* 'for-linus' of git://xenbits.xen.org/people/sstabellini/linux-pvhvm:
xen: suspend: remove xen_hvm_suspend
xen: suspend: pull pre/post suspend hooks out into suspend_info
xen: suspend: move arch specific pre/post suspend hooks into generic hooks
xen: suspend: refactor non-arch specific pre/post suspend hooks
xen: suspend: add "arch" to pre/post suspend hooks
xen: suspend: pass extra hypercall argument via suspend_info struct
xen: suspend: refactor cancellation flag into a structure
xen: suspend: use HYPERVISOR_suspend for PVHVM case instead of open coding
xen: switch to new schedop hypercall by default.
xen: use new schedop interface for suspend
xen: do not respond to unknown xenstore control requests
xen: fix compile issue if XEN is enabled but XEN_PVHVM is disabled
xen: PV on HVM: support PV spinlocks and IPIs
xen: make the ballon driver work for hvm domains
xen-blkfront: handle Xen major numbers other than XENVBD
xen: do not use xen_info on HVM, set pv_info name to "Xen HVM"
xen: no need to delay xen_setup_shutdown_event for hvm guests anymore
rdma_destroy_id currently uses the global rdma cm 'lock' to test if an
rdma_cm_id has been bound to a device. This prevents an active
address resolution callback handler from assigning a device to the
rdma_cm_id after rdma_destroy_id checks for one.
Instead, we can replace the use of the global lock around the check to
the rdma_cm_id device pointer by setting the id state to destroying,
then flushing all active callbacks. The latter is accomplished by
acquiring and releasing the handler_mutex. Any active handler will
complete first, and any newly scheduled handlers will find the
rdma_cm_id in an invalid state.
In addition to optimizing the current locking scheme, the use of the
rdma_cm_id mutex is a more intuitive synchronization mechanism than
that of the global lock. These changes are based on feedback from
Doug Ledford <dledford@redhat.com> while he was trying to debug a
crash in the rdma cm destroy path.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This problem was reported by Moni Shoua <monis@mellanox.com> and Amir
Vadai <amirv@mellanox.com>:
When destroying a cm_id from a context of a work queue and if
the lap_state of this cm_id is IB_CM_LAP_SENT, we need to
release the reference of this id that was taken upon the send
of the LAP message. Otherwise, if the expected APR message
gets lost, it is only after a long time that the reference
will be released, while during that the work handler thread is
not available to process other things.
It turns out that we need to cancel any pending LAP messages whenever
we transition out of the IB_CM_ESTABLISH state. This occurs when
disconnecting - either sending or receiving a DREQ. It can also
happen in a corner case where we receive a REJ message after sending
an RTU, followed by a LAP. Add checks and cancel any outstanding LAP
messages in these three cases.
Canceling the LAP when sending a DREQ fixes the destroy problem
reported by Moni. When a cm_id is destroyed in the IB_CM_ESTABLISHED
state, it sends a DREQ to the remote side to notify the peer that the
connection is going away.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
When processing a SIDR REQ, the ib_cm allocates a new cm_id. The
refcount of the cm_id is initialized to 1. However, cm_process_work
will decrement the refcount after invoking all callbacks. The result
is that the cm_id will end up with refcount set to 0 by the end of the
sidr req handler.
If a user tries to destroy the cm_id, the destruction will proceed,
under the incorrect assumption that no other threads are referencing
the cm_id. This can lead to a crash when the cm callback thread tries
to access the cm_id.
This problem was noticed as part of a larger investigation with kernel
crashes in the rdma_cm when running on a real time OS.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Acked-by: Doug Ledford <dledford@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds DDR mode support to dw_mmc.
If we set any bit in UHS_REG bit[16:31], the card of that slot is
supported for DDR mode. For example, if UHS_REG[16] is set, card
number 0 is DDR mode.
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
drivers/mmc/host/via-sdmmc.c: In function ‘via_reset_pcictrl’:
drivers/mmc/host/via-sdmmc.c:805:8: warning: variable ‘addrbase’ set but
not used [-Wunused-but-set-variable]
Signed-off-by: Chris Ball <cjb@laptop.org>
Cc: Harald Welte <HaraldWelte@viatech.com>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Fixes:
drivers/mmc/host/cb710-mmc.c: In function ‘cb710_wait_while_busy’:
drivers/mmc/host/cb710-mmc.c:182:6: warning: variable ‘err’ set but not
used [-Wunused-but-set-variable]
Signed-off-by: Chris Ball <cjb@laptop.org>
Acked-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
drivers/mmc/host/sdhci-pci.c: In function ‘sdhci_pci_probe_slot’:
drivers/mmc/host/sdhci-pci.c:913:18: warning: variable ‘addr’ set but
not used [-Wunused-but-set-variable]
Signed-off-by: Chris Ball <cjb@laptop.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
This adds the mmc host driver for Freescale MXS-based SoC i.MX23/28.
The driver calls into mxs-dma via generic dmaengine api for both pio
and data transfer.
Thanks Chris Ball for the indentation patch.
Signed-off-by: Shawn Guo <shawn.guo@freescale.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Tested-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
This switches the mxcmmc driver to use the dmaengine API. Unlike
the old one this one is always present in the tree, even if no DMA
is implemented, hence we can remove all the #ifdefs in from the driver.
The driver automatically switches to PIO mode if no DMA support or no
suitable channel is available.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
This patch adds support for 8-bit buswidth.
dw_mmc can use 8-bit buswidth and set to CTYPE_8BIT in card-type register.
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
If we need some quirks, maybe add quirks in future
But now, quirks value set to integer..later we should be confused..
So I think that need bit-shift control.
And If we need not any quirks, we didn't set anything..
(Need not DW_MCI_QUIRK_NONE)
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Acked-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Use the new dmaengine helper functions, and drop the error check
on the returned cookier from the dmaengine - we recently
established that this is really not allowed to fail.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
This variable doesn't seem to be used for anything after the
other patches so just drop it.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
According to the DMA-API you shall unmap the sglists with the same
sglist length as passed into the mapping function, not the
returned value from the mapping function.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
As established for the MMCI, it is proper to map the DMA buffers
on the DMA engine which is the one actually performing the DMA.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
* 'stable/ia64' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen: ia64 build broken due to "xen: switch to new schedop hypercall by default."
* 'stable/blkfront-cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen: Union the blkif_request request specific fields
* 'stable/cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen: annotate functions which only call into __init at start of day
xen p2m: annotate variable which appears unused
xen: events: mark cpu_evtchn_mask_p as __refdata
Use the new dmaengine helper functions, and drop the error check
on the returned cookier from the dmaengine - we recently
established that this is really not allowed to fail.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
The host_sglen is now actually used to keep track of whether DMA
is active or not, so rename and retype it to bool.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
According to the DMA-API you shall unmap the sglists with the same
sglist length as passed into the mapping function, not the
returned value from the mapping function.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
As established for the MMCI, it is proper to map the DMA buffers
on the DMA engine which is the one actually performing the DMA.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
Use the new dmaengine helpers to make the code more readable.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Fixes the following:
- It is perfectly legal for the dma_map_sg() to return fewer
entries than were passed in.
- Supply the returned numer of (possibly coalesced) entries to
the device_pre_slave_sg() function.
- Use the proper original sg_len when unmapping the sglist
in the error path.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
As established for the MMCI, it is proper to map the DMA buffers
on the DMA engine which is the one actually performing the DMA.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
So we know the implementation and prototypes agree with each other.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
The test file is created under debugfs, not sysfs. Also remove
the unnecessary default n.
Signed-off-by: Will Newton <will.newton@imgtec.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
Setting this bit in the clock enable register will stop the clock
when the card is in the IDLE state.
Signed-off-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
We need to run the card detect tasklet at the end of slot initialisation
as it is possible that a card has been inserted prior to boot, so we don't
see an insertion interrupt and now the card is sitting there inserted but
with no power to it.
Signed-off-by: Neil Jones <neil.jones@imgtec.com>
Signed-off-by: Will Newton <will.newton@imgtec.com>
Reviewed-by: Matt Fleming <matt@console-pimps.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
Currently MMC_MXC driver can be selected by all i.MX devices.
Restrict its use only for the appropriate processors.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
Add two large sequential I/O performance tests:
35. Large sequential read into scattered pages
36. Large sequential write from scattered pages
The tests measure transfer times for 10MiB, 100MiB, 1000MiB.
Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Existing performance tests measure single or sequential I/O speed.
Add two random I/O tests:
33. Random read performance by transfer size
34. Random write performance by transfer size
Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
The test area size was set to the preferred erase size but for comparison
purposes it is better if it is the same size for different devices. Make
it a multiple of preferred erase size that is greater than or equal to 4MiB.
Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
This sdio card supports having its sdio clock shutdown.
It is also not using the SDIO IRQ, but rather uses a side gpio irq.
Signed-off-by: Pierre Tardy <tardyp@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Some sdio card are not following sdio standard, and do not work
when the sdio bus's clock is gated.
To keep functionnality for all legacy driver, we turn this quirk on
for every sdio card.
Drivers needs to disable the quirk manually when someone verifies that
their supported card works with clock gating.
Signed-off-by: Pierre Tardy <tardyp@gmail.com>
Acked-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Some cards have quirks valid for every platforms using current
platform quirk hooks leads to a lot of code and debug duplication.
So we inspire a bit from what exists in PCI subsystem and do our own
per vendorid/deviceid quirk. We still drop the complexity of the pci
quirk system (with special section tables, and so on).
That can be added later if needed.
Signed-off-by: Pierre Tardy <pierre.tardy@intel.com>
Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Acked-by: Ohad Ben-Cohen <ohad@wizery.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Since mmc clock gating can also be used as a power gating
tip, it's better to put the led blinking after having
ungated the clock.
Signed-off-by: Pierre Tardy <pierre.tardy@intel.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
If the MMC host controller does not support waiting for card signaling
busy state (MMC_CAP_WAIT_WHILE_BUSY cap), there is no point in prining
the relevant warning message.
Signed-off-by: Pawel Moll <pawel.moll@arm.com>
Reviewed-by: Matt Fleming <matt@console-pimps.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
Samsung SDHCI host controller supports the Auto CMD12.
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
While the MMC handled the card's read only flag correctly on open,
it did not setup the flag in the allocated disk structure. The
consequence being that probing the /sys/class/block/mmcblkX/ro
attribute always reported 0.
Signed-off-by: Marc-Andre Hebert <hebert.marcandre@gmail.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Tested-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
Enhanced area feature is a new feature defined in eMMC4.4 standard. This
user data area provides higher performance/reliability, at the expense
of using twice the effective media space due to the area using SLC.
The MMC driver now reads out the enhanced area offset and size and adds
them to the device attributes in sysfs. Enabling the enhanced area can
only be done once, and should be done in manufacturing. To use this
feature, bit ERASE_GRP_DEF should also be set.
Documentation/ABI/testing/sysfs-devices-mmc describes the two new
attributes.
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Reviewed-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
* 'stable/irq.cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen: events: remove dom0 specific xen_create_msi_irq
xen: events: use xen_bind_pirq_msi_to_irq from xen_create_msi_irq
xen: events: push set_irq_msi down into xen_create_msi_irq
xen: events: update pirq_to_irq in xen_create_msi_irq
xen: events: refactor xen_create_msi_irq slightly
xen: events: separate MSI PIRQ allocation from PIRQ binding to IRQ
xen: events: assume PHYSDEVOP_get_free_pirq exists
xen: pci: collapse apic_register_gsi_xen_hvm and xen_hvm_register_pirq
xen: events: return irq from xen_allocate_pirq_msi
xen: events: drop XEN_ALLOC_IRQ flag to xen_allocate_pirq_msi
xen: events: do not leak IRQ from xen_allocate_pirq_msi when no pirq available.
xen: pci: only define xen_initdom_setup_msi_irqs if CONFIG_XEN_DOM0
* 'stable/irq.rework' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen/irq: Cleanup up the pirq_to_irq for DomU PV PCI passthrough guests as well.
xen: Use IRQF_FORCE_RESUME
xen/timer: Missing IRQF_NO_SUSPEND in timer code broke suspend.
xen: Fix compile error introduced by "switch to new irq_chip functions"
xen: Switch to new irq_chip functions
xen: Remove stale irq_chip.end
xen: events: do not free legacy IRQs
xen: events: allocate GSIs and dynamic IRQs from separate IRQ ranges.
xen: events: add xen_allocate_irq_{dynamic, gsi} and xen_free_irq
xen:events: move find_unbound_irq inside CONFIG_PCI_MSI
xen: handled remapped IRQs when enabling a pcifront PCI device.
genirq: Add IRQF_FORCE_RESUME
* 'stable/pcifront-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
pci/xen: When free-ing MSI-X/MSI irq->desc also use generic code.
pci/xen: Cleanup: convert int** to int[]
pci/xen: Use xen_allocate_pirq_msi instead of xen_allocate_pirq
xen-pcifront: Sanity check the MSI/MSI-X values
xen-pcifront: don't use flush_scheduled_work()
* 'stable/p2m-identity.v4.9.1' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen/m2p: Check whether the MFN has IDENTITY_FRAME bit set..
xen/m2p: No need to catch exceptions when we know that there is no RAM
xen/debug: WARN_ON when identity PFN has no _PAGE_IOMAP flag set.
xen/debugfs: Add 'p2m' file for printing out the P2M layout.
xen/setup: Set identity mapping for non-RAM E820 and E820 gaps.
xen/mmu: WARN_ON when racing to swap middle leaf.
xen/mmu: Set _PAGE_IOMAP if PFN is an identity PFN.
xen/mmu: Add the notion of identity (1-1) mapping.
xen: Mark all initial reserved pages for the balloon as INVALID_P2M_ENTRY.
* 'stable/e820' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen/e820: Don't mark balloon memory as E820_UNUSABLE when running as guest and fix overflow.
xen/setup: Inhibit resource API from using System RAM E820 gaps as PCI mem gaps.
Doug Ledford and Red Hat reported a crash when running the rdma_cm on
a real-time OS. The crash has the following call trace:
cm_process_work
cma_req_handler
cma_disable_callback
rdma_create_id
kzalloc
init_completion
cma_get_net_info
cma_save_net_info
cma_any_addr
cma_zero_addr
rdma_translate_ip
rdma_copy_addr
cma_acquire_dev
rdma_addr_get_sgid
ib_find_cached_gid
cma_attach_to_dev
ucma_event_handler
kzalloc
ib_copy_ah_attr_to_user
cma_comp
[ preempted ]
cma_write
copy_from_user
ucma_destroy_id
copy_from_user
_ucma_find_context
ucma_put_ctx
ucma_free_ctx
rdma_destroy_id
cma_exch
cma_cancel_operation
rdma_node_get_transport
rt_mutex_slowunlock
bad_area_nosemaphore
oops_enter
They were able to reproduce the crash multiple times with the
following details:
Crash seems to always happen on the:
mutex_unlock(&conn_id->handler_mutex);
as conn_id looks to have been freed during this code path.
An examination of the code shows that a race exists in the request
handlers. When a new connection request is received, the rdma_cm
allocates a new connection identifier. This identifier has a single
reference count on it. If a user calls rdma_destroy_id() from another
thread after receiving a callback, rdma_destroy_id will proceed to
destroy the id and free the associated memory. However, the request
handlers may still be in the process of running. When control returns
to the request handlers, they can attempt to access the newly created
identifiers.
Fix this by holding a reference on the newly created rdma_cm_id until
the request handler is through accessing it.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Acked-by: Doug Ledford <dledford@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
d_move puts the renamed dentry at the end of d_subdirs, screwing with our
cached dentry directory offsets. We were just clearing I_COMPLETE to avoid
any possibility of trouble. However, assigning the renamed dentry an
offset at the end of the directory (to match it's new d_subdirs position)
is sufficient to maintain correct behavior and hold onto I_COMPLETE.
This is especially important for workloads like rsync, which renames files
into place. Before, we would lose I_COMPLETE and do MDS lookups for each
file. With this patch we only talk to the MDS on create and rename.
Signed-off-by: Sage Weil <sage@newdream.net>
