Aaargh! My own followup got mis-routed!
Begin forwarded message: Date: Fri, 22 Jul 2011 11:15:46 +0100 From: Nick Kew <[email protected]> On 22 Jul 2011, at 09:38, Nick Kew wrote: > I take it this is way too unpredictable to reproduce to order, so an > interactive gdb > session would be pointless? Oh, hang on ... that's nonsense based on what you said! So, looks like you can get a whole lot more useful information: 1. Compile APR with -g -O0 2. Run httpd with -X, so you retain it in an interactive session 3. Get a backtrace on segfault. Note that a simple "bt" in gdb may report the wrong thread, so some poking around needed. One more quick check: if you rebuilt APR, did you rebuild APR-UTIL and HTTPD on top of it, just in case your build options affected binary compatibility somewhere? It shouldn't happen, but you never know, especially if your packager has made their own changes - e.g. optimised something. -- Nick Kew Available for work, contract or permanent http://www.webthing.com/~nick/cv.html
