New trace can't generate in persistent memory file system
Using the new feature Recording trace data on persistent memory file systems in LTTng 2.7.
After the device restart due to system crash, we can read the crash log in the specific shm-path directory. When we want to trace the new log, the following debug log reported and the new data can't be traced to the specific shm-path.
DEBUG3 - 20:07:54.353680 [286/291]: mkdir() recursive /shm/lttng/mysession-20160302-182255/ust/uid/0/32-bit with mode 504 for uid 0 and gid 0 (in run_as_mkdir_recursive() at runas.c:468)
DEBUG1 - 20:07:54.353726 [286/291]: Using run_as worker (in run_as() at runas.c:449)
DEBUG3 - 20:07:54.354141 [286/291]: open() /shm/lttng/mysession-20160302-182255/ust/uid/0/32-bit/metadata with flags C1 mode 384 for uid 0 and gid 0 (in run_as_open() at runas.c:498)
DEBUG1 - 20:07:54.354202 [286/291]: Using run_as worker (in run_as() at runas.c:449)
PERROR - 20:07:54.354724 [286/291]:
DEBUG3 - 20:07:54.354801 [286/291]: rmdir_recursive() /shm/lttng/mysession-20160302-182255 with for uid 0 and gid 0 (in run_as_rmdir_recursive() at runas.c:524)
DEBUG1 - 20:07:54.354847 [286/291]: Using run_as worker (in run_as() at runas.c:449)
DEBUG3 - 20:07:54.355554 [287/287]: Attempting rmdir /shm/lttng/mysession-20160302-182255 (in utils_recursive_rmdir() at utils.c:1247)
DEBUG3 - 20:07:54.356905 [286/291]: Buffer registry per UID destroy with id: 0, ABI: 32, uid: 0 (in buffer_reg_uid_destroy() at buffer-registry.c:641)
For detail info, please see my attachment.