sbp2_host_reset and sbp2_handle_status_write are not serialized against
sbp2_alloc_device and sbp2_remove_device.
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
x86 uses target specific assignment of EXTRA_AFLAGS,
EXTRA_CFLAGS - this caused troubles with
introducing asflags-y, ccflags-y.
Fixed the target specific assignments in arch/x86/boot/Makefile
and auditted the rest of the kernel for similar usage.
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
This belongs to user-space (and only if really needed).
text data bss dec hex filename
6723 2000 28 8751 222f drivers/ide/pci/amd74xx.o.before
3833 2000 16 5849 16d9 drivers/ide/pci/amd74xx.o.after
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
drive->dn is initialized by ide-probe.c::probe_hwif() so no need to do it
in ->init_hwif method.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
This belongs to user-space (and only if really needed).
text data bss dec hex filename
7129 404 8 7541 1d75 drivers/ide/pci/sis5513.o.before
3916 404 1 4321 10e1 drivers/ide/pci/sis5513.o.after
Additionaly to being bloat the code contained two bugs:
- wrong cable bit was checked (0x0001 instead of 0x8000) on ATA_133 chipsets
- incorrect UDMA cycle time was reported on ATA_100a/ATA_133 chipsets
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
ATAPI devices with broken DMA support should be handled by DMA blacklist
and for debugging purposes we have new shiny "hdx=nodma" kernel paremeter.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* Add "hdx=nodma" option allowing user to disallow DMA for a given device.
* Obsolete "ide=nodma" option.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* hpt34x.c: disable DMA masks for HPT345
(hwif->autodma is zero so DMA won't be enabled anyway).
* trm290.c: disable IDE_HFLAG_TRUST_BIOS_FOR_DMA flag
(hwif->autodma is zero so DMA won't be enabled anyway).
* Check noautodma global variable instead of drive->autodma in ide_tune_dma().
This fixes handling of "ide=nodma" kernel parameter for icside, ide-cris,
au1xxx-ide, pmac, it821x, jmicron, sgiioc4 and siimage host drivers.
* Remove hwif->autodma (it was not checked by IDE core code anyway) and
drive->autodma (was set by all host drivers - except HPT345/TRM290 special
cases - unless "ide=nodma" was used).
While at it:
- remove needless printk() from icside.c
- remove stale FIXME/comment from ide-probe.c
- don't force DMA off if PCI bus-mastering had to be enabled in setup-pci.c
(this setting was always later over-ridden by host drivers anyway)
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
It has been obsoleted since 1 Nov 2004, is valid only when used together with
"ide=nodma" kernel parameter (which makes no sense) and only for aec62xx,
cmd64x, cs5220, generic, hpt366, sl82c105, tc86c001 and triflex host drivers.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
This code doesn't use pci_assign_resource() and ide_get_or_set_dma_base() uses
pci_resource_start() to get DMA base address so it has no chance of working.
Moreover this belongs to PCI quirks (in case somebody would like to fix it).
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
While at it:
- make struct pci_device_id tables const
- use PCI_DEVICE_ID_ITE_8213 define in it8213.c
- fix comment in generic.c
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Clear prefetch and postwrite for ATAPI devices instead of depending on BIOS.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Don't force PIO4 when programming UDMA mode (suggested by Sergei).
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
noautodma can now be unexported.
Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* Add IDE_HFLAG_TRUST_BIOS_FOR_DMA host flag for host drivers that depend
on BIOS for programming device/controller for DMA. Set it in cy82c693,
generic, ns87415, opti621 and trm290 host drivers.
* Add IDE_HFLAG_VDMA host flag for host drivers using VDMA. Set it in cs5520
host driver.
* Teach ide_tune_dma() about IDE_HFLAG_TRUST_BIOS_FOR_DMA flag.
* Add generic ide_dma_check() helper and remove all open coded ->ide_dma_check
implementations. Fix all places checking for presence of ->ide_dma_check
hook to check for ->ide_dma_on instead.
* Remove no longer needed code from config_drive_for_dma().
* Make ide_tune_dma() static.
v2:
* Fix config_drive_for_dma() return values.
* Fix ide-dma.c build for CONFIG_BLK_DEV_IDEDMA_PCI=n by adding
dummy config_drive_for_dma() inline.
* Fix IDE_HFLAG_TRUST_BIOS_FOR_DMA handling in ide_dma_check().
* Fix init_hwif_it8213() comment.
There should be no functionality changes caused by this patch.
Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Since ide-pmac allows separate PIO and DMA timings, and ide-pmac always sets
->autotune this change shouldn't have any other effect than bringing ide-pmac
driver in-line with other host drivers wrt ->ide_dma_check implementation.
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Since ide-cris allows separate PIO and DMA timings, and ide-cris always sets
->autotune this change shouldn't have any other effect than bringing ide-cris
driver in-line with other host drivers wrt ->ide_dma_check implementation.
Cc: Mikael Starvik <starvik@axis.com>
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Since sl82c105 re-programs DMA/PIO timings for DMA enable/disable, and sets
->autotune this change shouldn't have any other effect than bringing sl82c105
driver in-line with other host drivers wrt ->ide_dma_check implementation.
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* Add PIO autotune fallback to ->ide_dma_check.
Since cs5530/sc1200 allow separate PIO and DMA timings, and cs5530/sc1200
always set ->autotune this change shouldn't have any other effect than
bringing these drivers in-line with other host drivers wrt ->ide_dma_check
implementations.
* Fix ->ide_dma_check to return "-1" if DMA cannot be set.
* Bump driver version.
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Remove ide_use_fast_pio() and just re-tune PIO unconditionally if DMA tuning
has failed in ->ide_dma_check. All host drivers using ide_use_fast_pio() set
drive->autotune so PIO is always tuned anyway and in some cases we _really_
need to re-tune PIO because PIO and DMA timings are shared.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Several host drivers used to reset drive->init_speed in their ->ide_dma_check
implementations which resulted in incorrect init speed being reported to the
user, fix it.
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* Remove needless setting of drive->using_dma from auide_dma_check().
* Split off auide_mdma_filter() from auide_dma_check().
* Use ide_tune_dma() in auide_dma_check(), this fixes following issues:
- device's DMA capability bit not being checked
- device not being checked against generic DMA blacklist
- transfer mode not being set on device/host
* Add PIO autotune fallback to auide_dma_check().
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* Fix handling of unsupported/invalid modes in atiixp_set_dma_mode().
* ATI controllers allow separate PIO and DMA timings so remove programming
of PIO modes from atiixp_set_dma_mode(). Also remove no longer needed
atiixp_dma_2_pio() helper.
* SWDMA timings are not programmed by atiixp_set_dma_mode() and if SWDMA
mode is chosen atiixp_dma_host_on() erroneously enables UDMA, just disable
support for SWDMA modes for now.
* Enable support for MWDMA0 mode.
* Bump driver version.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* Always set ->autotune in init_hwif_sis5513(), this means practically
no change in behavior since PIO was always tuned in ->ide_dma_check
and >autotune was always set for ->dma_base == 0 case.
* Bump driver version.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
* Add sis_ata133_get_base() helper function for obtaining the address of
the drive control registers on chipset_family == ATA_133 chipsets.
* Add three helper functions for programming PIO/MWDMA timings:
- sis_ata16_program_timings() (for ATA_16/33/66/100a chipset families)
- sis_ata100_program_timings() (for ATA_100/133a chipset families)
- sis_ata133_program_timings() (for ATA_133 chipset family)
MWDMA timings are taken from datasheets and they match ATA spec.
* Add generic helper function sis_program_timings() and use it in
->set_pio_mode and ->set_dma_mode methods (previously the driver
depended on BIOS to program the correct MWDMA timings).
* Remove redundant !chipset_family check from init_hwif_sis5513()
(init_chipset_sis5513() guarantees that we will never get here if
chipset_family cannot be determined).
* SWDMA seems to be unsupported by SiS chipsets (no info about SWDMA in
datasheets and for SWDMA0 mode timing requirements are impossible to
fulfill) so remove ->swdma_mask from init_hwif_sis5513() and handling
of SWDMA modes from sis_set_dma_mode().
* Enable DMA support for chipset_family == ATA_16.
* Bump driver version.
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
This patch adds id strings for Corsair 1GB (identified as Hyperstone Model1)
inside legacy/ide-cs.c.
It also includes some minor whitespace cleanups.
Signed-off-by: Kristoffer Ericson <kristoffer.ericson@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
The patch described by the following excerpt from ChangeLog-2.6.22 makes
it impossible to use UDMA on a Tyan S2707 motherboard (SvrWks CSB5):
commit 2d5eaa6dd7
Author: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Date: Thu May 10 00:01:08 2007 +0200
ide: rework the code for selecting the best DMA transfer mode (v3)
...
This one-line patch against 2.6.23 fixes the problem.
Signed-off-by: Tony Battersby <tonyb@cybernetics.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
d5a7430ddc missed a spot where we
use cpu_sibling_map and cpu_core_map. These don't exist on a
uni-processor build. Wrap #ifdef CONFIG_SMP ... #endif around it.
Signed-off-by: Tony Luck <tony.luck@intel.com>
htab_bolt_mapping takes another argument now the 1TB code has been
merged. Update vmemmap_populate to match.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 78bdc3106a ("PPC: sg chaining
support") looks to have removed some unrelated ppc code. Lets put it
back in.
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Use round_jiffies() to align the 1 second ah_reap_task with other work
and potentially save power by sleeping cores for longer.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Deadlock condition reported by Kanoj Sarcar <kanoj@netxen.com>.
The deadlock occurs when a connection request arrives at the same
time that a wildcard listen is being destroyed.
A wildcard listen maintains per device listen requests for each
RDMA device in the system. The per device listens are automatically
added and removed when RDMA devices are inserted or removed from
the system.
When a wildcard listen is destroyed, rdma_destroy_id() acquires
the rdma_cm's device mutex ('lock') to protect against hot-plug
events adding or removing per device listens. It then tries to
destroy the per device listens by calling ib_destroy_cm_id() or
iw_destroy_cm_id(). It does this while holding the device mutex.
However, if the underlying iw/ib CM reports a connection request
while this is occurring, the rdma_cm callback function will try
to acquire the same device mutex. Since we're in a callback,
the ib_destroy_cm_id() or iw_destroy_cm_id() calls will block until
their callback thread returns, but the callback is blocked waiting for
the device mutex.
Fix this by re-working how per device listens are destroyed. Use
rdma_destroy_id(), which avoids the deadlock, in place of
cma_destroy_listen(). Additional synchronization is added to handle
device hot-plug events and ensure that the id is not destroyed twice.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
If a user allocates a QP on an rdma_cm_id, the rdma_cm will automatically
transition the QP through its states (RTR, RTS, error, etc.) While the
QP state transitions are occurring, the QP itself must remain valid.
Provide locking around the QP pointer to prevent its destruction while
accessing the pointer.
This fixes an issue reported by Olaf Kirch from Oracle that resulted in
a system crash:
"An incoming connection arrives and we decide to tear down the nascent
connection. The remote ends decides to do the same. We start to shut
down the connection, and call rdma_destroy_qp on our cm_id. ... Now
apparently a 'connect reject' message comes in from the other host,
and cma_ib_handler() is called with an event of IB_CM_REJ_RECEIVED.
It calls cma_modify_qp_err, which for some odd reason tries to modify
the exact same QP we just destroyed."
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
XFS leaves stray mappings around when it vmaps memory to make it
virtually contigious. This upsets Xen if one of those pages is being
recycled into a pagetable, since it finds an extra writable mapping of
the page.
This patch solves the problem in a brute force way, by making XFS
always eagerly unmap its mappings. David Chinner says this shouldn't
have any performance impact on filesystems with default block sizes;
it will only affect filesystems with large block sizes.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Acked-by: David Chinner <dgc@sgi.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: XFS masters <xfs-masters@oss.sgi.com>
Cc: Stable kernel <stable@kernel.org>
Cc: Morten =?utf-8?q?B=C3=B8geskov?= <xen-users@morten.bogeskov.dk>
Cc: Mark Williamson <mark.williamson@cl.cam.ac.uk>
Multicalls are expected to never fail, and the normal response to a
failed multicall is very terse. In the interests of better
debuggability, add some more verbose output. It may be worth turning
this off once it all seems more tested.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
The kernel's copy of struct vcpu_register_vcpu_info was out of date,
at best causing the hypercall to fail and the guest kernel to fall
back to the old mechanism, or worse, causing random memory corruption.
[ Stable folks: applies to 2.6.23 ]
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Stable Kernel <stable@kernel.org>
Cc: Morten =?utf-8?q?B=C3=B8geskov?= <xen-users@morten.bogeskov.dk>
Cc: Mark Williamson <mark.williamson@cl.cam.ac.uk>
Ask the hypervisor how much space it needs reserved, since 32-on-64
doesn't need any space, and it may change in future.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
When a pagetable is created, it is made globally visible in the rmap
prio tree before it is pinned via arch_dup_mmap(), and remains in the
rmap tree while it is unpinned with arch_exit_mmap().
This means that other CPUs may race with the pinning/unpinning
process, and see a pte between when it gets marked RO and actually
pinned, causing any pte updates to fail with write-protect faults.
As a result, all pte pages must be properly locked, and only unlocked
once the pinning/unpinning process has finished.
In order to avoid taking spinlocks for the whole pagetable - which may
overflow the PREEMPT_BITS portion of preempt counter - it locks and pins
each pte page individually, and then finally pins the whole pagetable.
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Hugh Dickens <hugh@veritas.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andi Kleen <ak@suse.de>
Cc: Keir Fraser <keir@xensource.com>
Cc: Jan Beulich <jbeulich@novell.com>
