Bug 44091

Created attachment 54744 [details]
Xorg.0.log

2.17.0-216-g56e9727, what is that? Can you confirm using upstream whilst I try to reproduce?

I was in different branch, but it was tracking master and I had no modifications except disabling legacy i810.

latest commit was 84d97bdba02b909369b54de21425ffc9f6ad581a

Also xorg-x11-server-Xorg-1.11.99.1-8 is extremely buggy. Is there an update in fedora yet?

It is unfortunate it is extremely buggy still after 25 years, but here the results for xserver git e7df42ab68e30588a5e32ed543b0711821daf009:

with -vo gl, screen does not get corrupted anymore.

however, when I was just using gnome-terminal, I got this:

[ 1039.951478] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 1039.951545] [drm] capturing error event; look for more information in /debug/dri/0/i915_error_state
[ 1039.953924] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 41006 at 41004, next 41007)

also, screen got corrupted while I was only typing in gnome-terminal. (jee.png)

I now had wmcpu running, maybe it freaked out the driver(s).

I killed it, hoping I don't need to reboot so often.

Created attachment 54747 [details]
intel_reg_dumper after gpu hang

Created attachment 54748 [details]
i915_error_state after gpu hang

Created attachment 54749 [details]
screen corruption

If you attach the /sys/kernel/debug/dri/0/i915_error_state I can tell you which bug you hit next...

2011-12-23 16:24:57.731957973 <3>[ 3316.523105] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
2011-12-23 16:24:57.731959168 <3>[ 3316.523123] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 172391 at 172387, next 172392)
2011-12-23 16:26:26.411963994 <3>[ 3405.141176] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
2011-12-23 16:26:26.411965043 <3>[ 3405.141198] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 177735 at 177731, next 177736)

intel_reg_dumper and error state output is the same as in the attached files

Sorry missed the error state you posted. That looks like the new code to produce more compact batch buffers is at fault.

This should fix that last GPU hang:

commit 98f15fc61361b7f1e01969f8d4237c13e93e3fb0
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Fri Dec 23 17:25:42 2011 +0000

    sna: Don't align pwrite to cachelines for doing discontiguous copies
    
    The batch compaction breaks the 1:1 mapping between the cpu buffer and
    the bo, so we can no longer safely align the transfer to whole
    cachelines.
    
    References: https://bugs.freedesktop.org/show_bug.cgi?id=44091
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

That was some debugging from a bunch of hex numbers ;)

That fix definitely made a stability improvement, thank you.

There's a new drawing bug, though, please see sna-drawing-chrome.png
(xrefresh does not fix it)

Created attachment 54753 [details]
sna-drawing-chrome.png

with sna , xf86-video-intel 98f15fc61361b7f1e01969f8d4237c13e93e3fb0

this with efc8d04fc114e9966e5ca00600f9663ecf03a5ca

Backtrace:
0: X (xorg_backtrace+0x26) [0x472d26]
1: X (0x400000+0x78229) [0x478229]
2: /lib64/libpthread.so.0 (0x300c800000+0xf4f0) [0x300c80f4f0]
3: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7fe110268000+0x50a57) [0x7fe1102b8a57]
4: X (miCopyRegion+0x18a) [0x58339a]
5: X (miDoCopy+0x392) [0x583892]
6: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7fe110268000+0x50312) [0x7fe1102b8312]
7: X (0x400000+0x12923f) [0x52923f]
8: X (0x400000+0x31923) [0x431923]
9: X (0x400000+0x358a1) [0x4358a1]
10: X (0x400000+0x22eca) [0x422eca]
11: /lib64/libc.so.6 (__libc_start_main+0xed) [0x3004c2169d]
12: X (0x400000+0x2322d) [0x42322d]
Segmentation fault at address 0x20

Fatal server error:
Caught signal 11 (Segmentation fault). Server aborting

hmm not very useful because I had optimizations enabled

$ addr2line -e /usr/lib64/xorg/modules/drivers/intel_drv.so.2.17.0-efc8d04fc114e9966e5ca00600f9663ecf03a5ca 0x50312
sna_accel.c:0

Ok, that should be fixed with

commit b86e4f59299f935d5a0ea8375da97e6fc57571f9
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Sat Dec 24 11:45:27 2011 +0000

    sna: Check that the copy dst is attached before replacing damage
    
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Created attachment 54848 [details]
sna-git-655a96cd5f-xpdf.png

SNA git-655a96cd5f seems pretty stable now, but xpdf has some oddities (and space is missing between the words in the menus) and e.g. xscreensaver has also spaces missing (it says "Pleaseenteryourpassword.").
I did not try does UXA work.

Oops, discarding spaces is not a good idea after all.

commit 8fc21328a0bdf87fde35d68d2b27834011acde7b
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Tue Dec 27 15:26:46 2011 +0000

    sna: Don't discard empty glyphs, just skip them
    
    A space is encoded as a 1x1 blank glyph, but we still need to advance by
    its character width and so we cannot simply discard the glyph.
    
    References: https://bugs.freedesktop.org/show_bug.cgi?id=44091
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

8fc21328a0bdf87fde35d68d2b27834011acde7b fixed xpdf drawing issues, except that the spaces are still missing (also for xscreensaver).

Also, when I do "mplayer -vo gl -fs *" , and play each video 1s, after a little time every gnome-terminal goes (almost totally) black, and dockapps all go black.
when I move another app over the gnome-terminals, underneath is uncovered correctly rendered text output....  but it goes almost black again when I type some commands.  xterm works ok, however. 
I can't fix this with xrefresh, also, new gnome-terminals are also screwed.
Note that this (gnome-terminal getting screwed) happened also with older SNA version from last Saturday.

regs are the same as in attachment 3 [details] [review], except
-                      DSPASURF: 0x04ea7000
+                      DSPASURF: 0x04396000

Created attachment 54862 [details]
sna-gnome-terminal-ls.png SNA 8fc21328a0

ignore the auto-generated link to attachment 3 [details] [review]...  I meant attachment number three in this bug report. (attachment 54747 [details])

I tried mplayer again with 337635ab97bbfc9b4455eadb63214783bb90bb2b

It showed some movies OK, but then froze in black screen, sysrq did not work and I had to power cycle computer.  after reboot Xorg.0.log was 0 bytes long.

BTW, when I exited Xorg when using intel git-8fc21328a0, it whined
X: free(): invalid pointer ... X [0x441c07] [0x422f41]

(gdb) disass 0x441c07
Dump of assembler code for function CloseDownEvents:
   0x0000000000441bf0 <+0>:	push   %rbx
   0x0000000000441bf1 <+1>:	callq  0x44ea80 <GetMaximumEventsNum>
   0x0000000000441bf6 <+6>:	mov    0x3b8fcb(%rip),%rbx        # 0x7fabc8
   0x0000000000441bfd <+13>:	mov    %eax,%esi
   0x0000000000441bff <+15>:	mov    (%rbx),%rdi
   0x0000000000441c02 <+18>:	callq  0x44f070 <FreeEventList>
   0x0000000000441c07 <+23>:	movq   $0x0,(%rbx)
   0x0000000000441c0e <+30>:	pop    %rbx
   0x0000000000441c0f <+31>:	retq   
End of assembler dump.

...


   0x0000000000422f23 <+995>:	add    $0x8,%rax
   0x0000000000422f27 <+999>:	cmp    %rcx,%rax
   0x0000000000422f2a <+1002>:	movq   $0x0,0xb0(%rdx)
   0x0000000000422f35 <+1013>:	jne    0x422f20 <main+992>
   0x0000000000422f37 <+1015>:	callq  0x42ac20 <CloseDownDevices>
   0x0000000000422f3c <+1020>:	callq  0x441bf0 <CloseDownEvents>
   0x0000000000422f41 <+1025>:	mov    0x2c(%rbx),%r13d
   0x0000000000422f45 <+1029>:	sub    $0x1,%r13d
   0x0000000000422f49 <+1033>:	js     0x422fae <main+1134>
   0x0000000000422f4b <+1035>:	mov    %r14d,%r15d
   0x0000000000422f4e <+1038>:	xchg   %ax,%ax
   0x0000000000422f50 <+1040>:	mov    %r13d,%edi
   0x0000000000422f53 <+1043>:	movslq %r13d,%r14
   0x0000000000422f56 <+1046>:	callq  0x454dd0 <FreeScratchPixmapsForScreen>
   0x0000000000422f5b <+1051>:	mov    %r13d,%edi

337635ab97:

[ 24361.644] (WW) intel(0): sna_dri_get_msc:1633 get vblank counter failed: Invalid argument
[ 24368.038] (II) AIGLX: Suspending AIGLX clients for VT switch
[ 24368.100] sna_page_flip: failed to add fb: 1920x1200 depth=24, bpp=32, pitch=7680
[ 24374.665] (II) AIGLX: Resuming AIGLX clients after VT switch
[ 24381.244] (WW) intel(0): sna_dri_get_msc:1633 get vblank counter failed: Invalid argument

[24344.710131] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[24344.710189] [drm] capturing error event; look for more information in /debug/dri/0/i915_error_state
[24344.712681] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 1324359 at 1324358, next 1324364)

[24351.033712] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[24351.033746] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 1324497 at 1324365, next 1324498)
[24351.395505] usb 2-1.4: unlink qh1-0e01/ffff8803b8ab5180 start 0 [1/3 us]

[24364.304438] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[24364.304457] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 1324802 at 1324800, next 1324804)
[24370.608033] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[24370.608050] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 1324929 at 1324800, next 1324930)

[24377.059537] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[24377.059572] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 1325056 at 1324800, next 1325057)

Created attachment 54909 [details]
i915_error_state.337635ab97.txt

after comment #25 I tried to play a movie, but X crashed and gpu wedged.

# addr2line -e /usr/lib64/xorg/modules/drivers/intel_drv.so.337635ab97 0xdd289 0xdd2bd 0xde459 
/usr/include/xorg/list.h:183
/usr/include/xorg/list.h:205
/wrk/safari/cvs/xf86-video-intel/src/sna/sna_dri.c:608

line 608:
                info->drawable_id = None;



(WW) intel(0): flip queue failed: Input/output error
AUDIT: Wed Dec 28 20:19:13 2011: 5671: client 31 disconnected

Backtrace:
0: X (xorg_backtrace+0x26) [0x472d26]
1: X (0x400000+0x78229) [0x478229]
2: /lib64/libpthread.so.0 (0x300c800000+0xf4f0) [0x300c80f4f0]
3: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f75c1788000+0xdd289) [0x7f75c1865289]
4: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f75c1788000+0xdd2bd) [0x7f75c18652bd]
5: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f75c1788000+0xde459) [0x7f75c1866459]
6: X (FreeClientResources+0xf5) [0x45cc45]
7: X (CloseDownClient+0x5a) [0x434d2a]
8: X (0x400000+0x358d8) [0x4358d8]
9: X (0x400000+0x22eca) [0x422eca]
10: /lib64/libc.so.6 (__libc_start_main+0xed) [0x3004c2169d]
11: X (0x400000+0x2322d) [0x42322d]
Segmentation fault at address (nil)

Fatal server error:
Caught signal 11 (Segmentation fault). Server aborting



..............


then when I tried to restart X, it segfaulted on startup.  had to powercycle again.

(WW) intel(0): cannot enable XVideo whilst the GPU is wedged
(II) intel(0): Overlay video not supported on this hardware
(WW) intel(0): Disabling Xv because no adaptors could be initialized.
(WW) intel(0): cannot enable DRI2 whilst the GPU is wedged
(==) intel(0): hotplug detection: "enabled"
(--) RandR disabled
(II) SELinux: Disabled on system
(II) AIGLX: Screen 0 is not DRI2 capable
(II) AIGLX: Screen 0 is not DRI capable
(II) AIGLX: Loaded and initialized swrast
(II) GLX: Initialized DRISWRAST GL provider for screen 0

Backtrace:
0: X (xorg_backtrace+0x26) [0x472d26]
1: X (0x400000+0x78229) [0x478229]
2: /lib64/libpthread.so.0 (0x300c800000+0xf4f0) [0x300c80f4f0]
3: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f506ef1b000+0x814ab) [0x7f506ef9c4ab]
4: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f506ef1b000+0x839e9) [0x7f506ef9e9e9]
5: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f506ef1b000+0x6eab9) [0x7f506ef89ab9]
6: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7f506ef1b000+0x7eba1) [0x7f506ef99ba1]
7: X (0x400000+0xc7b47) [0x4c7b47]
8: X (0x400000+0x22d64) [0x422d64]
9: /lib64/libc.so.6 (__libc_start_main+0xed) [0x3004c2169d]
10: X (0x400000+0x2322d) [0x42322d]
Segmentation fault at address 0x52

Fatal server error:
Caught signal 11 (Segmentation fault). Server aborting



# addr2line -e /usr/lib64/xorg/modules/drivers/intel_drv.so.337635ab97 0x814ab 0x839e9 0x6eab9 0x7eba1
/wrk/safari/cvs/xf86-video-intel/src/sna/sna_glyphs.c:176
/wrk/safari/cvs/xf86-video-intel/src/sna/sna_glyphs.c:926
/wrk/safari/cvs/xf86-video-intel/src/sna/sna_accel.c:9392
/wrk/safari/cvs/xf86-video-intel/src/sna/sna_driver.c:212

line 176:
                sna_pixmap(pixmap)->pinned = true;

The hangs are nothing to do with the ddx, but mesa.

Great.. so do I change Product to "Mesa"?

what does this crash have to do with?
xserver, libdrm, kernel, pixman, mesa, xf86-video-intel?

intel git-c9f7f10b

[mi] EQ overflow continuing.  200 events have been dropped.

Backtrace:
0: X (xorg_backtrace+0x26) [0x472d26]
1: X (QueuePointerEvents+0x5b) [0x44f7bb]
2: /usr/lib64/xorg/modules/input/evdev_drv.so (0x7ff24b792000+0x5881) [0x7ff24b797881]
3: X (0x400000+0x91957) [0x491957]
4: X (0x400000+0xb975e) [0x4b975e]
5: /lib64/libpthread.so.0 (0x300c800000+0xf4f0) [0x300c80f4f0]
6: /lib64/libc.so.6 (ioctl+0x7) [0x3004ce7b67]
7: /usr/lib64/libdrm.so.2 (drmIoctl+0x28) [0x7ff24cf70398]
8: /usr/lib64/libdrm.so.2 (drmCommandNone+0x16) [0x7ff24cf72bd6]
9: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7ff24ca44000+0x4bbbf) [0x7ff24ca8fbbf]

this is in kgem_throttle:
  kgem->wedged |= drmCommandNone(kgem->fd, DRM_I915_GEM_THROTTLE) == -EIO;

10: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7ff24ca44000+0x6ec71) [0x7ff24cab2c71]
11: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7ff24ca44000+0x6ecb4) [0x7ff24cab2cb4]
12: X (BlockHandler+0x4a) [0x439aaa]
13: X (WaitForSomething+0x269) [0x47b899]
14: X (0x400000+0x355b2) [0x4355b2]
15: X (0x400000+0x22eca) [0x422eca]
16: /lib64/libc.so.6 (__libc_start_main+0xed) [0x3004c2169d]
17: X (0x400000+0x2322d) [0x42322d]
(EE) intel(0): Detected a hung GPU, disabling acceleration.
(EE) intel(0): When reporting this, please include i915_error_state from debugfs and the full dmesg.
[mi] Increasing EQ size to 512 to prevent dropped events.
[mi] EQ processing has resumed after 281 dropped events.
[mi] This may be caused my a misbehaving driver monopolizing the server's resources.
AUDIT: Fri Dec 30 00:25:54 2011: 15911: client 33 disconnected
AUDIT: Fri Dec 30 00:25:54 2011: 15911: client 33 connected from local host ( uid=500 gid=509 pid=22546 )
AUDIT: Fri Dec 30 00:27:17 2011: 15911: client 33 disconnected
AUDIT: Fri Dec 30 00:27:17 2011: 15911: client 33 connected from local host ( uid=500 gid=509 pid=22601 )


Backtrace:
0: X (xorg_backtrace+0x26) [0x472d26]
1: X (0x400000+0x78229) [0x478229]
2: /lib64/libpthread.so.0 (0x300c800000+0xf4f0) [0x300c80f4f0]
3: /usr/lib64/libpixman-1.so.0 (0x3025000000+0x5f668) [0x302505f668]
4: /usr/lib64/libpixman-1.so.0 (0x3025000000+0x7d032) [0x302507d032]
5: /usr/lib64/libpixman-1.so.0 (pixman_fill+0x31) [0x302500a081]
6: /usr/lib64/xorg/modules/libfb.so (fbFill+0x30d) [0x7ff24c01458d]
7: /usr/lib64/xorg/modules/drivers/intel_drv.so (0x7ff24ca44000+0x697c4) 
[0x7ff24caad7c4]

this is
src/sna/sna_accel.c:7628
} while (--n);

(after if (region.data == NULL) { )

8: X (0x400000+0x1273d0) [0x5273d0]
9: X (miPaintWindow+0x1b0) [0x585ee0]
10: X (miWindowExposures+0xc2) [0x586092]
11: X (0x400000+0xa86f0) [0x4a86f0]
12: X (miHandleValidateExposures+0x68) [0x59ea78]
13: X (MapWindow+0x2b7) [0x46c857]
14: X (0x400000+0x30160) [0x430160]
15: X (0x400000+0x358a1) [0x4358a1]
16: X (0x400000+0x22eca) [0x422eca]
17: /lib64/libc.so.6 (__libc_start_main+0xed) [0x3004c2169d]
18: X (0x400000+0x2322d) [0x42322d]
Segmentation fault at address (nil)

Fatal server error:
Caught signal 11 (Segmentation fault). Server aborting


........

I now am running latest mesa git, my goal is to get 24h uptime!

Created attachment 54961 [details]
i915_error_state.c9f7f10b.txt

<3>[68076.585833] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
<6>[68076.585899] [drm] capturing error event; look for more information in /debug/dri/0/i915_error_state
<3>[68076.588216] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 2700529 at 2700527, next 2700531)

Created attachment 54979 [details]
i915_error_state.0dc5c0651cb69.txt

Closing as a dupe of the residual bug.

*** This bug has been marked as a duplicate of bug 44364 ***