I'm back :-/
I just triggered the kswapd bug on 2.6.39.1, which has the
cond_resched in shrink_slab. This time my system's still usable (I'm
tying this email on it), but kswapd0 is taking 100% cpu. It *does*
schedule (tested by setting its affinity the same as another CPU hog
and confirming that each one gets 50%).
It appears to be calling i915_gem_inactive_shrink in a loop. I have
probes on entry and return of i915_gem_inactive_shrink and on return
of shrink_slab. I see:
kswapd0 47 [000] 59599.956573: mm_vmscan_kswapd_wake: nid=0 order=0
kswapd0 47 [000] 59599.956575: shrink_zone:
(ffffffff810c848c) priority=12 zone=ffff8801005fe000
kswapd0 47 [000] 59599.956576: shrink_zone_return:
(ffffffff810c848c <- ffffffff810c96c6) arg1=0
kswapd0 47 [000] 59599.956578: i915_gem_inactive_shrink:
(ffffffffa0081e48) gfp_mask=d0 nr_to_scan=0
kswapd0 47 [000] 59599.956589: shrink_return:
(ffffffffa0081e48 <- ffffffff810c6a62) arg1=320
kswapd0 47 [000] 59599.956589: shrink_slab_return:
(ffffffff810c69f5 <- ffffffff810c96ec) arg1=0
kswapd0 47 [000] 59599.956592: i915_gem_inactive_shrink:
(ffffffffa0081e48) gfp_mask=d0 nr_to_scan=0
kswapd0 47 [000] 59599.956602: shrink_return:
(ffffffffa0081e48 <- ffffffff810c6a62) arg1=320
kswapd0 47 [000] 59599.956603: shrink_slab_return:
(ffffffff810c69f5 <- ffffffff810c96ec) arg1=0
kswapd0 47 [000] 59599.956605: shrink_zone:
(ffffffff810c848c) priority=12 zone=ffff8801005fee00
kswapd0 47 [000] 59599.956606: shrink_zone_return:
(ffffffff810c848c <- ffffffff810c96c6) arg1=0
kswapd0 47 [000] 59599.956608: i915_gem_inactive_shrink:
(ffffffffa0081e48) gfp_mask=d0 nr_to_scan=0
kswapd0 47 [000] 59599.956609: shrink_return:
(ffffffffa0081e48 <- ffffffff810c6a62) arg1=0
kswapd0 47 [000] 59599.956610: shrink_slab_return:
(ffffffff810c69f5 <- ffffffff810c96ec) arg1=0
kswapd0 47 [000] 59599.956611: mm_vmscan_kswapd_wake: nid=0 order=0
kswapd0 47 [000] 59599.956612: shrink_zone:
(ffffffff810c848c) priority=12 zone=ffff8801005fe000
kswapd0 47 [000] 59599.956614: shrink_zone_return:
(ffffffff810c848c <- ffffffff810c96c6) arg1=0
kswapd0 47 [000] 59599.956616: i915_gem_inactive_shrink:
(ffffffffa0081e48) gfp_mask=d0 nr_to_scan=0
kswapd0 47 [000] 59599.956617: shrink_return:
(ffffffffa0081e48 <- ffffffff810c6a62) arg1=0
kswapd0 47 [000] 59599.956618: shrink_slab_return:
(ffffffff810c69f5 <- ffffffff810c96ec) arg1=0
kswapd0 47 [000] 59599.956620: i915_gem_inactive_shrink:
(ffffffffa0081e48) gfp_mask=d0 nr_to_scan=0
kswapd0 47 [000] 59599.956621: shrink_return:
(ffffffffa0081e48 <- ffffffff810c6a62) arg1=0
kswapd0 47 [000] 59599.956621: shrink_slab_return:
(ffffffff810c69f5 <- ffffffff810c96ec) arg1=0
kswapd0 47 [000] 59599.956623: shrink_zone:
(ffffffff810c848c) priority=12 zone=ffff8801005fee00
kswapd0 47 [000] 59599.956624: shrink_zone_return:
(ffffffff810c848c <- ffffffff810c96c6) arg1=0
kswapd0 47 [000] 59599.956626: i915_gem_inactive_shrink:
(ffffffffa0081e48) gfp_mask=d0 nr_to_scan=0
kswapd0 47 [000] 59599.956627: shrink_return:
(ffffffffa0081e48 <- ffffffff810c6a62) arg1=0
kswapd0 47 [000] 59599.956628: shrink_slab_return:
(ffffffff810c69f5 <- ffffffff810c96ec) arg1=0
kswapd0 47 [000] 59599.956629: mm_vmscan_kswapd_wake: nid=0 order=0
The command was:
perf record -g -aR -p 47 -e probe:i915_gem_inactive_shrink -e
probe:shrink_return -e probe:shrink_slab_return -e probe:shrink_zone
-e probe:shrink_zone_return -e probe:kswapd_try_to_sleep -e
vmscan:mm_vmscan_kswapd_sleep -e vmscan:mm_vmscan_kswapd_wake -e
vmscan:mm_vmscan_wakeup_kswapd -e vmscan:mm_vmscan_lru_shrink_inactive
-e probe:wakeup_kswapd; perf script
(shrink_return is i915_gem_inactive_shrink's return. sorry, badly named.)
It looks like something kswapd_try_to_sleep is not getting called.
I do not know how to reproduce this, but I'll leave it running overnight.
--Andy
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>