kern/49089: Incorrect locking in i915_capture_error_state

[degroote%NetBSD.org@localhost]

>Number:         49089
>Category:       kern
>Synopsis:       Incorrect locking in i915_capture_error_state
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Aug 07 19:10:00 +0000 2014
>Originator:     Arnaud Degroote
>Release:        6.99.49
>Organization:
>Environment:
NetBSD hermes 6.99.49 NetBSD 6.99.49 (DRMKMS) #4: Tue Aug  5 22:10:37 CEST 2014 
 
zul@hermes:/home/zul/netbsd-dev/build_temp/amd64/objdir/sys/arch/amd64/compile/DRMKMS
 amd64

>Description:
On my IronLake card, when launching firefox, I often hit the following kernel 
panic 

#0  0xffffffff80678f6f in cpu_reboot (howto=howto@entry=260, 
bootstr=bootstr@entry=0x0)
    at /home/zul/netbsd-dev/src/sys/arch/amd64/amd64/machdep.c:671
#1  0xffffffff808c4634 in vpanic (fmt=0xffffffff80d31620 "kernel %sassertion 
\"%s\" failed: file \"%s\", line %d ", 
    ap=ap@entry=0xfffffe8043463d60) at 
/home/zul/netbsd-dev/src/sys/kern/subr_prf.c:284
#2  0xffffffff80a5f7d3 in kern_assert (fmt=fmt@entry=0xffffffff80d31620 "kernel 
%sassertion \"%s\" failed: file \"%s\", line %d ")
    at /home/zul/netbsd-dev/src/sys/lib/libkern/kern_assert.c:51
#3  0xffffffff803b14ec in DRM_SPIN_WAITERS_P (interlock=0xffff80000360cca0, 
q=0xffff80000360c688)
    at 
/home/zul/netbsd-dev/src/sys/external/bsd/drm2/include/drm/drm_wait_netbsd.h:89
#4  i915_record_ring_state (ering=0xfffffe8019fa7210, ring=0xffff80000360c560, 
dev=0xfffffe80435dd810)
    at 
/home/zul/netbsd-dev/src/sys/external/bsd/drm2/dist/drm/i915/i915_gpu_error.c:812
#5  i915_gem_record_rings (error=0xfffffe8019fa7010, dev=0xfffffe80435dd810)
    at 
/home/zul/netbsd-dev/src/sys/external/bsd/drm2/dist/drm/i915/i915_gpu_error.c:924
#6  i915_capture_error_state (dev=0xfffffe80435dd810, wedged=<optimized out>, 
error_msg=0xfffffe8043463e88 "Ring hung")
    at 
/home/zul/netbsd-dev/src/sys/external/bsd/drm2/dist/drm/i915/i915_gpu_error.c:1191
#7  0xffffffff803b633f in i915_handle_error (dev=0xfffffe80435dd810, 
wedged=<optimized out>, fmt=<optimized out>)
    at 
/home/zul/netbsd-dev/src/sys/external/bsd/drm2/dist/drm/i915/i915_irq.c:2364
#8  0xffffffff80617732 in callout_softclock (v=<optimized out>) at 
/home/zul/netbsd-dev/src/sys/kern/kern_timeout.c:736
#9  0xffffffff8060c208 in softint_execute (l=<optimized out>, s=2, 
si=0xffff8000224090c0)
    at /home/zul/netbsd-dev/src/sys/kern/kern_softint.c:589


>How-To-Repeat:

>Fix:
Adding some lock dance around DRM_SPIN_WAITERS_P in i915_record_ring_state 
should be sufficient, but I'm not familiar enough with the code to know if it 
is the good solution.