Re: 2.4 testing ?

From: Adrian Chadd <adrian@dont-contact.us>
Date: Mon, 20 Nov 2000 22:19:42 +0800

On Sat, Nov 18, 2000, Andres Kroonmaa wrote:
>
> I'm seeing this difficult crashing. I think this is what Henrik meant.
> The patch to ftp.c was another issue. I'm having real trouble to track
> it, both because it is pretty rare, and I'm not able to produce it on
> will. Still, from several crashes I'm seeing the same pattern: they
> are related to client_side.c
>
> assert crashes occur in 2 scenarios:
> (ALL related to mempool: MEM_CLIENT_SOCK_BUF)
> 1) memFree from clientSendMoreData() after CommWriteStateCallbackAndFree()
> 2) memAlloc from clientWriteComplete after CommWriteStateCallbackAndFree()
> 3) memAlloc from random places after a corruption...

Ok. this is a little strange ..

Take a look at clientHandleIMSReply() . There's some code which
handles size >= CLIENT_SOCK_SZ . size shouln't do that, because
if it ever gets there, the storeClientCopy() which initiated
it has already trashed our buffer

That is one potential killer, which shouldn't happen. Andres,
can you please put an assert(1==0); inside that bit of code
and see if your caches get triggered?

If by some fluke it *is* being triggered, I'd be very surprised,
because storeClientCopy() is being passed an explicit copy
size limit, which it should not overflow.. :)

Adrian

-- 
Adrian Chadd			"God: Damn! I left pot everywhere!
<adrian@creative.net.au>	  Now I'll have to create Republicans!"
				    - Bill Hicks
Received on Mon Nov 20 2000 - 07:19:54 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:12:59 MST