[Bug general/3667] fcore crashes on large (eclipse) java process

["pmuldoon at redhat dot com" <sourceware-bugzilla at sourceware dot org>]

------- Additional Comments From pmuldoon at redhat dot com  2006-12-06 22:24 -------
I added some comments to the uslurp, and it seems to be dying on realloc. What
is puzzling is I read the maps twice: once to forward count the number of maps,
and then again to actually read the maps. The first time around the reallocs
pass, but the second time around they fail on the first realloc. Also this seems
limited to large maps files.

[pmuldoon@localhost frysk_bin]$ ./frysk-core/frysk/bindir/fcore 31807
Counting maps
Inside slurp
Read 4006 bytes, read buffer should be 8191
realloc size 12198
realloc completed
Read 4061 bytes, read buffer should be 8191
realloc size 16259
realloc completed
Read 4021 bytes, read buffer should be 8191
realloc size 20280
realloc completed
Read 4045 bytes, read buffer should be 8191
realloc size 24325
realloc completed
Read 4074 bytes, read buffer should be 8191
realloc size 28399
realloc completed
Read 4037 bytes, read buffer should be 8191
realloc size 32436
realloc completed
Read 2121 bytes, read buffer should be 8191
realloc size 34557
realloc completed
Read 0 bytes, read buffer should be 8191
realloc size 34557
realloc completed
Building program header

Building notes
Building program segments
Running segment construct

Inside slurp
Read 4006 bytes, read buffer should be 8191
realloc size 12198
*** glibc detected *** ./frysk-core/frysk/bindir/fcore: realloc(): invalid next
size: 0x095f8bd8 ***


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=3667

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.