Inode cache bugs
Mar 23, 2009
by
Both this and the one below look to be caused by the reentrancy problem fixed in Zest about 2 weeks ago.
[root@castor mount_slash]# fg PSC_LOG_LEVEL=2 SLASH_MDS_NID="128.182.58.80@tcp10" LNET_NETWORKS="tcp10(eth1)" USOCK_CPORT=1300 gdb ./mount_slash #0 0x0000000000404e0c in atomic_dec (v=0x28) at ..//..//psc_fsutil_libs/include/psc_util/atomic.h:393 #1 0x000000000040531a in slash2fuse_access (req=0x7fafb0, ino=15294024, mask=1) at main.c:229 #2 0x0000000000403ee2 in slash2fuse_listener_loop (arg=0x0) at fuse_listener.c:260 #3 0x0000003e8ce073da in start_thread () from /lib64/libpthread.so.0 #4 0x0000003e8c2e62bd in clone () from /lib64/libc.so.6 (gdb) up #1 0x000000000040531a in slash2fuse_access (req=0x7fafb0, ino=15294024, mask=1) at main.c:229 229 fidc_membh_dropref(c); (gdb) print *c Cannot access memory at address 0x0 (gdb)