On Aug 30, 2012, at 18:51 , Andrew Deason <[email protected]> wrote:
> On Thu, 30 Aug 2012 15:19:16 +0200 > Stephan Wiesand <[email protected]> wrote: > >> I fstraced a read getting stuck. This is what it looks like: > > What about a kernel stack trace? Could you get one with 'crash', or at > least 'echo t > /proc/sysrq-trigger' ? In crash it would just be like > 'set 1234' and 'bt', iirc, where 1234 is the pid of a hanging proc. I'm afraid Simon is right about all this. I'm still curious though, so I tried the crash exercise (sysrq looks similar): crash> set 4479 PID: 4479 COMMAND: "cfagent" TASK: c24d2000 [THREAD_INFO: eb4a4000] CPU: 0 STATE: (unknown) crash> bt PID: 4479 TASK: c24d2000 CPU: 0 COMMAND: "cfagent" #0 [eb4a5ce4] schedule at c083c6a3 #1 [eb4a5da8] io_schedule at c083ce34 #2 [eb4a5db4] sync_page at c04e0e57 #3 [eb4a5db8] sync_page_killable at c04e0e70 #4 [eb4a5dbc] __wait_on_bit_lock at c083d50b #5 [eb4a5dd8] __lock_page_killable at c04e0d7b #6 [eb4a5e00] generic_file_aio_read at c04e2608 #7 [eb4a5e78] do_sync_read at c052d3b2 #8 [eb4a5f20] afs_linux_read at edaef53b [openafs] #9 [eb4a5f74] vfs_read at c052de3b #10 [eb4a5f94] sys_read at c052df6c #11 [eb4a5fb0] system_call at c083ec8d EAX: ffffffda EBX: 00000003 ECX: b77c1000 EDX: 00001000 DS: 007b ESI: 09b38420 ES: 007b EDI: 00000400 SS: 007b ESP: bf8f6c84 EBP: bf8f6cb4 GS: 0033 CS: 0073 EIP: 00fd5416 ERR: 00000003 EFLAGS: 00200246 But unless this gets us somewhere soon, I'm not planning to do much more research on this issue. Thanks again for all the feedback. Stephan -- Stephan Wiesand DESY - DV - Platanenallee 6 15732 Zeuthen, Germany _______________________________________________ OpenAFS-devel mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-devel
