Re: 5.9.1-rt18: issues with Firewire card on AMD hardware

From: David Runge <dave@sleepmap.de>
To: linux-rt-users@vger.kernel.org
Subject: Re: 5.9.1-rt18: issues with Firewire card on AMD hardware
Date: Mon, 26 Oct 2020 01:37:08 +0100	[thread overview]
Message-ID: <20201026003708.GA21902@hmbx> (raw)
In-Reply-To: <20201021175059.GA4989@hmbx>

[-- Attachment #1: Type: text/plain, Size: 5277 bytes --]

On 2020-10-21 19:50:59 (+0200), David Runge wrote:
> P.S.: I was trying the 5.6.x series of realtime kernels on this machine
> as well, but although the hardware gets initialized properly (at least
> without crashes) my system completely freezes as soon as I start jackd
> on the firewire interface (this does not happen with a non-rt kernel).

Thanks! 5.9.1-rt19 fixes the crash for me.
Unfortunately I now have the ca. 5.6.x-rt* behavior back. The Firewire
card initializes properly, but as soon as I start jackd using the device
and initialize playback, firewire_core blocks forever and a reboot is
required:

Oct 26 01:13:35 hmbx kernel: firewire_core 0000:05:00.0: created device
fw1: GUID 000a3500ada83262, S800
Oct 26 01:14:47 hmbx kernel: logitech-hidpp-device 0003:046D:101A.0006:
HID++ 1.0 device connected.
Oct 26 01:20:11 hmbx kernel: INFO: task jackd:6873 blocked for more than
122 seconds.
Oct 26 01:20:11 hmbx kernel:       Not tainted 5.9.1-rt19-1-rt #1
Oct 26 01:20:11 hmbx kernel: "echo 0 >
/proc/sys/kernel/hung_task_timeout_secs" disables this message.
Oct 26 01:20:11 hmbx kernel: task:jackd           state:D stack:    0
pid: 6873 ppid:  1664 flags:0x80004086
Oct 26 01:20:11 hmbx kernel: Call Trace:
Oct 26 01:20:11 hmbx kernel:  __schedule+0x2c0/0x8f0
Oct 26 01:20:11 hmbx kernel:  schedule+0x60/0x100
Oct 26 01:20:11 hmbx kernel:  fw_device_op_release+0x230/0x290
[firewire_core]
Oct 26 01:20:11 hmbx kernel:  ? wait_woken+0x80/0x80
Oct 26 01:20:11 hmbx kernel:  __fput+0x8e/0x240
Oct 26 01:20:11 hmbx kernel:  task_work_run+0x5c/0x90
Oct 26 01:20:11 hmbx kernel:  do_exit+0x383/0xaf0
Oct 26 01:20:11 hmbx kernel:  ? finish_task_switch.isra.0+0x91/0x4c0
Oct 26 01:20:11 hmbx kernel:  do_group_exit+0x39/0xb0
Oct 26 01:20:11 hmbx kernel:  get_signal+0x14f/0x990
Oct 26 01:20:11 hmbx kernel:  ? preempt_count_add+0x68/0xa0
Oct 26 01:20:11 hmbx kernel:  ? _raw_spin_lock_irqsave+0x26/0x50
Oct 26 01:20:11 hmbx kernel:  arch_do_signal+0x3d/0x750
Oct 26 01:20:11 hmbx kernel:  ? do_epoll_wait+0xda/0x670
Oct 26 01:20:11 hmbx kernel:  exit_to_user_mode_prepare+0x12d/0x1a0
Oct 26 01:20:11 hmbx kernel:  syscall_exit_to_user_mode+0x2c/0x1b0
Oct 26 01:20:11 hmbx kernel:  entry_SYSCALL_64_after_hwframe+0x44/0xa9
Oct 26 01:20:11 hmbx kernel: RIP: 0033:0x7f42186895de
Oct 26 01:20:11 hmbx kernel: Code: Bad RIP value.
Oct 26 01:20:11 hmbx kernel: RSP: 002b:00007fff3dcdea40 EFLAGS: 00000246
ORIG_RAX: 00000000000000e8
Oct 26 01:20:11 hmbx kernel: RAX: fffffffffffffffc RBX: 00005596b98d6310
RCX: 00007f42186895de
Oct 26 01:20:11 hmbx kernel: RDX: 0000000000000020 RSI: 00007fff3dcdea70
RDI: 000000000000000d
Oct 26 01:20:11 hmbx kernel: RBP: 00005596b98d6310 R08: 0000000000000002
R09: 00007fff3dcded80
Oct 26 01:20:11 hmbx kernel: R10: 00000000ffffffff R11: 0000000000000246
R12: 00005596b98f5040
Oct 26 01:20:11 hmbx kernel: R13: 00007fff3dcdea70 R14: 00000000801c0000
R15: 0000000000000001
Oct 26 01:22:13 hmbx kernel: INFO: task jackd:6873 blocked for more than
245 seconds.
Oct 26 01:22:13 hmbx kernel:       Not tainted 5.9.1-rt19-1-rt #1
Oct 26 01:22:13 hmbx kernel: "echo 0 >
/proc/sys/kernel/hung_task_timeout_secs" disables this message.
Oct 26 01:22:13 hmbx kernel: task:jackd           state:D stack:    0
pid: 6873 ppid:  1664 flags:0x80004086
Oct 26 01:22:13 hmbx kernel: Call Trace:
Oct 26 01:22:13 hmbx kernel:  __schedule+0x2c0/0x8f0
Oct 26 01:22:13 hmbx kernel:  schedule+0x60/0x100
Oct 26 01:22:13 hmbx kernel:  fw_device_op_release+0x230/0x290
[firewire_core]
Oct 26 01:22:13 hmbx kernel:  ? wait_woken+0x80/0x80
Oct 26 01:22:13 hmbx kernel:  __fput+0x8e/0x240
Oct 26 01:22:13 hmbx kernel:  task_work_run+0x5c/0x90
Oct 26 01:22:13 hmbx kernel:  do_exit+0x383/0xaf0
Oct 26 01:22:13 hmbx kernel:  ? finish_task_switch.isra.0+0x91/0x4c0
Oct 26 01:22:13 hmbx kernel:  do_group_exit+0x39/0xb0
Oct 26 01:22:13 hmbx kernel:  get_signal+0x14f/0x990
Oct 26 01:22:13 hmbx kernel:  ? preempt_count_add+0x68/0xa0
Oct 26 01:22:13 hmbx kernel:  ? _raw_spin_lock_irqsave+0x26/0x50
Oct 26 01:22:13 hmbx kernel:  arch_do_signal+0x3d/0x750
Oct 26 01:22:13 hmbx kernel:  ? do_epoll_wait+0xda/0x670
Oct 26 01:22:13 hmbx kernel:  exit_to_user_mode_prepare+0x12d/0x1a0
Oct 26 01:22:13 hmbx kernel:  syscall_exit_to_user_mode+0x2c/0x1b0
Oct 26 01:22:13 hmbx kernel:  entry_SYSCALL_64_after_hwframe+0x44/0xa9
Oct 26 01:22:13 hmbx kernel: RIP: 0033:0x7f42186895de
Oct 26 01:22:13 hmbx kernel: Code: Bad RIP value.
Oct 26 01:22:13 hmbx kernel: RSP: 002b:00007fff3dcdea40 EFLAGS: 00000246
ORIG_RAX: 00000000000000e8
Oct 26 01:22:13 hmbx kernel: RAX: fffffffffffffffc RBX: 00005596b98d6310
RCX: 00007f42186895de
Oct 26 01:22:13 hmbx kernel: RDX: 0000000000000020 RSI: 00007fff3dcdea70
RDI: 000000000000000d
Oct 26 01:22:13 hmbx kernel: RBP: 00005596b98d6310 R08: 0000000000000002
R09: 00007fff3dcded80
Oct 26 01:22:13 hmbx kernel: R10: 00000000ffffffff R11: 0000000000000246
R12: 00005596b98f5040
Oct 26 01:22:13 hmbx kernel: R13: 00007fff3dcdea70 R14: 00000000801c0000
R15: 0000000000000001

This does not happen on the vanilla 5.9.x kernel.

Any hints on what to try or which patch to omit would be greatly
appreciated!

Best,
David

-- 
https://sleepmap.de

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]