Kevin Hilman [Tue, 10 Jul 2012 23:40:56 +0000 (16:40 -0700)]
mmc: omap_hsmmc: ensure probe returns error upon resource failure
If platform_get_resource_by_name() fails, driver probe is aborted an
should return an error so the driver is not bound to the device.
However, in the current error path of platform_get_resource_by_name(),
probe returns zero since the return value (ret) is not properly set.
With a zero return value, the driver core assumes probe was successful
and will bind the driver to the device.
Fix this by ensuring that probe returns an error code in this failure
path.
Signed-off-by: Kevin Hilman <khilman@ti.com> Acked-by: Venkatraman S <svenkatr@ti.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Aaron Lu [Tue, 10 Jul 2012 08:55:37 +0000 (16:55 +0800)]
mmc: core: reset signal voltage on power up
Add a call to mmc_set_signal_voltage() to set signal voltage to 3.3v in
mmc_power_up so that we do not need to touch signal voltage setting in
mmc/sd/sdio init functions and rescan function.
For mmc/sd cards, when doing a suspend/resume cycle, consider the unsafe
resume case, the card will lose its power and when powered on again, we
will set signal voltage to 3.3v in mmc_power_up before its resume function
gets called, which will re-init the card.
And for sdio cards, when doing a suspend/resume cycle, consider the unsafe
resume case, the card will either lose its power or not depending on if it
wants to wakeup the host. If power is not maintained, it is the same case as
mmc/sd cards. If power is maintained, mmc_power_up will not be called and
the card's signal voltage will remain at the last setting.
Signed-off-by: Aaron Lu <aaron.lu@amd.com> Tested-by: Venkatraman S <svenkatr@ti.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Aaron Lu [Wed, 4 Jul 2012 05:31:48 +0000 (13:31 +0800)]
mmc: sd: Fix sd current limit setting
Host has different current capabilities at different voltages, we need
to record these settings seperately. The defined voltages are 1.8/3.0/3.3.
For other voltages, we do not touch current limit setting.
Before we set the current limit for the sd card, find out the host's
operating voltage first and then find out the current capabilities of
the host at that voltage to set the current limit.
Signed-off-by: Aaron Lu <aaron.lu@amd.com> Reviewed-by: Philip Rakity <prakity@marvell.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Nicolas Ferre [Fri, 6 Jul 2012 10:11:51 +0000 (12:11 +0200)]
mmc: atmel-mci: modify CLKDIV displaying in debugfs
Modify clock division displaying in debugfs for matching
the new CLKDIV,CLKODD user interface arrangement.
Is using the has_odd_clk_div property to choose the proper format.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com> Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Nicolas Ferre [Fri, 6 Jul 2012 09:58:33 +0000 (11:58 +0200)]
mmc: atmel-mci: fix incorrect setting of host->data to NULL
Setting host->data to NULL is incorrect sequence in STATE_SENDING_STOP
state of FSM: This early setting leads to the skip of dma_unmap_sg()
in atmci_dma_cleanup() which is a bug.
Idea taken from dw_mmc by Seungwon Jeon.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com> Cc: Seungwon Jeon <tgih.jun@samsung.com> Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Daniel Drake [Thu, 5 Jul 2012 21:06:13 +0000 (22:06 +0100)]
mmc: sdhci: poll for card even when card is logically unremovable
The Marvell CaFe is now marked as having bad card detection to fix
a problem during system resume.
Now on the OLPC XO-1 we are facing the issue that the card is marked
as logically unremovable (via MMC_UNSAFE_RESUME), which means that
mmc_card_is_removable considers the card non-removable. The existing
code logic decides not to poll for card presence in this case, and
card detection is also disabled because of the quirk being set.
This means that no SD cards are detected when inserted after boot.
Refine the logic to enable card presence polling in the case when
a card is logically unremovable, only avoiding the poll in the case
when the card is physically non-removable (denoted with
MMC_CAP_NONREMOVABLE).
Signed-off-by: Daniel Drake <dsd@laptop.org> Signed-off-by: Chris Ball <cjb@laptop.org>
Aaron Lu [Wed, 4 Jul 2012 05:29:09 +0000 (13:29 +0800)]
mmc: sdhci: Introduce new flag SDHCI_USING_RETUNING_TIMER
Add a new flag of SDHCI_USING_RETUNING_TIMER to represent if the host
is using a retuning timer for the card inserted.
This flag is set when the host does tuning the first time for the card
and the host's retuning mode is 1. This flag is used afterwards whenever
needs to decide if the host is currently using a retuning timer.
This flag is cleared when the card is removed in sdhci_reinit.
The set/clear of the flag and the start/stop of the retuning timer is
associated with the card's init/remove time, so there is no need to
touch it when the host is to be removed as at that time the card should
have already been removed.
Signed-off-by: Aaron Lu <aaron.lu@amd.com> Reviewed-by: Girish K S <girish.shivananjappa@linaro.org> Reviewed-by: Philip Rakity <prakity@marvell.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Aaron Lu [Tue, 3 Jul 2012 06:16:11 +0000 (14:16 +0800)]
mmc: core: Simplify and fix for SD switch processing
In mmc_read_switch, just do a one time mode 0 switch command to get the
support bits information, no need to do multiple times as the support
bits do not change with different arguments.
And no need to check current limit support bits, as these bits are
fixed according to the signal voltage. If the signal voltage is 1.8V,
the support bits would be 0xf and if the signal voltage is 3.3V, the
support bits would be 0x01. We will check host's ability to set the
current limit.
Signed-off-by: Aaron Lu <aaron.lu@amd.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Aaron Lu [Fri, 29 Jun 2012 08:17:31 +0000 (16:17 +0800)]
mmc: sdhci: restore host settings when card is removed
Some of the host settings are affected by different cards inserted, e.g.
when an UHS-I card is inserted, the SDHCI_NEEDS_RETUING flag might be
set when the tuning timer expired and host's max_blk_count will be
reduced to make sure the data transfer for a command does not exceed 4MiB
to meet the retuning mode 1's requirement.
When the card is removed, we should restore the original setting of the
host since we can't be sure the next card being inserted will still be
an UHS-I card that needs tuning. The original setting include its
max_blk_count and no set of the flag of SDHCI_NEEDS_RETUNING.
Signed-off-by: Aaron Lu <aaron.lu@amd.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Aaron Lu [Tue, 3 Jul 2012 09:27:49 +0000 (17:27 +0800)]
mmc: sdhci: fix incorrect command used in tuning
For SD hosts using retuning mode 1, when retuning timer expired, it will
need to do retuning in sdhci_request before processing the actual
request. But the retuning command is fixed: cmd19 for SD card and cmd21
for eMMC card, so we can't use the original request's command to do the
tuning.
And since the tuning command depends on the card type attached to the
host, we will need to know the card type to use the correct tuning
command.
Signed-off-by: Aaron Lu <aaron.lu@amd.com> Reviewed-by: Philip Rakity <prakity@marvell.com> Cc: stable <stable@vger.kernel.org> [3.3+] Signed-off-by: Chris Ball <cjb@laptop.org>
Daniel Drake [Tue, 3 Jul 2012 22:13:39 +0000 (23:13 +0100)]
mmc: sdhci-pci: CaFe has broken card detection
At http://dev.laptop.org/ticket/11980 we have determined that the
Marvell CaFe SDHCI controller reports bad card presence during
resume. It reports that no card is present even when it is.
This is a regression -- resume worked back around 2.6.37.
Around 400ms after resuming, a "card inserted" interrupt is
generated, at which point it starts reporting presence.
Work around this hardware oddity by setting the
SDHCI_QUIRK_BROKEN_CARD_DETECTION flag.
Thanks to Chris Ball for helping with diagnosis.
Signed-off-by: Daniel Drake <dsd@laptop.org>
[stable@: please apply to 3.0+] Cc: <stable@vger.kernel.org> Signed-off-by: Chris Ball <cjb@laptop.org>
Mark Brown [Mon, 2 Jul 2012 13:26:15 +0000 (14:26 +0100)]
mmc: sdhci: Report failure reasons for all cases in sdhci_add_host()
For most error conditions sdhci_add_host() will print a diagnostic
message indicating why it failed but there are a few cases where this
does not happen. Add error messages in these cases to aid diagnosis.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Philip Rakity [Thu, 28 Jun 2012 04:49:27 +0000 (21:49 -0700)]
mmc: sdhci: Allow caps[1] to be set via SDHCI_QUIRK_MISSING_CAPS
Currently only the capability_0 register can be set if
SDHCI_QUIRK_MISSING_CAPS is defined. This is a problem when
the capability_1 register also needs changing. Use the quirk
SDHCI_QUIRK_MISSING_CAPS to allow both registers to be set.
Redefining caps[1] is useful when the board design does not
support 1.8v vccq so UHS modes are not available. The code that
calls sdhci_add_host can then detect this condition and adjust
the caps so the UHS mode will not be attempted on UHS cards.
Signed-off-by: Philip Rakity <prakity@marvell.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Philip Rakity [Fri, 8 Jun 2012 19:26:13 +0000 (12:26 -0700)]
mmc: only support voltage (vdd) that regulator agrees with
If we are using a regulator the SD Host Controller and the
regulator should agree about the voltages supported. Use
the common subset that is supported.
Signed-off-by: Philip Rakity <prakity@marvell.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Philip Rakity [Mon, 28 May 2012 01:36:33 +0000 (18:36 -0700)]
mmc: sdhci: only set 200mA support for 1.8v if 200mA is available
max_current_caps can return 0 if not available from the sd controller.
If no regulator is present or the regulator specifies a current
less then 200ma, we no longer still set the 200mA caps bit anyway.
Signed-off-by: Philip Rakity <prakity@marvell.com> Reviewed-by: Aaron Lu <aaron_lu@amd.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Philip Rakity [Mon, 28 May 2012 01:36:44 +0000 (18:36 -0700)]
mmc: sdhci: if MAX_CURRENT is 0, try getting current from regulator
The sd host controller spec indicates the the MAX_CURRENT value may
be returned as 0. In this case other methods need to be used to
return the current. If 0 is returned and there is a regulator,
ask the regulator for how much current is available.
Signed-off-by: Philip Rakity <prakity@marvell.com> Signed-off-by: Mark F. Brown <mark.brown314@gmail.com> Reviewed-by: Aaron Lu <aaron.lu@amd.com> Signed-off-by: Chris Ball <cjb@laptop.org>
Extend the sh_mmcif driver to support GPIO card detection, provided by the
slot function module. The original .get_cd() platform callback is also
preserved for now.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: prohibit card detection when host is not ready
Currently mmc host drivers have to decide whether to enable card
detection before calling mmc_add_host() -- in which case a card
insertion event can arrive before the host has been completely
initialised -- or after mmc_add_host(), in which case the initial
card detection can be problematic.
This patch adds an explicit indication of when card detection should
not be carried out. With it in place enabling card detection before
calling mmc_add_host() should be safe. Similarly, disabling it again
after calling mmc_remove_host() will avoid any races.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
The tmio-mmc driver is already using the generic GPIO CD handler in IRQ
mode. This patch adds support for CD polling mode and also checks for
availability of a WP GPIO.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: add CD GPIO polling support to slot functions
A simple extension of mmc slot functions add support for CD GPIO polling
for cases where the GPIO cannot produce interrupts, or where this is not
desired for some reason.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
Add regulator support to the sh_mmcif driver, but also preserve the current
power-callback.
Also note, that the card power is not switched off during clock gating
periods, hence there's no need to power it on every time the card is
re-activated.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Reviwed-by: Mark Brown <broonie@opensource.wolfsonmicro.com> Reviewed-by: Simon Horman <horms@verge.net.au> Signed-off-by: Chris Ball <cjb@laptop.org>
From the original version of sh_mmcif the .set_pwr() callback has only been
used to turn the card's power on, and the .down_pwr() callback has been
used to turn it off. .set_pwr() can be used for both these tasks, which is
also how it is implemented by the only user of this API: the SH7724 ecovec
board.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Reviewed-by: Simon Horman <horms@verge.net.au> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: sh_mmcif: re-read the clock frequency every time it is turned on
With aggressive clock gating the clock can be disabled during interface
inactivity. During this time its frequency can be changed by another its
user. Therefore when the interface is activated again and the clock is
re-enabled, its frequency has to be re-read.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Reviewed-by: Simon Horman <horms@verge.net.au> Signed-off-by: Chris Ball <cjb@laptop.org>
Regardless of whether the MMC bus clock is the same as the PM clock on
this specific interface, it has to be managed separately. Its proper
management should also include enabling and disabling of the clock,
whenever the interface is becoming active or going idle respectively.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Reviewed-by: Simon Horman <horms@verge.net.au> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: sh_mmcif: simplify and use meaningful label names in error-handling
A check for NULL platform data can be conveniently made in the very
beginning of probing. Replace numbered error-handling labels in .probe()
with meaningful names to make any future reorganisation simpler.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Reviewed-by: Simon Horman <horms@verge.net.au> Signed-off-by: Chris Ball <cjb@laptop.org>
Currently the SDHI glue for the TMIO MMC driver installs dummy .get_cd() and
.set_pwr() callbacks even if the platform didn't supply them. This is not
necessary, since the TMIO MMC driver itself checks for NULL callbacks. This
is also dubious if the platform provides a regulator for SD-card power
switching. It is better to only install those callbacks, if they are really
provided by the platform.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
Currently the TMIO MMC driver derives the OCR mask from the platform data
and uses a platform callback to turn card power on and off. This patch adds
regulator support to the driver.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Liam Girdwood <lrg@ti.com> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: sdhi: implement tmio-mmc clock enable-update and disable callbacks
Instead of delivering one static clock frequency value, read from the
hardware during probing, enable the tmio-mmc driver to re-read the
frequency dynamically.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: tmio: don't needlessly enable interrupts during probing
We don't have to force-enable MMC interrupts in our .probe() method,
mmc_add_host() will trigger card detection, that will enable all the
necessary interrupts.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: dw_mmc: Fix null dma_ops access when use_dma is false
host->dma_ops is not valid if host->usa dma is 0 so protect
host->dma_ops reference in dw_mci_resume
Signed-off-by: Jonathan Kliegman <kliegs@chromium.org> Acked-by: Jaehoon Chung <jh80.chung@samsung.com> Acked-by: Will Newton <will.newton@imgtec.com> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: add a function to get regulators, supplying card's power
Add a function to get regulators, supplying card's Vdd and Vccq on a
specific host. If a Vdd supplying regulator is found, the function checks,
whether a valid OCR mask can be obtained from it. The Vccq regulator is
optional. A failure to get it is not fatal.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Reviwed-by: Mark Brown <broonie@opensource.wolfsonmicro.com> Acked-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
Laurent Pinchart [Tue, 12 Jun 2012 21:29:35 +0000 (23:29 +0200)]
mmc: tmio: Don't access hardware registers after stopping clocks
The tmio_mmc_set_ios() function configures the MMC power, clock and bus
width. When the mmc core requests the driver to power off the card, we
inform runtime PM, that the controller can be suspended. This can lead
to the MSTP clock being turned off.
Writing to any 16-bit hardware registers with the MSTP clock off leads
to timeouts and errors being printed to the kernel log. This can occur
both when stopping the MMC clock and when configuring the bus width.
To fix this, stop the MMC clock before calling put_runtime_pm(), and
skip bus width configuration when power is off.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com> Tested-by: Simon Horman <horms@verge.net.au> Acked-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
Laurent Pinchart [Tue, 12 Jun 2012 20:56:09 +0000 (22:56 +0200)]
mmc: sh_mmcif: Support MMC_SLEEP_AWAKE command
The MMC_SLEEP_AWAKE and SD_IO_SEND_OP_COND commands share the same
opcode. SD_IO_SEND_OP_COND isn't supported by the SH MMCIF, but
MMC_SLEEP_AWAKE is. Discriminate between the two commands using the
command flags, and reject SD_IO_SEND_OP_COND only.
Girish K S [Tue, 12 Jun 2012 09:58:22 +0000 (15:28 +0530)]
mmc: dw_mmc: Add check for IDMAC configuration
In the current dwmmc driver there is support for selecting IDMAC from
the menu config option. If the support for IDMAC is enabled in the menu
config and the hardware configuration register's DMA_INTERFACE field is
0, the driver will still try to do the DMA initialization.
The dw_mci_idmac_init function currently implemented returns only success
indicating that the DMA initialization is always successful. This patch
adds a check for existence of the DMA IP to allow the DMA initialization.
Signed-off-by: Girish K S <girish.shivananjappa@linaro.org> Acked-by: Will Newton <will.newton@imgtec.com> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: block: replace __blk_end_request() with blk_end_request()
For completing any block request, MMC block driver is calling:
spin_lock_irq(queue)
__blk_end_request()
spin_unlock_irq(queue)
But if we analyze the sources of latency in kernel using ftrace,
__blk_end_request() function at times may take up to 6.5ms with
spinlock held and irq disabled.
__blk_end_request() calls couple of functions and ftrace output
shows that blk_update_bidi_request() function is almost taking 6ms.
There are 2 function to end the current request: ___blk_end_request()
and blk_end_request(). Both these functions do same thing except
that blk_end_request() function doesn't take up the spinlock
while calling the blk_update_bidi_request().
This patch replaces all __blk_end_request() calls with
blk_end_request() and __blk_end_request_all() calls with
blk_end_request_all().
Testing done: 20 process concurrent read/write on sd card
and eMMC. Ran this test for almost a day on multicore system
and no errors observed.
This change is not meant for improving MMC throughput; it's basically
about becoming fair to other threads/interrupts in the system. By
holding spin lock and interrupts disabled for longer duration, we
won't allow other threads/interrupts to run at all. Actually slight
performance degradation at file system level can be expected as we
are not holding the spin lock during blk_update_bidi_request() which
means our mmcqd thread may get preempted for other high priority
thread or any interrupt in the system.
These are performance numbers (100MB file write) with eMMC running
in DDR mode:
Without this patch:
Name of the Test Value Unit
LMDD Read Test 53.79 MBPS
LMDD Write Test 18.86 MBPS
IOZONE Read Test 51.65 MBPS
IOZONE Write Test 24.36 MBPS
With this patch:
Name of the Test Value Unit
LMDD Read Test 52.94 MBPS
LMDD Write Test 16.70 MBPS
IOZONE Read Test 52.08 MBPS
IOZONE Write Test 23.29 MBPS
Read numbers are fine. Write numbers are bit down (especially LMDD
write), may be because write requests normally have large transfer
size and which means there are chances that while mmcq is executing
blk_update_bidi_request(), it may get interrupted by interrupts or
other high priority thread.
Venkatraman S [Fri, 22 Jun 2012 06:12:36 +0000 (11:42 +0530)]
mmc: core: Fix the HPI execution sequence
mmc_execute_hpi should send the HPI command only once, and only
if the card is in PRG state.
According to eMMC spec, the command's completion time is
not dependent on OUT_OF_INTERRUPT_TIME. Only the transition
out of PRG STATE is guarded by OUT_OF_INTERRUPT_TIME - which
is defined to begin at the end of sending the command itself.
Specify the default timeout for the actual sending of HPI
command, and then use OUT_OF_INTERRUPT_TIME to wait for
the transition out of PRG state.
Reported-by: Alex Lemberg <Alex.Lemberg@sandisk.com> Signed-off-by: Venkatraman S <svenkatr@ti.com> Reviewed-by: Namjae Jeon <linkinjeon@gmail.com> Reviewed-by: Subhash Jadavani <subhashj@codeaurora.org> Acked-by: Jaehoon Chung <jh80.chung@samsung.com> Signed-off-by: Chris Ball <cjb@laptop.org>
mmc: extend and rename cd-gpio helpers to handle more slot GPIO functions
GPIOs can be used in MMC/SD-card slots not only for hotplug detection, but
also to implement the write-protection pin. Rename cd-gpio helpers to
slot-gpio to make addition of further slot GPIO functions possible.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by: Chris Ball <cjb@laptop.org>
Venkatraman S [Mon, 2 Jul 2012 06:56:34 +0000 (12:26 +0530)]
mmc: core: Revert "skip card initialization if power class selection fails"
This reverts commit 3d93576e(skip card initialization if
power class selection fails).
Problem has been reported when this is used with eMMC4.41
card with Tegra Platform. Till the issue is root caused,
bus width selection failure should not be treated as fatal.
Reported-by: Marc Dietrich <marvin24@gmx.de> Signed-Off-by: Venkatraman S <svenkatr@ti.com> CC: Ulf Hansson <ulf.hansson@stericsson.com> CC: Subhash Jadavani <subhashj@codeaurora.org> CC: Saugata Das <saugata.das@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
48ddbe1946 "cgroup: make css->refcnt clearing on cgroup removal
optional" allowed a css to linger after the associated cgroup is
removed. As a css holds a reference on the cgroup's dentry, it means
that cgroup dentries may linger for a while.
Destroying a superblock which has dentries with positive refcnts is a
critical bug and triggers BUG() in vfs code. As each cgroup dentry
holds an s_active reference, any lingering cgroup has both its dentry
and the superblock pinned and thus preventing premature release of
superblock.
Unfortunately, after 48ddbe1946, there's a small window while
releasing a cgroup which is directly under the root of the hierarchy.
When a cgroup directory is released, vfs layer first deletes the
corresponding dentry and then invokes dput() on the parent, which may
recurse further, so when a cgroup directly below root cgroup is
released, the cgroup is first destroyed - which releases the s_active
it was holding - and then the dentry for the root cgroup is dput().
This creates a window where the root dentry's refcnt isn't zero but
superblock's s_active is. If umount happens before or during this
window, vfs will see the root dentry with non-zero refcnt and trigger
BUG().
Before 48ddbe1946, this problem didn't exist because the last dentry
reference was guaranteed to be put synchronously from rmdir(2)
invocation which holds s_active around the whole process.
Fix it by holding an extra superblock->s_active reference across
dput() from css release, which is the dput() path added by 48ddbe1946
and the only one which doesn't hold an extra s_active ref across the
final cgroup dput().
Revert "cgroup: superblock can't be released with active dentries"
This reverts commit fa980ca87d15bb8a1317853f257a505990f3ffde. The
commit was an attempt to fix a race condition where a cgroup hierarchy
may be unmounted with positive dentry reference on root cgroup. While
the commit made the race condition slightly more difficult to trigger,
the race was still there and could be reliably triggered using a
different test case.
Revert the incorrect fix. The next commit will describe the race and
fix it correctly.
Merge branch 'fixes' of git://git.linaro.org/people/rmk/linux-arm
Pull ARM fixes from Russell King:
"Last merge window, we had some updates from Al cleaning up the signal
restart handling. These have caused some problems on ARM, and while
Al has some fixes, we have some concerns with Al's patches but we've
been unsuccesful with discussing this.
We have got to the point where we need to do something, and we've
decided that the best solution is to revert the appropriate commits
until Al is able to reply to us.
Also included here are four patches to fix warnings that I've noticed
in my build system, and one fix for kprobes test code."
* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
ARM: fix warning caused by wrongly typed arm_dma_limit
ARM: fix warnings about atomic64_read
ARM: 7440/1: kprobes: only test 'sub pc, pc, #1b-2b+8-2' on ARMv6
ARM: 7441/1: perf: return -EOPNOTSUPP if requested mode exclusion is unavailable
ARM: 7443/1: Revert "new way of handling ERESTART_RESTARTBLOCK"
ARM: 7442/1: Revert "remove unused restart trampoline"
ARM: fix set_domain() macro
ARM: fix mach-versatile/pci.c warning
Merge tag 'ecryptfs-3.5-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs
Pull eCryptfs fixes from Tyler Hicks:
"Fixes an incorrect access mode check when preparing to open a file in
the lower filesystem. This isn't an urgent fix, but it is simple and
the check was obviously incorrect.
Also fixes a couple important bugs in the eCryptfs miscdev interface.
These changes are low risk due to the small number of users that use
the miscdev interface. I was able to keep the changes minimal and I
have some cleaner, more complete changes queued up for the next merge
window that will build on these patches."
* tag 'ecryptfs-3.5-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs:
eCryptfs: Gracefully refuse miscdev file ops on inherited/passed files
eCryptfs: Fix lockdep warning in miscdev operations
eCryptfs: Properly check for O_RDONLY flag before doing privileged open
Pull target fixes from Nicholas Bellinger:
"Two minor target fixes. There is really nothing exciting and/or
controversial this time around.
There's one fix from MDR for a RCU debug warning message within tcm_fc
code (CC'ed to stable), and a small AC fix for qla_target.c based upon
a recent Coverity static report.
Also, there is one other outstanding virtio-scsi LUN scanning bugfix
that has been uncovered with the in-flight tcm_vhost driver over the
last days, and that needs to make it into 3.5 final too. This patch
has been posted to linux-scsi again here:
and I've asked James to include it in his next PULL request."
* git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending:
qla2xxx: print the right array elements in qlt_async_event
tcm_fc: Resolve suspicious RCU usage warnings
Tyler Hicks [Mon, 11 Jun 2012 16:24:11 +0000 (09:24 -0700)]
eCryptfs: Gracefully refuse miscdev file ops on inherited/passed files
File operations on /dev/ecryptfs would BUG() when the operations were
performed by processes other than the process that originally opened the
file. This could happen with open files inherited after fork() or file
descriptors passed through IPC mechanisms. Rather than calling BUG(), an
error code can be safely returned in most situations.
In ecryptfs_miscdev_release(), eCryptfs still needs to handle the
release even if the last file reference is being held by a process that
didn't originally open the file. ecryptfs_find_daemon_by_euid() will not
be successful, so a pointer to the daemon is stored in the file's
private_data. The private_data pointer is initialized when the miscdev
file is opened and only used when the file is released.
Alan Cox [Wed, 4 Jul 2012 15:35:35 +0000 (16:35 +0100)]
qla2xxx: print the right array elements in qlt_async_event
Based upon Alan's patch from Coverity scan id 793583, these debug
messages in qlt_async_event() should be starting from byte 0, which is
always the Asynchronous Event Status Code from the parent switch statement.
Also, rename reason_code -> login_code following the language used in
2500 FW spec for Port Database Changed (0x8014) -> Port Database Changed
Event Mailbox Register for mailbox[2].
Signed-off-by: Alan Cox <alan@linux.intel.com> Cc: Chad Dupuis <chad.dupuis@qlogic.com> Cc: Giridhar Malavali <giridhar.malavali@qlogic.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Mark Rustad [Tue, 26 Jun 2012 22:57:30 +0000 (15:57 -0700)]
tcm_fc: Resolve suspicious RCU usage warnings
Use rcu_dereference_protected to tell rcu that the ft_lport_lock
is held during ft_lport_create. This resolved "suspicious RCU usage"
warnings when debugging options are turned on.
Signed-off-by: Mark Rustad <mark.d.rustad@intel.com> Tested-by: Ross Brattain <ross.b.brattain@intel.com> Cc: <stable@vger.kernel.org> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Merge tag 'for-linus-20120706' of git://git.infradead.org/linux-mtd
Pull two MTD fixes from David Woodhouse:
- Fix a logic error in OLPC CAFÉ NAND ready() function.
- Fix regression due to bitflip handling changes.
* tag 'for-linus-20120706' of git://git.infradead.org/linux-mtd:
mtd: cafe_nand: fix an & vs | mistake
mtd: nand: initialize bitflip_threshold prior to BBT scanning
Merge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jlbec/ocfs2
Pull ocfs2 fixes from Joel Becker.
* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jlbec/ocfs2:
aio: make kiocb->private NUll in init_sync_kiocb()
ocfs2: Fix bogus error message from ocfs2_global_read_info
ocfs2: for SEEK_DATA/SEEK_HOLE, return internal error unchanged if ocfs2_get_clusters_nocache() or ocfs2_inode_lock() call failed.
ocfs2: use spinlock irqsave for downconvert lock.patch
ocfs2: Misplaced parens in unlikley
ocfs2: clear unaligned io flag when dio fails
* git://git.samba.org/sfrench/cifs-2.6:
cifs: when server doesn't set CAP_LARGE_READ_X, cap default rsize at MaxBufferSize
cifs: fix parsing of password mount option
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Pull input layer fixes from Dmitry Torokhov:
"Two fixes for regressions in Wacom driver and fixes for drivers using
threaded IRQ framework without specifying IRQF_ONESHOT."
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: request threaded-only IRQs with IRQF_ONESHOT
Input: wacom - don't retrieve touch_max when it is predefined
Input: wacom - fix retrieving touch_max bug
Input: fix input.h kernel-doc warning
Dan Carpenter [Sat, 9 Jun 2012 16:08:25 +0000 (19:08 +0300)]
mtd: cafe_nand: fix an & vs | mistake
The intent here was clearly to set result to true if the 0x40000000 flag
was set. But instead there was a | vs & typo and we always set result
to true.
Artem: check the spec at
wiki.laptop.org/images/5/5c/88ALP01_Datasheet_July_2007.pdf
and this fix looks correct.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Cc: stable@vger.kernel.org Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Pull ARM SoC fixes from Arnd Bergmann:
"Small fixes on multiple ARM platforms
- A build regression from a previous fix on dove and mv78xx0
- Two fixes for recently (3.5-rc1) changed mmp/pxa code
- multiple omap2+ bug fixes
- two trivial fixes for i.MX
- one v3.5 regression for mxs"
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
ARM: apx4devkit: fix FEC enabling PHY clock
ARM: OMAP2+: hwmod data: Fix wrong McBSP clock alias on OMAP4
ARM: OMAP4: hwmod data: temporarily comment out data for the usb_host_fs and aess IP blocks
ARM: Orion: Fix WDT compile for Dove and MV78xx0
ARM: mmp: remove mach/gpio-pxa.h
ARM: imx: assert SCC gate stays enabled
ARM: OMAP4: TWL6030: ensure sys_nirq1 is mux'd and wakeup enabled
ARM: OMAP2: Overo: init I2C before MMC to fix MMC suspend/resume failure
ARM: imx27_visstrim_m10: Do not include <asm/system.h>
ARM: pxa: hx4700: Fix basic suspend/resume
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs
Pull btrfs updates from Chris Mason:
"I held off on my rc5 pull because I hit an oops during log recovery
after a crash. I wanted to make sure it wasn't a regression because
we have some logging fixes in here.
It turns out that a commit during the merge window just made it much
more likely to trigger directory logging instead of full commits,
which exposed an old bug.
The new backref walking code got some additional fixes. This should
be the final set of them.
Josef fixed up a corner where our O_DIRECT writes and buffered reads
could expose old file contents (not stale, just not the most recent).
He and Liu Bo fixed crashes during tree log recover as well.
Ilya fixed errors while we resume disk balancing operations on
readonly mounts."
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
Btrfs: run delayed directory updates during log replay
Btrfs: hold a ref on the inode during writepages
Btrfs: fix tree log remove space corner case
Btrfs: fix wrong check during log recovery
Btrfs: use _IOR for BTRFS_IOC_SUBVOL_GETFLAGS
Btrfs: resume balance on rw (re)mounts properly
Btrfs: restore restriper state on all mounts
Btrfs: fix dio write vs buffered read race
Btrfs: don't count I/O statistic read errors for missing devices
Btrfs: resolve tree mod log locking issue in btrfs_next_leaf
Btrfs: fix tree mod log rewind of ADD operations
Btrfs: leave critical region in btrfs_find_all_roots as soon as possible
Btrfs: always put insert_ptr modifications into the tree mod log
Btrfs: fix tree mod log for root replacements at leaf level
Btrfs: support root level changes in __resolve_indirect_ref
Btrfs: avoid waiting for delayed refs when we must not
Merge branch 'fixes-for-grant' of git://sources.calxeda.com/kernel/linux
Pull DT fixes from Rob Herring:
"Mainly some documentation updates and 2 fixes:
- An export symbol fix for of_platform_populate from Stephen W.
- A fix for the order compatible entries are matched to ensure the
first compatible string is matched when there are multiple matches."
Normally these would go through Grant Likely (thus the "fixes-for-grant"
branch name), but Grant is in the middle of moving to Scotland, and is
practically offline until sometime in August. So pull directly from Rob.
* 'fixes-for-grant' of git://sources.calxeda.com/kernel/linux:
of: match by compatible property first
dt: mc13xxx.txt: Fix gpio number assignment
dt: fsl-fec.txt: Fix gpio number assignment
dt: fsl-mma8450.txt: Add missing 'reg' description
dt: fsl-imx-esdhc.txt: Fix gpio number assignment
dt: fsl-imx-cspi.txt: Fix comment about GPIOs used for chip selects
of: Add Avionic Design vendor prefix
of: export of_platform_populate()
Russell King [Thu, 5 Jul 2012 12:11:31 +0000 (13:11 +0100)]
ARM: fix warning caused by wrongly typed arm_dma_limit
arch/arm/mm/init.c: In function 'arm_memblock_init':
arch/arm/mm/init.c:380: warning: comparison of distinct pointer types lacks a cast
by fixing the typecast in its definition when DMA_ZONE is disabled.
This was missed in 4986e5c7c (ARM: mm: fix type of the arm_dma_limit
global variable).
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Russell King [Thu, 5 Jul 2012 12:06:32 +0000 (13:06 +0100)]
ARM: fix warnings about atomic64_read
Fix:
net/netfilter/xt_connbytes.c: In function 'connbytes_mt':
net/netfilter/xt_connbytes.c:43: warning: passing argument 1 of 'atomic64_read' discards qualifiers from pointer target type
...
by adding the missing const.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Rabin Vincent [Wed, 4 Jul 2012 06:37:37 +0000 (07:37 +0100)]
ARM: 7440/1: kprobes: only test 'sub pc, pc, #1b-2b+8-2' on ARMv6
'sub pc, pc, #1b-2b+8-2' results in address<1:0> == '10'.
sub pc, pc, #const (== ADR pc, #const) performs an interworking branch
(BXWritePC()) on ARMv7+ and a simple branch (BranchWritePC()) on earlier
versions.
In ARM state, BXWritePC() is UNPREDICTABLE when address<1:0> == '10'.
In ARM state on ARMv6+, BranchWritePC() ignores address<1:0>. Before
ARMv6, BranchWritePC() is UNPREDICTABLE if address<1:0> != '00'
So the instruction is UNPREDICTABLE both before and after v6.
Acked-by: Jon Medhurst <tixy@yxit.co.uk> Signed-off-by: Rabin Vincent <rabin.vincent@stericsson.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Merge tag 'omap-fixes-for-v3.5-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap into fixes
PM related fixes for omaps mostly to get suspend/resume
working again.
* tag 'omap-fixes-for-v3.5-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap:
ARM: OMAP2+: hwmod data: Fix wrong McBSP clock alias on OMAP4
ARM: OMAP4: hwmod data: temporarily comment out data for the usb_host_fs and aess IP blocks
ARM: OMAP4: TWL6030: ensure sys_nirq1 is mux'd and wakeup enabled
ARM: OMAP2: Overo: init I2C before MMC to fix MMC suspend/resume failure
Will Deacon [Wed, 4 Jul 2012 17:15:42 +0000 (18:15 +0100)]
ARM: 7441/1: perf: return -EOPNOTSUPP if requested mode exclusion is unavailable
We currently return -EPERM if the user requests mode exclusion that is
not supported by the CPU. This looks pretty confusing from userspace
and is inconsistent with other architectures (ppc, x86).
This patch returns -EOPNOTSUPP instead.
Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The new syscall restarting code can lead to problems if we take an
interrupt in userspace just before restarting the svc instruction. If
a signal is delivered when returning from the interrupt, the
TIF_SYSCALL_RESTARTSYS will remain set and cause any syscalls executed
from the signal handler to be treated as a restart of the previously
interrupted system call. This includes the final sigreturn call, meaning
that we may fail to exit from the signal context. Furthermore, if a
system call made from the signal handler requires a restart via the
restart_block, it is possible to clear the thread flag and fail to
restart the originally interrupted system call.
The right solution to this problem is to perform the restarting in the
kernel, avoiding the possibility of handling a further signal before the
restart is complete. Since we're almost at -rc6, let's revert the new
method for now and aim for in-kernel restarting at a later date.
Acked-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
We need the restart trampoline back so that we can revert a related
problematic patch 6b5c8045ecc7e726cdaa2a9d9c8e5008050e1252 ("arm: new
way of handling ERESTART_RESTARTBLOCK").
Acked-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Junxiao Bi [Wed, 27 Jun 2012 09:09:54 +0000 (17:09 +0800)]
aio: make kiocb->private NUll in init_sync_kiocb()
Ocfs2 uses kiocb.*private as a flag of unsigned long size. In
commit a11f7e6 ocfs2: serialize unaligned aio, the unaligned
io flag is involved in it to serialize the unaligned aio. As
*private is not initialized in init_sync_kiocb() of do_sync_write(),
this unaligned io flag may be unexpectly set in an aligned dio.
And this will cause OCFS2_I(inode)->ip_unaligned_aio decreased
to -1 in ocfs2_dio_end_io(), thus the following unaligned dio
will hang forever at ocfs2_aiodio_wait() in ocfs2_file_aio_write().
Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com> Cc: stable@vger.kernel.org Acked-by: Jeff Moyer <jmoyer@redhat.com> Signed-off-by: Joel Becker <jlbec@evilplan.org>
Input: request threaded-only IRQs with IRQF_ONESHOT
Since commit 1c6c69525b ("genirq: Reject bogus threaded irq requests")
threaded IRQs without a primary handler need to be requested with
IRQF_ONESHOT, otherwise the request will fail. This patch adds the
IRQF_ONESHOT to input drivers where it is missing. Not modified by
this patch are those drivers where the requested IRQ will always be a
nested IRQ (e.g. because it's part of an MFD), since for this special
case IRQF_ONESHOT is not required to be specified when requesting the
IRQ.
Paul Walmsley [Wed, 4 Jul 2012 12:55:29 +0000 (06:55 -0600)]
ARM: OMAP4: hwmod data: temporarily comment out data for the usb_host_fs and aess IP blocks
The OMAP4 usb_host_fs (OHCI) and AESS IP blocks require some special
programming for them to enter idle. Without this programming, they
will prevent the rest of the chip from entering full chip idle.
To implement the idle programming cleanly, this will take some
coordination between maintainers. This is likely to take some time,
so it is probably best to leave this for 3.6 or 3.7. So, in the
meantime, prevent these IP blocks from being registered.
Later, once the appropriate support is available, this patch can be
reverted.
This second version comments out the IP block data since Benoît didn't
like removing it.
Signed-off-by: Paul Walmsley <paul@pwsan.com> Cc: Benoît Cousson <b-cousson@ti.com>