wordpress開(kāi)發(fā)視頻網(wǎng)站模板下載/免費(fèi)二級(jí)域名平臺(tái)
目錄
前言
問(wèn)題分析
page buffers創(chuàng)建
page buffers丟失
Write-Protect
Dirty Page w/o Buffers
問(wèn)題解決
前言
這個(gè)問(wèn)題發(fā)生在3.10.0-514.el7上，并且在RHEL的知識(shí)庫(kù)中快速找到了對(duì)應(yīng)的案例以及解決方案，但是，理解問(wèn)題如何發(fā)生和解決則著實(shí)費(fèi)了些功夫。
RHEL的鏈接如下：
RHEL7: kernel crash in xfs_vm_writepage - kernel BUG at fs/xfs/xfs_aops.c:1062! - Red Hat Customer Portal
調(diào)用棧為：
[1004630.854317] kernel BUG at fs/xfs/xfs_aops.c:1062!
[1004630.854894] invalid opcode: 0000 [#1] SMP
[1004630.861333] CPU: 6 PID: 56715 Comm: kworker/u48:4 Tainted: G W ------------ 3.10.0-514.el7.x86_64 #1
[1004630.862046] Hardware name: HP ProLiant BL460c Gen9, BIOS I36 12/28/2015
[1004630.862703] Workqueue: writeback bdi_writeback_workfn (flush-253:28)
[1004630.863414] task: ffff881f8436de20 ti: ffff881f23a4c000 task.ti: ffff881f23a4c000
[1004630.864117] RIP: 0010:[<ffffffffa083f2fb>] [<ffffffffa083f2fb>] xfs_vm_writepage+0x58b/0x5d0 [xfs]
[1004630.864860] RSP: 0018:ffff881f23a4f948 EFLAGS: 00010246
[1004630.865749] RAX: 002fffff00040029 RBX: ffff881bedd50308 RCX: 000000000000000c
[1004630.866466] RDX: 0000000000000008 RSI: ffff881f23a4fc40 RDI: ffffea00296b7800
[1004630.867218] RBP: ffff881f23a4f9f0 R08: fffffffffffffffe R09: 000000000001a098
[1004630.867941] R10: ffff88207ffd6000 R11: 0000000000000000 R12: ffff881bedd50308
[1004630.868656] R13: ffff881f23a4fc40 R14: ffff881bedd501b8 R15: ffffea00296b7800
[1004630.869399] FS: 0000000000000000(0000) GS:ffff881fff180000(0000) knlGS:0000000000000000
[1004630.870147] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[1004630.870868] CR2: 0000000000eb3d30 CR3: 0000001ff79dc000 CR4: 00000000001407e0
[1004630.871610] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[1004630.872349] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[1004630.873072] Stack:
[1004630.873749] 0000000000008000 ffff880070b03644 ffff881f23a4fc40 ffff881f23a4fa68
[1004630.874480] ffff881f23a4fa80 ffffea00296b7800 0000000000001000 000000000000000e
[1004630.875223] 0000000000001000 ffffffff81180981 0000000000000000 ffff881bedd50310
[1004630.875957] Call Trace:
[1004630.876665] [<ffffffff81180981>] ? find_get_pages_tag+0xe1/0x1a0
[1004630.877417] [<ffffffff8118b3b3>] __writepage+0x13/0x50
[1004630.878173] [<ffffffff8118bed1>] write_cache_pages+0x251/0x4d0
[1004630.878915] [<ffffffffa00c170a>] ? enqueue_cmd_and_start_io+0x3a/0x40 [hpsa]
[1004630.879626] [<ffffffff8118b3a0>] ? global_dirtyable_memory+0x70/0x70
[1004630.880368] [<ffffffff8118c19d>] generic_writepages+0x4d/0x80
[1004630.881157] [<ffffffffa083e063>] xfs_vm_writepages+0x53/0x90 [xfs]
[1004630.881907] [<ffffffff8118d24e>] do_writepages+0x1e/0x40
[1004630.882643] [<ffffffff81228730>] __writeback_single_inode+0x40/0x210
[1004630.883403] [<ffffffff8122941e>] writeback_sb_inodes+0x25e/0x420
[1004630.884141] [<ffffffff8122967f>] __writeback_inodes_wb+0x9f/0xd0
[1004630.884863] [<ffffffff81229ec3>] wb_writeback+0x263/0x2f0
[1004630.885610] [<ffffffff810ab776>] ? set_worker_desc+0x86/0xb0
[1004630.886378] [<ffffffff8122bd05>] bdi_writeback_workfn+0x115/0x460
[1004630.887142] [<ffffffff810c4cf8>] ? try_to_wake_up+0x1c8/0x330
[1004630.887875] [<ffffffff810a7f3b>] process_one_work+0x17b/0x470
[1004630.888638] [<ffffffff810a8d76>] worker_thread+0x126/0x410
[1004630.889389] [<ffffffff810a8c50>] ? rescuer_thread+0x460/0x460
[1004630.890126] [<ffffffff810b052f>] kthread+0xcf/0xe0
[1004630.890816] [<ffffffff810b0460>] ? kthread_create_on_node+0x140/0x140
[1004630.891521] [<ffffffff81696418>] ret_from_fork+0x58/0x90
[1004630.892229] [<ffffffff810b0460>] ? kthread_create_on_node+0x140/0x140
[1004630.892877] Code: e0 80 3d 4d b4 06 00 00 0f 85 a4 fe ff ff be d7 03 00 00 48 c7 c7 4a e0 88 a0 e8 61 66 84 e0 c6 05 2f b4 06 00 01 e9 87 fe ff ff <0f> 0b 8b 4d a4 e9 e8 fb ff ff 41 b9 01 00 00 00 e9 69 fd ff ff
[1004630.894245] RIP [<ffffffffa083f2fb>] xfs_vm_writepage+0x58b/0x5d0 [xfs]
[1004630.894890] RSP <ffff881f23a4f94問(wèn)題發(fā)生的位置為：
xfs_vm_writepage()
---...bh = head = page_buffers(page);...
---#define page_buffers(page) \({ \BUG_ON(!PagePrivate(page)); \((struct buffer_head *)page_private(page)); \})問(wèn)題分析
page buffers創(chuàng)建
在Linux內(nèi)核中，buffer head的目的是為了對(duì)接在存儲(chǔ)子系統(tǒng)和內(nèi)存子系統(tǒng)的兩個(gè)基本單位：
- sector，這是存儲(chǔ)的基本單位，512字節(jié)
- page，這個(gè)是內(nèi)存的基本單位，4096字節(jié)
文件緩存，即page cache，是存儲(chǔ)子系統(tǒng)和內(nèi)存子系統(tǒng)的結(jié)合部，buffer_head對(duì)應(yīng)的就是page cache中的一個(gè)sector；當(dāng)我們格式化文件系統(tǒng)，把fsblock設(shè)置為512、1024、2048字節(jié)時(shí)，或者操作raw block設(shè)備時(shí)，每個(gè)page就會(huì)對(duì)應(yīng)多個(gè)buffer_head；
近些年，512字節(jié)的fsblock基本被主流文件系統(tǒng)拋棄，雖然還支持，但是都以支持4K為主要的優(yōu)化方向，xfs甚至拋棄了buffer_head，直接使用page作為IO操作的基本單位；
那么，buffer_head都是在哪些契機(jī)被創(chuàng)建呢？
3.10.0-514.el7
【A】
generic_perform_write()-> aops->write_begin()xfs_vm_write_begin()-> grab_cache_page_write_begin()-> __block_write_begin()-> create_page_buffers()
【B】
do_shared_fault()-> __do_fault()-> vma->vm_ops->page_mkwrite()xfs_filemap_page_mkwrite()-> __block_page_mkwrite()-> __block_write_begin()-> create_page_buffers()do_mpage_readpage() -> block_read_full_page() //如果page中各個(gè)bh的狀態(tài)不一致，不如有些map有些unmap，會(huì)進(jìn)入到此路徑，分別對(duì)bh進(jìn)行操作-> create_page_buffers()通常來(lái)講【A】和【B】路徑可以保證，在對(duì)page做寫(xiě)操作之前，保證page buffers已經(jīng)創(chuàng)建，它們分別代表是通過(guò)系統(tǒng)調(diào)用和mmap對(duì)文件進(jìn)行寫(xiě)的場(chǎng)景；
page buffers丟失
那本問(wèn)題中，被設(shè)置了dirty標(biāo)記的page的buffer是如何丟失的呢？
page buffer被釋放的典型場(chǎng)景：
3.10.0-514.el7shrink_page_list()-> try_to_unmap()-> try_to_release_page()-> aops->releasepage()xfs_vm_releasepage()-> try_to_free_buffers()-> __remove_mapping()在try_to_release_page()之前，try_to_unmap()會(huì)被調(diào)用，它會(huì)清理掉pte，并且檢測(cè)是否需要page dirty
3.10.0-514.el7try_to_unmap()-> try_to_unmap_file()-> try_to_unmap_one()-> set_page_dirty() //pte_dirty()這樣就可以保證，在執(zhí)行try_to_release_page()之前，給page及其buffer設(shè)置dirty 標(biāo)記，
3.10.0-514.el7xfs_vm_set_page_dirty()
---spin_lock(&mapping->private_lock);if (page_has_buffers(page)) {struct buffer_head *head = page_buffers(page);struct buffer_head *bh = head;do {if (offset < end_offset)set_buffer_dirty(bh);bh = bh->b_this_page;offset += 1 << inode->i_blkbits;} while (bh != head);}
---try_to_free_buffers()-> drop_buffers()-> buffer_busy()-> atomic_read(&bh->b_count) | (bh->b_state & ((1 << BH_Dirty) | (1 << BH_Lock)))如果buffer有dirty標(biāo)記就不會(huì)被釋放。另外，有truncate和invalidate的場(chǎng)景，也是類(lèi)似的操作。
但是有一個(gè)機(jī)器特殊的場(chǎng)景：
3.10.0-514.el7shrink_active_list()
---if (unlikely(buffer_heads_over_limit)) {if (page_has_private(page) && trylock_page(page)) {if (page_has_private(page))try_to_release_page(page, 0);unlock_page(page);}}
---這里對(duì)buffers直接進(jìn)行釋放。
Write-Protect
用對(duì)mmap的page的寫(xiě)操作，是通過(guò)下面的機(jī)器觸發(fā)的page fault并給page設(shè)置dirty的
3.10.0-514.el7generic_writepages()-> clear_page_dirty_for_io()-> page_mkclean()-> page_mkclean_file()-> page_mkclean_one()-> pte_wrprotect()-> pte_mkclean()handle_pte_fault()-> do_wp_page() // pte_present() && !pte_write()-> wp_page_shared()-> do_page_mkwrite()-> xfs_filemap_page_mkwrite()-> __block_page_mkwrite()-> lock_page()-> __block_write_begin()-> block_commit_write()-> set_page_dirty(page);-> wait_for_stable_page(page);-> wp_page_reuse()-> set_page_dirty()-> unlock_page()在執(zhí)行writepage之前，在page_lock的保護(hù)之下，通過(guò)clean_page_dirty_for_io()清除page的dirty? flags以及mmap的pte的寫(xiě)權(quán)限，將相關(guān)page變?yōu)閣rite-protect，這樣，下次用戶寫(xiě)這個(gè)page的時(shí)候，就會(huì)觸發(fā)pagefault，內(nèi)核在這里將相關(guān)的page設(shè)置為dirty，在此期間，會(huì)給page創(chuàng)建buffers；這樣，就可以保證任何對(duì)mmap的寫(xiě)操作，都可以通過(guò)page fault提交到writeback子系統(tǒng)中。
write-protect page fault發(fā)生時(shí)，寫(xiě)操作還沒(méi)有發(fā)生，所以，dirty bit并不會(huì)被設(shè)置；而一旦寫(xiě)操作發(fā)生，那么上面的代碼所代表的過(guò)程，一定會(huì)發(fā)生，那么buffer也一定是具備的。
Dirty Page w/o Buffers
這里我們對(duì)比下ext4和xfs的writepages操作的調(diào)用棧：
ext4_writepages()-> mpage_prepare_extent_to_map()-> pagevec_lookup_tag()-> lock_page()-> wait_on_page_writeback()-> mpage_process_page_bufs()-> mpage_submit_page()-> clear_page_dirty_for_io(page)-> ext4_bio_write_page()-> set_page_writeback()-> io_submit_add_bh()-> clean_buffer_dirty()-> unlock_page()xfs_vm_writepages()-> generic_writepages()-> write_cache_pages()-> pagevec_lookup_tag()-> lock_page()-> xfs_vm_writepage()-> lock_buffer()-> xfs_add_to_ioend()-> xfs_start_page_writeback()-> clear_page_dirty_for_io()-> set_page_writeback()-> unlock_page() <<-----------------------HERE!!!! -> xfs_submit_ioend()-> xfs_start_page_writeback()-> mark_buffer_async_write()-> set_buffer_uptodate()-> clear_buffer_dirty()在ext4調(diào)用棧中，Page Dirty和Buffer Dirty的清理都是在page_lock下進(jìn)行的；而xfs中，buffer的清理是在page_lock之外，這時(shí)，我們?nèi)绻雙age fault過(guò)程中的page_mkwrite調(diào)用鏈，就會(huì)產(chǎn)生以下競(jìng)態(tài)，
writeback workqueue user page fault()
xfs_vm_writepages() xfs_filemap_page_mkwrite()
lock_page() __block_page_mkwrite()
clear_page_dirty_for_io()
unlock_page()lock_page()xfs_vm_set_page_dirty()set_buffer_dirty()TestSetPageDirty()
clear_buffer_dirty()end_page_writeback()于是這里，我們得到了一個(gè)page有dirty flags，但是buffer全是clean的；如果將此場(chǎng)景帶入到ext4，就不會(huì)有這種問(wèn)題，因?yàn)橛衟age_lock的保護(hù)，最終的結(jié)果，要么是page buffer全部dirty，要是全是clean。
到這一步，產(chǎn)生了兩個(gè)關(guān)鍵點(diǎn)：
- page dirty + buffer clean
- dirty bit，因?yàn)閣rite-protect page fault已經(jīng)發(fā)生過(guò)，所以，寫(xiě)操作已經(jīng)完成
其中page dirty + buffer clean將繼續(xù)推進(jìn)問(wèn)題的發(fā)生；
我們?cè)倩氐絪hrink_active_list()，它可能會(huì)調(diào)用try_to_release_page()，
try_to_free_buffers()-> drop_buffers()-> buffer_busy() // dirty or lock-> cancel_dirty_page()page dirty + buffer clean，在這里，因?yàn)閎uffer是clean的，所以，它可以被釋放，然后page的dirty也被清除了；
但是，此時(shí)pte中的dirty bit是存在的，于是在后續(xù)的shrink_page_list()中：
tshrink_page_list()-> try_to_unmap()-> try_to_unmap_file()-> try_to_unmap_one()-> set_page_dirty() //pte_dirty()page被設(shè)置dirty，然后回收中止；得到了一個(gè)page dirty + no buffers
所以，問(wèn)題的關(guān)鍵是，xfs_vm_writepage()中，對(duì)page dirty和buffer dirty的clean操作并沒(méi)有在page_lock的保護(hù)下。
問(wèn)題解決
在搜索社區(qū)代碼和Commit記錄之后，該問(wèn)題在以下commit解決：
commit e10de3723c53378e7cf441529f563c316fdc0dd3
Author: Dave Chinner <dchinner@redhat.com>
Date: Mon Feb 15 17:23:12 2016 +1100xfs: don't chain ioends during writepage submission@@ -565,6 +539,7 @@ xfs_add_to_ioend(bh->b_private = NULL;wpc->ioend->io_size += bh->b_size;wpc->last_block = bh->b_blocknr;
+ xfs_start_buffer_writeback(bh);}在該修改之后，buffer clean操作也放到了page lock之下。