svn commit: r249939 - head/sys/cam/scsi

[Edward Tomasz Napierała]

Wiadomość napisana przez Steven Hartland w dniu 30 kwi 2013, o godz. 05:16:

[..]

> If anyone's interested the trace was:-
> #0  doadump (textdump=0) at pcpu.h:231
> #1  0xffffffff802f6d6e in db_dump (dummy=<value optimized out>, dummy2=0, dummy3=0, dummy4=0x0) at /usr/home/smh/freebsd/base/head/sys/ddb/db_command.c:543
> #2  0xffffffff802f683d in db_command (cmd_table=<value optimized out>) at /usr/home/smh/freebsd/base/head/sys/ddb/db_command.c:449
> #3  0xffffffff802f65b4 in db_command_loop () at /usr/home/smh/freebsd/base/head/sys/ddb/db_command.c:502
> #4  0xffffffff802f8f50 in db_trap (type=<value optimized out>, code=0) at /usr/home/smh/freebsd/base/head/sys/ddb/db_main.c:231
> #5  0xffffffff805c0df3 in kdb_trap (type=9, code=0, tf=<value optimized out>) at /usr/home/smh/freebsd/base/head/sys/kern/subr_kdb.c:654
> #6  0xffffffff8075580a in trap_fatal (frame=0xffffff823b8dc790, eva=<value optimized out>) at /usr/home/smh/freebsd/base/head/sys/amd64/amd64/trap.c:867
> #7  0xffffffff807554b7 in trap (frame=<value optimized out>) at /usr/home/smh/freebsd/base/head/sys/amd64/amd64/trap.c:224
> #8  0xffffffff8073e1f2 in calltrap () at /usr/home/smh/freebsd/base/head/sys/amd64/amd64/exception.S:228
> #9  0xffffffff8029c860 in cam_periph_alloc (periph_ctor=0xffffffff802af410 <proberegister>, periph_oninvalidate=0xfffffe0019cfa200, periph_dtor=<value optimized out>, periph_start=0xfffffe0015980a90, name=<value optimized out>, type=2159638184,
>   path=0xfffffe0015ad79a0, ac_callback=<value optimized out>, code=<value optimized out>) at /usr/home/smh/freebsd/base/head/sys/cam/cam_periph.c:227
> #10 0xffffffff802aed6b in scsi_scan_lun (request_ccb=<value optimized out>) at /usr/home/smh/freebsd/base/head/sys/cam/scsi/scsi_xpt.c:2266
> #11 0xffffffff802b2859 in scsi_scan_bus (periph=0x0, request_ccb=0xfffffe00234df000) at /usr/home/smh/freebsd/base/head/sys/cam/scsi/scsi_xpt.c:1969
> #12 0xffffffff802a66c5 in xpt_scanner_thread (dummy=<value optimized out>) at /usr/home/smh/freebsd/base/head/sys/cam/cam_xpt.c:2411

I've seen a panic which is slightly different, but perhaps could be related:

#11 0xffffffff8085fde2 in __mtx_assert (c=0xffffff81aeccd8c0, 
    what=<value optimized out>, 
    file=0xffffffff80e6b0aa "/home/trasz/p4/iscsi/sys/kern/kern_cons.c", 
    line=128) at /home/trasz/p4/iscsi/sys/kern/kern_mutex.c:788
#12 0xffffffff802cc8e1 in xpt_compile_path (new_path=0xfffffe00150d7180, 
    perph=0xfffffe00112e1e00, path_id=<value optimized out>, target_id=0, 
    lun_id=0) at /home/trasz/p4/iscsi/sys/cam/cam_xpt.c:4664
#13 0xffffffff802cba1b in xpt_create_path (new_path_ptr=0xffffff81aeccdb50, 
    perph=0xfffffe00112e1e00, path_id=2, target_id=0, lun_id=0)
    at /home/trasz/p4/iscsi/sys/cam/cam_xpt.c:3398
#14 0xffffffff802dcca3 in scsi_scan_bus (periph=<value optimized out>, 
    request_ccb=0xfffffe00157e4800)
    at /home/trasz/p4/iscsi/sys/cam/scsi/scsi_xpt.c:1955
#15 0xffffffff802d0ca4 in xpt_scanner_thread (dummy=<value optimized out>)
    at /home/trasz/p4/iscsi/sys/cam/cam_xpt.c:841

It looks like it's impossible for the lock to be not owned in that code
path, so it suggests some kind of memory corruption.  It can be easily
reproduced by creating several SIMs at the exact same time.

[..]

-- 
If you cut off my head, what would I say?  Me and my head, or me and my body?