Bug 15206

Summary: qemumips BUG using smp_processor_id() in preemptible
Product: [Build System, Metadata & Runtime] OE-Core Reporter: Ross Burton <ross.burton>
Component: kernelAssignee: Bruce Ashfield <bruce.ashfield>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: Medium+ CC: bruce.ashfield, paulg, randy.macleod, richard.purdie, tom.zanussi
Version: 4.3   
Target Milestone: 4.3 M4   
Hardware: x86   
OS: Multiple   
Whiteboard:
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: No (bug/feature does not impact docs)

Description Ross Burton 2023-09-04 11:54:05 UTC 
With current master, qemumips has frequent BUG alerts:

[  130.730686] BUG: using smp_processor_id() in preemptible [00000000] code: systemd/1
[  130.738542] caller is blast_dcache32+0x40/0x190
[  130.745908] CPU: 0 PID: 1 Comm: systemd Not tainted 6.4.11-yocto-standard #1
[  130.753928] Stack : 0000000000000001 0000000000000000 0000000000000018 9000000004937798
[  130.762192]         9000000004937798 90000000049378c8 0000000000000000 0000000000000000
[  130.770458]         008fac6060240fea 0000000000014f20 9000000004937728 0000000000000000
[  130.778826]         00000000000b61b4 0000000000000030 ffffffff80e8eed0 0000000000000004
[  130.787111]         9000000084937577 0000000000000000 0000000000000000 ffffffff810c3b18
[  130.795440]         ffffffff811f0000 0000000000000000 ffffffff81440000 0000000000000000
[  130.803830]         0000000000000000 0000000000000000 ffffffff809842e8 0000000000175097
[  130.812272]         0000000000000001 9000000004934000 9000000004937790 c0000000005d9000
[  130.820722]         ffffffff8010b6fc 0000000000000000 0000000000000000 0000000000000000
[  130.829134]         0000000000000000 0000000000000000 ffffffff8010b71c 008fac6060240fea
[  130.837649]         ...
[  130.845556] Call Trace:
[  130.853432] [<ffffffff8010b71c>] show_stack+0x64/0x158
[  130.861534] [<ffffffff80ea1790>] dump_stack_lvl+0x5c/0x7c
[  130.869565] [<ffffffff80ea29c4>] check_preemption_disabled+0x10c/0x118
[  130.877605] [<ffffffff8012d7c8>] blast_dcache32+0x40/0x190
[  130.885585] [<ffffffff8039f648>] __vmalloc_node_range+0x3e0/0x780
[  130.893527] [<ffffffff8039fbbc>] vzalloc+0x6c/0x78
[  130.901378] [<ffffffff8095d6c4>] n_tty_open+0x1c/0xb8
[  130.909252] [<ffffffff8095f5a8>] tty_ldisc_open+0x50/0xd0
[  130.917053] [<ffffffff80960150>] tty_ldisc_setup+0x20/0x70
[  130.924927] [<ffffffff80958694>] tty_init_dev.part.0+0xbc/0x2c0
[  130.932742] [<ffffffff80958da8>] tty_open+0x4b8/0x730
[  130.940486] [<ffffffff803e6960>] chrdev_open+0xe0/0x250
[  130.948211] [<ffffffff803da7f4>] do_dentry_open+0x274/0x520
[  130.955960] [<ffffffff803f642c>] path_openat+0xbc4/0x10d8
[  130.963718] [<ffffffff803f7460>] do_filp_open+0x148/0x190
[  130.971482] [<ffffffff803dc264>] do_sys_openat2+0xcc/0x1b0
[  130.979288] [<ffffffff803dc710>] sys_openat+0x50/0xa8
[  130.987077] [<ffffffff80119490>] syscall_common+0x34/0x58

Kernel 6.4.11.
Comment 1 Ross Burton 2023-09-05 09:27:49 UTC 
Ah this only happens with DEBUG_PREEMPT enabled.  Did we leave that on?
Comment 2 Richard Purdie 2023-09-06 10:51:19 UTC 
That is set, yes.
Comment 3 Bruce Ashfield 2023-09-28 22:02:05 UTC 
This is also in 6.5, I've managed to brute force make this stop happening.

Working on a better fix now.

I can confirm that this isn't something that we've caused as it happens with v6.5/base as well as our branches.
Comment 4 Bruce Ashfield 2023-09-29 00:17:02 UTC 
I have it fixed. It will be part of the upcoming 6.5 reference kernel.
Comment 5 Bruce Ashfield 2023-10-23 14:48:18 UTC 
I fixed this in the 6.5 kernel series.