48510 - assertion if get PickleSecurityStatus before connection shutdown

Status: VERIFIED WONTFIX

(Core :: Security: PSM, defect, P3)

Description

[John G. Myers]

This is related to bug 46739.

Running with the patch attached to bug 31982, I frequently get asserts on 
sslconn.c:262.  One thread is trying to destroy the data connection while 
another is still in SSM_SSLDataServiceThread on the connection.  This happens 
most, but not all of the time, so timing likely has something to do with this.

Comment 1

[Javier Delgadillo]

Putting on nsbeta2 radar

Comment 2

[Javier Delgadillo]

jgmyers:

Can you check in your patch to some branch you create?  The diff's way to large 
to apply by hand and there won't be any errors if I do a cvs up -j instead of 
manually patching my tree from your diff.

Comment 3

[John G. Myers]

I am working on verifying and attaching a new patch with the conflicts resolved.

Comment 4

[Javier Delgadillo]

My point is that your patch is so big that applying it by hand will surely lead 
to an error on my part.  If you create a branch that was used only as a vehicle 
for deilvering your patch to my tree, that would reduce the chance of me 
mis-applying the patch.  It's a simple cvs tag -b   ;-)

Comment 5

[John G. Myers]

Once an updated patch is created, you should apply it with the "patch -p0" 
command, not by hand.  Creating a branch is a nontrivial undertaking since there 
are changes to netwerk and conflicts are freqent.

Comment 6

[John G. Myers]

My test case is no longer crashing, I suspect the timing has changed.

Comment 7

[Javier Delgadillo]

Removing [nsbeta3+] statys since this no longer happens.  Will investigate once 
again if ther problem comes up again.

Comment 8

[Michael La Guardia]

marking nsbeta3- per Javi's comment

Comment 9

[John Unruh]

Changing QA contact to nitinp

Comment 10

[John Unruh]

Mass reassigning nitinp's bugs to me.

Comment 11

[John G. Myers]

This bug is no longer relevant.

Comment 12

[John Unruh]

Verified wontfix based on jgmyers's comment.