Bug 56831

Summary: RHEL Linux Data Corruption found with EXT4 system
Product: File System Reporter: searchfordave
Component: ext4Assignee: fs_ext4 (fs_ext4)
Status: CLOSED OBSOLETE    
Severity: blocking CC: alan, lalwanihiro
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.32-220 Subsystem:
Regression: No Bisected commit-id:
Attachments: File "random_multi_write_read.c" gives data corruption on ext4

Description searchfordave 2013-04-19 13:41:01 UTC 
The attached file when compiled on a 32GB Mem system, gives a data corruption if run with number of threads greater than 15

How to run:
1. compile: # gcc random_multi_write_read.c -o test -lpthread
2. Usage:   # ./test <file path> <no. of threads>
3. Run:     # ./test file_1 15

Sample Output:
Starting 15 write threads to file_1 at different offsets and write sizes...
Parallelly reading the file for any data corruption...
DATA CORRUPTION HIT
Actual: '', Expected: 'K'd
, start:4581061500, end:5039167650
...

For More info, refer file "V_loop<%d>char<%c>" which gets created newly to capture the corrupted data as soon as corruption is hit.

Test system details:
#CPUS: 2
#CORES: 6
SYSTEM MEMORY: 15 GB
SWAP MEMORY: 17 GB
Comment 1 searchfordave 2013-04-19 13:42:27 UTC 
Created attachment 99321 [details]
File "random_multi_write_read.c" gives data corruption on ext4
Comment 2 Alan 2013-11-13 21:03:12 UTC 
2.6.32 is obsolete, please take this up with your Red Hat support