Thank you for the bug report.

The crash.txt file you attached shows Chromium crashed after
getting a handshake failure:

[21868:21882:336417980872:ERROR:net/socket/ssl_client_socket_nss.cc(1418)] handshake failed; NSS error code -12229, net_error -107

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffea364710 (LWP 21882)]
0x0000000000000000 in ?? ()
(gdb) bt full
#0  0x0000000000000000 in ?? ()
No symbol table info available.
#1  0x0000000000c49490 in net::HttpNetworkTransaction::DoReadHeadersComplete (this=0x4570a80, result=0) at net/http/http_network_transaction.cc:1379
        ssl_socket = 0x4ebd340
        rv = <value optimized out>

In r51387 and r52001 of http_network_transaction.cc, line 1379 is:

1376   if (using_ssl_) {
1377     SSLClientSocket* ssl_socket =
1378         reinterpret_cast<SSLClientSocket*>(connection_->socket());
1379     ssl_socket->GetSSLInfo(&response_.ssl_info);
1380   }

Most likely, connection_->socket() is not an SSLClientSocket at that point.

It sounds like it requires an ISA server to reproduce (directly) ?

BUGabundo: can you help debug this problem by checking wtc's assumption as to the cause?  After the segfault enter these commands into gdb and report the output:

p /a (* (void***)ssl_socket)[0]
p /a (* (void***)ssl_socket)[1]
p /a (* (void***)ssl_socket)[2]
p /a (* (void***)ssl_socket)[3]
p /a (* (void***)ssl_socket)[4]

vandebo: yes, I'm afraid that an ISA server is the best way to reproduce this crash.
We can also try configure a Squid proxy server to request Negotiate or NTLM auth.
I tried that once before, but couldn't get it to work.

One special property of Negotiate/NTLM is that it is connection-based, which
means only the very first HTTP request on the connection needs authentication.
Subsequent HTTP requests on the same connection don't need any authentication
header.  This may be relevant to this crash.

Would a combination of HttpNetworkTransactionTest.RestartAfterTLSDecompressionFailure and HttpProxyClientSocketPoolTest.HaveAuth be likely to trigger the same code paths in HttpNetworkTransaction?

vandebo sorry for the lack of early reply. i was without access to may email client.
its weekend, so i dont have access to the ISA server.
will try to retest on Monday.
Sorry once again for the delay, and thanks for the quick reply

i tried very hard, but could not reproduce this but with
6.0.471.0 (52879) Built on Ubuntu 9.10, running on Debian unstable
with both ISA server 2003 and 2010, today.

BUGabundo: did you build Chromium from source code yourself?
I'm curious why you're running on Debian unstable Chromium
binaries built on Ubuntu 9.10.

 Issue 45987  has been merged into this issue.

wtc i have not.
I'm just using Ubuntu daily PPA at work (as I do at home, in ubuntu 10.10).
I know now theres a repo for debian, but there wasnt one when i first installed chromium in debian.
if you think thats the root cause for some of my problems, i'll gladly follow your instructions to run a trunk version appropriated to my office OS.

BUGabundo: it should be fine to run Ubuntu daily PPA on Debian unstable.
I'm going to mark this bug WontFix (for "cannot reproduce") based on your
comment 9.  Please post a comment if this crash comes back.  Thanks!

This issue has been closed for some time. No one will pay attention to new comments.
If you are seeing this bug or have new data, please click New Issue to start a new bug.

Issue 505988 has been merged into this issue.