On Thu, Jun 07, 2012 at 06:51:55AM -0400, Richard Hipp wrote: > On Thu, Jun 7, 2012 at 6:46 AM, Lluís Batlle i Rossell > <vi...@viric.name>wrote: > > > On Thu, Jun 07, 2012 at 06:36:23AM -0400, Richard Hipp wrote: > > > On Thu, Jun 7, 2012 at 3:51 AM, Christopher Vance <cjsva...@gmail.com > > >wrote: > > > > > > > Firefox gave the message I quoted. > > > > Chrome said that the server returned no data. > > > > Safari just shows an empty page. > > > > Telnet to the relevant port returns no data. > > > > > > > > > > Ok. Good clue. > > > > > > It appears that the Fossil binary is segfaulting while it is trying to > > > generate the page. What version of Fossil are you running? Have you > > tried > > > updating to the tip of trunk? > > > > As of debugging this, you (Christopher) can look at the PID of the fossil > > server, attach to it with gdb (gdb fossil PID), then run "set > > follow-fork-mode > > child", and tell gdb "contiue". Then try to visit the page, and gdb should > > stop > > at the segfaulting point. > > > > Not that easy, unfortunately. Fossil starts a new instance of itself to > handle each incoming page request, leaving the debugger looking at the > original instance, which is not the one that is crashing. > > The way to debug this is: > > (1) Start your debugger > (2) Invoke fossil with a single argument "test-http" > (3) Manually type in an HTTP header. This can be minimal: "GET > /timeline". You'll need to press enter twice to signal the end-of-header. > > With any luck, you'll hit the segfault that way. But sometimes, the > problem is due to a specific login or user permission, in which case a more > elaborate HTTP request has to be entered.
Well, the key was "set follow-fork-mode child". But that requires the user to have the browser ready to access the faulting link as the *first* fossil request after continuing gdb. :) I find the core dump way easier, sure. I didn't know the test-http approach. Thank you, Lluís. _______________________________________________ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
