As a result of a checkin sometime last night, Linux viewer started crashing when visiting any external URLs (e.g. type in http://www.mozilla.org in the urlbar and hit return), with the message: Assertion failure: 0, at mkgeturl.c:3246 That assert in the code is accompanied by a trace message: call to processnet while doing gethostbyname call
I checked in some code on monday night (i.e. TWO nights ago) that puts nsSocketTransport protocols (currently only NNTP) on a seperate thread. This is defined with -DNETLIB_THREAD. I tested it after I turned on NETLIB_THREAD: It worked for me, and it appearantly worked for everyone yesterday because nobody complained and it passed the smoketests. It's possible my checkin did cause this, though I doubt it. But if it did, here are some possibilities: - the smoketests and regular builds aren't quitting on PR_ASSERT - there's some thread related race condition that didn't pop up in previous builds, and something else actually changed the timing. I'm doing a build right now so I can try with and without NETLIB_THREAD.
I checked my build from yesterday, and it can visit mozilla.org. So it's not Alec's change specifically which caused this; but it may be something else checked in last night which isn't compatible with Alec's changes (maybe someone made a change on an old tree and didn't test after updating netlib?)
Status: NEW → RESOLVED
Last Resolved: 19 years ago
Resolution: --- → FIXED
This was fixed when alecf backed out the NETLIB_THREAD change. The reason the smoketests work is because they are built without debug so the asserts do nothing.
seems to not crash anymore...in 2/5 build.
Changing all Networking Library/Browser bugs to Networking-Core component for Browser. Occasionally, Bugzilla will burp and cause Verified bugs to reopen when I do this in a bulk change. If this happens, I will fix. ;-)
Bulk move of all Networking-Core (to be deleted component) bugs to new Networking component.
You need to log in before you can comment on or make changes to this bug.