Hi David, Looking at inode.c in 2.6.23-mm1, in generic_forget_inode, I see code: if (!hlist_unhashed(&inode->i_hash)) { if (!(inode->i_state & (I_DIRTY|I_SYNC))) list_move(&inode->i_list, &inode_unused); so it looks to me like: If the inode is hashed and dirty, then move it (off the s_dirty list) to inode_unused. So it seems to me that generic_forget_inode also finds it needs to remove the inode from the dirty list when setting I_WILL_FREE. Maybe we are looking at ...

If we are in the process of dropping an inode and it is hashed, generic_forget_inode() will mark it I_WILL_FREE and drop the inode_lock before calling write_inode_now(). However, at this point, the inode is still on the sb->s_dirty_list so sync_sb_inodes() could see it and try to write it back. i.e.: generic_forget_inode sync_sb_inodes i_state |= ...

Thanks for the fix! Looks right to me. Yes, somehow I missed this one when you sent it privately. Applied and pushed out to git://linux-nfs.org/~bfields/linux.git nfs-server-stable and I'll submit it for 2.6.25. Minor nit: your editor is messing up the whitespace; keep the indents all tabs. Also, for future patches, could you see Documentation/SubmittingPatches? In particular, see "12) Sign your work". And it'd be better to inline the patch if at all possible. Thanks ...

The problem seems to be with the fact that the client and server are on the same machine. This test work fine with or without an underlaying fs that supports locking when the client and the server are on a different machines. Like you said the server is trying to send the grant message to the client but for some reason it fails when the client is on the same machine. Marc. -

Distributed storage. I'm pleased to announce the 9'th release of the distributed storage subsystem (DST). This is maintenance release and include bug fixing only. DST allows to form a storage on top of local and remote nodes and combine them into linear or mirroring setup, which in turn can be exported to remote nodes. Short changelog: * use node's size in sectors instead of bytes * fixed old/new ages for the first node. Error spotted by Matthew Hodgson <matthew@mxtelecom.com> * ...

Distributed storage documentation. Algorithms used in the system, userspace interfaces (sysfs dirs and files), design and implementation details are described here. Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru> diff --git a/Documentation/dst/algorithms.txt b/Documentation/dst/algorithms.txt new file mode 100644 index 0000000..1437a6a --- /dev/null +++ b/Documentation/dst/algorithms.txt @@ -0,0 +1,115 @@ +Each storage by itself is just a set of contiguous logical blocks, ...

Core distributed storage files. Include userspace interfaces, initialization, block layer bindings and other core functionality. Signed-off-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru> diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig index b4c8319..ca6592d 100644 --- a/drivers/block/Kconfig +++ b/drivers/block/Kconfig @@ -451,6 +451,8 @@ config ATA_OVER_ETH This driver provides Support for ATA over Ethernet block devices like the Coraid EtherDrive (R) Storage Blade. ...

Use page_cache_xxx in drivers/block/rd.c Signed-off-by: Christoph Lameter <clameter@sgi.com> --- drivers/block/rd.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) Index: mm/drivers/block/rd.c =================================================================== --- mm.orig/drivers/block/rd.c 2007-11-28 12:19:49.673905513 -0800 +++ mm/drivers/block/rd.c 2007-11-28 14:13:01.076977633 -0800 @@ -122,7 +122,7 @@ static void make_page_uptodate(struct pa } } while ((bh = ...

Works for me. Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Results in a gcc warning about the possible use of an unitialized variable. How about this? fs/buffer.c enhancements and fixes Signed-off-by: Christoph Lameter <clameter@sgi.com> --- fs/buffer.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) Index: mm/fs/buffer.c =================================================================== --- mm.orig/fs/buffer.c 2007-11-28 19:39:23.606383803 -0800 +++ mm/fs/buffer.c 2007-11-28 19:46:10.238382715 -0800 @@ -914,11 ...

We use the macros PAGE_CACHE_SIZE PAGE_CACHE_SHIFT PAGE_CACHE_MASK and PAGE_CACHE_ALIGN in various places in the kernel. Many times common operations like calculating the offset or the index are coded using shifts and adds. This patch provides inline functions to get the calculations accomplished without having to explicitly shift and add constants. All functions take an address_space pointer. The address space pointer will be used in the future to eventually support a larger ...

return page_cache_index(a, pos + page_cache_mask(a)); Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

In other words the following patch? Fixes to the use of page_cache_xx functions in xfs Signed-off-by: Christoph Lameter <clameter@sgi.com> --- fs/xfs/linux-2.6/xfs_aops.c | 17 ++++++----------- fs/xfs/linux-2.6/xfs_lrw.c | 2 +- 2 files changed, 7 insertions(+), 12 deletions(-) Index: mm/fs/xfs/linux-2.6/xfs_aops.c =================================================================== --- mm.orig/fs/xfs/linux-2.6/xfs_aops.c 2007-11-28 19:13:13.323382722 -0800 +++ ...

Use page_cache_xxx for fs/xfs Signed-off-by: Christoph Lameter <clameter@sgi.com> --- fs/xfs/linux-2.6/xfs_aops.c | 55 +++++++++++++++++++++++--------------------- fs/xfs/linux-2.6/xfs_lrw.c | 4 +-- 2 files changed, 31 insertions(+), 28 deletions(-) Index: mm/fs/xfs/linux-2.6/xfs_aops.c =================================================================== --- mm.orig/fs/xfs/linux-2.6/xfs_aops.c 2007-11-28 12:25:38.768212813 -0800 +++ mm/fs/xfs/linux-2.6/xfs_aops.c 2007-11-28 ...

That's not right. Should be Hmmmm. p_offset = min(val & 4095, 4096)? I think that should just be: Again, that can be: p_offset = page_cache_offset(map, end_offset); offset = page_cache_index(map, Kill the "within page" comment. Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Yup, looks good now. Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

unsigned long offset = page_cache_offset(p_s_inode->i_mapping, Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Reiserfs: Wrong type of inline function Signed-off-by: Christoph Lameter <clameter@sgi.com> --- fs/reiserfs/inode.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Index: mm/fs/reiserfs/inode.c =================================================================== --- mm.orig/fs/reiserfs/inode.c 2007-11-28 19:59:41.083133259 -0800 +++ mm/fs/reiserfs/inode.c 2007-11-28 20:00:23.317882809 -0800 @@ -2006,7 +2006,7 @@ static int grab_tail_page(struct inode * unsigned long pos ...

These three should use the pagesize variable. Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

ok. Cheers, dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Looks good. Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Ok. So this? ext2: Simplify some functions Signed-off-by: Christoph Lameter <clameter@sgi.com> --- fs/ext2/dir.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) Index: mm/fs/ext2/dir.c =================================================================== --- mm.orig/fs/ext2/dir.c 2007-11-28 20:13:07.387132777 -0800 +++ mm/fs/ext2/dir.c 2007-11-28 20:14:35.739632586 -0800 @@ -63,8 +63,7 @@ static inline void ext2_put_page(struct static inline unsigned long ...

I don't think that gives the same return value. The return value is supposed to be clamped at a maximum of page_cache_size(mapping). Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Ok. Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

in->f_ra.prev_pos = page_cache_pos(mapping, index, 0); Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Use page_cache_xxx in fs/sync. Signed-off-by: Christoph Lameter <clameter@sgi.com> --- fs/sync.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) Index: mm/fs/sync.c =================================================================== --- mm.orig/fs/sync.c 2007-11-16 21:16:36.000000000 -0800 +++ mm/fs/sync.c 2007-11-28 14:10:56.269227507 -0800 @@ -260,8 +260,8 @@ int do_sync_mapping_range(struct address ret = 0; if (flags & SYNC_FILE_RANGE_WAIT_BEFORE) { ret = ...

Use page_cache_xxx in mm/truncate.c Signed-off-by: Christoph Lameter <clameter@sgi.com> --- mm/truncate.c | 35 ++++++++++++++++++----------------- 1 file changed, 18 insertions(+), 17 deletions(-) Index: mm/mm/truncate.c =================================================================== --- mm.orig/mm/truncate.c 2007-11-28 12:27:32.480099915 -0800 +++ mm/mm/truncate.c 2007-11-28 14:10:39.013977394 -0800 @@ -46,9 +46,10 @@ void do_invalidatepage(struct page *page ...

Use page_cache_xxx in fs/libfs.c Signed-off-by: Christoph Lameter <clameter@sgi.com> --- fs/libfs.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) Index: mm/fs/libfs.c =================================================================== --- mm.orig/fs/libfs.c 2007-11-28 12:24:57.449215408 -0800 +++ mm/fs/libfs.c 2007-11-28 14:10:51.773477763 -0800 @@ -17,7 +17,8 @@ int simple_getattr(struct vfsmount *mnt, { struct inode *inode = dentry->d_inode; ...

page_cache_offset(mapping, isize - 1) + 1; or: page_cache_next(mapping, isize); Cheers, Fengguang -

Use page_cache_xxx in mm/rmap.c Signed-off-by: Christoph Lameter <clameter@sgi.com> --- mm/rmap.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) Index: mm/mm/rmap.c =================================================================== --- mm.orig/mm/rmap.c 2007-11-28 12:27:32.312059099 -0800 +++ mm/mm/rmap.c 2007-11-28 14:10:42.758227810 -0800 @@ -190,9 +190,14 @@ static void page_unlock_anon_vma(struct static inline unsigned long vma_address(struct page *page, ...

Yes that is much simpler rmap: simplify page_referenced_file use of page cache inlines Signed-off-by: Christoph Lameter <clameter@sgi.com> --- mm/rmap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) Index: mm/mm/rmap.c =================================================================== --- mm.orig/mm/rmap.c 2007-11-28 19:28:45.689883608 -0800 +++ mm/mm/rmap.c 2007-11-28 19:29:35.090382690 -0800 @@ -350,7 +350,7 @@ static int page_referenced_file(struct p { unsigned ...

Based on the first hunk, shouldn't this be: pgoff_t pgoff = page->index << mapping_order(mapping); Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group -

Convert the uses of PAGE_CACHE_xxx to use page_cache_xxx instead. Signed-off-by: Christoph Lameter <clameter@sgi.com> --- mm/filemap.c | 91 +++++++++++++++++++++++++++++------------------------------ 1 file changed, 46 insertions(+), 45 deletions(-) Index: mm/mm/filemap.c =================================================================== --- mm.orig/mm/filemap.c 2007-11-28 12:27:32.155962689 -0800 +++ mm/mm/filemap.c 2007-11-28 14:10:29.408977142 -0800 @@ -314,8 +314,8 @@ ...

This patchset cleans up page cache handling by replacing open coded shifts and adds with inline function calls. The ultimate goal is to replace all uses of PAGE_CACHE_xxx in the kernel through the use of these functions. All the functions take a mapping parameter. The mapping parameter is required if we want to support large block sizes in filesystems and block devices. Patchset against 2.6.24-rc3-mm2. -- -

thanks ;) I'll merge version 2.. -

I'd have to reread the original proposal, but I remember FIEMAP as being a generalized way of getting information about a files extents. I think the original proposal only dealt with mapping file offsets to physical extents, but IIRC the interface was flexible enough to implement a "where are the holes" request. Regardless, SEEK_HOLE/SEEK_DATA being a better suited interface for the needs of logfs doesn't make it the best interface for that need. -- Nicholas Miell ...

The FIEMAP interface in some ways well suited to your needs, because it uses byte offsets instead of FIBMAP, which uses block offsets. I now see that one hole in the specification is that your physical extent is not the same length as the logical extent that the data represents. I'm not sure that is in itself a reason not to use FIEMAP. There is already a provision for logical extents that do not map directly to the physical layout (i.e. FIEMAP_EXTENT_NO_DIRECT flag on the extent). In the ...