Deadock Xorg runing strace on a client (from within a client terminal)

[Peter Clifton]

Hi,

I was trying to debug a program earlier with strace, and quickly ran
"strace pcb" from within a gnome-terminal on my desktop.

After a short while mousing over the GTK menus, the X11 server
deadlocked and I had to SSH in remotely to kill strace before I could
continue.

I just wanted to ask:

1. Is this an expected problem?
2. Is it a bug?

(If it isn't expected I'd be happy to assist debugging. If it is useful,
I saved the contents of terminal's scrollback buffer after killing
strace.

It ends:

gettimeofday({1219921218, 507707}, NULL) = 0
writev(12, [{"GIOP\1\2\1\0\20\f\0\0", 12}, {"\360\376\205\277\3\0\0\0\0\0\0\0\34\0\0\0\0\0\0\0a\270"..., 2036}, {"ROLE_TOOL_TIP\0\0\0\n\0\0\0ROLE_TREE\0\0\0"..., 1052}], 3) = 3100
poll([{fd=7, events=POLLIN}, {fd=12, events=POLLIN|POLLPRI, revents=POLLIN}, {fd=13, events=POLLIN|POLLPRI}, {fd=15, events=POLLIN|POLLPRI}], 4, -1) = 1
read(12, "GIOP\1\2\1\1$\0\0\0", 12)     = 12
read(12, "\360\376\205\277\0\0\0\0\1\0\0\0\1\0\0\0\f\0\0\0\1\1\1"..., 36) = 36
select(5, [4], [4], NULL, NULL)         = 1 (out [4])
writev(4, [{"5\30\4\0E\33@\7X\0@\7\1\0Q\1\230\4\5\0F\33@\7E\33@\7(\0"..., 3248}], 1) = 3248
select(5, [4], [], NULL, NULL)          = 1 (in [4])
read(4, "\34%Ef/\33@\7\16\1\0\0/\255\266\3\0\210\322\277\332\23"..., 4096) = 1468
read(4, 0x8f9bb44, 4096)                = -1 EAGAIN (Resource temporarily unavailable)
ioctl(14, 0x4008642a, 0xbf860858)       = 0
ioctl(14, 0xc0046444, 0xbf8607d0)       = 0
ioctl(14, 0x40046445, 0xbf860788)       = 0
ioctl(14, 0xc0046444, 0xbf860810)       = 0
ioctl(14, 0x40046445, 0xbf8607c8)       = 0
ioctl(14, 0x40186443, 0xbf860894)       = 0
ioctl(14, 0xc0046444, 0xbf860840)       = 0
ioctl(14, 0x4008642b, 0xbf860890)       = 0
ioctl(14, 0x4008642a, 0xbf8608c8)       = 0
ioctl(14, 0x4008642b, 0xbf860900)       = 0
ioctl(14, 0x4008642a, 0xbf860868)       = 0
ioctl(14, 0xc0046444, 0xbf860820)       = 0
ioctl(14, 0x40046445, 0xbf8607d8)       = 0
ioctl(14, 0x40186443, 0xbf8608a4)       = 0
ioctl(14, 0xc0046444, 0xbf860850)       = 0
ioctl(14, 0x4008642b, 0xbf8608a0)       = 0
ioctl(14, 0x4008642a, 0xbf8608f8)       = 0
ioctKilled

(IIRC, it stalled on a read call, can't be 100% sure it is the last one
above, but it looks likely).

If it is of relevance, the program under test uses OpenGL. I have a
recent git version of mesa, libdrm, and the 2.4.1 Intel Xorg driver. GEM
isn't working (in spite having tried to patch the kernel). I suspect
made a mistake backporting the pci_read_base patches from the LKML.

[   44.080535] [drm:i915_gem_detect_bit_6_swizzle] *ERROR* pci_read_base failed: -19


Best regards,

-- 
Peter Clifton

Electrical Engineering Division,
Engineering Department,
University of Cambridge,
9, JJ Thomson Avenue,
Cambridge
CB3 0FA

Tel: +44 (0)7729 980173 - (No signal in the lab!)