Use ftrace to trace block event, and found the blow when bad read performance[1] is observed:
mmcqd/0-57 [000] 8827.689789: block_rq_issue: 179,0 W 0 () 5653017 + 8 [mmcqd/0] mmcqd/0-57 [000] 8828.408600: block_rq_complete: 179,0 W () 5653017 + 8 [0] Seems the bad read performance is caused by slow write operation, as traced by above: over 700ms is spent to write 8 sectors. The slow write can be often happened, but not each time. [1], use dd to test read performance on mmc card: #dd if=/dev/mmcblk0p1 iflag=direct of=/dev/null bs=65536 count=40 40+0 records in 40+0 records out 2621440 bytes (2.6 MB) copied, 10.0263 s, 261 kB/s [2], block event trace log, see attachment. ** Attachment added: "mmc-tr-dd-rd-261K.trace" https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap4/+bug/784421/+attachment/2176909/+files/mmc-tr-dd-rd-261K.trace -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/784421 Title: natty slow boot on pandaboard(rev A1) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux-ti-omap4/+bug/784421/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs