Re: [Bacula-users] bacula-sd segfaulting on Linux

[Jeffrey L. <jtl...@po...>]

On Wed, 2004-03-17 at 03:16, Kern Sibbald wrote:
> Hello,
> 
> Hmmm. It is hard to know if the problem is because of the drive or the
> 2.6 kernel.  In principle, there is no reason for Bacula to die where it
> did even if there is a problem with the drive unless something is
> *seriously* wrong.  I have two suggestions:
> 
> 1. Run Bacula under the debugger again and when it crashes, enter:
> 
>    print *rec
>    print FI_to_ascii(rec->FileIndex)
>    print stream_to_ascii(rec->Stream, rec->FileIndex)
> 
> I've never tried printing a subroutine call, so the last two may die. If
> they do, you might try investigating a bit further.
> 
> 2. Please download the latest 1.33 from the CVS, if possible. If not,
> get the latest devel 1.33 tar. Then build (not necessary to install) and
> run btape "test" on your tape drive. That should give us more
> information on whether or not Bacula should work with your drive.
> 
> Best regards, Kern
> 
> 

Ok, I downloaded the CVS tree and built it using the same configure
options as the debian source package. I then ran the btape test command
and it segfaulted soon after running.

So I then ran it under GDB, and did a little (inept) poking around.
Anything I should look at in particular with GDB? Here's the output from
btape under gdb:

Many thanks for the help, Kern!
-- Jeff

-----[snip]-----

(gdb) run -c /etc/bacula/bacula-sd.conf /dev/nst0
Starting program: /home/jtlayton/svn/bacula/src/stored/btape -c
/etc/bacula/bacula-sd.conf /dev/nst0
[New Thread 1076171424 (LWP 31369)]
Tape block granularity is 1024 bytes.
btape: butil.c:146 Using device: /dev/nst0 for writing.
btape: btape.c:305 open_dev /dev/nst0 OK
*test
 
=== Write, rewind, and re-read test ===
 
I'm going to write 1000 records and an EOF
then write 1000 records and an EOF, then rewind,
and re-read the data to verify that it is correct.
 
This is an *essential* feature ...
 
 
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1076171424 (LWP 31369)]
write_block_to_dev(DCR*, DEV_BLOCK*) (dcr=0x0, block=0x80a4498) at
block.c:377
377        JCR *jcr = dcr->jcr;
Current language:  auto; currently c++
(gdb) thread apply all bt
 
Thread 1 (Thread 1076171424 (LWP 31369)):
#0  write_block_to_dev(DCR*, DEV_BLOCK*) (dcr=0x0, block=0x80a4498) at
block.c:377
#1  0x0804be55 in write_read_test () at btape.c:751
#2  0x0804d6de in testcmd () at btape.c:1270
#3  0x0805094d in do_tape_cmds () at btape.c:2418
#4  0x0804a536 in main (margc=0, margv=0xbffffbe0) at btape.c:241
(gdb) print *dcr
Cannot access memory at address 0x0
(gdb) print *block
$1 = {next = 0x0, dev = 0x80a3d88, binbuf = 64448, block_len = 64512,
buf_len = 64512, BlockNumber = 0, read_len = 0, VolSessionId = 0,
  VolSessionTime = 0, read_errors = 0, BlockVer = 2, write_failed =
false, block_read = false, FirstIndex = 0, LastIndex = 0,
  bufp = 0x80b40c8 'U' <repeats 64 times>, "%", buf = 0x80a4508 'U'
<repeats 24 times>}
(gdb)