The dump does indicate that the thread is dying (hence the call to NOTIFY_KnlFinalize). I suspect that this is happening because your application is somehow generating a seg fault or something that's getting caught by the DSPLink default signal handler, then resulting in calling the function that does the cleanup. When cleanup happens, this crash dump is expected, and you'll see that you can continue and restart your application.
Can you put prints inside DSPLINK_atExitHandler and DSPLINK_sigHandler functions (dsplink/gpp/src/api/Linux/drv_api.c), rebuild DSPLink user-side and see if this print comes up? If yes, then this is probably what's happening, and you need to find out which signal is getting caught, and what's the cause of that signal (may be completely DSPLink-unrelated, since we have stress tests that run for several hours and we have never seen this issue in a normal run). Regards, Mugdha -----Original Message----- From: Çağlar AKYÜZ [mailto:[email protected]] Sent: Wednesday, March 24, 2010 12:52 PM To: [email protected] Cc: Yuvraj Pasi; Kamoolkar, Mugdha Subject: Re: kernel crash while running h264 codec on DM6446 based board On Wednesday 24 March 2010 06:36:58 am Yuvraj Pasi wrote: > Hi, Hi, > Thanks for the reply. No I'm not calling kill inside the application & the > kernel does not hang after the crash. I'm able to restart > the application again after the crash. & every time it runs smoothly for > some period before it crashes. > > thanks & regards > yuvraj pasi > [...] > > [<bf00d4b0>] (SYNC_WaitSEM+0x0/0x260 [dsplinkk]) from [<bf00c4d4>] By looking at oops dump I remembered a smilar problem related to sync in Dsplink. I don't remember my exact oops dump but maybe your problem is somehow related to it. Have you checked the thread at [1] ? Secondly, this maybe a memory leak issue. Have you checked your free memory before and after running your application? Regards, Caglar [1] http://www.mail-archive.com/[email protected]/msg11540.html > > (UEVENT_AddBufByPid+0x150/0x17c [dsplinkk]) > > [<bf00c384>] (UEVENT_AddBufByPid+0x0/0x17c [dsplinkk]) from [<bf00c714>] > > (NOTIFY_KnlFinalize+0x14c/0x16c [dsplinkk]) > > [<bf00c5c8>] (NOTIFY_KnlFinalize+0x0/0x16c [dsplinkk]) from [<bf008930>] > > (DRV_CallAPI+0x778/0x8f0 [dsplinkk]) > > r6:00008000 r5:00008000 r4:c14edf00 > > [<bf0081b8>] (DRV_CallAPI+0x0/0x8f0 [dsplinkk]) from [<bf008138>] > > (DRV_Ioctl+0x90/0x110 [dsplinkk]) > > r7:00007302 r6:00000000 r5:4391c0b0 r4:00000000 > > [<bf0080a8>] (DRV_Ioctl+0x0/0x110 [dsplinkk]) from [<c0088f10>] > > (do_ioctl+0x74/0x84) > > r7:0000000b r6:4391c0b0 r5:ffffffe7 r4:c6fed160 > > [<c0088e9c>] (do_ioctl+0x0/0x84) from [<c00891ac>] > > (vfs_ioctl+0x28c/0x2ac) r6:00000000 r5:4391c0b0 r4:c6fed160 > > [<c0088f20>] (vfs_ioctl+0x0/0x2ac) from [<c0089210>] > > (sys_ioctl+0x44/0x68) r7:c6fed160 r6:00007302 r5:4391c0b0 r4:fffffff7 > > [<c00891cc>] (sys_ioctl+0x0/0x68) from [<c0027e20>] > > (ret_fast_syscall+0x0/0x2c) > > > > According to the FAQ it is a problem with syncronisation between threads. > > How can I find out which thread calls are causing this problem because my > > application uses different threads > > for video encode & decoding , audio encoding & decoding. > > The backtrace given above is of no help!!! > > > > > > On Mon, Mar 22, 2010 at 11:25 PM, Uppal, Deepali <[email protected]> wrote: > > > > Hello, > > > > > > > > Please check here: > > > > > > > > > > http://wiki.davincidsp.com/index.php/DSPLink_FAQs#Why_does_the_MSGQ_get_A > >PI_call_in_my_application_crash.3F > > > > > > > > Does this FAQ apply to your scenario? > > > > > > > > Thanks and Regards, > > Deepali Uppal > > DSP/BIOS Link > > ------------------------------ > > > > *From:* davinci-linux-open-source-bounces+deepali=ti.com@ > > linux.davincidsp.com > > [mailto:davinci-linux-open-source-bounces+deepali<davinci-linux-open-sour > >ce-bounces%2Bdeepali> [email protected]] *On Behalf Of *Yuvraj > > Pasi > > *Sent:* Friday, March 19, 2010 2:35 PM > > *To:* [email protected] > > *Subject:* kernel crash while running h264 codec on DM6446 based board > > > > > > > > Hi all, > > we are using our own custom made board with DM6446. I have written a > > camera capture application which > > captures images encodes it in H264, decodes it & then display it on fb. > > It runs smoothly for some time & then crashes . > > > > How do I debug this crash dump so that i can find the source of the > > problem. > > > > ortp-m, *pte=00000000essage-get_pictu, *ppte=00000000re_buffer_size 1 > > 07 > > ortp-messageInternal error: Oops: 7 [#1] > > Modules linked in: dsplinkk cmemk > > CPU: 0 Not tainted (2.6.23-davinci1 #219) > > PC is at SYNC_WaitSEM+0x1d0/0x260 [dsplinkk] > > LR is at __atomic_notifier_call_chain+0x1c/0x24 > > pc : [<bf00d680>] lr : [<c004959c>] psr: 50000013 > > sp : c1975dd0 ip : 00000000 fp : c1975e44 > > r10: c0038e14 r9 : 00008000 r8 : c7930008 > > r7 : c1974000 r6 : c1975de8 r5 : c7930000 r4 : c1975de0 > > r3 : 00000001 r2 : c04bb900 r1 : 00000002 r0 : c04bb900 > > Flags: nZcV IRQs on FIQs on Mode SVC_32 ISA ARM Segment user > > Control: 0005317f Table: 81794000 DAC: 00000015 > > Process linphonec (pid: 1118, stack limit = 0xc1974260) > > Stack: (0xc1975dd0 to 0xc1976000) > > 5dc0: 00000000 c04bb900 c0038e14 > > 00000000 > > 5de0: 00000000 00000000 00000001 c04bb900 c0038e14 c7930008 c7930008 > > c0038edc > > 5e00: 60000013 ffffffff 00000000 bf02211c 00000000 c7932008 c1975e44 > > 00008000 > > 5e20: c15174c0 bf02211c c67526e0 bf02210c 00000000 00000000 c1975e7c > > c1975e48 > > 5e40: bf00c4d4 bf00d4c0 bf00b8d8 ffffffff c1975e8c c1975e8c 00000000 > > 00008000 > > 5e60: 00007302 c002861c c1974000 00900036 c1975eac c1975e80 bf00c714 > > bf00c394 > > 5e80: 00000000 00000000 c1975f0c 00000000 c0027ab4 c1975f00 00008000 > > 00008000 > > 5ea0: c1975efc c1975eb0 bf008930 bf00c5d8 40020000 c6687288 00000000 > > 40020000 > > 5ec0: 4002b000 40020000 c748d23c c1975f1c c1975f0c c1975ee0 c006e18c > > 43f60170 > > 5ee0: 00000000 43f60170 00000000 00007302 c1975f34 c1975f00 bf008138 > > bf0081c8 > > 5f00: 00008000 00000000 40020480 00008200 001eff10 00008000 c75758a0 > > ffffffe7 > > 5f20: 43f60170 00000011 c1975f54 c1975f38 c008b490 bf0080b8 c1975f84 > > c75758a0 > > 5f40: 43f60170 00000000 c1975f7c c1975f58 c008b72c c008b42c 00008680 > > c6752714 > > 5f60: fffffff7 43f60170 00007302 c75758a0 c1975fa4 c1975f80 c008b790 > > c008b4b0 > > 5f80: c0073d9c 00000001 001f02b0 43f61000 00215d70 00000036 00000000 > > c1975fa8 > > 5fa0: c0027e20 c008b75c 001f02b0 43f61000 00000011 00007302 43f60170 > > 00213e60 > > 5fc0: 001f02b0 43f61000 00215d70 00000108 00215d34 000005a0 00000000 > > 43f6016c > > 5fe0: 001d6024 43f60110 000221e4 401d2294 80000010 00000011 020030fe > > e2200000 > > Backtrace: > > [<bf00d4b0>] (SYNC_WaitSEM+0x0/0x260 [dsplinkk]) from [<bf00c4d4>] > > (UEVENT_AddBufByPid+0x150/0x17c [dsplinkk]) > > [<bf00c384>] (UEVENT_AddBufByPid+0x0/0x17c [dsplinkk]) from [<bf00c714>] > > (NOTIFY_KnlFinalize+0x14c/0x16c [dsplinkk]) > > [<bf00c5c8>] (NOTIFY_KnlFinalize+0x0/0x16c [dsplinkk]) from [<bf008930>] > > (DRV_CallAPI+0x778/0x8f0 [dsplinkk]) > > r6:00008000 r5:00008000 r4:c1975f00 > > [<bf0081b8>] (DRV_CallAPI+0x0/0x8f0 [dsplinkk]) from [<bf008138>] > > (DRV_Ioctl+0x90/0x110 [dsplinkk]) > > r7:00007302 r6:00000000 r5:43f60170 r4:00000000 > > [<bf0080a8>] (DRV_Ioctl+0x0/0x110 [dsplinkk]) from [<c008b490>] > > (do_ioctl+0x74/0x84) > > r7:00000011 r6:43f60170 r5:ffffffe7 r4:c75758a0 > > [<c008b41c>] (do_ioctl+0x0/0x84) from [<c008b72c>] > > (vfs_ioctl+0x28c/0x2ac) r6:00000000 r5:43f60170 r4:c75758a0 > > [<c008b4a0>] (vfs_ioctl+0x0/0x2ac) from [<c008b790>] > > (sys_ioctl+0x44/0x68) r7:c75758a0 r6:00007302 r5:43f60170 r4:fffffff7 > > [<c008b74c>] (sys_ioctl+0x0/0x68) from [<c0027e20>] > > (ret_fast_syscall+0x0/0x2c) > > r7:00000036 r6:00215d70 r5:43f61000 r4:001f02b0 > > Code: eb410c34 e597200c e3a03001 e5823000 (e5953010) > > > > > > -- > > Thanks & regards > > yuvraj pasi > > > > > > > > > > -- > > Thanks & regards > > yuvraj pasi > _______________________________________________ Davinci-linux-open-source mailing list [email protected] http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
