Re: [i915-rt v6 00/24] drm/i915/display: All patches to make PREEMPT_RT work on i915 + xe.

[Sebastian Andrzej Siewior <bigeasy@linutronix.de>]

On 2026-02-24 17:27:03 [+0100], To Maarten Lankhorst wrote:
> 
> There is nothing else, that popped up. So far no objections from my side
> ;) Thank you.

different HW.

|  i915 0000:00:02.0: [drm] *ERROR* [CRTC:170:pipe A] flip_done timed out
|  rcu: INFO: rcu_preempt self-detected stall on CPU
|  rcu: 	7-....: (1 GPs behind) idle=bc04/1/0x4000000000000000 softirq=0/0 fqs=3084 rcuc=6502 jiffies(starved)
|  rcu: 	(t=6500 jiffies g=46813 q=13322 ncpus=8)
|  CPU: 7 UID: 0 PID: 12616 Comm: irq/182-i915 Tainted: G     U      E       7.0.0-rc1-lockdep+ #2 PREEMPT_{RT,(lazy)}
|  Tainted: [U]=USER, [E]=UNSIGNED_MODULE
|  Hardware name: LENOVO 20TD00GLGE/20TD00GLGE, BIOS R1EET64W(1.64 ) 03/18/2025
|  RIP: 0010:intel_context_remove_breadcrumbs+0x14a/0x1a0 [i915]
|  Code: 00 00 48 8b 50 70 b8 ff ff ff ff f0 41 0f c1 06 83 f8 01 74 58 85 c0 7f 11 be 03 00 00 00 4c 89 f7 e8 fa 65 db e7 eb 02 f3 90 <41> 8b 84 24 a0 00 00 00 85 c0 75 f2 48 83 c4 08 5b 5d 41 5c 41 5d
|  RSP: 0018:ffffd2aa007d7c48 EFLAGS: 00000202
|  RAX: 0000000000000001 RBX: 0000000000000001 RCX: 0000000000000001
|  RDX: ffffffffc282ab20 RSI: ffffffffab33e33f RDI: 00000000ffffffff
|  RBP: ffff8f1247401680 R08: 0000000000000001 R09: 0000000000000000
|  R10: 0000000000000000 R11: 0000000000000000 R12: ffff8f104f3b6200
|  R13: ffff8f12474017e0 R14: ffff8f1247401798 R15: ffff8f119e89e3a8
|  FS:  0000000000000000(0000) GS:ffff8f140318b000(0000) knlGS:0000000000000000
|  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
|  CR2: 00007f5409dc6000 CR3: 000000029ea56006 CR4: 0000000000f72ef0
|  PKRU: 55555554
|  Call Trace:
|   <TASK>
|   __execlists_schedule_out+0x1a6/0x340 [i915]
|   execlists_submission_tasklet+0x178/0x1bb0 [i915]
|   tasklet_action_common+0x1e9/0x410
|   handle_softirqs.isra.0+0xbd/0x3f0
|   __local_bh_enable_ip+0xe0/0x170
|   irq_forced_thread_fn+0x41/0x50
|   irq_thread+0x20d/0x360
|   kthread+0x103/0x140
|   ret_from_fork+0x34d/0x480
|   ret_from_fork_asm+0x1a/0x30
|   </TASK>

It seems to be always stuck in intel_context_remove_breadcrumbs.

and top shows
|     42 root      20   0       0      0      0 R 100,0   0,0   3:21.62 ksoftirqd/2
|  12616 root     -51   0       0      0      0 R 100,0   0,0   3:34.61 irq/182-i915

seems busy. The NMI backtrace reports always
intel_context_remove_breadcrumbs() on CPU, don't see the irq handler.

This one I can reproduce while opening a browser and surfing a bit. Let
me try step by step without lockdep, without RT and maybe just play
v7.0-rc1 to figure out where this is from…

Sebastian