc37bef7b67 in MakeContextCurrent (dpy=0x25838a0,
draw=37748763, read=37748763, gc_user=0x25a4350) at glxcurrent.c:228
#16 0x00007fc3849763d8 in ?? () from
/usr/lib/qt5/plugins/xcbglintegrations/libqxcb-glx-integration.so
#17 0x00007fc37fd63870 in QOpenGLContext::makeCurrent(QSurface*) ()
from /usr/lib/libQt5Gui.so.5
#18 0x00007fc3829d1860 in ?? () from /usr/lib/libQt5Quick.so.5
Also mesa is at 13.0.2 and kernel is 4.8.10. Video card is an AMD
HD5670 which as far as I know is not faulty ( doesn't show any signs
of screen corruption or problems in Windows ).
I have to do some stuff to recreate this but I'm able to consistently
recreate it.
Post by Michel Dänzer
Post by Marius Cirsta
Hello,
I have a problem that came once I upgraded my Linux OS ( less known
one called Fruglware OS ).
For some reason with libxcb 1.12 libx11 1.6.4 I'm having this weird
#0 0x00007fc2cf9cd3e1 in poll () from /lib/libc.so.6
#3 0x00007fc2d3984d31 in xcb_wait_for_reply64 (c=0x117aaf0,
request=10682, e=0x7fff869b5d98) at xcb_in.c:560
#4 0x00007fc2d3bea2e8 in _XReply (dpy=0x11798a0, rep=0x7fff869b5e40,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc2cc781cd2 in ?? () from /usr/lib/libGL.so.1
#6 0x00007fc2cc781ff7 in ?? () from /usr/lib/libGL.so.1
#7 0x00007fc2216f9239 in ?? () from /usr/lib/dri/r600_dri.so
#8 0x00007fc2216f3ab4 in ?? () from /usr/lib/dri/r600_dri.so
#9 0x00007fc2215dd92c in ?? () from /usr/lib/dri/r600_dri.so
#10 0x00007fc2215de84c in ?? () from /usr/lib/dri/r600_dri.so
#11 0x00007fc22158a67b in ?? () from /usr/lib/dri/r600_dri.so
#12 0x00007fc221592eb4 in ?? () from /usr/lib/dri/r600_dri.so
Please get a backtrace with debugging symbols available for
/usr/lib/libGL.so.1 and /usr/lib/dri/r600_dri.so as well.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Marius Cirsta
2016-12-02 23:59:34 UTC
Permalink
Turns out it was xf86-video-ati after all. Went through the updates I
made to my distro and managed to find this. Using the latest from git
works like a charm. Thanks for the fixes there Michel.

Maybe release something with at least some of these fixes included ?

I can test if you want.
Post by Marius Cirsta
https://bugs.freedesktop.org/show_bug.cgi?id=98645
Funny thing is that I have an X99 chipset too, I actually have this
motherboard: http://www.asrock.com/mb/Intel/X99%20Extreme4/?cat=Specifications
I only have a single screen though and it's all it takes to reproduce
this apparently.
I think maybe we should continue the discussion on that bug, right ?
I'll also try with radeon.msi=0 but need to do it tomorrow.
These problems only started appearing recently though so I suspect
maybe a kernel update or something. I'll could try with an older
version of the kernel I suppose and maybe even bisect this.
Thanks.
Post by Marius Cirsta
Googling a bit ( like a good engineer does, right ? ) I found this
https://lkml.org/lkml/2016/1/18/261
https://lists.freedesktop.org/archives/dri-devel/2016-January/099151.html
http://code.metager.de/source/history/linux/stable/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
Post by Marius Cirsta
#0 0x00007fc37f1673dd in poll () from /lib/libc.so.6
#3 0x00007fc38311ed31 in xcb_wait_for_reply64 (c=0x2584af0,
request=65199, e=0x7ffe6ac998b8) at xcb_in.c:560
#4 0x00007fc3833842e8 in _XReply (dpy=0x25838a0, rep=0x7ffe6ac99960,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc37bf1b772 in DRI2GetBuffersWithFormat (dpy=0x25838a0,
outCount=0x7ffe6ac99ad0) at dri2.c:491
#6 0x00007fc37bf1ba97 in dri2GetBuffersWithFormat
(driDrawable=<optimized out>, width=0x4f54578, height=0x4f5457c,
attachments=<optimized out>, count=<optimized out>,
out_count=0x7ffe6ac99ad0, loaderPrivate=0x5260100) at dri2_glx.c:894
#7 0x00007fc358d6ac20 in dri2_drawable_get_buffers (count=<synthetic
pointer>, atts=0x41f3338, drawable=0x438fbe0) at dri2.c:285
#8 dri2_allocate_textures (ctx=0x25a7270, drawable=0x438fbe0,
statts=0x41f3338, statts_count=2) at dri2.c:480
#9 0x00007fc358d63d94 in dri_st_framebuffer_validate
(stctx=<optimized out>, stfbi=<optimized out>, statts=0x41f3338,
count=2, out=0x7ffe6ac99c60) at dri_drawable.c:83
#10 0x00007fc358c3948c in st_framebuffer_validate (stfb=0x41f2ed0,
#11 0x00007fc358c3a7de in st_api_make_current (stapi=<optimized out>,
stctxi=0x2d1f250, stdrawi=0x438fbe0, streadi=0x438fbe0) at
state_tracker/st_manager.c:781
#12 0x00007fc358d636a1 in dri_make_current (cPriv=<optimized out>,
driDrawPriv=0x4f54550, driReadPriv=0x4f54550) at dri_context.c:245
#13 0x00007fc358d62606 in driBindContext (pcp=<optimized out>,
pdp=<optimized out>, prp=<optimized out>) at dri_util.c:553
#14 0x00007fc37bf1d41b in dri2_bind_context (context=0x25a4350,
old=<optimized out>, draw=37748763, read=37748763) at dri2_glx.c:154
#15 0x00007fc37bef7b67 in MakeContextCurrent (dpy=0x25838a0,
draw=37748763, read=37748763, gc_user=0x25a4350) at glxcurrent.c:228
#16 0x00007fc3849763d8 in ?? () from
/usr/lib/qt5/plugins/xcbglintegrations/libqxcb-glx-integration.so
#17 0x00007fc37fd63870 in QOpenGLContext::makeCurrent(QSurface*) ()
from /usr/lib/libQt5Gui.so.5
#18 0x00007fc3829d1860 in ?? () from /usr/lib/libQt5Quick.so.5
Also mesa is at 13.0.2 and kernel is 4.8.10. Video card is an AMD
HD5670 which as far as I know is not faulty ( doesn't show any signs
of screen corruption or problems in Windows ).
I have to do some stuff to recreate this but I'm able to consistently
recreate it.
Post by Michel Dänzer
Post by Marius Cirsta
Hello,
I have a problem that came once I upgraded my Linux OS ( less known
one called Fruglware OS ).
For some reason with libxcb 1.12 libx11 1.6.4 I'm having this weird
#0 0x00007fc2cf9cd3e1 in poll () from /lib/libc.so.6
#3 0x00007fc2d3984d31 in xcb_wait_for_reply64 (c=0x117aaf0,
request=10682, e=0x7fff869b5d98) at xcb_in.c:560
#4 0x00007fc2d3bea2e8 in _XReply (dpy=0x11798a0, rep=0x7fff869b5e40,
extra=0, discard=0) at xcb_io.c:596
#5 0x00007fc2cc781cd2 in ?? () from /usr/lib/libGL.so.1
#6 0x00007fc2cc781ff7 in ?? () from /usr/lib/libGL.so.1
#7 0x00007fc2216f9239 in ?? () from /usr/lib/dri/r600_dri.so
#8 0x00007fc2216f3ab4 in ?? () from /usr/lib/dri/r600_dri.so
#9 0x00007fc2215dd92c in ?? () from /usr/lib/dri/r600_dri.so
#10 0x00007fc2215de84c in ?? () from /usr/lib/dri/r600_dri.so
#11 0x00007fc22158a67b in ?? () from /usr/lib/dri/r600_dri.so
#12 0x00007fc221592eb4 in ?? () from /usr/lib/dri/r600_dri.so
Please get a backtrace with debugging symbols available for
/usr/lib/libGL.so.1 and /usr/lib/dri/r600_dri.so as well.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Michel Dänzer
2016-12-05 07:42:59 UTC
Permalink
Post by Marius Cirsta
Turns out it was xf86-video-ati after all. Went through the updates I
made to my distro and managed to find this.
Was only xf86-video-ati updated, or Xorg as well? Which versions were
they updated from and to?
Post by Marius Cirsta
Using the latest from git works like a charm. Thanks for the fixes
there Michel.
You're welcome, but there hasn't been any fix since the 7.8.0 release
which could explain this. I suspect that you're simply using DRI3 now,
but that the problem only occurs with DRI2.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Marius Cirsta
2016-12-05 22:20:27 UTC
Permalink
Post by Michel Dänzer
Post by Marius Cirsta
Turns out it was xf86-video-ati after all. Went through the updates I
made to my distro and managed to find this.
Was only xf86-video-ati updated, or Xorg as well? Which versions were
they updated from and to?
Only xf86-video-ati was updated updated from 7.8.0 to include the
patches from git after 7.8.0 ( 11 patches to be found here :
http://git.frugalware.org/gitweb/gitweb.cgi?p=frugalware-current.git;a=tree;f=source/x11/xf86-video-ati
) I'm not sure which one these actually solved the problem ( maybe
more than one ? ).
Post by Michel Dänzer
Post by Marius Cirsta
Using the latest from git works like a charm. Thanks for the fixes
there Michel.
You're welcome, but there hasn't been any fix since the 7.8.0 release
which could explain this. I suspect that you're simply using DRI3 now,
but that the problem only occurs with DRI2.
Judging only by their explanations I'd say one could have easily fixed
my problem. Other than applying these 11 patches ( 10 if you don't
count the build fix for old Xorg ) I really did nothing and not I have
no more freezes.

Let me know if you need additional details for this. I could do some
more testing if needed.
Post by Michel Dänzer
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Michel Dänzer
2016-12-06 00:51:50 UTC
Permalink
Post by Marius Cirsta
Post by Michel Dänzer
Post by Marius Cirsta
Turns out it was xf86-video-ati after all. Went through the updates I
made to my distro and managed to find this.
Was only xf86-video-ati updated, or Xorg as well? Which versions were
they updated from and to?
Only xf86-video-ati was updated updated from 7.8.0 to include the
patches from git after 7.8.0 [...]
I mean which versions the xf86-video-ati and xserver packages were
updated from and to when the problem started.
Post by Marius Cirsta
Post by Michel Dänzer
Post by Marius Cirsta
Using the latest from git works like a charm. Thanks for the fixes
there Michel.
You're welcome, but there hasn't been any fix since the 7.8.0 release
which could explain this. I suspect that you're simply using DRI3 now,
but that the problem only occurs with DRI2.
Judging only by their explanations I'd say one could have easily fixed
my problem.
I'm afraid not. Please provide Xorg log files for the bad and good cases.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Marius Cirsta
2016-12-14 20:04:18 UTC
Permalink
Post by Michel Dänzer
Post by Marius Cirsta
Post by Michel Dänzer
Post by Marius Cirsta
Turns out it was xf86-video-ati after all. Went through the updates I
made to my distro and managed to find this.
Was only xf86-video-ati updated, or Xorg as well? Which versions were
they updated from and to?
Only xf86-video-ati was updated updated from 7.8.0 to include the
patches from git after 7.8.0 [...]
I mean which versions the xf86-video-ati and xserver packages were
updated from and to when the problem started.
Sorry for being late to reply to this. The problem started happening
when I upgraded these:

[11/21/16 20:01] upgraded xorg-server (1.18.4-7 -> 1.19.0-2)
[11/21/16 20:01] upgraded xf86-video-ati (7.7.1-2 -> 7.8.0-1)

But as soon as I downgraded just the xf86-video-ati back to 7.7.1 all
was fine again. Also I've said with the latest patches from Git all is
working fine too.
Post by Michel Dänzer
Post by Marius Cirsta
Post by Michel Dänzer
Post by Marius Cirsta
Using the latest from git works like a charm. Thanks for the fixes
there Michel.
You're welcome, but there hasn't been any fix since the 7.8.0 release
which could explain this. I suspect that you're simply using DRI3 now,
but that the problem only occurs with DRI2.
Judging only by their explanations I'd say one could have easily fixed
my problem.
I'm afraid not. Please provide Xorg log files for the bad and good cases.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Michel Dänzer
2016-12-15 01:55:08 UTC
Permalink
Post by Marius Cirsta
Post by Michel Dänzer
Post by Marius Cirsta
Post by Michel Dänzer
Post by Marius Cirsta
Turns out it was xf86-video-ati after all. Went through the updates I
made to my distro and managed to find this.
Was only xf86-video-ati updated, or Xorg as well? Which versions were
they updated from and to?
Only xf86-video-ati was updated updated from 7.8.0 to include the
patches from git after 7.8.0 [...]
I mean which versions the xf86-video-ati and xserver packages were
updated from and to when the problem started.
Sorry for being late to reply to this. The problem started happening
[11/21/16 20:01] upgraded xorg-server (1.18.4-7 -> 1.19.0-2)
[11/21/16 20:01] upgraded xf86-video-ati (7.7.1-2 -> 7.8.0-1)
But as soon as I downgraded just the xf86-video-ati back to 7.7.1 all
was fine again.
7.7.1 doesn't even build against xorg-server 1.19, let alone run with
it. So I suspect downgrading the driver to 7.7.1 also downgraded
xorg-server to 1.18.4.

Your problem is probably the same as https://bugs.debian.org/846779 , a
xorg-server regression.
Post by Marius Cirsta
Also I've said with the latest patches from Git all is working fine too.
That probably avoids the problem because DRI3 is enabled by default now.
--
Earthling Michel Dänzer | http://www.amd.com
Libre software enthusiast | Mesa and X developer
Loading...