Martin Stjernholm wrote:
"Stephen R. van den Berg" srb@cuci.nl wrote:
Some others:
#3 0xb7cebace in *__GI___assert_fail ( assertion=0x82f1303 "!gc_mark_queue.first",
Ok, let's take a couple more steps from the mark queue garbage angle then. I've added a few more checks now which hopefully nails that bug at the source. It doesn't need GC_MARK_DEBUG, but if you can, please define that anyway since it might improve the debug info a little.
It seems like I spoke too soon. I doublechecked previously, but apparently confused binaries. The server has been running with the latest, without problems for a day or so, and I've downgraded to the version with just the fix which I thought was still crashing, and it doesn't crash anymore (about this). So it seems like it is solved after all.
Sorry for the confusion.