[LU-7423] slab corruption from sanity 300k

Thu, 12 Nov 2015 22:58:31 +0000

Running sanity 300k with MDSCOUNT=4 introduces slab corruption.

# export MDSCOUNT=4
# llmount.sh
...
# ONLY=300k sh lustre/tests/sanity.sh
...
== sanity test 300k: test large striped directory == 14:29:59 (1447360199)
fail_loc=0x1703
fail_loc=0
/mnt/lustre/d300k.sanity/striped_dir
lmv_stripe_count: 512 lmv_stripe_offset: 0
mdtidx		 FID[seq:oid:ver]
     0		 [0x200000400:0x2:0x0]		
     1		 [0x240000403:0x2:0x0]		
...



[15335.512746] Slab corruption (Tainted: G        W  ---------------   ): size-32768 start=ffff88007c5b0000, len=32768
[15335.514968] 000: 00 00 00 00 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b
[15335.516752] 010: 00 00 00 00 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b
[15335.518440] 020: 00 00 00 00 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b
[15335.520005] 030: 00 00 00 00 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b
[15335.521703] 040: 00 00 00 00 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b
[15335.523398] 050: 00 00 00 00 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b

The slab corruption started after http://review.whamcloud.com/15899. The fix for ~~LU-7414~~ does not fix the corruption. But restoring the old OUT_UPDATE_REPLY_SIZE of 16384 does fix it.

Whamcloud Community JIRA

[LU-7423] slab corruption from sanity 300k