On Mon, Apr 30, 2007 at 09:42:19AM -0400, Stephen Isard wrote: > > Wow, that was fast. Yes. No segfault now when I index with the binary > compiled with the newer gcc 4.1.1.
Well, my reply was fast. The actual patch was constructed a few days ago, arising from a similar bug report via the Debian package maintainer. > From a quick look at your patch, it appears that you now release some > memory in a situation where before you didn't. Can you explain what > has changed to make the new compiler require that, where the older > versions didn't? Or point me at a relevant piece of documentation? No, the releasing of memory is a secondary issue. The problem before was that if the name-value parser was confused by one of the headers it was parsing, it resulted in current_state becoming -1, and this variable is used to index into arrays. It was something around here that was causing the segfault. The calls to free() are to clean up the partially built information when a parse is being abandoned. I presume that the old compiler was laying out the data in memory differently, so that the bogus array reference somehow "got away with it". Also, this parser was new in 0.20, so previous mairix versions would not have had this bug. Perhaps your compiler upgrade coincided with building a new mairix version. ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Mairix-users mailing list Mairix-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mairix-users
