Bug 11003 - nwamd cores when trying to print a string
: nwamd cores when trying to print a string
Status: REOPENED
Product: opensolaris
networking
: unspecified
: i86pc/i386 OpenSolaris
: P2 normal (vote)
: ---
Assigned To: Watcher account for OpenSolaris networking
:
:
:
:
:
:
  Show dependency treegraph
 
Reported: 2009-08-28 04:48 UTC by Sanjay Nadkarni
Modified: 2009-11-12 02:35 UTC (History)
2 users (show)

See Also:


Attachments
core from nwamd (10.71 MB, application/octet-stream)
2009-08-28 04:51 UTC, Sanjay Nadkarni
no flags Details


Note

You need to log in before you can comment on or make changes to this bug.


Description Sanjay Nadkarni 2009-08-28 04:48:36 UTC
I am running build 118.  I see a number of cores with the following traceback
libc_hwcap1.so.1`strlen+0x30(80641fc, fe2a0b18, fe2a06d4, 0)
libc_hwcap1.so.1`vsnprintf+0x65(fe2a0704, 400, 80641fc, fe2a0b18)
dprintf+0x2a(80641fc, 0, 0, feef578a)
clear_unscanned_entries+0x310(fe2a0df4, 2, fe2a0df4, 805e17c)
scan_wireless_nets+0xf4(fe2a0df4)
rescan_wifi_no_lock+0x88(0, 0, 2, 2, 306b7769, 0)
periodic_wireless_scan+0x1d7(0, fef80000, fe2a0fec, feefce2e)
libc_hwcap1.so.1`_thrp_setup+0x7e(fed42200)
libc_hwcap1.so.1`_lwp_start(fed42200, 0, 0, feefce2e, 0, 0)

dprint is trying to print:
80641fc/s
0x80641fc:      keeping unscanned but connected AP %s %s
which is fairly reasonable except that the strings passed appear to be null.
but strlen of 0x80641fc should be reasonable.

since this is failing in strlen, let's see if there are leaks.

running ::findleaks 

 ::findleaks
BYTES             LEAKED VMEM_SEG CALLER
73728                  3 fdc8d000 MMAP
8192                   1 fe499000 MMAP
77824                  1 fde3e000 MMAP
------------------------------------------------------------------------
           Total       3 oversized leaks, 159744 bytes

CACHE     LEAKED   BUFFER CALLER
0807f290       1 0809ef10 ?
0807f510       1 08087ea8 ?
08082510       1 080a9a40 ?
------------------------------------------------------------------------
   Total       3 buffers, 680 bytes

So there are some leaks but there is no stack associated with it.
Comment 1 Sanjay Nadkarni 2009-08-28 04:51:13 UTC
Created an attachment (id=2533) [details]
core from nwamd

attaching the core
Comment 2 Michael Hunter 2009-08-31 08:41:55 UTC
Bumped prio to P2.  If it causes interruptions in service (as a core does) it
is _at least_ a P2.
Comment 3 Michael Hunter 2009-08-31 08:56:23 UTC
This code was recently changed so that these log messages are not emitted so
this bug doesn't happen.
Comment 4 Michael Hunter 2009-08-31 15:40:34 UTC
This should be against Phase 0.5 (picea).  I can't figure out how to
recatagorize it in bugzilla?!?!
Comment 5 Renee Danson Sommerfeld 2009-11-12 02:35:53 UTC
As Michael noted in comment 4, this is a bug against nwam phase 0.5.  It should
be in opensolaris/networking.  The nwam product under the development
classification is for bugs against the not-yet-integrated nwam phase 1.