On Mon, May 09, 2016 at 10:47:03AM +0200, Christoph Hellwig wrote: > This series add a new file system I/O path that uses the iomap structure > introduced for the pNFS support and support multi-page buffered writes. > > This was first started by Dave Chinner a long time ago, then I did beat > it into shape for production runs in a very constrained ARM NAS > enviroment for Tuxera almost as long ago, and now half a dozen rewrites > later it's back. > > The basic idea is to avoid the per-block get_blocks overhead > and make use of extents in the buffered write path by iterating over > them instead. > > Note that patch 1 conflicts with Vishals dax error handling series. > It would be great to have a stable branch with it so that both the > XFS and nvdimm tree could pull it in before the other changes in this > area. I just pulled this forward to 4.7-rc1, and I get an immediate failure in generic/346: [ 70.701300] ------------[ cut here ]------------ [ 70.702029] kernel BUG at fs/xfs/xfs_aops.c:1253! [ 70.702778] invalid opcode: 0000 [#1] PREEMPT SMP [ 70.703484] Modules linked in: [ 70.703952] CPU: 2 PID: 5374 Comm: holetest Not tainted 4.7.0-rc1-dgc+ #812 [ 70.704991] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Debian-1.8.2-1 04/01/2014 [ 70.706285] task: ffff8801365e23c0 ti: ffff8800b0698000 task.ti: ffff8800b0698000 [ 70.707395] RIP: 0010:[<ffffffff814f5ba7>] [<ffffffff814f5ba7>] __xfs_get_blocks+0x597/0x6b0 [ 70.708768] RSP: 0000:ffff8800b069b990 EFLAGS: 00010246 [ 70.709518] RAX: ffff88013ac283c0 RBX: 000000000005c000 RCX: 000000000000000c [ 70.710527] RDX: 000000000005d000 RSI: 0000000000000008 RDI: ffff8800b3fc1b90 [ 70.711579] RBP: ffff8800b069ba18 R08: 000000000000006b R09: ffff8800b069b914 [ 70.712626] R10: 0000000000000000 R11: 000000000000006b R12: ffff8800b3fc1ce0 [ 70.713656] R13: 0000000000001000 R14: ffff8800b069bb38 R15: ffff8800b9442000 [ 70.714653] FS: 00007ff002a27700(0000) GS:ffff88013fd00000(0000) knlGS:0000000000000000 [ 70.715820] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 70.716669] CR2: 00007ff00436ec00 CR3: 00000000ae8c1000 CR4: 00000000000006e0 [ 70.717656] Stack: [ 70.717940] ffff8800b3fc1b40 ffff8800b3fc1b60 ffff880000000000 000000000000005c [ 70.719062] ffff880100000000 ffff8800b3fc1b00 0000000000000000 00000001b069b9d8 [ 70.720199] 0000000000000000 ffffffffffffffff 000000000000005d 0000000000000000 [ 70.721294] Call Trace: [ 70.721644] [<ffffffff814f5cd7>] xfs_get_blocks+0x17/0x20 [ 70.722401] [<ffffffff812368f4>] do_mpage_readpage+0x3d4/0x710 [ 70.723250] [<ffffffff811ab61e>] ? lru_cache_add+0xe/0x10 [ 70.724013] [<ffffffff81236d28>] mpage_readpages+0xf8/0x150 [ 70.724828] [<ffffffff814f5cc0>] ? __xfs_get_blocks+0x6b0/0x6b0 [ 70.725654] [<ffffffff814f5cc0>] ? __xfs_get_blocks+0x6b0/0x6b0 [ 70.726504] [<ffffffff811e544c>] ? alloc_pages_current+0x8c/0x110 [ 70.727365] [<ffffffff814f38d8>] xfs_vm_readpages+0x38/0xa0 [ 70.728177] [<ffffffff811a97f2>] __do_page_cache_readahead+0x192/0x230 [ 70.729107] [<ffffffff8119e030>] filemap_fault+0x440/0x4b0 [ 70.729881] [<ffffffff81e39080>] ? down_read+0x20/0x40 [ 70.730616] [<ffffffff815007cf>] xfs_filemap_fault+0x5f/0x110 [ 70.731456] [<ffffffff811c2907>] __do_fault+0x67/0xf0 [ 70.732205] [<ffffffff811c6aa9>] handle_mm_fault+0x239/0x1460 [ 70.733015] [<ffffffff810a2403>] __do_page_fault+0x1c3/0x4f0 [ 70.733821] [<ffffffff810a27f3>] trace_do_page_fault+0x43/0x140 [ 70.734654] [<ffffffff8109cc8a>] do_async_page_fault+0x1a/0xa0 [ 70.735493] [<ffffffff81e3d018>] async_page_fault+0x28/0x30 [ 70.736500] Code: 41 ff d2 4d 8b 16 4d 85 d2 75 dd 4c 8b 65 98 4c 8b 75 80 65 ff 0d 82 69 b1 7e 74 11 e9 e6 fb ff ff e8 76 c4 b0 ff e9 e9 fd ff ff <0f> 0b e8 6a c4 [ 70.740283] RIP [<ffffffff814f5ba7>] __xfs_get_blocks+0x597/0x6b0 [ 70.741157] RSP <ffff8800b069b990> [ 70.742097] ---[ end trace aeed47f2452ca28a ]--- Maybe I screwed up the forward merge sorting out all the bits that conflicted with what went into 4.7-rc1. Perhaps it would be best if you rebased and reposted, Christoph? Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
