I've been doing some more real-time device testing with XFS, and I was able to get past the BUG I reported here: http://www.spinics.net/lists/xfs/msg37445.html by setting CONFIG_XFS_DEBUG=n as Dave suggested. After that I/O seems to work, although it doesn't look like the DAX mount option is applied to I/Os that go to the real-time device. After the first I/O, though, the following warning is hit: [ 413.086897] XFS (ram1): _xfs_buf_ioapply: no ops on block 0xa0/0x8 [ 413.091102] ffff8805038e8000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ [ 413.093495] ffff8805038e8010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ [ 413.095855] ffff8805038e8020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ [ 413.098221] ffff8805038e8030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ [ 413.100576] CPU: 0 PID: 1362 Comm: xfsaild/ram1 Not tainted 4.5.0-rc2 #20 [ 413.102244] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.2-20150714_191134- 04/01/2014 [ 413.104494] 0000000000000000 0000000004f1f04f ffff8800981abbe0 ffffffff8155d6b2 [ 413.106378] 0000000000000001 ffff8800981abc98 ffffffff814780d2 ffffffff81a50737 [ 413.108253] ffffe8ffffa00c00 ffff8800981abc50 ffffffff810c1947 00000000001d12d0 [ 413.110071] Call Trace: [ 413.110635] [<ffffffff8155d6b2>] dump_stack+0x44/0x62 [ 413.111807] [<ffffffff814780d2>] _xfs_buf_ioapply+0x432/0x480 [ 413.113106] [<ffffffff81a50737>] ? _raw_spin_unlock+0x27/0x30 [ 413.114406] [<ffffffff810c1947>] ? __queue_work+0x167/0x3a0 [ 413.115788] [<ffffffff810f8a64>] ? lockdep_init_map+0x64/0x6b0 [ 413.117116] [<ffffffff810fbc7d>] ? trace_hardirqs_on+0xd/0x10 [ 413.118421] [<ffffffff810d6a20>] ? wake_up_q+0x70/0x70 [ 413.119589] [<ffffffff81479964>] ? __xfs_buf_delwri_submit+0x1f4/0x2e0 [ 413.120984] [<ffffffff81479683>] xfs_buf_submit+0x73/0x160 [ 413.122175] [<ffffffff81479964>] __xfs_buf_delwri_submit+0x1f4/0x2e0 [ 413.123520] [<ffffffff8147a7cf>] ? xfs_buf_delwri_submit_nowait+0x2f/0x50 [ 413.124967] [<ffffffff8147a7cf>] ? xfs_buf_delwri_submit_nowait+0x2f/0x50 [ 413.126417] [<ffffffff8147a7cf>] xfs_buf_delwri_submit_nowait+0x2f/0x50 [ 413.127815] [<ffffffff814a7734>] xfsaild+0x264/0x6c0 [ 413.128871] [<ffffffff814a74d0>] ? xfs_trans_ail_cursor_first+0x90/0x90 [ 413.130222] [<ffffffff814a74d0>] ? xfs_trans_ail_cursor_first+0x90/0x90 [ 413.131529] [<ffffffff810c99a6>] kthread+0xf6/0x110 [ 413.132510] [<ffffffff810fbbe9>] ? trace_hardirqs_on_caller+0x129/0x1b0 [ 413.133805] [<ffffffff810c98b0>] ? kthread_create_on_node+0x250/0x250 [ 413.135036] [<ffffffff81a514df>] ret_from_fork+0x3f/0x70 [ 413.136084] [<ffffffff810c98b0>] ? kthread_create_on_node+0x250/0x250 The steps to reproduce this are as follows: # fdisk -l /dev/ram* Disk /dev/ram0: 4 GiB, 4294967296 bytes, 8388608 sectors Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 4096 bytes I/O size (minimum/optimal): 4096 bytes / 4096 bytes Disk /dev/ram1: 4 GiB, 4294967296 bytes, 8388608 sectors Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 4096 bytes I/O size (minimum/optimal): 4096 bytes / 4096 bytes # mkfs.xfs -f -r rtdev=/dev/ram0 /dev/ram1 meta-data=/dev/ram1 isize=512 agcount=4, agsize=262144 blks = sectsz=4096 attr=2, projid32bit=1 = crc=1 finobt=1, sparse=0 data = bsize=4096 blocks=1048576, imaxpct=25 = sunit=0 swidth=0 blks naming =version 2 bsize=4096 ascii-ci=0 ftype=1 log =internal log bsize=4096 blocks=2560, version=2 = sectsz=4096 sunit=1 blks, lazy-count=1 realtime =/dev/ram0 extsz=4096 blocks=1048576, rtextents=1048576 # mount -t xfs -o rtdev=/dev/ram0 /dev/ram1 /mnt # mount | grep '/mnt' /dev/ram1 on /mnt type xfs (rw,relatime,seclabel,attr2,inode64,rtdev=/dev/ram0,noquota) # xfs_rtcp ~/data /mnt/ xfs_rtcp: /root/data is not a realtime file. Thanks, - Ross _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
