[PATCH] md: fix some small races in bitmap plugging in raid5
The comment gives more details, but I didn't quite have the sequencing write,
so there was room for races to leave bits unset in the on-disk bitmap for
short periods of time.
Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
When a device is unplugged, requests are moved from one or two (depending on
whether a bitmap is in use) queues to the main request queue.
So whenever requests are put on either of those queues, we should make sure
the raid5 array is 'plugged'. However we don't. We currently plug the raid5
queue just before putting requests on queues, so there is room for a race. If
something unplugs the queue at just the wrong time, requests will be left on
the queue and nothing will want to unplug them. Normally something else will
plug and unplug the queue fairly soon, but there is a risk that nothing will.
Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] md: fix resync speed calculation for restarted resyncs
We introduced 'io_sectors' recently so we could count the sectors that causes
io during resync separate from sectors which didn't cause IO - there can be a
difference if a bitmap is being used to accelerate resync.
However when a speed is reported, we find the number of sectors processed
recently by subtracting an oldish io_sectors count from a current
'curr_resync' count. This is wrong because curr_resync counts all sectors,
not just io sectors.
So, add a field to mddev to store the curren io_sectors separately from
curr_resync, and use that in the calculations.
Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] md: delay starting md threads until array is completely setup
When an array is started we start one or two threads (two if there is a
reshape or recovery that needs to be completed).
We currently start these *before* the array is completely set up and in
particular before queue->queuedata is set. If the thread actually starts
very quickly on another CPU, we can end up dereferencing queue->queuedata
and oops.
This patch also makes sure we don't try to start a recovery if a reshape is
being restarted.
Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
I have reports of a problem with raid5 which turns out to be because the raid5
device gets stuck in a 'plugged' state. This shouldn't be able to happen as
3msec after it gets plugged it should get unplugged. However it happens
none-the-less. This patch fixes the problem and is a reasonable thing to do,
though it might hurt performance slightly in some cases.
Until I can find the real problem, we should probably have this workaround in
place.
Signed-off-by: Neil Brown <neilb@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jon Smirl [Mon, 10 Jul 2006 11:44:13 +0000 (04:44 -0700)]
[PATCH] tty: Remove include of screen_info.h from tty.h
screen_info.h doesn't have anything to do with the tty layer and shouldn't be
included by tty.h. This patches removes the include and modifies all users to
directly include screen_info.h. struct screen_info is mainly used to
communicate with the console drivers in drivers/video/console. Note that this
patch touches every arch and I have no way of testing it. If there is a
mistake the worst thing that will happen is a compile error.
[akpm@osdl.org: fix arm build]
[akpm@osdl.org: fix alpha build] Signed-off-by: Jon Smirl <jonsmir@gmail.com> Signed-off-by: Antonino Daplas <adaplas@pol.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jon Smirl [Mon, 10 Jul 2006 11:44:12 +0000 (04:44 -0700)]
[PATCH] vt: Remove VT-specific declarations and definitions from tty.h
MAX_NR_CONSOLES, fg_console, want_console and last_console are more of a
function of the VT layer than the TTY one. Moving these to vt.h and vt_kern.h
allows all of the framebuffer and VT console drivers to remove their
dependency on tty.h.
[akpm@osdl.org: fix alpha build] Signed-off-by: Jon Smirl <jonsmir@gmail.com> Signed-off-by: Antonino Daplas <adaplas@pol.net> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Luiz Fernando N. Capitulino <lcapitulino@mandriva.com.br> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Magnus Damm [Mon, 10 Jul 2006 11:44:09 +0000 (04:44 -0700)]
[PATCH] release_firmware() fixes
Use release_firmware() to free requested resources.
According to Documentation/firmware_class/README the request_firmware()
call should be followed by a release_firmware(). Some drivers do not
however free the firmware previously allocated with request_firmware().
This patch tries to fix this by making sure that release_firmware() is used
as expected.
Signed-off-by: Magnus Damm <magnus@valinux.co.jp> Acked-by: Marcel Holtmann <marcel@holtmann.org> Cc: Mauro Carvalho Chehab <mchehab@infradead.org> Cc: "John W. Linville" <linville@tuxdriver.com> Cc: Greg KH <greg@kroah.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Jonathan Corbet [Mon, 10 Jul 2006 11:44:07 +0000 (04:44 -0700)]
[PATCH] VFS documentation tweak
As I was looking over the get_sb() changes, I stumbled across a little
mistake in the documentation updates. Unless we're getting into an
interesting new object-oriented realm, I doubt that get_sb() should really
return "struct int"...
Signed-off-by: Jonathan Corbet <corbet@lwn.net> Acked-by: David Howells <dhowells@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
lockdep_map is embedded into every lock, which blows up data structure
sizes all around the kernel. Reduce the class-cache to be for the default
class only - that is used in 99.9% of the cases and even if we dont have a
class cached, the lookup in the class-hash is lockless.
This change reduces the per-lock dep_map overhead by 56 bytes on 64-bit
platforms and by 28 bytes on 32-bit platforms.
Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Arjan van de Ven [Mon, 10 Jul 2006 11:44:03 +0000 (04:44 -0700)]
[PATCH] lockdep: improve debug output
Make lockdep print which lock is held, in the "kfree() of a live lock"
scenario.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] partitions: let partitions inherit policy from disk
Change the partition code in fs/partitions/check.c to initialize a newly
detected partition's policy field with that of the containing block device
(see patch below).
My reasoning is that function set_disk_ro() in block/genhd.c modifies the
policy field (read-only indicator) of a disk and all contained partitions.
When a partition is detected after the call to set_disk_ro(), the policy
field of this partition will currently not inherit the disk's policy field.
This behavior poses a problem in cases where a block device can be
'logically de- and reactivated' like e.g. the s390 DASD driver because
partition detection may run after the policy field has been modified.
Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com> Acked-by: Al Viro <viro@ftp.linux.org.uk> Makes-sense-to: Jens Axboe <axboe@suse.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
When write() extends a file(i_size is increased) and fsync() is called,
change of inode must be written to journaling area through fsync().
But,currently the i_trans_id is not correctly updated when i_size is
increased. So fsync() does not kick the journal writer.
Reiserfs_file_write() already updates the transaction when blocks are
allocated, but the case when i_size increases and new blocks are not added
is not correctly treated.
Following patch fix this bug.
Signed-off-by: Hisashi Hifumi <hifumi.hisashi@oss.ntt.co.jp> Cc: Jeff Mahoney <jeffm@suse.com> Cc: Chris Mason <mason@suse.com> Cc: Hans Reiser <reiser@namesys.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Eric Paris [Mon, 10 Jul 2006 11:43:55 +0000 (04:43 -0700)]
[PATCH] SELinux: add rootcontext= option to label root inode when mounting
Introduce a new rootcontext= option to FS mounting. This option will allow
you to explicitly label the root inode of an FS being mounted before that
FS or inode because visible to userspace. This was found to be useful for
things like stateless linux, see
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=190001
Signed-off-by: Eric Paris <eparis@parisplace.org> Acked-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Eric Paris [Mon, 10 Jul 2006 11:43:53 +0000 (04:43 -0700)]
[PATCH] SELinux: decouple fscontext/context mount options
Remove the conflict between fscontext and context mount options. If
context= is specified without fscontext it will operate just as before, if
both are specified we will use mount point labeling and all inodes will get
the label specified by context=. The superblock will be labeled with the
label of fscontext=, thus affecting operations which check the superblock
security context, such as associate permissions.
Signed-off-by: Eric Paris <eparis@parisplace.org> Acked-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: James Morris <jmorris@namei.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Peter Williams [Mon, 10 Jul 2006 11:43:51 +0000 (04:43 -0700)]
[PATCH] sched: fix bug in __migrate_task()
Problem:
In the function __migrate_task(), deactivate_task() followed by
activate_task() is used to move the task from one run queue to
another. This has two undesirable effects:
1. The task's priority is recalculated. (Nowhere else in the
scheduler code is the priority recalculated for a change of CPU.)
2. The task's time stamp is set to the current time. At the very least,
this makes the adjustment of the time stamp before the call to
deactivate_task() redundant but I believe the problem is more serious
as the time stamp now holds the time of the queue change instead of
the time at which the task was woken. In addition, unless dest_rq is
the same queue as "current" is on the time stamp could be inaccurate
due to inter CPU drift.
Solution:
Replace the call to activate_task() with one to __activate_task().
Signed-off-by: Peter Williams <pwil3058@bigpond.net.au> Cc: Ingo Molnar <mingo@elte.hu> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Andrew Morton [Mon, 10 Jul 2006 11:43:48 +0000 (04:43 -0700)]
[PATCH] don't select CONFIG_HOTPLUG
It's useful to be able to turn off CONFIG_HOTPLUG for compile-coverage testing
and for section-checking coverage. But a few things go and select
CONFIG_HOTPLUG, making it a royal PITA to turn the thing off.
It's only turnable offable if CONFIG_EMBEDDED anyway. So let's make those
things depend on HOTPLUG, not select it.
[PATCH] i386: use thread_info flags for debug regs and IO bitmaps
Use thread info flags to track use of debug registers and IO bitmaps.
- add TIF_DEBUG to track when debug registers are active
- add TIF_IO_BITMAP to track when I/O bitmap is used
- modify __switch_to() to use the new TIF flags
Performance tested on Pentium II, ten runs of LMbench context switch
benchmark (smaller is better:)
Merge branch 'blktrace' of git://brick.kernel.dk/data/git/linux-2.6-block
* 'blktrace' of git://brick.kernel.dk/data/git/linux-2.6-block:
[PATCH] Only the first two bits in bio->bi_rw and rq->flags match
[PATCH] blktrace: readahead support
[PATCH] blktrace: fix barrier vs sync typo
...
CC net/atm/clip.o
net/atm/clip.c: In function ‘atm_clip_init’:
net/atm/clip.c:975: error: ‘atm_proc_root’ undeclared (first use in this function)
net/atm/clip.c:975: error: (Each undeclared identifier is reported only once
net/atm/clip.c:975: error: for each function it appears in.)
net/atm/clip.c:977: error: ‘arp_seq_fops’ undeclared (first use in this function)
make[2]: *** [net/atm/clip.o] Error 1
<-- snip -->
Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: David S. Miller <davem@davemloft.net>
i386: improve and correct inline asm memory constraints
Use "+m" rather than a combination of "=m" and "m" for improved clarity
and consistency.
This also fixes some inlines that incorrectly didn't tell the compiler
that they read the old value at all, potentially causing the compiler to
generate bogus code. It appear that all of those potential bugs were
hidden by the use of extra "volatile" specifiers on the data structures
in question, though.
Herbert Xu [Sat, 8 Jul 2006 20:34:56 +0000 (13:34 -0700)]
[NET] gso: Fix up GSO packets with broken checksums
Certain subsystems in the stack (e.g., netfilter) can break the partial
checksum on GSO packets. Until they're fixed, this patch allows this to
work by recomputing the partial checksums through the GSO mechanism.
Once they've all been converted to update the partial checksum instead of
clearing it, this workaround can be removed.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Sat, 8 Jul 2006 20:34:32 +0000 (13:34 -0700)]
[NET] gso: Add skb_is_gso
This patch adds the wrapper function skb_is_gso which can be used instead
of directly testing skb_shinfo(skb)->gso_size. This makes things a little
nicer and allows us to change the primary key for indicating whether an skb
is GSO (if we ever want to do that).
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
The Coverity checker spotted, that from the changes from commit 898b1d16f8230fb912a0c2248df685735c6ceda3 the
if (ret)
platform_driver_unregister(&ali_ircc_driver);
was dead code.
This patch changes this function to what seems to have been the
intention.
Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: David S. Miller <davem@davemloft.net>
[ATM]: fix possible recursive locking in skb_migrate()
ok this is a real potential deadlock in a way, it takes two locks of 2
skbuffs without doing any kind of lock ordering; I think the following
patch should fix it. Just sort the lock taking order by address of the
skb.. it's not pretty but it's the best this can do in a minimally
invasive way.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Chas Williams <chas@cmf.nrl.navy.mil> Signed-off-by: David S. Miller <davem@davemloft.net>
Matt LaPlante [Sat, 8 Jul 2006 20:30:09 +0000 (13:30 -0700)]
[ATM]: Typo in drivers/atm/Kconfig...
From: Matt LaPlante <kernel1@cyberdogtech.com> Signed-off-by: Charles Williams <chas@cmf.nrl.navy.mil> Signed-off-by: David S. Miller <davem@davemloft.net>
Add the AMD 8131 bridge to the list of chipsets that reorder writes.
Signed-off-by: John W. Linville <linville@tuxdriver.com> Acked-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Dave Jones [Fri, 7 Jul 2006 16:31:27 +0000 (12:31 -0400)]
[PATCH] Fix cpufreq vs hotplug lockdep recursion.
[ There's some not quite baked bits in cpufreq-git right now
so sending this on as a patch instead ]
On Thu, 2006-07-06 at 07:58 -0700, Tom London wrote:
> After installing .2356 I get this each time I boot:
> =======================================================
> [ INFO: possible circular locking dependency detected ]
> -------------------------------------------------------
> S06cpuspeed/1620 is trying to acquire lock:
> (dbs_mutex){--..}, at: [<c060d6bb>] mutex_lock+0x21/0x24
>
> but task is already holding lock:
> (cpucontrol){--..}, at: [<c060d6bb>] mutex_lock+0x21/0x24
>
> which lock already depends on the new lock.
>
make sure the cpu hotplug recursive mutex (yuck) is taken early in the
cpufreq codepaths to avoid a AB-BA deadlock.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Dave Jones <davej@redhat.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Thomas Graf [Thu, 6 Jul 2006 03:47:28 +0000 (20:47 -0700)]
[PKT_SCHED]: Fix error handling while dumping actions
"return -err" and blindly inheriting the error code in the netlink
failure exception handler causes errors codes to be returned as
positive value therefore making them being ignored by the caller.
May lead to sending out incomplete netlink messages.
Signed-off-by: Thomas Graf <tgraf@suug.ch> Signed-off-by: David S. Miller <davem@davemloft.net>
[SPARC64]: Fix stack overflow checking in modular non-SMP kernels.
The sparc64 kernel's EXPORT_SYMBOL(_mcount) is inside an
#ifdef CONFIG_SMP. This breaks modules in non-SMP kernels
built with stack overflow checking (CONFIG_STACK_DEBUG=y),
as modules_install reports:
WARNING: /lib/modules/2.6.17/kernel/drivers/ide/ide-cd.ko needs unknown symbol _mcount
Trivially fixed by moving EXPORT_SYMBOL(_mcount) outside of
the #ifdef CONFIG_SMP.
Signed-off-by: Mikael Pettersson <mikpe@it.uu.se> Signed-off-by: David S. Miller <davem@davemloft.net>
[PATCH] ahci: Ensure that we don't grab both functions
When we force the chip into dual fn mode so we get PATA and AHCI we must
be sure we don't then do anything dumb like try and grab both with the AHCI
driver.
Signed-off-by: Alan Cox <alan@redhat.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Reimplement controller-wide PM. ata_host_set_suspend/resume() are
defined to suspend and resume a host_set. While suspended, EHs for
all ports in the host_set are pegged using ATA_FLAG_SUSPENDED and
frozen.
Because SCSI device hotplug is done asynchronously against the rest of
libata EH and the same mutex is used when adding new device, suspend
cannot wait for hotplug to complete. So, if SCSI device hotplug is in
progress, suspend fails with -EBUSY.
In most cases, host_set resume is followed by device resume. As each
resume operation requires a reset, a single host_set-wide resume
operation may result in multiple resets. To avoid this, resume waits
upto 1 second giving PM to request resume for devices.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Reimplement per-dev PM. The original implementation directly put the
device into suspended mode and didn't synchronize w/ EH operations
including hotplug. This patch reimplements ata_scsi_device_suspend()
and ata_scsi_device_resume() such that they request EH to perform the
respective operations. Both functions synchronize with hotplug such
that it doesn't operate on detached devices.
Suspend waits for completion but resume just issues request and
returns. This allows parallel wake up of devices and thus speeds up
system resume.
Due to sdev detach synchronization, it's not feasible to separate out
EH requesting from sdev handling; thus, ata_device_suspend/resume()
are removed and everything is implemented in the respective
libata-scsi functions.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Implement two PM per-dev EH actions - ATA_EH_SUSPEND and
ATA_EH_RESUME. Each action puts the target device into suspended mode
and resumes from it respectively.
Once a device is put to suspended mode, no EH operations other than
RESUME is allowed on the device. The device will stay suspended till
it gets resumed and thus reset and revalidated. To implement this, a
new device state helper - ata_dev_ready() - is implemented and used in
EH action implementations to make them operate only on attached &
running devices.
If all possible devices on a port are suspended, reset is skipped too.
This prevents spurious events including hotplug events from disrupting
suspended devices.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
[PATCH] libata: separate out __ata_ehi_hotplugged()
Separate out __ata_ehi_hotplugged() from ata_ehi_hotplugged(). The
underscored version doesn't set AC_ERR_ATA_BUS. This will be used for
resume which is a hotplug event but not an ATA bus error.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
[PATCH] libata: implement ATA_EHI_NO_AUTOPSY and QUIET
Implement ATA_EHI_NO_AUTOPSY and QUIET. These used to be implied by
ATA_PFLAG_LOADING, but new power management and PMP support need to
use these separately. e.g. Suspend/resume operations shouldn't print
full EH messages and resume shouldn't be recorded as an error.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
[PATCH] libata: clean up debounce parameters and improve parameter selection
The names of predefined debounce timing parameters didn't exactly
match their usages. Rename to more generic names and implement param
selection helper sata_ehc_deb_timing() which uses EHI_HOTPLUGGED to
select params.
Combined with the previous EHI_RESUME_LINK differentiation, this makes
parameter selection accurate. e.g. user scan resumes link but normal
deb param is used instead of hotplug param.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Implement ATA_EHI_RESUME_LINK, which indicates that the link needs to
be resumed. This used to be implied by ATA_EHI_HOTPLUGGED. However,
hotplug isn't the only event which requires link resume and separating
this out allows other places to request link resume. This
differentiation also allows better debounce timing selection.
This patch converts user scan to use ATA_EHI_RESUME_LINK.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
[PATCH] libata: replace ap_lock w/ ap->lock in ata_scsi_error()
ap_lock was used because &ap->host_set->lock was too long and used a
lot. Now that &ap->host_set->lock is replaced with ap->lock, there's
no reason to keep ap_lock.
[ed. note: that's not entirely true. ap_lock is a local variable,
caching the results of a de-ref. In theory, if the compiler is smart
enough, this patch is cosmetic. However, since this is not a fast
path (it is the error path), this patch is nonetheless acceptable,
even though it _may_ introduce a performance regression.]
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
[PATCH] libata: fix ehc->i.action setting in ata_eh_autopsy()
ata_eh_autopsy() used to directly assign determined action mask to
ehc->i.action thus overriding actions set by some of nested analyze
functions. This patch makes ata_eh_autopsy() add action masks just as
it's done in other places.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Tejun Heo [Wed, 28 Jun 2006 16:29:30 +0000 (01:29 +0900)]
[PATCH] libata: add ap->pflags and move core dynamic flags to it
ap->flags is way too clamped. Separate out core dynamic flags to
ap->pflags. ATA_FLAG_DISABLED is a dynamic flag but left alone as
it's referenced by a lot of LLDs and it's gonna be removed once all
LLDs are converted to new EH.
Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
* git://git.linux-nfs.org/pub/linux/nfs-2.6:
NLM,NFSv4: Wait on local locks before we put RPC calls on the wire
VFS: Add support for the FL_ACCESS flag to flock_lock_file()
NFSv4: Ensure nfs4_lock_expired() caches delegated locks
NLM,NFSv4: Don't put UNLOCK requests on the wire unless we hold a lock
VFS: Allow caller to determine if BSD or posix locks were actually freed
NFS: Optimise away an excessive GETATTR call when a file is symlinked
This fixes a panic doing the first READDIR or READDIRPLUS call when:
NFS: Fix NFS page_state usage
Revert "Merge branch 'odirect'"
Merge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6
* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6: (39 commits)
[PATCH] myri10ge - Export more parameters to ethtool
[PATCH] myri10ge - Use dev_info() when printing parameters after probe
[PATCH] myri10ge - Drop ununsed nvidia chipset id
[PATCH] myri10ge - Drop unused pm_state
[PATCH] Fix freeing of net device
[PATCH] remove dead entry in net wan Kconfig
[PATCH] NI5010 netcard cleanup
[PATCH] lock validator: fix ns83820.c irq-flags bug
[PATCH] pcnet32: Cleanup rx buffers after loopback test.
[PATCH] pcnet32: Suspend the chip rather than restart when changing multicast/promisc
[PATCH] pcnet32: Handle memory allocation failures cleanly when resizing tx/rx rings
[PATCH] pcnet32: Use kcalloc instead of kmalloc and memset
[PATCH] pcnet32: Fix off-by-one in get_ringparam
[PATCH] pcnet32: Use PCI_DEVICE macro
[PATCH] pcnet32: Fix Section mismatch error
[PATCH] Add support for the Cicada 8201 PHY
[PATCH] zd1211rw: disable TX queue during stop
[PATCH] ZyDAS ZD1211 USB-WLAN driver
[PATCH] softmac: fix build-break from 881ee6999d66c8fc903b429b73bbe6045b38c549
[PATCH] CONFIG_WIRELESS_EXT is neccessary after all
...
So netpoll_send_skb takes the _xmit_lock, which is all nitty gritty
but then rtl8139_start_xmit comes around while that lock is taken, and
does
spin_unlock_irq(&tp->lock);
which.. enables interrupts and softirqs; this is quite bad because the
xmit lock is taken in softirq context for the watchdog like this:
[<c1200376>] _spin_lock+0x23/0x32
[<c11af282>] dev_watchdog+0x14/0xb1
[<c101dab2>] run_timer_softirq+0xf2/0x14a
[<c101a691>] __do_softirq+0x55/0xb0
[<c1004a8d>] do_softirq+0x58/0xbd
Which would deadlock now that the spin_unlock_irq() has enabled
irqs/softirqs while the _xmit_lock is still held.
The patch below turns this into a irqsave/irqrestore pair so that
interrupts don't get enabled unconditionally.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Jeff Garzik <jeff@garzik.org>
[PATCH] myri10ge - Use dev_info() when printing parameters after probe
Displaying the interface name when listing the device parameters
at the end of myri10ge_probe is not a good idea since udev might
rename the interface soon afterwards.
Print the bus id instead, using dev_info().
Signed-off-by: Brice Goglin <brice@myri.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Paul Fulghum [Fri, 30 Jun 2006 09:27:19 +0000 (02:27 -0700)]
[PATCH] remove dead entry in net wan Kconfig
Remove dead entry from net wan Kconfig and net wan Makefile.. This entry is
left over from 2.4 where synclink used syncppp driver directly. synclink
drivers now use generic HDLC
Signed-off-by: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
Andreas Mohr [Fri, 30 Jun 2006 09:25:07 +0000 (02:25 -0700)]
[PATCH] NI5010 netcard cleanup
- updated MAINTAINERS entry to new format
- updated Jan-Pascal's (ACKed) and my email address
- driver cleanup/modernization (runtime-, not hardware-tested)
[bunk@stusta.de: build fix] Signed-off-by: Andreas Mohr <andi@lisas.de> Cc: Jeff Garzik <jeff@garzik.org> Cc: Jan-Pascal van Best <jvbest@qv3pluto.leidenuniv.nl> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
this is caused because the ns83820 driver re-enables irq flags
in hardirq context.
While legal in theory, in practice it should only be done if the
hardware is really old and has some very high overhead in its ISR.
(such as PIO IDE)
For modern hardware, running ISRs with irqs enabled is discouraged,
because 1) new hardware is fast enough to not cause latency problems
2) allowing the nesting of hardware interrupts only 'spreads out'
the handling of the current ISR, causing extra cachemisses that would
otherwise not happen. Furthermore, on architectures where ISRs share
the kernel stacks, enabling interrupts in ISRs introduces a much
higher kernel-stack-nesting and thus kernel-stack-overflow risk.
3) not managing irq-flags via the _irqsave / _irqrestore variants
is dangerous: it's easy to forget whether one function nests inside
another, and irq flags might be mismanaged.
In the few cases where re-enabling interrupts in an ISR is considered
useful (and unavoidable), it has to be taught to the lock validator
explicitly (because the lock validator needs the "no ISR ever enables
hardirqs" artificial simplification to keep the IRQ/softirq locking
dependencies manageable).
This teaching is done via the explicit use local_irq_enable_in_hardirq().
On a stock kernel this maps to local_irq_enable(). If the lock validator
is enabled then this does not enable interrupts.
Now, the analysis of drivers/net/ns83820.c's irq flags use: the
irq-enabling in irq context seems intentional, but i dont think it's
justified. Furthermore, the driver suffers from problem #3 above too,
in ns83820_tx_timeout() it disables irqs via local_irq_save(), but
then it calls do_tx_done() which does a spin_unlock_irq(),
re-enabling for a function that does not expect it! While currently
this bug seems harmless (only some debug printout seems to be
affected by it), it's nevertheless something to be fixed.
So this patch makes the ns83820 ISR irq-flags-safe, and cleans up
do_tx_done() use and locking to avoid the ns83820_tx_timeout() bug.
From: Arjan van de Ven <arjan@linux.intel.com>
ns83820_mib_isr takes the misc_lock in IRQ context. All other places that
do this in the ISR already use _irqsave versions, make this consistent at
least. At some point in the future someone should audit the driver to see
if all _irqsave's in the ISR can go away, this is generally an iffy/fragile
proposition though; for now get it safe, simple and consistent.
From: Arjan van de Ven <arjan@linux.intel.com>
ok this is a real driver deadlock:
The ns83820 driver enabled interrupts (by unlocking the misc_lock with
_irq) while still holding the rx_info.lock, which is required to be irq
safe since it's used in the ISR like this:
writel(1, dev->base + IER);
spin_unlock_irq(&dev->misc_lock);
kick_rx(ndev);
spin_unlock_irq(&dev->rx_info.lock);
This is can cause a deadlock if an irq was pending at the first
spin_unlock_irq already, or if one would hit during kick_rx().
Simply remove the first _irq solves this
Signed-off-by: Ingo Molnar <mingo@elte.hu> Cc: Benjamin LaHaise <bcrl@kvack.org> Cc: Jeff Garzik <jeff@garzik.org> Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
projects / linux-beck.git / log