Currently we schedule to the leftmost task in the runqueue. When the
runtimes are very short because of some server/client ping-pong,
especially in over-saturated workloads, this will cycle through all
tasks trashing the cache.
Reduce cache trashing by keeping dependent tasks together by running
newly woken tasks first. However, by not running the leftmost task first
we could starve tasks because the wakee can gain unlimited runtime.
Therefore we only run the wakee if its within a small
(wakeup_granularity) window of the leftmost task. This preserves
fairness, but does alternate server/client task groups.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Clear the cached inverse value when updating load. This is needed for
calc_delta_mine() to work correctly when using the rq load.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Current min_vruntime tracking is incorrect and will cause serious
problems when we don't run the leftmost task for some reason.
min_vruntime does two things; 1) it's used to determine a forward
direction when the u64 vruntime wraps, 2) it's used to track the
leftmost vruntime to position newly enqueued tasks from.
The current logic advances min_vruntime whenever the current task's
vruntime advance. Because the current task may pass the leftmost task
still waiting we're failing the second goal. This causes new tasks to be
placed too far ahead and thus penalizes their runtime.
Fix this by making min_vruntime the min_vruntime of the waiting tasks by
tracking it in enqueue/dequeue, and compare against current's vruntime
to obtain the absolute minimum when placing new tasks.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Fix a hard to trigger crash seen in the -rt kernel that also affects
the vanilla scheduler.
There is a race condition between schedule() and some dequeue/enqueue
functions; rt_mutex_setprio(), __setscheduler() and sched_move_task().
When scheduling to idle, idle_balance() is called to pull tasks from
other busy processor. It might drop the rq lock. It means that those 3
functions encounter on_rq=0 and running=1. The current task should be
put when running.
Here is a possible scenario:
CPU0 CPU1
| schedule()
| ->deactivate_task()
| ->idle_balance()
| -->load_balance_newidle()
rt_mutex_setprio() |
| --->double_lock_balance()
*get lock *rel lock
* on_rq=0, ruuning=1 |
* sched_class is changed |
*rel lock *get lock
: |
:
->put_prev_task_rt()
->pick_next_task_fair()
=> panic
The current process of CPU1(P1) is scheduling. Deactivated P1, and the
scheduler looks for another process on other CPU's runqueue because CPU1
will be idle. idle_balance(), load_balance_newidle() and
double_lock_balance() are called and double_lock_balance() could drop
the rq lock. On the other hand, CPU0 is trying to boost the priority of
P1. The result of boosting only P1's prio and sched_class are changed to
RT. The sched entities of P1 and P1's group are never put. It makes
cfs_rq invalid, because the cfs_rq has curr and no leaf, but
pick_next_task_fair() is called, then the kernel panics.
Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This bug was always here, but before my commit 6fa02839bf
("recheck for secure ports in fh_verify"), it could only be triggered by
failure of a kmalloc(). After that commit it could be triggered by a
client making a request from a non-reserved port for access to an export
marked "secure". (Exports are "secure" by default.)
The result is a struct svc_export with a reference count one too low,
resulting in likely oopses next time the export is accessed.
The reference counting here is not straightforward; a later patch will
clean up fh_verify().
Thanks to Lukas Hejtmanek for the bug report and followup.
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Cc: Lukas Hejtmanek <xhejtman@ics.muni.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The comments in the definition of struct export_operations don't match the
current members.
Add a comment for the 2 new functions and remove 2 comments for unused ones.
Signed-off-by: Marc Dionne <marc.c.dionne@gmail.com>
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Shut up "may be used uninitialised in this function" warnings due to
PPC32's implementation of dma_alloc_coherent().
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Currently, we do nothing to guarantee we have a consistent DMA buffer for
asynchronous receive packets. Rather than doing several sync's following a
dma_map_single() to get consistent buffers, just switch to using
dma_alloc_coherent().
Resolves constant buffer failures on my own x86_64 laptop w/4GB of RAM and
likely to fix a number of other failures witnessed on x86_64 systems with
4GB of RAM or more.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Fix I/O errors due to SYM13FW500's inability to handle larger request
sizes. Reported by Piergiorgio Sartor <piergiorgio.sartor@nexgo.de> for
firewire-sbp2 in https://bugzilla.redhat.com/show_bug.cgi?id=436879
This fix is necessary because sbp2's default request size limit has been
lifted since 2.6.25-rc1.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Fix I/O errors due to SYM13FW500's inability to handle larger request
sizes. Reported by Piergiorgio Sartor <piergiorgio.sartor@nexgo.de> in
https://bugzilla.redhat.com/show_bug.cgi?id=436879
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Remove some less necessary information, point out that video1394 and
dv1394 should be blacklisted along with ohci1394.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Per the SBP-2 specification, all SBP-2 target devices must have a BUSY_TIMEOUT
register. Per the 1394-1995 specification, the retry_limt portion of the
register should be set to 0x0 initially, and set on the target by a logged in
initiator (i.e., a Linux host w/firewire controller(s)).
Well, as it turns out, lots of devices these days have actually moved on to
starting to implement SBP-3 compliance, which says that retry_limit should
default to 0xf instead (yes, SBP-3 stomps directly on 1394-1995, oops).
Prior to this change, the firewire driver stack didn't touch retry_limit, and
any SBP-3 compliant device worked fine, while SBP-2 compliant ones were unable
to retransmit when the host returned an ack_busy_X, which resulted in stalled
out I/O, eventually causing the SCSI layer to give up and offline the device.
The simple fix is for us to set retry_limit to 0xf in the register for all
devices (which actually matches what the old ieee1394 stack did).
Prior to this change, a hard disk behind an SBP-2 Prolific PL-3507 bridge chip
would routinely encounter buffer I/O errors and wind up offlined by the SCSI
layer. With this change, I've encountered zero I/O failures moving tens of GB
of data around.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Mostly copied from ohci1394.c. Necessary for some older Macs, e.g.
PowerBook G3 Pismo and early PowerBook G4 Titanium.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Copied from ohci1394.c. This code is necessary to prevent machine check
exceptions when reloading or resuming the driver.
Tested on a 1st generation PowerBook G4 Titanium, which also needs the
pci_probe() hunk.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
I was able to reproduce the system exception on resume with a 3rd-gen
Titanium PowerBook G4 667, and this patch does let the system resume
successfully now.
Not quite clear if there was possibly an updated version coming using
pci_enable_device() instead of the pair of pmac_call_feature() calls,
but either way, this is a definite must-have, at least for older ppc
macs -- my Aluminum PowerBook G4/1.67 suspends and resumes without this
patch just fine.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Kills warnings from 'make C=1 CHECKFLAGS="-D__CHECK_ENDIAN__" modules':
drivers/firewire/fw-transaction.c:771:10: warning: incorrect type in assignment (different base types)
drivers/firewire/fw-transaction.c:771:10: expected unsigned int [unsigned] [usertype] <noident>
drivers/firewire/fw-transaction.c:771:10: got restricted unsigned int [usertype] <noident>
drivers/firewire/fw-transaction.h:93:10: warning: incorrect type in assignment (different base types)
drivers/firewire/fw-transaction.h:93:10: expected unsigned int [unsigned] [usertype] <noident>
drivers/firewire/fw-transaction.h:93:10: got restricted unsigned int [usertype] <noident>
drivers/firewire/fw-ohci.c:1490:8: warning: restricted degrades to integer
drivers/firewire/fw-ohci.c:1490:35: warning: restricted degrades to integer
drivers/firewire/fw-ohci.c:1516:5: warning: cast to restricted type
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
The generation of incoming requests was filled in in wrong byte order on
machines with big endian CPU.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
IPoIB: Allocate priv->tx_ring with vmalloc()
IPoIB/cm: Set tx_wr.num_sge in connected mode post_send()
IPoIB: Don't drop multicast sends when they can be queued
IB/ipath: Reset the retry counter for RDMA_READ_RESPONSE_MIDDLE packets
IB/ipath: Fix error completion put on send CQ instead of recv CQ
IB/ipath: Fix RC QP initialization
IB/ipath: Fix potentially wrong RNR retry counter returned in ipath_query_qp()
IB/ipath: Fix IB compliance problems with link state vs physical state
Commit 681cc5cd3e ("iommu sg merging:
swiotlb: respect the segment boundary limits") introduced two
possibilities for entering an endless loop in lib/swiotlb.c:
- if max_slots is zero (possible if mask is ~0UL)
- if the number of slots requested fits into a swiotlb segment, but is
too large for the part of a segment which remains after considering
offset_slots
This fixes them
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (32 commits)
ACPI: thermal: show temperature in millidegree Celsius
thermal: fix generic thermal I/F for hwmon
acer-wmi: build depends on i8042
documentation: Move power-related files to Documentation/power/
ACPI: buffer array too short in drivers/acpi/system.c
acer-wmi: Add DMI quirk for mail LED support on Acer Aspire 3610/ 5610
acer-wmi: Fix DSDT path in documentation
acer-wmi: Make device detection error messages more descriptive
laptops: move laptop-mode.txt to Documentation/laptops/
ACPICA: Warn if packages with invalid references are evaluated
ACPI: add _PRT quirks to work around broken firmware
Hibernation: Fix mark_nosave_pages()
ACPI: Ignore _BQC object when registering backlight device
ACPI: WMI: Clean up handling of spec violating data blocks
acer-wmi: Don't warn if mail LED cannot be detected
acer-wmi: Rename mail LED correctly & remove hardcoded colour
ACPI: use ACPI_DEBUG_PRINT instead of printk in acpi_processor_hotplug_notify()
ACPI: button: make real parent for input devices in device tree
toshiba_acpi: Enable autoloading
ACPI: EC: Handle IRQ storm on Acer laptops
...
Floppy rmmod locks up when no such hardware was initialized, since there is
nobody to wake the remove code up. Remove the completion, because release is
called during platform_unregister anyway.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Honor the environment variable "KBUILD_VERBOSE=1" (as set by make V=1) to
enable verbose mode in scripts/kernel-doc. Useful for getting more info and
warnings from kernel-doc.
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
iommu_is_span_boundary in lib/iommu-helper.c was exported for PARISC IOMMUs
(commit 3715863aa1). Alpha's IOMMU can use it.
This removes the check on the boundary size alignment because
iommu_is_span_boundary does.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Richard Henderson <rth@twiddle.net>
Acked-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In riscom8_init_module, rc_board should be indexed by i, not by 0, otherwise
the loop is useless.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
SPI controller drivers return number of bytes actually transfered from
bitbang->txrx_bufs() method. This updates handling of short transfers (where
the transfer size is less than requested):
- Even zero byte short transfers should report errors;
- Include short transfers in the total of transferred bytes;
- Use EREMOTEIO (like USB) not EMSGSIZE to report short transfers
Short transfers don't normally mean invalid message sizes, but if the
underlying controller driver needs to use EMSGSIZE it can still do so.
[db: fix two more minor issues]
Signed-off-by: Jan Nikitenko <jan.nikitenko@gmail.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Remove struct pt_regs *regs from prototype of au1550_spi_irq handler to avoid
warning in request_irq(). Also fix comment type leadin.
Signed-off-by: Jan Nikitenko <jan.nikitenko@gmail.com>
Signed-off-by: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Al Viro wrote:
>
> After that commit in asm-h8300/uaccess.h we have
>
> #define get_user(x, ptr) \
> ({ \
> int __gu_err = 0; \
> uint32_t __gu_val = 0; \
> ^^^^^^^^^^^^^^^^^
> switch (sizeof(*(ptr))) { \
> case 1: \
> case 2: \
> case 4: \
> __gu_val = *(ptr); \
> break; \
> case 8: \
> memcpy(&__gu_val, ptr, sizeof (*(ptr))); \
> ^^^^^^^^^^^^^^^^
>
> which, of course, is FUBAR whenever we actually hit that case - memcpy of
> 8 bytes into uint32_t is obviously wrong. Why don't we simply do
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Smackfs write() implementation does not put a higher bound on the number of
bytes to copy from user-space. This may lead to a DOS attack if a malicious
`count' field is given.
Assure that given `count' is exactly the length needed for a /smack/load rule.
In case of /smack/cipso where the length is relative, assure that `count'
does not exceed the size needed for a buffer representing maximum possible
number of CIPSO 2.2 categories.
Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Acked-by: Casey Schaufler <casey@schaufler-ca.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
__FUNCTION__ is gcc-specific, use __func__
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The patch 'fsldma: do not cleanup descriptors in hardirq context'
(commit 222ccf9ab8) removed descriptors
cleanup function to tasklet but the completed cookie do not updated.
Thus, the DMA controller will get lots of duplicated transfer
interrupts. Just make a completed cookie update in interrupt handler.
And keep other cleanup jobs in tasklet function.
Tested-by: Sebastian Siewior <bigeasy@linutronix.de>
Signed-off-by: Zhang Wei <wei.zhang@freescale.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
This is a bug that I assigned DMA_INTERRUPT capability to fsldma
but missing device_prep_dma_interrupt function. For a bug in
dmaengine.c the driver passed BUG_ON() checking. The patch fixes it.
Signed-off-by: Zhang Wei <wei.zhang@freescale.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The device->device_prep_dma_interrupt function is used by
DMA_INTERRUPT capability, not DMA_ZERO_SUM.
Signed-off-by: Zhang Wei <wei.zhang@freescale.com>
Acked-by: Maciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
There are warning messages reported by Stephen Rothwell with
ARCH=powerpc allmodconfig build:
drivers/dma/fsldma.c: In function 'fsl_dma_prep_memcpy':
drivers/dma/fsldma.c:439: warning: comparison of distinct pointer types
lacks a cast
drivers/dma/fsldma.c: In function 'fsl_chan_xfer_ld_queue':
drivers/dma/fsldma.c:584: warning: format '%016llx' expects type 'long long
unsigned int', but argument 4 has type 'dma_addr_t'
drivers/dma/fsldma.c: In function 'fsl_dma_chan_do_interrupt':
drivers/dma/fsldma.c:668: warning: format '%x' expects type 'unsigned int',
but argument 5 has type 'dma_addr_t'
drivers/dma/fsldma.c:684: warning: format '%016llx' expects type 'long long
unsigned int', but argument 4 has type 'dma_addr_t'
drivers/dma/fsldma.c:684: warning: format '%016llx' expects type 'long long
unsigned int', but argument 5 has type 'dma_addr_t'
drivers/dma/fsldma.c:701: warning: format '%02x' expects type 'unsigned
int', but argument 4 has type 'dma_addr_t'
drivers/dma/fsldma.c: In function 'fsl_dma_self_test':
drivers/dma/fsldma.c:840: warning: format '%d' expects type 'int', but
argument 5 has type 'size_t'
drivers/dma/fsldma.c: In function 'of_fsl_dma_probe':
drivers/dma/fsldma.c:1010: warning: format '%08x' expects type 'unsigned
int', but argument 5 has type 'resource_size_t'
This patch fixed the above warning messages.
Signed-off-by: Zhang Wei <wei.zhang@freescale.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
PCI busses can be registered multiple times, so we need to detect if we
have registered our bus structure in sysfs already. If so, don't do it
again.
Thanks to Guennadi Liakhovetski <g.liakhovetski@gmx.de> for reporting
the problem, and to Linus for poking me to get me to believe that it was
a real problem.
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Comparing with kernel 2.6.24, tbench result has regression with
2.6.25-rc1.
1) On 2 quad-core processor stoakley: 4%.
2) On 4 quad-core processor tigerton: more than 30%.
bisect located below patch.
b4ce92775c is first bad commit
commit b4ce92775c
Author: Herbert Xu <herbert@gondor.apana.org.au>
Date: Tue Nov 13 21:33:32 2007 -0800
[IPV6]: Move nfheader_len into rt6_info
The dst member nfheader_len is only used by IPv6. It's also currently
creating a rather ugly alignment hole in struct dst. Therefore this patch
moves it from there into struct rt6_info.
Above patch changes the cache line alignment, especially member
__refcnt. I did a testing by adding 2 unsigned long pading before
lastuse, so the 3 members, lastuse/__refcnt/__use, are moved to next
cache line. The performance is recovered.
I created a patch to rearrange the members in struct dst_entry.
With Eric and Valdis Kletnieks's suggestion, I made finer arrangement.
1) Move tclassid under ops in case CONFIG_NET_CLS_ROUTE=y. So
sizeof(dst_entry)=200 no matter if CONFIG_NET_CLS_ROUTE=y/n. I
tested many patches on my 16-core tigerton by moving tclassid to
different place. It looks like tclassid could also have impact on
performance. If moving tclassid before metrics, or just don't move
tclassid, the performance isn't good. So I move it behind metrics.
2) Add comments before __refcnt.
On 16-core tigerton:
If CONFIG_NET_CLS_ROUTE=y, the result with below patch is about 18%
better than the one without the patch;
If CONFIG_NET_CLS_ROUTE=n, the result with below patch is about 30%
better than the one without the patch.
With 32bit 2.6.25-rc1 on 8-core stoakley, the new patch doesn't
introduce regression.
Thank Eric, Valdis, and David!
Signed-off-by: Zhang Yanmin <yanmin.zhang@intel.com>
Acked-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
as now required by the generic thermal I/F
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Carlos Corbacho <carlos@strangeworlds.co.uk>
Signed-off-by: Len Brown <len.brown@intel.com>
Peter Zijlstra