dtrace discuss - Nov 2005 - re an issue with streams and Dtrace

Hi
Would someopne be able to look at bug for me:

6332949 "V240 panics with kmem_flags set, buffer corruption in 
streams/Dtrace while testing arbitrary patches"

The panic occurs with kmem_flags set to 0xf on any SPARC FCS box.

But only occurs when using out test harness ( a large set of ksh scripts 
), unfortunately.
basically we use ksh to rsh to a target machine and then do certain 
actions, ie add patches, remove patches etc, all remotely via rsh.

but we consistently see panics when using kmem_flags and the stack seems 
to indiacte something related to Dtrace and streams

Now I have not been able to reproduce outside of our test harness,
ie tried to cut down the test harness but the panic never reproduces, ( 
some timing issue maybe )

seems to indicate some threading issue where a buffer is being written 
after being freed

Any help appreciated, core dump location is in the comment section of 
the bug.

Currently this reproduces on every sparc box we use and is somewhat 
hindering our testing.
We use Dtrace to try and hit as many probes as possible on the running 
processes on the system etc, but it seems to panic the system fairly 
consistently, maybe not every patch but certainly every couple, and is 
not related to patches being tested ( most are just user land stuff, 
like patchadd etc )

thanks
regards
Enda