Thanks, I'm giving it a spin. Will report back in a couple of days.
On Wed, Jul 16, 2025 at 11:02:01PM +1000, Jonathan Gray wrote:
> On Wed, Jul 16, 2025 at 12:04:12PM +0000, Lucas Gabriel Vuotto wrote:
> > >Synopsis: splassert: pool_do_put: want 0 have 1
> > >Category: kernel
> > >Environment:
> > System : OpenBSD 7.7
> > Details : OpenBSD 7.7-current (GENERIC.MP) #53: Fri Jul 11 07:40:44
> > MDT 2025
> >
> > [email protected]:/usr/src/sys/arch/amd64/compile/GENERIC.MP
> >
> > Architecture: OpenBSD.amd64
> > Machine : amd64
> > >Description:
> > Yesterday, this happened. The OS didn't crash and I didn't
> > notice anything in particular while it happened; it was just printed in
> > xconsole. It didn't happen before since installing this snapshot from
> > last Saturday, despite doing the same things with the computer.
> >
> > Jul 16 01:36:57 nuc /bsd: splassert: pool_do_put: want 0 have 1
> > Jul 16 01:36:57 nuc /bsd: Starting stack trace...
> > Jul 16 01:36:57 nuc /bsd: pool_do_put(ffffffff82ac3468,fffffd86b1cee448) at
> > pool_do_put+0x43
> > Jul 16 01:36:57 nuc /bsd: pool_put(ffffffff82ac3468,fffffd86b1cee448) at
> > pool_put+0x79
> > Jul 16 01:36:57 nuc /bsd: i915_fence_release(ffff80002e7d75c0) at
> > i915_fence_release+0x86
> > Jul 16 01:36:57 nuc /bsd: dma_fence_chain_release(ffff80002e81c800) at
> > dma_fence_chain_release+0x104
> > Jul 16 01:36:57 nuc /bsd: dma_fence_chain_walk(ffff800003274900) at
> > dma_fence_chain_walk+0x1e5
> > Jul 16 01:36:57 nuc /bsd:
> > dma_fence_chain_enable_signaling(ffff800003274900) at
> > dma_fence_chain_enable_signaling+0x97
> > Jul 16 01:36:57 nuc /bsd: dma_fence_chain_timo(ffff800003274900) at
> > dma_fence_chain_timo+0x21
> > Jul 16 01:36:57 nuc /bsd:
> > softclock_process_tick_timeout(ffff800003274970,1) at
> > softclock_process_tick_timeout+0x10f
> > Jul 16 01:36:57 nuc /bsd: softclock(0) at softclock+0x10e
> > Jul 16 01:36:57 nuc /bsd: softintr_dispatch(0) at softintr_dispatch+0x101
> > Jul 16 01:36:57 nuc /bsd: dosoftint(0) at dosoftint+0x3f
> > Jul 16 01:36:57 nuc /bsd: Xsoftclock() at Xsoftclock+0x27
> > Jul 16 01:36:57 nuc /bsd: acpicpu_idle() at acpicpu_idle+0x228
> > Jul 16 01:36:57 nuc /bsd: sched_idle(ffffffff829daff0) at sched_idle+0x24b
> > Jul 16 01:36:57 nuc /bsd: end trace frame: 0x0, count: 243
> > Jul 16 01:36:57 nuc /bsd: End of stack trace.
> > Jul 16 01:47:49 nuc /bsd: splassert: pool_do_put: want 0 have 1
> > Jul 16 01:47:49 nuc /bsd: Starting stack trace...
> > Jul 16 01:47:49 nuc /bsd: pool_do_put(ffffffff82ac3468,fffffd856305e880) at
> > pool_do_put+0x43
> > Jul 16 01:47:49 nuc /bsd: pool_put(ffffffff82ac3468,fffffd856305e880) at
> > pool_put+0x79
> > Jul 16 01:47:49 nuc /bsd: i915_fence_release(ffff80000352b040) at
> > i915_fence_release+0x4c
> > Jul 16 01:47:49 nuc /bsd: dma_fence_chain_release(ffff80000355e900) at
> > dma_fence_chain_release+0x104
> > Jul 16 01:47:49 nuc /bsd: dma_fence_chain_walk(ffff800003471300) at
> > dma_fence_chain_walk+0x1e5
> > Jul 16 01:47:49 nuc /bsd:
> > dma_fence_chain_enable_signaling(ffff800003471300) at
> > dma_fence_chain_enable_signaling+0x97
> > Jul 16 01:47:49 nuc /bsd: dma_fence_chain_timo(ffff800003471300) at
> > dma_fence_chain_timo+0x21
> > Jul 16 01:47:49 nuc /bsd:
> > softclock_process_tick_timeout(ffff800003471370,1) at
> > softclock_process_tick_timeout+0x10f
> > Jul 16 01:47:49 nuc /bsd: softclock(0) at softclock+0x10e
> > Jul 16 01:47:49 nuc /bsd: softintr_dispatch(0) at softintr_dispatch+0x101
> > Jul 16 01:47:49 nuc /bsd: dosoftint(0) at dosoftint+0x3f
> > Jul 16 01:47:49 nuc /bsd: Xsoftclock() at Xsoftclock+0x27
> > Jul 16 01:47:49 nuc /bsd: acpicpu_idle() at acpicpu_idle+0x228
> > Jul 16 01:47:49 nuc /bsd: sched_idle(ffffffff829daff0) at sched_idle+0x24b
> > Jul 16 01:47:49 nuc /bsd: end trace frame: 0x0, count: 243
> > Jul 16 01:47:49 nuc /bsd: End of stack trace.
> > Jul 16 01:47:49 nuc /bsd: splassert: pool_do_put: want 0 have 1
> > Jul 16 01:47:49 nuc /bsd: Starting stack trace...
> > Jul 16 01:47:49 nuc /bsd: pool_do_put(ffffffff82ac3468,fffffd875c911b58) at
> > pool_do_put+0x43
> > Jul 16 01:47:49 nuc /bsd: pool_put(ffffffff82ac3468,fffffd875c911b58) at
> > pool_put+0x79
> > Jul 16 01:47:49 nuc /bsd: i915_fence_release(ffff80000352b040) at
> > i915_fence_release+0x86
> > Jul 16 01:47:49 nuc /bsd: dma_fence_chain_release(ffff80000355e900) at
> > dma_fence_chain_release+0x104
> > Jul 16 01:47:49 nuc /bsd: dma_fence_chain_walk(ffff800003471300) at
> > dma_fence_chain_walk+0x1e5
> > Jul 16 01:47:49 nuc /bsd:
> > dma_fence_chain_enable_signaling(ffff800003471300) at
> > dma_fence_chain_enable_signaling+0x97
> > Jul 16 01:47:49 nuc /bsd: dma_fence_chain_timo(ffff800003471300) at
> > dma_fence_chain_timo+0x21
> > Jul 16 01:47:49 nuc /bsd:
> > softclock_process_tick_timeout(ffff800003471370,1) at
> > softclock_process_tick_timeout+0x10f
> > Jul 16 01:47:49 nuc /bsd: softclock(0) at softclock+0x10e
> > Jul 16 01:47:49 nuc /bsd: softintr_dispatch(0) at softintr_dispatch+0x101
> > Jul 16 01:47:49 nuc /bsd: dosoftint(0) at dosoftint+0x3f
> > Jul 16 01:47:49 nuc /bsd: Xsoftclock() at Xsoftclock+0x27
> > Jul 16 01:47:49 nuc /bsd: acpicpu_idle() at acpicpu_idle+0x228
> > Jul 16 01:47:49 nuc /bsd: sched_idle(ffffffff829daff0) at sched_idle+0x24b
> > Jul 16 01:47:49 nuc /bsd: end trace frame: 0x0, count: 243
> > Jul 16 01:47:49 nuc /bsd: End of stack trace.
> > Jul 16 02:03:25 nuc /bsd: splassert: pool_do_put: want 0 have 1
> > Jul 16 02:03:25 nuc /bsd: Starting stack trace...
> > Jul 16 02:03:25 nuc /bsd: pool_do_put(ffffffff82ac3468,fffffd84207675a8) at
> > pool_do_put+0x43
> > Jul 16 02:03:25 nuc /bsd: pool_put(ffffffff82ac3468,fffffd84207675a8) at
> > pool_put+0x79
> > Jul 16 02:03:25 nuc /bsd: i915_fence_release(ffff800003687200) at
> > i915_fence_release+0x86
> > Jul 16 02:03:25 nuc /bsd: dma_fence_chain_release(ffff800003509600) at
> > dma_fence_chain_release+0x104
> > Jul 16 02:03:25 nuc /bsd: dma_fence_chain_walk(ffff800003684e00) at
> > dma_fence_chain_walk+0x1e5
> > Jul 16 02:03:25 nuc /bsd:
> > dma_fence_chain_enable_signaling(ffff800003684e00) at
> > dma_fence_chain_enable_signaling+0x97
> > Jul 16 02:03:25 nuc /bsd: dma_fence_chain_timo(ffff800003684e00) at
> > dma_fence_chain_timo+0x21
> > Jul 16 02:03:25 nuc /bsd:
> > softclock_process_tick_timeout(ffff800003684e70,1) at
> > softclock_process_tick_timeout+0x10f
> > Jul 16 02:03:25 nuc /bsd: softclock(0) at softclock+0x10e
> > Jul 16 02:03:25 nuc /bsd: softintr_dispatch(0) at softintr_dispatch+0x101
> > Jul 16 02:03:25 nuc /bsd: dosoftint(0) at dosoftint+0x3f
> > Jul 16 02:03:25 nuc /bsd: Xsoftclock() at Xsoftclock+0x27
> > Jul 16 02:03:25 nuc /bsd: acpicpu_idle() at acpicpu_idle+0x228
> > Jul 16 02:03:25 nuc /bsd: sched_idle(ffffffff829daff0) at sched_idle+0x24b
> > Jul 16 02:03:25 nuc /bsd: end trace frame: 0x0, count: 243
> > Jul 16 02:03:25 nuc /bsd: End of stack trace.
> > Jul 16 02:42:54 nuc /bsd: splassert: pool_do_put: want 0 have 1
> > Jul 16 02:42:54 nuc /bsd: Starting stack trace...
> > Jul 16 02:42:54 nuc /bsd: pool_do_put(ffffffff82ac3468,fffffd8555f12e20) at
> > pool_do_put+0x43
> > Jul 16 02:42:54 nuc /bsd: pool_put(ffffffff82ac3468,fffffd8555f12e20) at
> > pool_put+0x79
> > Jul 16 02:42:54 nuc /bsd: i915_fence_release(ffff80000302a000) at
> > i915_fence_release+0x86
> > Jul 16 02:42:54 nuc /bsd: dma_fence_chain_release(ffff8000030aab00) at
> > dma_fence_chain_release+0x104
> > Jul 16 02:42:54 nuc /bsd: dma_fence_chain_walk(ffff8000034b3900) at
> > dma_fence_chain_walk+0x1e5
> > Jul 16 02:42:54 nuc /bsd:
> > dma_fence_chain_enable_signaling(ffff8000034b3900) at
> > dma_fence_chain_enable_signaling+0x97
> > Jul 16 02:42:54 nuc /bsd: dma_fence_chain_timo(ffff8000034b3900) at
> > dma_fence_chain_timo+0x21
> > Jul 16 02:42:54 nuc /bsd:
> > softclock_process_tick_timeout(ffff8000034b3970,1) at
> > softclock_process_tick_timeout+0x10f
> > Jul 16 02:42:54 nuc /bsd: softclock(0) at softclock+0x10e
> > Jul 16 02:42:54 nuc /bsd: softintr_dispatch(0) at softintr_dispatch+0x101
> > Jul 16 02:42:54 nuc /bsd: dosoftint(0) at dosoftint+0x3f
> > Jul 16 02:42:54 nuc /bsd: Xsoftclock() at Xsoftclock+0x27
> > Jul 16 02:42:54 nuc /bsd: acpicpu_idle() at acpicpu_idle+0x228
> > Jul 16 02:42:54 nuc /bsd: sched_idle(ffffffff829daff0) at sched_idle+0x24b
> > Jul 16 02:42:54 nuc /bsd: end trace frame: 0x0, count: 243
> > Jul 16 02:42:54 nuc /bsd: End of stack trace.
> >
> > >How-To-Repeat:
> > Unknown. It just happened, and I can't pinpoint the timestamps
> > to specific actions.
> > >Fix:
> > Unknown.
>
> diff to do the pool_put() from process context
>
> Index: sys/dev/pci/drm/i915/i915_request.c
> ===================================================================
> RCS file: /cvs/src/sys/dev/pci/drm/i915/i915_request.c,v
> diff -u -p -r1.17 i915_request.c
> --- sys/dev/pci/drm/i915/i915_request.c 7 Feb 2025 11:57:17 -0000
> 1.17
> +++ sys/dev/pci/drm/i915/i915_request.c 8 Apr 2025 06:01:53 -0000
> @@ -116,6 +116,13 @@ struct pool *i915_request_slab_cache(voi
> }
> #endif
>
> +static void
> +i915_rq_free(void *p)
> +{
> + struct i915_request *rq = p;
> + pool_put(&slab_requests, rq);
> +}
> +
> static void i915_fence_release(struct dma_fence *fence)
> {
> struct i915_request *rq = to_request(fence);
> @@ -177,7 +184,8 @@ static void i915_fence_release(struct dm
> #ifdef __linux__
> kmem_cache_free(slab_requests, rq);
> #else
> - pool_put(&slab_requests, rq);
> + task_set(&rq->free_task, i915_rq_free, rq);
> + task_add(systq, &rq->free_task);
> #endif
> }
>
> Index: sys/dev/pci/drm/i915/i915_request.h
> ===================================================================
> RCS file: /cvs/src/sys/dev/pci/drm/i915/i915_request.h,v
> diff -u -p -r1.6 i915_request.h
> --- sys/dev/pci/drm/i915/i915_request.h 16 Jan 2024 23:38:08 -0000
> 1.6
> +++ sys/dev/pci/drm/i915/i915_request.h 8 Apr 2025 06:00:15 -0000
> @@ -357,6 +357,8 @@ struct i915_request {
> struct list_head link;
> unsigned long delay;
> } mock;)
> +
> + struct task free_task;
> };
>
> #define I915_FENCE_GFP (GFP_KERNEL | __GFP_RETRY_MAYFAIL | __GFP_NOWARN)
>
