Dmitry Adamushko пишет:
Just to put aside a few (quite) weird assumptions :
(1) did you have another task that has been using this mutex and has
(somehow) exited by the moment of lockup?
there are only 2 task that has access to that mutex.
(2) bus_monitor_proc[0] is created once and never
exits_then_created_again while the program is running ?
all task starting from begining of program and never restart.
i renamed mutex name twice to avoid name misspelled access and move
throw the struct to avoid offset with another struct - nothing help.
i tried to increase stack size - not helped.
i try to move lock/unlock to macros and add some history logging.
Is possible next scenario:
1. task A lock mutex M
2. task A going to secondary mode
3. task A performing "long" io wait operation and some flags modified.
4. task B tries to lock mutex M
5. core find no task A, because its waiting in secondary mode or it
think that task A is zombie or due to flag modification it's do not
think task A is RT
6. mutex M is granted to task B
7. task A returns from secondary mode
8. task A unlock mutex M because it's owned
9. task B tries to unlock, and not properly cleared some flags due to
step 8.
10. task B tries to lock mutex
11. core see that task B currently not own mutex M and put it into wait
queue
12. task A tries to lock mutex.
13. core see that task B currently not own mutex M and put it into wait
queue
14. deadlock here.
BTW
at some moment i got a kernel oops, not very informative but...
Feb 9 23:26:02 dev-3 kernel: BUG: unable to handle kernel NULL pointer
dereference at virtual address 00000005
Feb 9 23:26:02 dev-3 kernel: printing eip:
Feb 9 23:26:02 dev-3 kernel: c01b53b2
Feb 9 23:26:02 dev-3 kernel: *pde = 0ae5a001
Feb 9 23:26:02 dev-3 kernel: Oops: 0000 [#1]
Feb 9 23:26:02 dev-3 kernel: Modules linked in: ipv6 sunrpc dm_mirror
dm_mod lp parport_pc parport ohci1394 ieee1394 floppy snd_intel8x0
snd_ac97_codec snd_ac97_bus snd_seq_dummy snd_seq_oss i2c_i801
snd_seq_midi_event snd_seq i2c_core snd_seq_device snd_pcm_oss
snd_mixer_oss snd_pcm serio_raw snd_timer e100 snd mxser mii soundcore
snd_page_alloc
Feb 9 23:26:02 dev-3 kernel: CPU: 0
Feb 9 23:26:02 dev-3 kernel: EIP: 0060:[<c01b53b2>] Not tainted VLI
Feb 9 23:26:02 dev-3 kernel: EFLAGS: 00210246 (2.6.17.14-xenomai #1)
Feb 9 23:26:02 dev-3 kernel: EIP is at remove_proc_entry+0x22/0x1b0
Feb 9 23:26:02 dev-3 kernel: eax: 00000000 ebx: c03a90f4 ecx:
ffffffff edx: ce2bfb40
Feb 9 23:26:02 dev-3 kernel: esi: 00000004 edi: 00000005 ebp:
00000000 esp: cfec9e78
Feb 9 23:26:02 dev-3 kernel: ds: 007b es: 007b ss: 0068
Feb 9 23:26:02 dev-3 kernel: Process events/0 (pid: 3,
threadinfo=cfec8000 task=c1259030)
Feb 9 23:26:02 dev-3 kernel: Stack: cafed7a0 c01b51f1 0000033d 00008000
ce2bfb40 00000005 c03a90f4 00000004
Feb 9 23:26:02 dev-3 kernel: ce2bfb40 00000000 c014476e ce2bfb40
cafeda20 c036352c c0363599 00000000
Feb 9 23:26:02 dev-3 kernel: cff254a0 c036352c c0363599 ce2bfcc0
67f8acfc eaeb76d2 75bf19d4 c1e67be3
Feb 9 23:26:02 dev-3 kernel: Call Trace:
Feb 9 23:26:02 dev-3 kernel: <c01b51f1> proc_register+0x31/0xe0
<c014476e> registry_proc_callback+0x17e/0x370
Feb 9 23:26:02 dev-3 kernel: <c01644d0> free_hot_cold_page+0xc0/0x130
<c022f956> memmove+0x36/0x40
Feb 9 23:26:02 dev-3 kernel: <c017b1da> drain_array+0x7a/0xe0
<c034a5a4> schedule+0x2e4/0x690
Feb 9 23:26:02 dev-3 kernel: <c034a5d8> schedule+0x318/0x690
<c0123930> run_workqueue+0x90/0x110
Feb 9 23:26:02 dev-3 kernel: <c01445f0>
registry_proc_callback+0x0/0x370 <c0123e97> worker_thread+0x117/0x140
Feb 9 23:26:02 dev-3 kernel: <c0111c30> default_wake_function+0x0/0x10
<c0123d80> worker_thread+0x0/0x140
Feb 9 23:26:02 dev-3 kernel: <c012676d> kthread+0x9d/0xd0 <c01266d0>
kthread+0x0/0xd0
Feb 9 23:26:02 dev-3 kernel: <c0100cb5> kernel_thread_helper+0x5/0x10
Feb 9 23:26:02 dev-3 kernel: Code: 76 00 8d bc 27 00 00 00 00 55 57 56
53 83 ec 18 85 d2 89 54 24 10 89 44 24 14 0f 84 35 01 00 00 8b 7c 24 14
31 c0 b9 ff ff ff ff <f2> ae f7 d1 49 8b 5c 24 10 89 cd 83 c3 38 8b 3b
85 ff 0f 84 c8
Feb 9 23:26:02 dev-3 kernel: EIP: [<c01b53b2>]
remove_proc_entry+0x22/0x1b0 SS:ESP 0068:cfec9e78
--
________________________________________
Maksym Veremeyenko
_______________________________________________
Xenomai-help mailing list
[email protected]
https://mail.gna.org/listinfo/xenomai-help
