Merge branch 'for-linus' of git://oss.sgi.com:8090/xfs/xfs-2.6
* 'for-linus' of git://oss.sgi.com:8090/xfs/xfs-2.6: (37 commits)
[XFS] Fix lockdep annotations for xfs_lock_inodes
[LIB]: export radix_tree_preload()
[XFS] Fix XFS_IOC_FSBULKSTAT{,_SINGLE} & XFS_IOC_FSINUMBERS in compat mode
[XFS] Compat ioctl handler for handle operations
[XFS] Compat ioctl handler for XFS_IOC_FSGEOMETRY_V1.
[XFS] Clean up function name handling in tracing code
[XFS] Quota inode has no parent.
[XFS] Concurrent Multi-File Data Streams
[XFS] Use uninitialized_var macro to stop warning about rtx
[XFS] XFS should not be looking at filp reference counts
[XFS] Use is_power_of_2 instead of open coding checks
[XFS] Reduce shouting by removing unnecessary macros from dir2 code.
[XFS] Simplify XFS min/max macros.
[XFS] Kill off xfs_count_bits
[XFS] Cancel transactions on xfs_itruncate_start error.
[XFS] Use do_div() on 64 bit types.
[XFS] Fix remount,readonly path to flush everything correctly.
[XFS] Cleanup inode extent size hint extraction
[XFS] Prevent ENOSPC from aborting transactions that need to succeed
[XFS] Prevent deadlock when flushing inodes on unmount
...
Al Viro [Sun, 15 Jul 2007 20:37:16 +0000 (21:37 +0100)]
make i2c-acorn tristate
It depends on tristate I2C and it's trivial to make modular. The
current Kconfig allows I2C=m, I2C_ACORN=y, which doesn't work at
all; alternatives are dependency on I2C=y and making I2C_ACORN
itself a tristate. The latter is the right thing to do...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Al Viro [Sun, 15 Jul 2007 20:01:32 +0000 (21:01 +0100)]
icside: devm_iounmap() needs linux/io.h
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
David Chinner [Sat, 14 Jul 2007 06:05:04 +0000 (16:05 +1000)]
[LIB]: export radix_tree_preload()
XFS filestreams functionality uses radix trees and the preload
functions. XFS can be built as a module and hence we need
radix_tree_preload() exported. radix_tree_preload_end() is a
static inline, so it doesn't need exporting.
Signed-Off-By: Dave Chinner <dgc@sgi.com> Signed-Off-By: Tim Shimmin <tes@sgi.com>
Michal Marek [Wed, 11 Jul 2007 01:10:19 +0000 (11:10 +1000)]
[XFS] Fix XFS_IOC_FSBULKSTAT{,_SINGLE} & XFS_IOC_FSINUMBERS in compat mode
* 32bit struct xfs_fsop_bulkreq has different size and layout of
members, no matter the alignment. Move the code out of the #else
branch (why was it there in the first place?). Define _32 variants of
the ioctl constants.
* 32bit struct xfs_bstat is different because of time_t and on
i386 because of different padding. Make xfs_bulkstat_one() accept a
custom "output formatter" in the private_data argument which takes care
of the xfs_bulkstat_one_compat() that takes care of the different
layout in the compat case.
* i386 struct xfs_inogrp has different padding.
Add a similar "output formatter" mecanism to xfs_inumbers().
Michal Marek [Wed, 11 Jul 2007 01:10:09 +0000 (11:10 +1000)]
[XFS] Compat ioctl handler for handle operations
32bit struct xfs_fsop_handlereq has different size and offsets (due to
pointers). TODO: case XFS_IOC_{FSSETDM,ATTRLIST,ATTRMULTI}_BY_HANDLE still
not handled.
David Chinner [Wed, 11 Jul 2007 01:09:33 +0000 (11:09 +1000)]
[XFS] Quota inode has no parent.
Avoid using a special "zero inode" as the parent of the quota inode as
this can confuse the filestreams code into thinking the quota inode has a
parent. We do not want the quota inode to follow filestreams allocation
rules, so pass a NULL as the parent inode and detect this condition when
doing stream associations.
David Chinner [Wed, 11 Jul 2007 01:09:12 +0000 (11:09 +1000)]
[XFS] Concurrent Multi-File Data Streams
In media spaces, video is often stored in a frame-per-file format. When
dealing with uncompressed realtime HD video streams in this format, it is
crucial that files do not get fragmented and that multiple files a placed
contiguously on disk.
When multiple streams are being ingested and played out at the same time,
it is critical that the filesystem does not cross the streams and
interleave them together as this creates seek and readahead cache miss
latency and prevents both ingest and playout from meeting frame rate
targets.
This patch set creates a "stream of files" concept into the allocator to
place all the data from a single stream contiguously on disk so that RAID
array readahead can be used effectively. Each additional stream gets
placed in different allocation groups within the filesystem, thereby
ensuring that we don't cross any streams. When an AG fills up, we select a
new AG for the stream that is not in use.
The core of the functionality is the stream tracking - each inode that we
create in a directory needs to be associated with the directories' stream.
Hence every time we create a file, we look up the directories' stream
object and associate the new file with that object.
Once we have a stream object for a file, we use the AG that the stream
object point to for allocations. If we can't allocate in that AG (e.g. it
is full) we move the entire stream to another AG. Other inodes in the same
stream are moved to the new AG on their next allocation (i.e. lazy
update).
Stream objects are kept in a cache and hold a reference on the inode.
Hence the inode cannot be reclaimed while there is an outstanding stream
reference. This means that on unlink we need to remove the stream
association and we also need to flush all the associations on certain
events that want to reclaim all unreferenced inodes (e.g. filesystem
freeze).
Signed-off-by: David Chinner <dgc@sgi.com> Signed-off-by: Barry Naujok <bnaujok@sgi.com> Signed-off-by: Donald Douwsma <donaldd@sgi.com> Signed-off-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: Tim Shimmin <tes@sgi.com> Signed-off-by: Vlad Apostolov <vapo@sgi.com>
[XFS] XFS should not be looking at filp reference counts
A check for file_count is always a bad idea. Linux has the ->release
method to deal with cleanups on last close and ->flush is only for the
very rare case where we want to perform an operation on every drop of a
reference to a file struct.
This patch gets rid of vop_close and surrounding code in favour of simply
doing the page flushing from ->release.
Eric Sandeen [Thu, 28 Jun 2007 06:43:30 +0000 (16:43 +1000)]
[XFS] Kill off xfs_count_bits
xfs_count_bits is only called once, and is then compared to 0. IOW, what
it really wants to know is, is the bitmap empty. This can be done more
simply, certainly.
David Chinner [Mon, 18 Jun 2007 06:50:48 +0000 (16:50 +1000)]
[XFS] Fix remount,readonly path to flush everything correctly.
The remount readonly path can fail to writeback properly because we still
have active transactions after calling xfs_quiesce_fs(). Further
investigation shows that this path is broken in the same ways that the xfs
freeze path was broken so fix it the same way.
David Chinner [Mon, 18 Jun 2007 06:50:27 +0000 (16:50 +1000)]
[XFS] Prevent ENOSPC from aborting transactions that need to succeed
During delayed allocation extent conversion or unwritten extent
conversion, we need to reserve some blocks for transactions reservations.
We need to reserve these blocks in case a btree split occurs and we need
to allocate some blocks.
Unfortunately, we've only ever reserved the number of data blocks we are
allocating, so in both the unwritten and delalloc case we can get ENOSPC
to the transaction reservation. This is bad because in both cases we
cannot report the failure to the writing application.
The fix is two-fold:
1 - leverage the reserved block infrastructure XFS already
has to reserve a small pool of blocks by default to allow
specially marked transactions to dip into when we are at
ENOSPC.
Default setting is min(5%, 1024 blocks).
2 - convert critical transaction reservations to be allowed
to dip into this pool. Spots changed are delalloc
conversion, unwritten extent conversion and growing a
filesystem at ENOSPC.
This also allows growing the filesytsem to succeed at ENOSPC.
David Chinner [Mon, 18 Jun 2007 06:50:17 +0000 (16:50 +1000)]
[XFS] Prevent deadlock when flushing inodes on unmount
When we are unmounting the filesystem, we flush all the inodes to disk.
Unfortunately, if we have an inode cluster that has just been freed and
marked stale sitting in an incore log buffer (i.e. hasn't been flushed to
disk), it will be holding all the flush locks on the inodes in that
cluster.
xfs_iflush_all() which is called during unmount walks all the inodes
trying to reclaim them, and it doing so calls xfs_finish_reclaim() on each
inode. If the inode is dirty, if grabs the flush lock and flushes it.
Unfortunately, find dirty inodes that already have their flush lock held
and so we sleep.
At this point in the unmount process, we are running single-threaded.
There is nothing more that can push on the log to force the transaction
holding the inode flush locks to disk and hence we deadlock.
The fix is to issue a log force before flushing the inodes on unmount so
that all the flush locks will be released before we start flushing the
inodes.
David Chinner [Mon, 18 Jun 2007 06:49:58 +0000 (16:49 +1000)]
[XFS] Map unwritten extents correctly for I/o completion processing
If we have multiple unwritten extents within a single page, we fail to
tell the I/o completion construction handlers we need a new handle for the
second and subsequent blocks in the page. While we still issue the I/O
correctly, we do not have the correct ranges recorded in the ioend
structures and hence when we go to convert the unwritten extents we screw
it up.
Make sure we start a new ioend every time the mapping changes so that we
convert the correct ranges on I/O completion.
David Chinner [Mon, 18 Jun 2007 06:49:44 +0000 (16:49 +1000)]
[XFS] Apply transaction delta counts atomically to incore counters
With the per-cpu superblock counters, batch updates are no longer atomic
across the entire batch of changes. This is not an issue if each
individual change in the batch is applied atomically. Unfortunately, free
block count changes are not applied atomically, and they are applied in a
manner guaranteed to cause problems.
Essentially, the free block count reservation that the transaction took
initially is returned to the in core counters before a second delta takes
away what is used. because these two operations are not atomic, we can
race with another thread that can use the returned transaction reservation
before the transaction takes the space away again and we can then get
ENOSPC being reported in a spot where we don't have an ENOSPC condition,
nor should we ever see one there.
Fix it up by rolling the two deltas into the one so it can be applied
safely (i.e. atomically) to the incore counters.
David Chinner [Tue, 5 Jun 2007 06:24:36 +0000 (16:24 +1000)]
[XFS] Block on unwritten extent conversion during synchronous direct I/O.
Currently we do not wait on extent conversion to occur, and hence we can
return to userspace from a synchronous direct I/O write without having
completed all the actions in the write. Hence a read after the write may
see zeroes (unwritten extent) rather than the data that was written.
Block the I/O completion by triggering a synchronous workqueue flush to
ensure that the conversion has occurred before we return to userspace.
David Chinner [Tue, 5 Jun 2007 06:24:15 +0000 (16:24 +1000)]
[XFS] xfs_bmapi fails to update the previous extent pointer
When processing multiple extent maps, xfs_bmapi needs to keep track of the
extent behind the one it is currently working on to be able to trim extent
ranges correctly. Failing to update the previous pointer can result in
corrupted extent lists in memory and this will result in panics or assert
failures.
Update the previous pointer correctly when we move to the next extent to
process.
David Chinner [Thu, 24 May 2007 05:26:31 +0000 (15:26 +1000)]
[XFS] Lazy Superblock Counters
When we have a couple of hundred transactions on the fly at once, they all
typically modify the on disk superblock in some way.
create/unclink/mkdir/rmdir modify inode counts, allocation/freeing modify
free block counts.
When these counts are modified in a transaction, they must eventually lock
the superblock buffer and apply the mods. The buffer then remains locked
until the transaction is committed into the incore log buffer. The result
of this is that with enough transactions on the fly the incore superblock
buffer becomes a bottleneck.
The result of contention on the incore superblock buffer is that
transaction rates fall - the more pressure that is put on the superblock
buffer, the slower things go.
The key to removing the contention is to not require the superblock fields
in question to be locked. We do that by not marking the superblock dirty
in the transaction. IOWs, we modify the incore superblock but do not
modify the cached superblock buffer. In short, we do not log superblock
modifications to critical fields in the superblock on every transaction.
In fact we only do it just before we write the superblock to disk every
sync period or just before unmount.
This creates an interesting problem - if we don't log or write out the
fields in every transaction, then how do the values get recovered after a
crash? the answer is simple - we keep enough duplicate, logged information
in other structures that we can reconstruct the correct count after log
recovery has been performed.
It is the AGF and AGI structures that contain the duplicate information;
after recovery, we walk every AGI and AGF and sum their individual
counters to get the correct value, and we do a transaction into the log to
correct them. An optimisation of this is that if we have a clean unmount
record, we know the value in the superblock is correct, so we can avoid
the summation walk under normal conditions and so mount/recovery times do
not change under normal operation.
One wrinkle that was discovered during development was that the blocks
used in the freespace btrees are never accounted for in the AGF counters.
This was once a valid optimisation to make; when the filesystem is full,
the free space btrees are empty and consume no space. Hence when it
matters, the "accounting" is correct. But that means the when we do the
AGF summations, we would not have a correct count and xfs_check would
complain. Hence a new counter was added to track the number of blocks used
by the free space btrees. This is an *on-disk format change*.
As a result of this, lazy superblock counters are a mkfs option and at the
moment on linux there is no way to convert an old filesystem. This is
possible - xfs_db can be used to twiddle the right bits and then
xfs_repair will do the format conversion for you. Similarly, you can
convert backwards as well. At some point we'll add functionality to
xfs_admin to do the bit twiddling easily....
David Chinner [Thu, 24 May 2007 05:22:19 +0000 (15:22 +1000)]
[XFS] Make hole punching at EOF atomic.
If hole punching at EOF is done as two steps (i.e. truncate then extend)
the file is in a transient state between the two steps where an
application can see the incorrect file size. Punching a hole to EOF needs
to be treated in teh same way as all other hole punching cases so that the
file size is never seen to change.
David Chinner [Thu, 24 May 2007 05:21:57 +0000 (15:21 +1000)]
[XFS] Fix vmalloc leak on mount/unmount.
When setting the length of the iclogbuf to write out we should just be
changing the desired byte count rather completely reassociating the buffer
memory with the buffer. Reassociating the buffer memory changes the
apparent length of the buffer and hence when we free the buffer, we don't
free all the vmap()d space we originally allocated.
David Chinner [Mon, 14 May 2007 08:24:09 +0000 (18:24 +1000)]
[XFS] Sleeping with the ilock waiting for I/O completion is Bad.
Recent fixes to the filesystem freezing code introduced a vn_iowait call
in the middle of the sync code. Unfortunately, at the point where this
call was added we are holding the ilock. The ilock is needed by I/O
completion for unwritten extent conversion and now updating the file size.
Hence I/o cannot complete if we hold the ilock while waiting for I/O
completion.
Nathan Scott [Mon, 14 May 2007 08:24:02 +0000 (18:24 +1000)]
[XFS] Don't grow filesystems past the size they can index.
When growing a filesystem we don't check to see if the new size overflows
the page cache index range, so we can do silly things like grow a
filesystem page 16TB on a 32bit. Check new filesystem sizes against the
limits the kernel can support.
Many block drivers (aoe, iscsi) really want refcountable pages in bios,
which is what almost everyone send down. XFS unfortunately has a few
places where it sends down buffers that may come from kmalloc, which
breaks them.
because it changed how CONFIG_NETLABEL worked, and broke older SElinux
policies.
As a result, quoth James Morris:
"Can you please revert this patch?
We thought it only affected people running MLS, but it will affect others.
Sorry for the hassle."
Cc: James Morris <jmorris@namei.org> Cc: Stephen Smalley <sds@tycho.nsa.gov> Cc: Michal Piotrowski <michal.k.k.piotrowski@gmail.com> Cc: Paul Moore <paul.moore@hp.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* git://git.linux-nfs.org/pub/linux/nfs-2.6: (122 commits)
sunrpc: drop BKL around wrap and unwrap
NFSv4: Make sure unlock is really an unlock when cancelling a lock
NLM: fix source address of callback to client
SUNRPC client: add interface for binding to a local address
SUNRPC server: record the destination address of a request
SUNRPC: cleanup transport creation argument passing
NFSv4: Make the NFS state model work with the nosharedcache mount option
NFS: Error when mounting the same filesystem with different options
NFS: Add the mount option "nosharecache"
NFS: Add support for mounting NFSv4 file systems with string options
NFS: Add final pieces to support in-kernel mount option parsing
NFS: Introduce generic mount client API
NFS: Add enums and match tables for mount option parsing
NFS: Improve debugging output in NFS in-kernel mount client
NFS: Clean up in-kernel NFS mount
NFS: Remake nfsroot_mount as a permanent part of NFS client
SUNRPC: Add a convenient default for the hostname when calling rpc_create()
SUNRPC: Rename rpcb_getport to be consistent with new rpcb_getport_sync name
SUNRPC: Rename rpcb_getport_external routine
SUNRPC: Allow rpcbind requests to be interrupted by a signal.
...
When nfsd was transitioned to use splice instead of sendfile() for data
transfers, a line setting the page index was lost. Restore it, so that
nfsd is functional when that path is used.
* master.kernel.org:/pub/scm/linux/kernel/git/davej/cpufreq:
[CPUFREQ] Fix typos in powernow-k8 printk's.
[CPUFREQ] Restore previously used governor on a hot-replugged CPU
[CPUFREQ] bugfix cpufreq in combination with performance governor
[CPUFREQ] powernow-k8 compile fix.
[CPUFREQ] the overdue removal of X86_SPEEDSTEP_CENTRINO_ACPI
[CPUFREQ] Longhaul - Option to disable ACPI C3 support
Fixed up arch/i386/kernel/cpu/cpufreq/powernow-k8.c due to revert that
got fixed differently in the cpufreq branch.
Merge branch 'ioat-md-accel-for-linus' of git://lost.foo-projects.org/~dwillia2/git/iop
* 'ioat-md-accel-for-linus' of git://lost.foo-projects.org/~dwillia2/git/iop: (28 commits)
ioatdma: add the unisys "i/oat" pci vendor/device id
ARM: Add drivers/dma to arch/arm/Kconfig
iop3xx: surface the iop3xx DMA and AAU units to the iop-adma driver
iop13xx: surface the iop13xx adma units to the iop-adma driver
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines
md: remove raid5 compute_block and compute_parity5
md: handle_stripe5 - request io processing in raid5_run_ops
md: handle_stripe5 - add request/completion logic for async expand ops
md: handle_stripe5 - add request/completion logic for async read ops
md: handle_stripe5 - add request/completion logic for async check ops
md: handle_stripe5 - add request/completion logic for async compute ops
md: handle_stripe5 - add request/completion logic for async write ops
md: common infrastructure for running operations with raid5_run_ops
md: raid5_run_ops - run stripe operations outside sh->lock
raid5: replace custom debug PRINTKs with standard pr_debug
raid5: refactor handle_stripe5 and handle_stripe6 (v3)
async_tx: add the async_tx api
xor: make 'xor_blocks' a library routine for use with async_tx
dmaengine: make clients responsible for managing channels
dmaengine: refactor dmaengine around dma_async_tx_descriptor
...
Merge branch 'splice-2.6.23' of git://git.kernel.dk/data/git/linux-2.6-block
* 'splice-2.6.23' of git://git.kernel.dk/data/git/linux-2.6-block:
splice: fix offset mangling with direct splicing (sendfile)
security: revalidate rw permissions for sys_splice and sys_vmsplice
relay: fixup kerneldoc comment
relay: fix bogus cast in subbuf_splice_actor()
Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
[MIPS] Workaround for a sparse warning in include/asm-mips/mach-tx4927/ioremap.h
[MIPS] Make show_code static and add __user tag
[MIPS] Workaround for a sparse warning in include/asm-mips/compat.h
[MIPS] Add some __user tags
[MIPS] math-emu minor cleanup
[MIPS] Kill CONFIG_TX4927BUG_WORKAROUND
[MIPS] Alchemy: Remove code wrapped by dead symbol CONFIG_FB_XPERT98
[MIPS] Alchemy: Remove code wrapped by dead symbol CONFIG_AU1000_SRC_CLK
[MIPS] Alchemy: Remove code wrapped by dead symbol CONFIG_AU1000_USE32K
[MIPS] Alchemy: Remove code wrapped by dead symbol CONFIG_AU1XXX_PSC_SPI
[CHAR] Delete leftovers of old Alchemy UART driver
- the fields were not aligned properly with the description
- get rid of the lazy-TLB output: it's been quite some time since
we last had a bug there, and when we had a bug it wasnt helped a
bit by this debug output.
Mike Galbraith [Wed, 11 Jul 2007 19:21:47 +0000 (21:21 +0200)]
[PATCH] sched: fix prio_to_wmult[] for nice 1
There's a typo in the values in prio_to_wmult[] for nice level 1. While
it did not cause bad CPU distribution, but caused more rescheduling
between nice-0 and nice-1 tasks than necessary.
Dan Williams [Fri, 13 Jul 2007 15:06:19 +0000 (08:06 -0700)]
ioatdma: add the unisys "i/oat" pci vendor/device id
Cc: John Magolan <john.magolan@unisys.com> Signed-off-by: Shannon Nelson <shannon.nelson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Tue, 2 Jan 2007 20:52:31 +0000 (13:52 -0700)]
iop3xx: surface the iop3xx DMA and AAU units to the iop-adma driver
Adds the platform device definitions and the architecture specific support
routines (i.e. register initialization and descriptor formats) for the
iop-adma driver.
Changelog:
* add support for > 1k zero sum buffer sizes
* added dma/aau platform devices to iq80321 and iq80332 setup
* fixed the calculation in iop_desc_is_aligned
* support xor buffer sizes larger than 16MB
* fix places where software descriptors are assumed to be contiguous, only
hardware descriptors are contiguous for up to a PAGE_SIZE buffer size
* convert to async_tx
* add interrupt support
* add platform devices for 80219 boards
* do not call platform register macros in driver code
* remove switch() statements for compatible register offsets/layouts
* change over to bitmap based capabilities
* remove unnecessary ARM assembly statement
* checkpatch.pl fixes
* gpl v2 only correction
* phys move to dma_async_tx_descriptor
Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Tue, 2 Jan 2007 20:52:31 +0000 (13:52 -0700)]
iop13xx: surface the iop13xx adma units to the iop-adma driver
Adds the platform device definitions and the architecture specific
support routines (i.e. register initialization and descriptor formats) for the
iop-adma driver.
Changelog:
* added 'descriptor pool size' to the platform data
* add base support for buffer sizes larger than 16MB (hw max)
* build error fix from Kirill A. Shutemov
* rebase for async_tx changes
* add interrupt support
* do not call platform register macros in driver code
* remove unnecessary ARM assembly statement
* checkpatch.pl fixes
* gpl v2 only correction
Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Tue, 2 Jan 2007 20:52:26 +0000 (13:52 -0700)]
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines
The Intel(R) IOP series of i/o processors integrate an Xscale core with
raid acceleration engines. The capabilities per platform are:
iop219:
(2) copy engines
iop321:
(2) copy engines
(1) xor and block fill engine
iop33x:
(2) copy and crc32c engines
(1) xor, xor zero sum, pq, pq zero sum, and block fill engine
iop34x (iop13xx):
(2) copy, crc32c, xor, xor zero sum, and block fill engines
(1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine
The driver supports the features of the async_tx api:
* asynchronous notification of operation completion
* implicit (interupt triggered) handling of inter-channel transaction
dependencies
The driver adapts to the platform it is running by two methods.
1/ #include <asm/arch/adma.h> which defines the hardware specific
iop_chan_* and iop_desc_* routines as a series of static inline
functions
2/ The private platform data attached to the platform_device defines the
capabilities of the channels
20070626: Callbacks are run in a tasklet. Given the recent discussion on
LKML about killing tasklets in favor of workqueues I did a quick conversion
of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so
the tasklet implementation remains until a generic softirq interface is
available.
Changelog:
* fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few
slots to be requested eventually leading to data corruption
* enabled the slot allocation routine to attempt to free slots before
returning -ENOMEM
* switched the cleanup routine to solely use the software chain and the
status register to determine if a descriptor is complete. This is
necessary to support other IOP engines that do not have status writeback
capability
* make the driver iop generic
* modified the allocation routines to understand allocating a group of
slots for a single operation
* added a null xor initialization operation for the xor only channel on
iop3xx
* support xor operations on buffers larger than the hardware maximum
* split the do_* routines into separate prep, src/dest set, submit stages
* added async_tx support (dependent operations initiation at cleanup time)
* simplified group handling
* added interrupt support (callbacks via tasklets)
* brought the pending depth inline with ioat (i.e. 4 descriptors)
* drop dma mapping methods, suggested by Chris Leech
* don't use inline in C files, Adrian Bunk
* remove static tasklet declarations
* make iop_adma_alloc_slots easier to read and remove chances for a
corrupted descriptor chain
* fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt
* convert capabilities over to dma_cap_mask_t
* fixup sparse warnings
* add descriptor flush before iop_chan_enable
* checkpatch.pl fixes
* gpl v2 only correction
* move set_src, set_dest, submit to async_tx methods
* move group_list and phys to async_tx
Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Tue, 2 Jan 2007 20:52:31 +0000 (13:52 -0700)]
md: handle_stripe5 - request io processing in raid5_run_ops
I/O submission requests were already handled outside of the stripe lock in
handle_stripe. Now that handle_stripe is only tasked with finding work,
this logic belongs in raid5_run_ops.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 20:52:31 +0000 (13:52 -0700)]
md: handle_stripe5 - add request/completion logic for async expand ops
When a stripe is being expanded bulk copying takes place to move the data
from the old stripe to the new. Since raid5_run_ops only operates on one
stripe at a time these bulk copies are handled in-line under the stripe
lock. In the dma offload case we poll for the completion of the operation.
After the data has been copied into the new stripe the parity needs to be
recalculated across the new disks. We reuse the existing postxor
functionality to carry out this calculation. By setting STRIPE_OP_POSTXOR
without setting STRIPE_OP_BIODRAIN the completion path in handle stripe
can differentiate expand operations from normal write operations.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 20:52:31 +0000 (13:52 -0700)]
md: handle_stripe5 - add request/completion logic for async read ops
When a read bio is attached to the stripe and the corresponding block is
marked R5_UPTODATE, then a read (biofill) operation is scheduled to copy
the data from the stripe cache to the bio buffer. handle_stripe flags the
blocks to be operated on with the R5_Wantfill flag. If new read requests
arrive while raid5_run_ops is running they will not be handled until
handle_stripe is scheduled to run again.
Changelog:
* cleanup to_read and to_fill accounting
* do not fail reads that have reached the cache
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 20:52:31 +0000 (13:52 -0700)]
md: handle_stripe5 - add request/completion logic for async check ops
Check operations are scheduled when the array is being resynced or an
explicit 'check/repair' command was sent to the array. Previously check
operations would destroy the parity block in the cache such that even if
parity turned out to be correct the parity block would be marked
!R5_UPTODATE at the completion of the check. When the operation can be
carried out by a dma engine the assumption is that it can check parity as a
read-only operation. If raid5_run_ops notices that the check was handled
by hardware it will preserve the R5_UPTODATE status of the parity disk.
When a check operation determines that the parity needs to be repaired we
reuse the existing compute block infrastructure to carry out the operation.
Repair operations imply an immediate write back of the data, so to
differentiate a repair from a normal compute operation the
STRIPE_OP_MOD_REPAIR_PD flag is added.
Changelog:
* remove test_and_set/test_and_clear BUG_ONs, Neil Brown
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 20:52:30 +0000 (13:52 -0700)]
md: handle_stripe5 - add request/completion logic for async compute ops
handle_stripe will compute a block when a backing disk has failed, or when
it determines it can save a disk read by computing the block from all the
other up-to-date blocks.
Previously a block would be computed under the lock and subsequent logic in
handle_stripe could use the newly up-to-date block. With the raid5_run_ops
implementation the compute operation is carried out a later time outside
the lock. To preserve the old functionality we take advantage of the
dependency chain feature of async_tx to flag the block as R5_Wantcompute
and then let other parts of handle_stripe operate on the block as if it
were up-to-date. raid5_run_ops guarantees that the block will be ready
before it is used in another operation.
However, this only works in cases where the compute and the dependent
operation are scheduled at the same time. If a previous call to
handle_stripe sets the R5_Wantcompute flag there is no facility to pass the
async_tx dependency chain across successive calls to raid5_run_ops. The
req_compute variable protects against this case.
Changelog:
* remove the req_compute BUG_ON
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 20:52:30 +0000 (13:52 -0700)]
md: handle_stripe5 - add request/completion logic for async write ops
After handle_stripe5 decides whether it wants to perform a
read-modify-write, or a reconstruct write it calls
handle_write_operations5. A read-modify-write operation will perform an
xor subtraction of the blocks marked with the R5_Wantprexor flag, copy the
new data into the stripe (biodrain) and perform a postxor operation across
all up-to-date blocks to generate the new parity. A reconstruct write is run
when all blocks are already up-to-date in the cache so all that is needed
is a biodrain and postxor.
On the completion path STRIPE_OP_PREXOR will be set if the operation was a
read-modify-write. The STRIPE_OP_BIODRAIN flag is used in the completion
path to differentiate write-initiated postxor operations versus
expansion-initiated postxor operations. Completion of a write triggers i/o
to the drives.
Changelog:
* make the 'rcw' parameter to handle_write_operations5 a simple flag, Neil Brown
* remove test_and_set/test_and_clear BUG_ONs, Neil Brown
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 20:52:30 +0000 (13:52 -0700)]
md: common infrastructure for running operations with raid5_run_ops
All the handle_stripe operations that are to be transitioned to use
raid5_run_ops need a method to coherently gather work under the stripe-lock
and hand that work off to raid5_run_ops. The 'get_stripe_work' routine
runs under the lock to read all the bits in sh->ops.pending that do not
have the corresponding bit set in sh->ops.ack. This modified 'pending'
bitmap is then passed to raid5_run_ops for processing.
The transition from 'ack' to 'completion' does not need similar protection
as the existing release_stripe infrastructure will guarantee that
handle_stripe will run again after a completion bit is set, and
handle_stripe can tolerate a sh->ops.completed bit being set while the lock
is held.
A call to async_tx_issue_pending_all() is added to raid5d to kick the
offload engines once all pending stripe operations work has been submitted.
This enables batching of the submission and completion of operations.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 20:52:30 +0000 (13:52 -0700)]
md: raid5_run_ops - run stripe operations outside sh->lock
When the raid acceleration work was proposed, Neil laid out the following
attack plan:
1/ move the xor and copy operations outside spin_lock(&sh->lock)
2/ find/implement an asynchronous offload api
The raid5_run_ops routine uses the asynchronous offload api (async_tx) and
the stripe_operations member of a stripe_head to carry out xor+copy
operations asynchronously, outside the lock.
To perform operations outside the lock a new set of state flags is needed
to track new requests, in-flight requests, and completed requests. In this
new model handle_stripe is tasked with scanning the stripe_head for work,
updating the stripe_operations structure, and finally dropping the lock and
calling raid5_run_ops for processing. The following flags outline the
requests that handle_stripe can make of raid5_run_ops:
STRIPE_OP_BIOFILL
- copy data into request buffers to satisfy a read request
STRIPE_OP_COMPUTE_BLK
- generate a missing block in the cache from the other blocks
STRIPE_OP_PREXOR
- subtract existing data as part of the read-modify-write process
STRIPE_OP_BIODRAIN
- copy data out of request buffers to satisfy a write request
STRIPE_OP_POSTXOR
- recalculate parity for new data that has entered the cache
STRIPE_OP_CHECK
- verify that the parity is correct
STRIPE_OP_IO
- submit i/o to the member disks (note this was already performed outside
the stripe lock, but it made sense to add it as an operation type
The flow is:
1/ handle_stripe sets STRIPE_OP_* in sh->ops.pending
2/ raid5_run_ops reads sh->ops.pending, sets sh->ops.ack, and submits the
operation to the async_tx api
3/ async_tx triggers the completion callback routine to set
sh->ops.complete and release the stripe
4/ handle_stripe runs again to finish the operation and optionally submit
new operations that were previously blocked
Note this patch just defines raid5_run_ops, subsequent commits (one per
major operation type) modify handle_stripe to take advantage of this
routine.
Changelog:
* removed ops_complete_biodrain in favor of ops_complete_postxor and
ops_complete_write.
* removed the raid5_run_ops workqueue
* call bi_end_io for reads in ops_complete_biofill, saves a call to
handle_stripe
* explicitly handle the 2-disk raid5 case (xor becomes memcpy), Neil Brown
* fix race between async engines and bi_end_io call for reads, Neil Brown
* remove unnecessary spin_lock from ops_complete_biofill
* remove test_and_set/test_and_clear BUG_ONs, Neil Brown
* remove explicit interrupt handling for channel switching, this feature
was absorbed (i.e. it is now implicit) by the async_tx api
* use return_io in ops_complete_biofill
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Mon, 9 Jul 2007 18:56:43 +0000 (11:56 -0700)]
raid5: replace custom debug PRINTKs with standard pr_debug
Replaces PRINTK with pr_debug, and kills the RAID5_DEBUG definition in
favor of the global DEBUG definition. To get local debug messages just add
'#define DEBUG' to the top of the file.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Mon, 9 Jul 2007 18:56:43 +0000 (11:56 -0700)]
raid5: refactor handle_stripe5 and handle_stripe6 (v3)
handle_stripe5 and handle_stripe6 have very deep logic paths handling the
various states of a stripe_head. By introducing the 'stripe_head_state'
and 'r6_state' objects, large portions of the logic can be moved to
sub-routines.
'struct stripe_head_state' consumes all of the automatic variables that previously
stood alone in handle_stripe5,6. 'struct r6_state' contains the handle_stripe6
specific variables like p_failed and q_failed.
One of the nice side effects of the 'stripe_head_state' change is that it
allows for further reductions in code duplication between raid5 and raid6.
The following new routines are shared between raid5 and raid6:
Changes:
* v2: fixed 'conf->raid_disk-1' for the raid6 'handle_stripe_expansion' path
* v3: removed the unused 'dirty' field from struct stripe_head_state
* v3: coalesced open coded bi_end_io routines into return_io()
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Tue, 2 Jan 2007 18:10:44 +0000 (11:10 -0700)]
async_tx: add the async_tx api
The async_tx api provides methods for describing a chain of asynchronous
bulk memory transfers/transforms with support for inter-transactional
dependencies. It is implemented as a dmaengine client that smooths over
the details of different hardware offload engine implementations. Code
that is written to the api can optimize for asynchronous operation and the
api will fit the chain of operations to the available offload resources.
I imagine that any piece of ADMA hardware would register with the
'async_*' subsystem, and a call to async_X would be routed as
appropriate, or be run in-line. - Neil Brown
async_tx exploits the capabilities of struct dma_async_tx_descriptor to
provide an api of the following general format:
async_tx_find_channel() returns a capable channel from its pool. The
channel pool is organized as a per-cpu array of channel pointers. The
async_tx_rebalance() routine is tasked with managing these arrays. In the
uniprocessor case async_tx_rebalance() tries to spread responsibility
evenly over channels of similar capabilities. For example if there are two
copy+xor channels, one will handle copy operations and the other will
handle xor. In the SMP case async_tx_rebalance() attempts to spread the
operations evenly over the cpus, e.g. cpu0 gets copy channel0 and xor
channel0 while cpu1 gets copy channel 1 and xor channel 1. When a
dependency is specified async_tx_find_channel defaults to keeping the
operation on the same channel. A xor->copy->xor chain will stay on one
channel if it supports both operation types, otherwise the transaction will
transition between a copy and a xor resource.
Currently the raid5 implementation in the MD raid456 driver has been
converted to the async_tx api. A driver for the offload engines on the
Intel Xscale series of I/O processors, iop-adma, is provided in a later
commit. With the iop-adma driver and async_tx, raid456 is able to offload
copy, xor, and xor-zero-sum operations to hardware engines.
On iop342 tiobench showed higher throughput for sequential writes (20 - 30%
improvement) and sequential reads to a degraded array (40 - 55%
improvement). For the other cases performance was roughly equal, +/- a few
percentage points. On a x86-smp platform the performance of the async_tx
implementation (in synchronous mode) was also +/- a few percentage points
of the original implementation. According to 'top' on iop342 CPU
utilization drops from ~50% to ~15% during a 'resync' while the speed
according to /proc/mdstat doubles from ~25 MB/s to ~50 MB/s.
The tiobench command line used for testing was: tiobench --size 2048
--block 4096 --block 131072 --dir /mnt/raid --numruns 5
* iop342 had 1GB of memory available
Details:
* if CONFIG_DMA_ENGINE=n the asynchronous path is compiled away by making
async_tx_find_channel a static inline routine that always returns NULL
* when a callback is specified for a given transaction an interrupt will
fire at operation completion time and the callback will occur in a
tasklet. if the the channel does not support interrupts then a live
polling wait will be performed
* the api is written as a dmaengine client that requests all available
channels
* In support of dependencies the api implicitly schedules channel-switch
interrupts. The interrupt triggers the cleanup tasklet which causes
pending operations to be scheduled on the next channel
* Xor engines treat an xor destination address differently than a software
xor routine. To the software routine the destination address is an implied
source, whereas engines treat it as a write-only destination. This patch
modifies the xor_blocks routine to take a an explicit destination address
to mirror the hardware.
Changelog:
* fixed a leftover debug print
* don't allow callbacks in async_interrupt_cond
* fixed xor_block changes
* fixed usage of ASYNC_TX_XOR_DROP_DEST
* drop dma mapping methods, suggested by Chris Leech
* printk warning fixups from Andrew Morton
* don't use inline in C files, Adrian Bunk
* select the API when MD is enabled
* BUG_ON xor source counts <= 1
* implicitly handle hardware concerns like channel switching and
interrupts, Neil Brown
* remove the per operation type list, and distribute operation capabilities
evenly amongst the available channels
* simplify async_tx_find_channel to optimize the fast path
* introduce the channel_table_initialized flag to prevent early calls to
the api
* reorganize the code to mimic crypto
* include mm.h as not all archs include it in dma-mapping.h
* make the Kconfig options non-user visible, Adrian Bunk
* move async_tx under crypto since it is meant as 'core' functionality, and
the two may share algorithms in the future
* move large inline functions into c files
* checkpatch.pl fixes
* gpl v2 only correction
Cc: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-By: NeilBrown <neilb@suse.de>
Dan Williams [Mon, 9 Jul 2007 18:56:42 +0000 (11:56 -0700)]
xor: make 'xor_blocks' a library routine for use with async_tx
The async_tx api tries to use a dma engine for an operation, but will fall
back to an optimized software routine otherwise. Xor support is
implemented using the raid5 xor routines. For organizational purposes this
routine is moved to a common area.
The following fixes are also made:
* rename xor_block => xor_blocks, suggested by Adrian Bunk
* ensure that xor.o initializes before md.o in the built-in case
* checkpatch.pl fixes
* mark calibrate_xor_blocks __init, Adrian Bunk
Cc: Adrian Bunk <bunk@stusta.de> Cc: NeilBrown <neilb@suse.de> Cc: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Mon, 9 Jul 2007 18:56:42 +0000 (11:56 -0700)]
dmaengine: make clients responsible for managing channels
The current implementation assumes that a channel will only be used by one
client at a time. In order to enable channel sharing the dmaengine core is
changed to a model where clients subscribe to channel-available-events.
Instead of tracking how many channels a client wants and how many it has
received the core just broadcasts the available channels and lets the
clients optionally take a reference. The core learns about the clients'
needs at dma_event_callback time.
In support of multiple operation types, clients can specify a capability
mask to only be notified of channels that satisfy a certain set of
capabilities.
Changelog:
* removed DMA_TX_ARRAY_INIT, no longer needed
* dma_client_chan_free -> dma_chan_release: switch to global reference
counting only at device unregistration time, before it was also happening
at client unregistration time
* clients now return dma_state_client to dmaengine (ack, dup, nak)
* checkpatch.pl fixes
* fixup merge with git-ioat
Cc: Chris Leech <christopher.leech@intel.com> Signed-off-by: Shannon Nelson <shannon.nelson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-by: David S. Miller <davem@davemloft.net>
Dan Williams [Tue, 2 Jan 2007 18:10:43 +0000 (11:10 -0700)]
dmaengine: refactor dmaengine around dma_async_tx_descriptor
The current dmaengine interface defines mutliple routines per operation,
i.e. dma_async_memcpy_buf_to_buf, dma_async_memcpy_buf_to_page etc. Adding
more operation types (xor, crc, etc) to this model would result in an
unmanageable number of method permutations.
Are we really going to add a set of hooks for each DMA engine
whizbang feature?
- Jeff Garzik
The descriptor creation process is refactored using the new common
dma_async_tx_descriptor structure. Instead of per driver
do_<operation>_<dest>_to_<src> methods, drivers integrate
dma_async_tx_descriptor into their private software descriptor and then
define a 'prep' routine per operation. The prep routine allocates a
descriptor and ensures that the tx_set_src, tx_set_dest, tx_submit routines
are valid. Descriptor creation and submission becomes:
tx = dev->device_prep_dma_<operation>(chan, len, int_flag)
tx->tx_set_src(dma_addr_t, tx, index /* for multi-source ops */)
tx->tx_set_dest(dma_addr_t, tx, index)
tx->tx_submit(tx)
In addition to the refactoring, dma_async_tx_descriptor also lays the
groundwork for definining cross-channel-operation dependencies, and a
callback facility for asynchronous notification of operation completion.
Changelog:
* drop dma mapping methods, suggested by Chris Leech
* fix ioat_dma_dependency_added, also caught by Andrew Morton
* fix dma_sync_wait, change from Andrew Morton
* uninline large functions, change from Andrew Morton
* add tx->callback = NULL to dmaengine calls to interoperate with async_tx
calls
* hookup ioat_tx_submit
* convert channel capabilities to a 'cpumask_t like' bitmap
* removed DMA_TX_ARRAY_INIT, no longer needed
* checkpatch.pl fixes
* make set_src, set_dest, and tx_submit descriptor specific methods
* fixup git-ioat merge
* move group_list and phys to dma_async_tx_descriptor
Cc: Jeff Garzik <jeff@garzik.org> Cc: Chris Leech <christopher.leech@intel.com> Signed-off-by: Shannon Nelson <shannon.nelson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Acked-by: David S. Miller <davem@davemloft.net>
projects / linux-2.6 / log