search for: truncate_inode_pages

Hi, We have investigated the possible causes of the severe throughput drop of file reading in OCFS2 that found in the iozone benchmark. We find the major cause is the inadequate inode page cache cleaning when open or close the last reference to an inode. In routine ocfs_file_release(), if the caller is the last reference to this inode, truncate_inode_page is called to invalidate all page

=20 >-----Original Message----- >From: ocfs2-devel-bounces@oss.oracle.com=20 >[mailto:ocfs2-devel-bounces@oss.oracle.com] On Behalf Of Wim Coekaerts >Sent: 2004=C4=EA6=D4=C222=C8=D5 16:01 >To: Zhang, Sonic > >the problem is, how can we notify. I think we don't want to=20 >notify every >node on every change othewise we overload the interconnect and we don't

...rivate / /keeps unchanged since last write./ /5) in cont_prepare_write() it dead loops since bytes(ip_mmu_private) is not / /increased in last prepare write./ BUG is described in s://bug.oraclecorp.com/pls/bug/webbug_print.show?c_rptno=7183894. solution: /1) moves the clearing of page cache truncate_inode_pages() from ocfs2_truncate_file() to/ /ocfs2_set_inode_size() after resizing inode size and i_blocks./ /2) in ocfs2_get_block(), add checks on iblock >= inode->i_blocks. if true, return -EIO. 3) //a kernel bug fix is needed for 2.6.9 kernel. //see https://bugzilla.redhat.com/show_bug.cgi?id=45...

...and splitting. + */ +static void evict_inode_truncate_pages(struct inode *inode) +{ + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; + struct extent_map_tree *map_tree = &BTRFS_I(inode)->extent_tree; + struct rb_node *node; + + ASSERT(inode->i_state & I_FREEING); + truncate_inode_pages(&inode->i_data, 0); + + write_lock(&map_tree->lock); + while (!RB_EMPTY_ROOT(&map_tree->map)) { + struct extent_map *em; + + node = rb_first(&map_tree->map); + em = rb_entry(node, struct extent_map, rb_node); + remove_extent_mapping(map_tree, em); + free_extent_map(...

...l, but thanks to a tip from Kevin Fenzi I learned how to use sysrq to find out where the process was, viz: 18:03:10 kernel: mutt R current 1024 8893 7929 (NOTLB) 18:03:10 kernel: Call Trace: [<c0127061>] truncate_list_pages [kernel] 0x79 18:03:10 kernel: [<c01271ff>] truncate_inode_pages [kernel] 0x3b 18:03:10 kernel: [<c0124f2e>] vmtruncate [kernel] 0x96 18:03:10 kernel: [<c01491f0>] inode_setattr [kernel] 0x24 18:03:10 kernel: [<d401f963>] ext3_setattr [ext3] 0x1c3 18:03:10 kernel: [<d401d810>] ext3_get_block [ext3] 0x0 18:03:10 kernel: [<c012...

...20:29:37 torsa kernel: CPU: 1 Nov 10 20:29:37 torsa kernel: EIP: 0010:[__remove_inode_page+80/112] <snip> Nov 10 20:29:37 torsa kernel: Call Trace: [remove_inode_page+103/128] [truncate_list_pages+560/704] [e1000:__insmod_e1000_O/lib/modules/2.4.7-10smp/kernel/drivers/net+-1258 886/96] [truncate_inode_pages+94/192] [vmtruncate+180/576] Nov 10 20:29:37 torsa kernel: Call Trace: [<c012b967>] [<c012bf30>] [<f8873a7a>] [<c012c01e>] [<c01298b4>] Nov 10 20:29:37 torsa kernel: [inode_setattr+38/224] [e1000:__insmod_e1000_O/lib/modules/2.4.7-10smp/kernel/drivers/net+-1259 208/...

...gt;ip_flags & OCFS2_INODE_SKIP_ORPHAN_DIR)) + mlog(0, "skip delete inode %llu because " + "umount has begun.\n", + (unsigned long long)OCFS2_I(inode)->ip_blkno); + else + delete = 1; + } + + if (delete) ocfs2_delete_inode(inode); - } else { + else truncate_inode_pages(&inode->i_data, 0); - } + ocfs2_clear_inode(inode); } diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h index 70354a5..0544b0c 100644 --- a/fs/ocfs2/ocfs2.h +++ b/fs/ocfs2/ocfs2.h @@ -261,7 +261,7 @@ enum ocfs2_mount_options #define OCFS2_OSB_SOFT_RO 0x0001 #define OCFS2_OSB_HARD_RO...

...if (filemap_fdatawrite(mapping)) { mlog(ML_ERROR, "Could not sync inode %llu for downconvert!", (unsigned long long)OCFS2_I(inode)->ip_blkno); @@ -2725,7 +2734,6 @@ static int ocfs2_data_convert_worker(str sync_mapping_buffers(mapping); if (blocking == LKM_EXMODE) { truncate_inode_pages(mapping, 0); - unmap_mapping_range(mapping, 0, 0, 0); } else { /* We only need to wait on the I/O if we're not also * truncating pages because truncate_inode_pages waits diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 9fd590b..10953a5 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/f...

...t;===== > Trace; c015eb6e <journal_flushpage+9e/140> > Trace; c0156ae2 <ext3_flushpage+22/30> > Trace; c0125738 <do_flushpage+18/30> > Trace; c0125762 <truncate_complete_page+12/50> > Trace; c01258c6 <truncate_list_pages+126/190> > Trace; c0125970 <truncate_inode_pages+40/70> > Trace; c014485e <iput+ae/200> > Trace; c0142e4c <d_delete+4c/70> > Trace; c013c69c <vfs_unlink+13c/170> > Trace; c013c778 <sys_unlink+a8/120> > Trace; c0106e8a <system_call+32/38> Well, that's a straight forward trace, and looks perfectl...

...l, but thanks to a tip from Kevin Fenzi I learned how to use sysrq to find out where the process was, viz: 18:03:10 kernel: mutt R current 1024 8893 7929 (NOTLB) 18:03:10 kernel: Call Trace: [<c0127061>] truncate_list_pages [kernel] 0x79 18:03:10 kernel: [<c01271ff>] truncate_inode_pages [kernel] 0x3b 18:03:10 kernel: [<c0124f2e>] vmtruncate [kernel] 0x96 18:03:10 kernel: [<c01491f0>] inode_setattr [kernel] 0x24 18:03:10 kernel: [<d401f963>] ext3_setattr [ext3] 0x1c3 18:03:10 kernel: [<d401d810>] ext3_get_block [ext3] 0x0 18:03:10 kernel: [<c012...

We are also thinking about locking for each read/write, but would its = overhead be too high? We have another idea that is extending the function of flock, lockf, = fcntl to distributed. So any application that need strict data consistent can do a lock = operation on the whole or part of the file. For ordinary application, maybe the current logic is enough. How about it? >-----Original

...0c2b Call Trace: [<ffffffff810a8c67>] remove_from_page_cache+0x2b/0x38 [<ffffffff810b0b27>] truncate_complete_page+0x4c/0x61 [<ffffffff810b0c2b>] truncate_inode_pages_range+0xef/0x36e [<ffffffff810b0eb7>] truncate_inode_pages+0xd/0x12 [<ffffffff810f0937>] dispose_list+0x3b/0xf8 [<ffffffff810f0ce1>] invalidate_inodes+0xdc/0xfa...

...hed to disc, or they can be invalidated. Either way they must be non-dirty before their location on disc is allocated to another file. If not, they might be flushed out to disc much later and overwrite something that has since been allocate the same location on disc. The pages are dealt with by truncate_inode_pages which ultimately calls journal_flushpage on each page which inturn calls journal_unmap_buffer. For data=writeback the data block will not be attached to the journal at all and the buffer is "Zapped" (marked as clean). For data=ordered the data block can only be dirty if it is part of t...

ocfs2 currently uses the Journaled Block Device (JBD) for its journaling. This is a very stable and tested codebase. However, JBD is limited by architecture to 32bit block numbers. This means an ocfs2 filesystem is limited to 2^32 blocks. With a 4K blocksize, that's 16TB. People want larger volumes. Fortunately, there is now JBD2. JBD2 adds 64bit block number support and some other

Hi, This is the second batch of Ocfs2 patches intended for the merge window. The 1st batch were sent out previously: http://lkml.org/lkml/2008/12/19/280 The bulk of this set is comprised of Jan Kara's patches to add quota support to Ocfs2. Many of the quota patches are to generic code, which I carried to make merging of the Ocfs2 support easier. All of the non-ocfs2 patches should have

The following series of patches comprises the bulk of our outstanding changes for Ocfs2. Aside from the usual set of cleanups and fixes that were inappropriate for 2.6.25, there are a few highlights: The '/sys/o2cb' directory has been moved to '/sys/fs/o2cb'. The new location meshes better with modern sysfs layout. A symbolic link has been placed in the old location so as to