Hide Forgot
Description of problem: The test-mpeg2 utility from libiec61883-utils segfaults when running with the new firewire stack. strace dump forthcoming. Version-Release number of selected component (if applicable): libiec61883-utils-1.1.0-1.fc7 kernel-2.6.21-1.3142.fc7 How reproducible: 1. run plugreport to see what devices are on the firewire bus: Host Adapter 0 ============== Node 0 GUID 0xf0219500009521f0 ------------------------------ libiec61883 error: error reading oMPR libiec61883 error: error reading iMPR Node 1 GUID 0x000ce5fffe719455 ------------------------------ oMPR n_plugs=1, data_rate=2, bcast_channel=63 oPCR[0] online=1, bcast_connection=0, n_p2p_connections=0 channel=0, data_rate=2, overhead_id=0, payload=376 iMPR n_plugs=0, data_rate=2 Node 2 GUID 0x0019e3fffe2da720 ------------------------------ libiec61883 error: error reading oMPR libiec61883 error: error reading iMPR 2. run 'test-mpeg2 -r 1 > testcap.ts' to try to grab video off device (firewire-enabled cable box). $ test-mpeg2 -r 1 > testcap.ts Segmentation fault Additional info: works fine under FC6
Created attachment 155104 [details] strace of test-mpeg2
With the latest released F7 kernel (2.6.21-1.3228.fc7) and all other bits at their latest versions, I'm not getting a segfault when running as a normal user. Instead, I get this: [jarod@ares ~]$ test-mpeg2 -r 1 > testcap.ts Failed to get libraw1394 handle When run as root however, much badness, even beyond just the segfault... [root@ares ~]# test-mpeg2 -r 1 > testcap.ts libiec61883 warning: iec61883_cmp_overlay_p2p_output: Failed to set the oPCR[0] plug for node 1. libiec61883 warning: Overlayed connection on channel -1. You may need to manually set the channel on the receiving node. Connect failed, reverting to broadcast channel 63. Segmentation fault [root@ares ~]# Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: Oops: 0000 [#2] Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: SMP Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: CPU: 1 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: EIP: 0060:[<f8d2bab8>] Tainted: GF VLI Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: EFLAGS: 00210282 (2.6.21-1.3228.fc7 #1) Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: EIP is at fw_iso_context_destroy+0x0/0x8 [fw_core] Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: eax: ffffffea ebx: 00000008 ecx: f8d2cab7 edx: f0d192c0 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: esi: f08443c0 edi: f7616258 ebp: f0d192c0 esp: e05acf60 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: ds: 007b es: 007b fs: 00d8 gs: 0033 ss: 0068 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: Process test-mpeg2 (pid: 17819, ti=e05ac000 task=f11638f0 task.ti=e05ac000) Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: Stack: f8d2cadc 00000008 f7616258 c047654f 00000000 00000000 f7545198 c20bf0c0 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: f0d192c0 f7c160c0 00000000 e05ac000 c047405d bfeae634 f7c160c0 0000000a Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: f0d192c0 c0474f8c 0000000a 09b3d008 00000001 c0404f70 0000000a ffffffc8 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: Call Trace: Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: [<f8d2cadc>] fw_device_op_release+0x25/0xa2 [fw_core] Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: [<c047654f>] __fput+0xb4/0x16a Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: [<c047405d>] filp_close+0x51/0x58 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: [<c0474f8c>] sys_close+0x6e/0xa5 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: [<c0404f70>] syscall_call+0x7/0xb Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: [<c0600000>] __sched_text_start+0x570/0x89e Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: ======================= Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: Code: 02 00 00 74 07 0f ac d0 19 c1 ea 19 0f c8 31 c9 89 45 00 89 fa 89 f0 5b 5e 5f 5d e9 ec f5 ff ff 0f 0b eb fe b9 07 00 00 00 eb e8 <8b> 10 8b 12 ff 52 28 c3 56 53 83 ec 04 8b 18 8b 33 8b 5c 24 10 Message from syslogd@ at Tue Jul 17 11:21:06 2007 ... ares kernel: EIP: [<f8d2bab8>] fw_iso_context_destroy+0x0/0x8 [fw_core] SS:ESP 0068:e05acf60 A bit over my head here, reassigning to krh, in hopes he has some ideas/suggestions... One thing I'm not certain of is whether or not the cable box is on at the moment, or if it makes a difference one way or the other. Derek, I don't suppose you're seeing similar behavior? (this bug seems it could be at least somewhat related to bug 241729).
Well, now I'm getting *** glibc detected *** test-mpeg2: double free or corruption (top): 0x0000000000608160 ***, and the backtrace indeed looks pretty much identical to those in bug 241729, so I'm going to close-dupe this one. *** This bug has been marked as a duplicate of 241729 ***
Not actually a duplicate. The double free has been fixed (problem was actually in libraw1394), but test-mpeg2 and test-dv from libiec61883-utils still segfault.
Changing version to '9' as part of upcoming Fedora 9 GA. More information and reason for this action is here: http://fedoraproject.org/wiki/BugZappers/HouseKeeping
Re comment 2: Inability to run as non-root should be fixed by the latest libraw1394 update to cope with device-type dependent /dev/fw* permissions or ACLs. The iso context destroy bug might be fixed in latest driver code. And then there is your iso start/stop/start patch for libraw1394 but I think that's not related to this bug. Re comment 0, comment 1: libiec61883 1.2.0 release notes sound promising.
Both F10 and rawhide are good to go now, no problems running as non-root. F9's older libraw makes backporting this support... Less-than-fun. I've not seen any segfaults on any of my F10 or rawhide systems in quite some time now w/their much-improved libraw1394. Oh, hey, I opened this bug. Guess I can close it too then... :)