firefly-linux-kernel-4.4.55.git
19 years ago[libata sata_promise] add pci id
Jeff Garzik [Fri, 28 Oct 2005 21:00:31 +0000 (17:00 -0400)]
[libata sata_promise] add pci id

Contributed by Daniel Mueller @ Siemens AG.

19 years ago[libata scsi] MODE SELECT, strengthen mode sense
Douglas Gilbert [Fri, 28 Oct 2005 19:58:28 +0000 (15:58 -0400)]
[libata scsi] MODE SELECT, strengthen mode sense

  - move default mode pages to the front of libata-scsi.c
    so various functions can access them
  - partial annotation of these pages, point out divergence
    from sat-r06
  - replace various mode page magic numbers with defines
  - strengthen MODE SENSE command decoding: handle DBD
    bit in cdb, yield block descriptor (per sat-r06) and
    handle mode sub pages

Signed-off-by: Douglas Gilbert <dougg@torque.net>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[libata pdc_adma] minor fixes and cleanups
Jeff Garzik [Fri, 28 Oct 2005 19:43:16 +0000 (15:43 -0400)]
[libata pdc_adma] minor fixes and cleanups

Changes mostly from Mark Lord.

- fix bugs in probe-time error handling
- only complete qc if not NULL
- check port-level polling flags

19 years agoMerge branch 'master'
Jeff Garzik [Fri, 28 Oct 2005 16:31:34 +0000 (12:31 -0400)]
Merge branch 'master'

19 years agoMerge branch 'upstream' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/libata-dev
Linus Torvalds [Fri, 28 Oct 2005 16:06:50 +0000 (09:06 -0700)]
Merge branch 'upstream' of /linux/kernel/git/jgarzik/libata-dev

19 years agoMerge branch 'upstream' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6
Linus Torvalds [Fri, 28 Oct 2005 16:05:25 +0000 (09:05 -0700)]
Merge branch 'upstream' of /linux/kernel/git/jgarzik/netdev-2.6

19 years agoMerge branch 'elevator-switch' of git://brick.kernel.dk/data/git/linux-2.6-block
Linus Torvalds [Fri, 28 Oct 2005 15:56:34 +0000 (08:56 -0700)]
Merge branch 'elevator-switch' of git://brick.kernel.dk/data/git/linux-2.6-block

Manual fixup for trivial "gfp_t" changes.

19 years agoMerge branch 'generic-dispatch' of git://brick.kernel.dk/data/git/linux-2.6-block
Linus Torvalds [Fri, 28 Oct 2005 15:53:49 +0000 (08:53 -0700)]
Merge branch 'generic-dispatch' of git://brick.kernel.dk/data/git/linux-2.6-block

19 years agoMerge branch 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block
Linus Torvalds [Fri, 28 Oct 2005 15:53:00 +0000 (08:53 -0700)]
Merge branch 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block

19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/acme/net-2.6.15
Linus Torvalds [Fri, 28 Oct 2005 15:50:37 +0000 (08:50 -0700)]
Merge /pub/scm/linux/kernel/git/acme/net-2.6.15

19 years ago[PATCH] gfp_t: the rest
Al Viro [Fri, 21 Oct 2005 07:22:44 +0000 (03:22 -0400)]
[PATCH] gfp_t: the rest

zone handling, mapping->flags handling

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: reiserfs mapping_set_gfp_mask() use
Al Viro [Fri, 21 Oct 2005 07:22:39 +0000 (03:22 -0400)]
[PATCH] gfp_t: reiserfs mapping_set_gfp_mask() use

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: remaining bits of drivers/*
Al Viro [Fri, 21 Oct 2005 07:22:34 +0000 (03:22 -0400)]
[PATCH] gfp_t: remaining bits of drivers/*

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: drivers/net
Al Viro [Fri, 21 Oct 2005 07:22:29 +0000 (03:22 -0400)]
[PATCH] gfp_t: drivers/net

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: remaining bits of arch/*
Al Viro [Fri, 21 Oct 2005 07:22:24 +0000 (03:22 -0400)]
[PATCH] gfp_t: remaining bits of arch/*

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: sound
Al Viro [Fri, 21 Oct 2005 07:22:18 +0000 (03:22 -0400)]
[PATCH] gfp_t: sound

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: drivers/infiniband
Al Viro [Fri, 21 Oct 2005 07:22:13 +0000 (03:22 -0400)]
[PATCH] gfp_t: drivers/infiniband

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: drivers/scsi
Al Viro [Fri, 21 Oct 2005 07:22:08 +0000 (03:22 -0400)]
[PATCH] gfp_t: drivers/scsi

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: kernel/*
Al Viro [Fri, 21 Oct 2005 07:22:03 +0000 (03:22 -0400)]
[PATCH] gfp_t: kernel/*

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: drivers/usb
Al Viro [Fri, 21 Oct 2005 07:21:58 +0000 (03:21 -0400)]
[PATCH] gfp_t: drivers/usb

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (simple cases)
Al Viro [Fri, 21 Oct 2005 07:21:53 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (simple cases)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (xtensa)
Al Viro [Fri, 21 Oct 2005 07:21:48 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (xtensa)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (amd64)
Al Viro [Fri, 21 Oct 2005 07:21:43 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (amd64)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (sh)
Al Viro [Fri, 21 Oct 2005 07:21:38 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (sh)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (ppc)
Al Viro [Fri, 21 Oct 2005 07:21:33 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (ppc)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (parisc)
Al Viro [Fri, 21 Oct 2005 07:21:28 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (parisc)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (mips)
Al Viro [Fri, 21 Oct 2005 07:21:23 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (mips)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (frv)
Al Viro [Fri, 21 Oct 2005 07:21:18 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (frv)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (cris)
Al Viro [Fri, 21 Oct 2005 07:21:13 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (cris)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (alpha)
Al Viro [Fri, 21 Oct 2005 07:21:08 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (alpha)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (ia64)
Al Viro [Fri, 21 Oct 2005 07:21:03 +0000 (03:21 -0400)]
[PATCH] gfp_t: dma-mapping (ia64)

... and related annotations for amd64 - swiotlb code is shared, but
prototypes are not.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: dma-mapping (arm)
Al Viro [Fri, 21 Oct 2005 07:20:58 +0000 (03:20 -0400)]
[PATCH] gfp_t: dma-mapping (arm)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: block layer core
Al Viro [Fri, 21 Oct 2005 07:20:53 +0000 (03:20 -0400)]
[PATCH] gfp_t: block layer core

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: fs/*
Al Viro [Fri, 21 Oct 2005 07:20:48 +0000 (03:20 -0400)]
[PATCH] gfp_t: fs/*

 - ->releasepage() annotated (s/int/gfp_t), instances updated
 - missing gfp_t in fs/* added
 - fixed misannotation from the original sweep caught by bitwise checks:
   XFS used __nocast both for gfp_t and for flags used by XFS allocator.
   The latter left with unsigned int __nocast; we might want to add a
   different type for those but for now let's leave them alone.  That,
   BTW, is a case when __nocast use had been actively confusing - it had
   been used in the same code for two different and similar types, with
   no way to catch misuses.  Switch of gfp_t to bitwise had caught that
   immediately...

One tricky bit is left alone to be dealt with later - mapping->flags is
a mix of gfp_t and error indications.  Left alone for now.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: net/*
Al Viro [Fri, 21 Oct 2005 07:20:43 +0000 (03:20 -0400)]
[PATCH] gfp_t: net/*

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: lib/*
Al Viro [Fri, 21 Oct 2005 07:18:50 +0000 (03:18 -0400)]
[PATCH] gfp_t: lib/*

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: mm/* (easy parts)
Al Viro [Fri, 21 Oct 2005 07:18:50 +0000 (03:18 -0400)]
[PATCH] gfp_t: mm/* (easy parts)

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gfp_t: infrastructure
Al Viro [Fri, 21 Oct 2005 06:55:38 +0000 (02:55 -0400)]
[PATCH] gfp_t: infrastructure

Beginning of gfp_t annotations:

 - -Wbitwise added to CHECKFLAGS
 - old __bitwise renamed to __bitwise__
 - __bitwise defined to either __bitwise__ or nothing, depending on
   __CHECK_ENDIAN__ being defined
 - gfp_t switched from __nocast to __bitwise__
 - force cast to gfp_t added to __GFP_... constants
 - new helper - gfp_zone(); extracts zone bits out of gfp_t value and casts
   the result to int

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[BLOCK] Leftover reference to ->max_back_kb
Jens Axboe [Fri, 28 Oct 2005 07:01:17 +0000 (09:01 +0200)]
[BLOCK] Leftover reference to ->max_back_kb

Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] elevator switch fixes/cleanup
Jens Axboe [Fri, 28 Oct 2005 06:30:39 +0000 (08:30 +0200)]
[BLOCK] elevator switch fixes/cleanup

- 100msec sleep is a little excessive, lots of requests can complete
  in that timeframe. Use 10msec instead.
- Rename QUEUE_FLAG_BYPASS to QUEUE_FLAG_ELVSWITCH to indicate what
  is going on.

Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] Reimplement elevator switch
Tejun Heo [Fri, 28 Oct 2005 06:29:39 +0000 (08:29 +0200)]
[BLOCK] Reimplement elevator switch

 This patch reimplements elevator switch.  This patch assumes generic
dispatch queue patchset is applied.

 * Each request is tagged with REQ_ELVPRIV flag if it has its elevator
   private data set.
 * Requests which doesn't have REQ_ELVPRIV flag set never enter
   iosched.  They are always directly back inserted to dispatch queue.
   Of course, elevator_put_req_fn is called only for requests which
   have its REQ_ELVPRIV set.
 * Request queue maintains the current number of requests which have
   its elevator data set (elevator_set_req_fn called) in
   q->rq->elvpriv.
 * If a request queue has QUEUE_FLAG_BYPASS set, elevator private data
   is not allocated for new requests.

 To switch to another iosched, we set QUEUE_FLAG_BYPASS and wait until
elvpriv goes to zero; then, we attach the new iosched and clears
QUEUE_FLAG_BYPASS.  New implementation is much simpler and main code
paths are less cluttered, IMHO.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] kill generic max_back_kb handling
Tejun Heo [Mon, 24 Oct 2005 06:35:58 +0000 (08:35 +0200)]
[BLOCK] kill generic max_back_kb handling

This patch kills max_back_kb handling from elv_dispatch_sort() and
kills max_back_kb field from struct request_queue.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[PATCH] 05/05 update biodoc to match new generic dispatch api
Tejun Heo [Thu, 20 Oct 2005 14:47:40 +0000 (16:47 +0200)]
[PATCH] 05/05 update biodoc to match new generic dispatch api

Updates biodoc to reflect changes in elevator API

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[PATCH] 04/05 remove last_merge handling from ioscheds
Tejun Heo [Thu, 20 Oct 2005 14:46:54 +0000 (16:46 +0200)]
[PATCH] 04/05 remove last_merge handling from ioscheds

Remove last_merge handling from all ioscheds.  This patch
removes merging capability of noop iosched.

Signed-off-by: Tejun Heo <htejun@gmail.com>
19 years ago[PATCH] 03/05 move last_merge handlin into generic elevator code
Tejun Heo [Thu, 20 Oct 2005 14:46:23 +0000 (16:46 +0200)]
[PATCH] 03/05 move last_merge handlin into generic elevator code

Currently, both generic elevator code and specific ioscheds
participate in the management and usage of last_merge.  This
and the following patches move last_merge handling into
generic elevator code.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[PATCH] 02/05: update ioscheds to use generic dispatch queue
Jens Axboe [Thu, 20 Oct 2005 14:42:29 +0000 (16:42 +0200)]
[PATCH] 02/05: update ioscheds to use generic dispatch queue

This patch updates all four ioscheds to use generic dispatch
queue.  There's one behavior change in as-iosched.

* In as-iosched, when force dispatching
  (ELEVATOR_INSERT_BACK), batch_data_dir is reset to REQ_SYNC
  and changed_batch and new_batch are cleared to zero.  This
  prevernts AS from doing incorrect update_write_batch after
  the forced dispatched requests are finished.

* In cfq-iosched, cfqd->rq_in_driver currently counts the
  number of activated (removed) requests to determine
  whether queue-kicking is needed and cfq_max_depth has been
  reached.  With generic dispatch queue, I think counting
  the number of dispatched requests would be more appropriate.

* cfq_max_depth can be lowered to 1 again.

Original from Tejun Heo, modified version applied.

Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[PATCH] elevator: leftover function declaration
Jens Axboe [Thu, 20 Oct 2005 14:37:38 +0000 (16:37 +0200)]
[PATCH] elevator: leftover function declaration

Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[PATCH] generic dispatch fixes
Jens Axboe [Thu, 20 Oct 2005 14:37:00 +0000 (16:37 +0200)]
[PATCH] generic dispatch fixes

- Split elv_dispatch_insert() into two functions
- Rename rq_last_sector() to rq_end_sector()

Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[PATCH] 01/05 Implement generic dispatch queue
Tejun Heo [Thu, 20 Oct 2005 14:23:44 +0000 (16:23 +0200)]
[PATCH] 01/05 Implement generic dispatch queue

Implements generic dispatch queue which can replace all
dispatch queues implemented by each iosched.  This reduces
code duplication, eases enforcing semantics over dispatch
queue, and simplifies specific ioscheds.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[PATCH] fix try_module_get race in elevator_find
Tejun Heo [Thu, 20 Oct 2005 08:56:41 +0000 (10:56 +0200)]
[PATCH] fix try_module_get race in elevator_find

This patch removes try_module_get race in elevator_find.
try_module_get should always be called with the spinlock protecting
what the module init/cleanup routines register/unregister to held. In
the case of elevators, we should be holding elv_list to avoid it going
away between spin_unlock_irq and try_module_get.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years agoFollowing the same idea, it occurs to me that we should only update
Chen, Kenneth W [Thu, 13 Oct 2005 19:49:29 +0000 (21:49 +0200)]
Following the same idea, it occurs to me that we should only update
disk stat when "now" is different from disk->stamp.  Otherwise, we
are again needlessly adding zero to the stats.

Signed-off-by: Ken Chen <kenneth.w.chen@intel.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[patch] remove gendisk->stamp_idle field
Chen, Kenneth W [Thu, 13 Oct 2005 19:48:42 +0000 (21:48 +0200)]
[patch] remove gendisk->stamp_idle field

struct gendisk has these two fields: stamp, stamp_idle.  Update to
stamp_idle is always in sync with stamp and they are always the same.
Therefore, it does not add any value in having two fields tracking
same timestamp.  Suggest to remove it.

Also, we should only update gendisk stats with non-zero value.
Advantage is that we don't have to needlessly calculate memory address,
and then add zero to the content.

Signed-off-by: Ken Chen <kenneth.w.chen@intel.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years agoMerge /home/trondmy/scm/kernel/git/torvalds/linux-2.6
Trond Myklebust [Fri, 28 Oct 2005 02:13:32 +0000 (22:13 -0400)]
Merge /home/trondmy/scm/kernel/git/torvalds/linux-2.6

19 years agoRPC: Ensure that nobody can queue up new upcalls after rpc_close_pipes()
Trond Myklebust [Fri, 28 Oct 2005 02:12:46 +0000 (22:12 -0400)]
RPC: Ensure that nobody can queue up new upcalls after rpc_close_pipes()

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Allow files that are open for write to invalidate caches
Trond Myklebust [Fri, 28 Oct 2005 02:12:45 +0000 (22:12 -0400)]
NFS: Allow files that are open for write to invalidate caches

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFSv4: Convert unnecessary XDR warning messages into dprintk()
Trond Myklebust [Fri, 28 Oct 2005 02:12:45 +0000 (22:12 -0400)]
NFSv4: Convert unnecessary XDR warning messages into dprintk()

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFSv4: Add post-op attributes to NFSv4 write and commit callbacks.
Trond Myklebust [Fri, 28 Oct 2005 02:12:44 +0000 (22:12 -0400)]
NFSv4: Add post-op attributes to NFSv4 write and commit callbacks.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFSv4: Add post-op attributes to nfs4_proc_remove()
Trond Myklebust [Fri, 28 Oct 2005 02:12:44 +0000 (22:12 -0400)]
NFSv4: Add post-op attributes to nfs4_proc_remove()

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFSv4: Add post-op attributes to nfs4_proc_rename()
Trond Myklebust [Fri, 28 Oct 2005 02:12:43 +0000 (22:12 -0400)]
NFSv4: Add post-op attributes to nfs4_proc_rename()

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFSv4: Add post-op attributes to nfs4_proc_link()
Trond Myklebust [Fri, 28 Oct 2005 02:12:42 +0000 (22:12 -0400)]
NFSv4: Add post-op attributes to nfs4_proc_link()

 Optimise attribute revalidation when hardlinking. Add post-op attributes
 for the directory and the original inode.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Ensure that nfs_link() instantiates the dentry correctly
Trond Myklebust [Fri, 28 Oct 2005 02:12:42 +0000 (22:12 -0400)]
NFS: Ensure that nfs_link() instantiates the dentry correctly

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Add optional post-op getattr instruction to the NFSv4 file close.
Trond Myklebust [Fri, 28 Oct 2005 02:12:41 +0000 (22:12 -0400)]
NFS: Add optional post-op getattr instruction to the NFSv4 file close.

 "Optional" means that the close call will not fail if the getattr
 at the end of the compound fails.
 If it does succeed, try to refresh inode attributes.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Optimise attribute revalidation on close().
Trond Myklebust [Fri, 28 Oct 2005 02:12:41 +0000 (22:12 -0400)]
NFS: Optimise attribute revalidation on close().

 Only force a getattr in nfs_file_flush() if the attribute
 cache is stale.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFSv4: Add directory post-op attributes to the CREATE operations.
Trond Myklebust [Fri, 28 Oct 2005 02:12:40 +0000 (22:12 -0400)]
NFSv4: Add directory post-op attributes to the CREATE operations.

 Since the directory attributes change every time we CREATE a file,
 we might as well pick up the new directory attributes in the same
 compound.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: nfs_lookup doesn't need to revalidate the parent directory's inode
Chuck Lever [Tue, 25 Oct 2005 15:48:36 +0000 (11:48 -0400)]
NFS: nfs_lookup doesn't need to revalidate the parent directory's inode

 nfs_lookup() used to consult a lookup cache before trying an actual wire
 lookup operation.  The lookup cache would be invalid, of course, if the
 parent directory's mtime had changed, so nfs_lookup performed an inode
 revalidation on the parent.

 Since nfs_lookup() doesn't use a cache anymore, the revalidation is no
 longer necessary.  There are cases where it will generate a lot of
 unnecessary GETATTR traffic.

 See http://bugzilla.linux-nfs.org/show_bug.cgi?id=9

 Test-plan:
 Use lndir and "rm -rf" and watch for excess GETATTR traffic or application
 level errors.

Signed-off-by: Chuck Lever <cel@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Don't let nfs_end_data_update() clobber attribute update information
Trond Myklebust [Fri, 28 Oct 2005 02:12:39 +0000 (22:12 -0400)]
NFS: Don't let nfs_end_data_update() clobber attribute update information

 Since we almost always call nfs_end_data_update() after we called
 nfs_refresh_inode(), we now end up marking the inode metadata
 as needing revalidation immediately after having updated it.

 This patch rearranges things so that we mark the inode as needing
 revalidation _before_ we call nfs_refresh_inode() on those operations
 that need it.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Optimise inode attribute cache updates
Trond Myklebust [Fri, 28 Oct 2005 02:12:39 +0000 (22:12 -0400)]
NFS: Optimise inode attribute cache updates

 Allow nfs_refresh_inode() also to update attributes on the inode if the
 RPC call was sent after the last call to nfs_update_inode().

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Convert cache_change_attribute into a jiffy-based value
Trond Myklebust [Fri, 28 Oct 2005 02:12:38 +0000 (22:12 -0400)]
NFS: Convert cache_change_attribute into a jiffy-based value

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoNFS: Cleanup initialisation of struct nfs_fattr
Trond Myklebust [Fri, 28 Oct 2005 02:12:38 +0000 (22:12 -0400)]
NFS: Cleanup initialisation of struct nfs_fattr

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoMerge branch 'master'
Jeff Garzik [Fri, 28 Oct 2005 00:43:20 +0000 (20:43 -0400)]
Merge branch 'master'

19 years agoMerge branch 'master'
Jeff Garzik [Fri, 28 Oct 2005 00:35:17 +0000 (20:35 -0400)]
Merge branch 'master'

19 years agoLinux v2.6.14
Linus Torvalds [Fri, 28 Oct 2005 00:02:08 +0000 (17:02 -0700)]
Linux v2.6.14

"Better late than never"

19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/acme/net-2.6
Linus Torvalds [Thu, 27 Oct 2005 23:58:20 +0000 (16:58 -0700)]
Merge /pub/scm/linux/kernel/git/acme/net-2.6

19 years ago[PATCH] cpufreq: SMP fix for conservative governor
Dave Jones [Thu, 27 Oct 2005 23:16:25 +0000 (16:16 -0700)]
[PATCH] cpufreq: SMP fix for conservative governor

Don't try to access not-present CPUs.  Conservative governor will always
oops on SMP without this fix.

Fixes http://bugzilla.kernel.org/show_bug.cgi?id=4781

Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoRevert "x86-64: Avoid unnecessary double bouncing for swiotlb"
Linus Torvalds [Thu, 27 Oct 2005 23:28:39 +0000 (16:28 -0700)]
Revert "x86-64: Avoid unnecessary double bouncing for swiotlb"

Commit id 6142891a0c0209c91aa4a98f725de0d6e2ed4918

Andi Kleen reports that it seems to break things for some people,
and since it's purely a small optimization, revert it for now.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge /home/trondmy/scm/kernel/git/torvalds/linux-2.6
Trond Myklebust [Thu, 27 Oct 2005 23:12:49 +0000 (19:12 -0400)]
Merge /home/trondmy/scm/kernel/git/torvalds/linux-2.6

19 years agoNFS: Fix a bad cast in nfs3_read_done
Trond Myklebust [Thu, 27 Oct 2005 23:10:09 +0000 (19:10 -0400)]
NFS: Fix a bad cast in nfs3_read_done

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
19 years agoRevert "RPC: stops the release_pipe() funtion from being called twice"
Trond Myklebust [Thu, 27 Oct 2005 23:08:18 +0000 (19:08 -0400)]
Revert "RPC: stops the release_pipe() funtion from being called twice"

This reverts 747c5534c9a6da4aa87e7cdc2209ea98ea27f381 commit.

19 years ago[TCP]: Clear stale pred_flags when snd_wnd changes
Herbert Xu [Thu, 27 Oct 2005 08:47:46 +0000 (18:47 +1000)]
[TCP]: Clear stale pred_flags when snd_wnd changes

This bug is responsible for causing the infamous "Treason uncloaked"
messages that's been popping up everywhere since the printk was added.
It has usually been blamed on foreign operating systems.  However,
some of those reports implicate Linux as both systems are running
Linux or the TCP connection is going across the loopback interface.

In fact, there really is a bug in the Linux TCP header prediction code
that's been there since at least 2.1.8.  This bug was tracked down with
help from Dale Blount.

The effect of this bug ranges from harmless "Treason uncloaked"
messages to hung/aborted TCP connections.  The details of the bug
and fix is as follows.

When snd_wnd is updated, we only update pred_flags if
tcp_fast_path_check succeeds.  When it fails (for example,
when our rcvbuf is used up), we will leave pred_flags with
an out-of-date snd_wnd value.

When the out-of-date pred_flags happens to match the next incoming
packet we will again hit the fast path and use the current snd_wnd
which will be wrong.

In the case of the treason messages, it just happens that the snd_wnd
cached in pred_flags is zero while tp->snd_wnd is non-zero.  Therefore
when a zero-window packet comes in we incorrectly conclude that the
window is non-zero.

In fact if the peer continues to send us zero-window pure ACKs we
will continue making the same mistake.  It's only when the peer
transmits a zero-window packet with data attached that we get a
chance to snap out of it.  This is what triggers the treason
message at the next retransmit timeout.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
19 years ago[PATCH] Yet more posix-cpu-timer fixes
Roland McGrath [Thu, 27 Oct 2005 10:16:42 +0000 (03:16 -0700)]
[PATCH] Yet more posix-cpu-timer fixes

This just makes sure that a thread's expiry times can't get reset after
it clears them in do_exit.

This is what allowed us to re-introduce the stricter BUG_ON() check in
a362f463a6d316d14daed0f817e151835ce97ff7.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoRevert "remove false BUG_ON() from run_posix_cpu_timers()"
Linus Torvalds [Thu, 27 Oct 2005 16:07:33 +0000 (09:07 -0700)]
Revert "remove false BUG_ON() from run_posix_cpu_timers()"

This reverts commit 3de463c7d9d58f8cf3395268230cb20a4c15bffa.

Roland has another patch that allows us to leave the BUG_ON() in place
by just making sure that the condition it tests for really is always
true.

That goes in next.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Fix cpu timers expiration time
Oleg Nesterov [Wed, 26 Oct 2005 16:26:53 +0000 (20:26 +0400)]
[PATCH] Fix cpu timers expiration time

There's a silly off-by-one error in the code that updates the expiration
of posix CPU timers, causing them to not be properly updated when they
hit exactly on their expiration time (which should be the normal case).

This causes them to then fire immediately again, and only _then_ get
properly updated.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge master.kernel.org:/home/rmk/linux-2.6-serial
Linus Torvalds [Wed, 26 Oct 2005 21:02:49 +0000 (14:02 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-serial

19 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Wed, 26 Oct 2005 21:01:57 +0000 (14:01 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

19 years agoposix cpu timers: fix timer ordering
Linus Torvalds [Wed, 26 Oct 2005 18:23:06 +0000 (11:23 -0700)]
posix cpu timers: fix timer ordering

Pointed out by Oleg Nesterov, who has been walking over the code
forwards and backwards.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] fix radeon_cp_init_ring_buffer()
Ivan Kokshaysky [Wed, 26 Oct 2005 10:05:25 +0000 (11:05 +0100)]
[PATCH] fix radeon_cp_init_ring_buffer()

I've seen similar failure on alpha.

Obviously, someone forgot to convert sg->handle stuff for
PCI gart case.

Signed-off-by: Dave Airlie <airlied@linux.ie>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] svcsock timestamp fix
Andrew Morton [Wed, 26 Oct 2005 08:59:03 +0000 (01:59 -0700)]
[PATCH] svcsock timestamp fix

Convert nanoseconds to microseconds correctly.

Spotted by Steve Dickson <SteveD@redhat.com>

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Fix HFS+ to free up the space when a file is deleted.
Peter Wainwright [Wed, 26 Oct 2005 08:59:02 +0000 (01:59 -0700)]
[PATCH] Fix HFS+ to free up the space when a file is deleted.

fsck_hfs reveals lots of temporary files accumulating in the hidden
directory "\000\000\000HFS+ Private Data".  According to the HFS+
documentation these are files which are unlinked while in use.  However,
there may be a bug in the Linux hfsplus implementation which causes this to
happen even when the files are not in use.  It looks like the "opencnt"
field is never initialized as (I think) it should be in hfsplus_read_inode.
 This means that a file can appear to be still in use when in fact it has
been closed.  This patch seems to fix it for me.

Signed-off-by: Anton Altaparmakov <aia21@cantab.net>
Cc: Roman Zippel <zippel@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] kill massive wireless-related log spam
Jeff Garzik [Wed, 26 Oct 2005 08:59:01 +0000 (01:59 -0700)]
[PATCH] kill massive wireless-related log spam

Although this message is having the intended effect of causing wireless
driver maintainers to upgrade their code, I never should have merged this
patch in its present form.  Leading to tons of bug reports and unhappy
users.

Some wireless apps poll for statistics regularly, which leads to a printk()
every single time they ask for stats.  That's a little bit _too_ much of a
reminder that the driver is using an old API.

Change this to printing out the message once, per kernel boot.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] export cpu_online_map
Andrew Morton [Wed, 26 Oct 2005 08:59:01 +0000 (01:59 -0700)]
[PATCH] export cpu_online_map

With CONFIG_SMP=n:

*** Warning: "cpu_online_map" [drivers/firmware/dcdbas.ko] undefined!

due to set_cpus_allowed().

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc64: Fix wrong register mapping in mpic driver
Benjamin Herrenschmidt [Wed, 26 Oct 2005 08:59:00 +0000 (01:59 -0700)]
[PATCH] ppc64: Fix wrong register mapping in mpic driver

The mpic interrupt controller driver (used on G5 and early pSeries among
others) has a bug where it doesn't get the right virtual address for the
timer registers.  It causes the driver to poke at the MMIO space of
whatever has been mapped just next to it (ouch !) when initializing and
causes boot failures on some IBM machines.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] NUMA: broken per cpu pageset counters
Magnus Damm [Wed, 26 Oct 2005 08:58:59 +0000 (01:58 -0700)]
[PATCH] NUMA: broken per cpu pageset counters

The NUMA counters in struct per_cpu_pageset (linux/mmzone.h) are never
cleared today.  This works ok for CPU 0 on NUMA machines because
boot_pageset[] is already zero, but for other CPU:s this results in
uninitialized counters.

Signed-off-by: Magnus Damm <magnus@valinux.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] md: make sure mdthreads will always respond to kthread_stop
NeilBrown [Wed, 26 Oct 2005 08:58:58 +0000 (01:58 -0700)]
[PATCH] md: make sure mdthreads will always respond to kthread_stop

There are still a couple of cases where md threads (the resync/recovery
thread) is not interruptible since the change to use kthreads.  All places
there it tests "signal_pending", it should also test kthread_should_stop,
as with this patch.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[libata] ata_timing fix
Alan Cox [Wed, 26 Oct 2005 16:17:46 +0000 (12:17 -0400)]
[libata] ata_timing fix

19 years ago[ARM] 3032/1: sparse: complains about generic_fls() prototype in asm-arm/bitops.h
Ian Campbell [Wed, 26 Oct 2005 14:04:21 +0000 (15:04 +0100)]
[ARM] 3032/1: sparse: complains about generic_fls() prototype in asm-arm/bitops.h

Patch from Ian Campbell

Sparse complains about the definition of generic_fls in asm-arm/bitops.h:
  CHECK   /home/icampbell/devel/kernel/2.6/arch/arm/mach-pxa/viper.c
include2/asm/bitops.h:350:34: error: marked inline, but without a definition

The definition is unnecessary since linux/bitops.h defines generic_fls before including asm/bitops.h and asm/bitops.h should not be included directly. There are still some places where asm/bitops.h is directly included, but I think that code should be fixed. I was a little wary of the patch for this reason but lubbock, mainstone and assabet all build OK and so do my in house boards...

ARM is the only arch with the generic_fls prototype in this way.

Signed-off-by: Ian Campbell <icampbell@arcom.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years agoMerge branch 'master'
Jeff Garzik [Wed, 26 Oct 2005 05:08:05 +0000 (01:08 -0400)]
Merge branch 'master'

19 years agoMerge branch 'master'
Jeff Garzik [Wed, 26 Oct 2005 05:06:45 +0000 (01:06 -0400)]
Merge branch 'master'

19 years ago[PATCH] ieee80211 build fix
James Ketrenos [Mon, 24 Oct 2005 15:27:46 +0000 (10:27 -0500)]
[PATCH] ieee80211 build fix

James Ketrenos wrote:
> [3/4] Use the tx_headroom and reserve requested space.

This patch introduced a compile problem; patch below corrects this.

Fixed compilation error due to not passing tx_headroom in
ieee80211_tx_frame.

Signed-off-by: James Ketrenos <jketreno@linux.intel.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[netdrvr forcedeth] scatter gather and segmentation offload support
Ayaz Abdulla [Wed, 26 Oct 2005 04:51:24 +0000 (00:51 -0400)]
[netdrvr forcedeth] scatter gather and segmentation offload support

also:
- eliminate use of pointless get_nvpriv() wrapper,
  and use netdev_priv() directly.
- use NETDEV_TX_xxx return codes

19 years agoPCI: be more verbose about resource quirks
Linus Torvalds [Wed, 26 Oct 2005 03:40:09 +0000 (20:40 -0700)]
PCI: be more verbose about resource quirks

When reserving an PCI quirk, note that in the kernel bootup messages.

Also, parse the strange PIIX4 device resources - they should get their
own PCI resource quirks, but for now just print out what it finds to
verify that the code does the right thing.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>