profile/ivi/kernel-x86-ivi.git
14 years agommc: add a file to debugfs for changing host clock at runtime
Andy Shevchenko [Wed, 13 Oct 2010 08:22:22 +0000 (11:22 +0300)]
mmc: add a file to debugfs for changing host clock at runtime

For debugging power management features it is convenient to have the
possibility of changing the MMC host controller clock at runtime.  This
patch adds a 'clock' file for this under the MMC host root of debugfs.

Usage is as follows:

# cat /sys/kernel/debug/mmc0/clock
52000000

# echo "1000000000" > /sys/kernel/debug/mmc0/clock
# cat /sys/kernel/debug/mmc0/clock
52000000

# echo "48000000" > /sys/kernel/debug/mmc0/clock
# cat /sys/kernel/debug/mmc0/clock
48000000

The middle example shows limits being applied by the host driver.

Signed-off-by: Andy Shevchenko <ext-andriy.shevchenko@nokia.com>
Cc: Adrian Hunter <adrian.hunter@nokia.com>
[cjb: modify changelog language]
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: allow for eMMC 74 clock generation by controller
Philip Rakity [Thu, 23 Sep 2010 15:24:32 +0000 (08:24 -0700)]
mmc: sdhci: allow for eMMC 74 clock generation by controller

Snippet of code for how adaptation layer should handle the call:
/*
 * eMMC spec calls for the host to send 74 clocks to the card
 * during initialization, right after voltage stabilization.
 * create the clocks manually right here.
 */
void generate_init_clocks_A0(struct sdhci_host *host, u8 power_mode)
{
struct sdhci_mmc_slot *slot = sdhci_priv(host);

if (slot->power_mode == MMC_POWER_UP &&
    power_mode == MMC_POWER_ON) {
/* controller specific code here */
/* slot->power_mode holds previous power setting */
}
slot->power_mode = power_mode;
}

Signed-off-by: Philip Rakity <prakity@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: highspeed: check for mmc as well as sd cards
Philip Rakity [Wed, 6 Oct 2010 18:57:23 +0000 (11:57 -0700)]
mmc: sdhci: highspeed: check for mmc as well as sd cards

The SD and MMC code set highspeed using different constants.
Change the sd driver to recognize this and switch to high speed.
Validated code when testing eMMC dual data rate.

Signed-off-by: Philip Rakity <prakity@marvell.com>
[cjb: changelog + indentation fixes]
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: Add Moorestown device support
Alan Cox [Mon, 4 Oct 2010 14:25:11 +0000 (15:25 +0100)]
mmc: sdhci: Add Moorestown device support

This adds the basic identifiers. Due to the various chip quirks it's not
enough to make MRST support very useful for earlier steppings but that can
follow.

(I'm currently trying to verify which steps actually matter outside Intel
 so I can avoid unneeded stuff going upstream)

[Extracted from original development]
Signed-off-by: JiebingLi <jiebing.li@intel.com>
[Folds in fixes]
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: Intel Medfield support
Xiaochen Shen [Mon, 4 Oct 2010 14:24:52 +0000 (15:24 +0100)]
mmc: sdhci: Intel Medfield support

Basic support for the Intel Medfield devices

Give them their own quirks as we will need to update this later.

Signed-off-by: Xiaochen Shen <xiaochen.shen@intel.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: Allow the probe handler to override slots
Alan Cox [Mon, 4 Oct 2010 14:24:21 +0000 (15:24 +0100)]
mmc: sdhci: Allow the probe handler to override slots

Currently we write it to the chip data, but if the probe handler overrides
it we ignore the new value and keep using our cached one. Fix this so that
a probe handler can adjust the slot count.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdio: support suspend/resume while runtime suspended
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:13 +0000 (13:54 +0200)]
mmc: sdio: support suspend/resume while runtime suspended

Bring SDIO devices back to full power before their suspend
handler is invoked.

Doing so ensures that SDIO suspend/resume semantics are
maintained (drivers still get to decide whether their
card should be removed or kept during system suspend,
and at what power state), and that SDIO suspend/resume
execution paths are unchanged.

This is achieved by resuming a runtime-suspended SDIO device
in its ->prepare() PM callback (similary to the PCI subsystem).

Since the PM core always increments the run-time usage
counter before calling the ->prepare() callback and decrements
it after calling the ->complete() callback, it is guaranteed
that when the system will come out of suspend, our device's
power state will reflect its runtime PM usage counter.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdio: ensure mmc_sdio_detect is powered
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:12 +0000 (13:54 +0200)]
mmc: sdio: ensure mmc_sdio_detect is powered

To prevent an erroneous removal of the card, make sure
the device is powered when it is mmc_sdio_detect()ed.

This is required since mmc_sdio_detect may be invoked
while the device is runtime suspended (e.g., MMC core
is rescanning when system comes out of suspend).

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdio: enable runtime PM for SDIO functions
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:11 +0000 (13:54 +0200)]
mmc: sdio: enable runtime PM for SDIO functions

Enable runtime PM for SDIO functions.

SDIO functions are initialized with a disabled runtime PM state,
and are set active (and their usage count is incremented)
only before potential drivers are probed.

SDIO function drivers that support runtime PM should call
pm_runtime_put_noidle() in their probe routine, and
pm_runtime_get_noresume() in their remove routine (very
similarly to PCI drivers).

In case a matching driver does not support runtime PM, power will
always be kept high (since the usage count is positive).

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdio: enable runtime PM for SDIO cards
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:10 +0000 (13:54 +0200)]
mmc: sdio: enable runtime PM for SDIO cards

Enable runtime PM for new SDIO cards.

As soon as the card will be added to the device tree, runtime PM core
will release its power, since it doesn't have any users yet.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdio: use the generic runtime PM handlers
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:09 +0000 (13:54 +0200)]
mmc: sdio: use the generic runtime PM handlers

Assign the generic runtime PM handlers for SDIO.

These handlers invoke the relevant SDIO function drivers'
handlers, if exist, otherwise they just return success
(so SDIO drivers don't have to define any runtime PM handlers
unless they need to).

Runtime PM is still disabled by default, so this patch alone
has no immediate effect.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: add runtime PM handlers
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:08 +0000 (13:54 +0200)]
mmc: add runtime PM handlers

Add MMC runtime PM handlers, which call mmc_power_save_host
and mmc_power_restore_host in response to runtime_suspend and
runtime_resume events.

Runtime PM is still disabled by default, so this patch alone
has no immediate effect.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdio: add power_restore support
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:07 +0000 (13:54 +0200)]
mmc: sdio: add power_restore support

Add a power_restore handler to the SDIO bus ops,
in order to support waking up SDIO cards that
were powered off by runtime pm.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: propagate power save/restore ops return value
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:06 +0000 (13:54 +0200)]
mmc: propagate power save/restore ops return value

Allow power save/restore and their relevant mmc_bus_ops handlers
exit with a return value.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdio: fully reconfigure oldcard on resume
Ohad Ben-Cohen [Sat, 2 Oct 2010 11:54:05 +0000 (13:54 +0200)]
mmc: sdio: fully reconfigure oldcard on resume

On resume, let mmc_sdio_init_card go all the way, instead
of skipping the reconfiguration of the card's speed and width.

This is needed to ensure cards wake up with their clock
reconfigured (otherwise it's kept low).

This patch also removes the explicit bus width reconfiguration
on resume, since now this is part of mmc_sdio_init_card.

Signed-off-by: Ohad Ben-Cohen <ohad@wizery.com>
Tested-by: Luciano Coelho <luciano.coelho@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: split up sdhci.h for sdhci-pltfm users
Giuseppe Cavallaro [Tue, 28 Sep 2010 08:41:29 +0000 (10:41 +0200)]
mmc: sdhci: split up sdhci.h for sdhci-pltfm users

Some platforms based on sdhci-pltfm need to set their own quirks.
Previously to this patch, the quirks were in drivers/mmc/host/sdhci.h.

This patch splits drivers/mmc/host/sdhci.h into two parts:

* drivers/mmc/host/sdhci.h  includes the HC registers and I/O accessors.
* include/linux/mmc/sdhci.h includes the sdhci structure and quirks.

Instead of including drivers/mmc/host/sdhci.h, -pltfm drivers should
now include include/linux/mmc/sdhci.h and include/linux/sdhci-pltfm.h.

This patch avoids adding/changing the calls/flags in the
sdhci_pltfm_data structure.  It has been tested on STM platforms
(e.g. STx7106, STx7108, STx5206) where the driver is configured
and used as shown in the example below:

[snip]
static int mmc_pad_resources(struct sdhci_host *sdhci)
{
if (!devm_stm_pad_claim(sdhci->mmc->parent,
&stx7108_mmc_pad_config,
dev_name(sdhci->mmc->parent)))
return -ENODEV;

return 0;
}

static struct sdhci_pltfm_data stx7108_mmc_platform_data = {
.init = mmc_pad_resources,
.quirks = SDHCI_QUIRK_NO_ENDATTR_IN_NOPDESC,
};

static struct platform_device stx7108_mmc_device = {
.name = "sdhci",
[snip]

Note: drivers/mmc/host/sdhci.h now also includes linux/mmc/sdhci.h,
and no modifications should be needed on other sdhci-<XXX> drivers.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: fix "pwr may be used uninitialized" warning
Giuseppe Cavallaro [Tue, 28 Sep 2010 08:41:28 +0000 (10:41 +0200)]
mmc: sdhci: fix "pwr may be used uninitialized" warning

This patch fixes a warning when compiling the sdhci driver:
  pwr may be used uninitialized in sdhci_set_power

Tested with the following compiler versions: 4.2.4 and 4.4.4

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci-pltfm: add suspend/resume functions
Giuseppe Cavallaro [Tue, 28 Sep 2010 08:41:27 +0000 (10:41 +0200)]
mmc: sdhci-pltfm: add suspend/resume functions

This patch adds the suspend and resume functions
in the sdhci-pltfm device driver.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: name mmc queue thread by host index
Ethan Du [Thu, 30 Sep 2010 22:40:27 +0000 (18:40 -0400)]
mmc: name mmc queue thread by host index

Usually there are multiple mmc host controllers; rename mmc queue thread
by host index so we can easily identify which controller it belongs to.

Signed-off-by: Ethan Du <ethan.too@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: refine DDR support
Adrian Hunter [Mon, 11 Oct 2010 09:43:50 +0000 (12:43 +0300)]
mmc: refine DDR support

One flaw with DDR support is that MMC core does not inform the driver
which DDR mode it has selected.  This patch expands the ios->ddr flag
to do that.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Fixes for Dual Data Rate (DDR) support
Adrian Hunter [Tue, 24 Aug 2010 10:20:26 +0000 (13:20 +0300)]
mmc: Fixes for Dual Data Rate (DDR) support

The DDR support patch needs the following fixes:

- The block driver does not need to know about DDR, any more
  than it needs to know about bus width.
- Not only the card must be switched to DDR mode.  The host
  controller must also be configured, which is done through
  the 'set_ios()' function.
- Do not set the DDR mode state until after the switch command
  is successful.
- Setting block length is not supported in DDR mode.  Make that
  a core function and change the other place it is used (mmc_test)
  also.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: MMC 4.4 DDR support
Hanumath Prasad [Thu, 30 Sep 2010 21:37:23 +0000 (17:37 -0400)]
mmc: MMC 4.4 DDR support

Add support for Dual Data Rate MMC cards as defined in the 4.4
specification.

Signed-off-by: Hanumath Prasad <hanumath.prasad@stericsson.com>
Cc: linux-mmc@vger.kernel.org
Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Tested-by Zhangfei Gao <zhangfei.gao@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Move regulator handling closer to core
Linus Walleij [Wed, 29 Sep 2010 05:08:27 +0000 (01:08 -0400)]
mmc: Move regulator handling closer to core

After discovering a problem in regulator reference counting I took Mark
Brown's advice to move the reference count into the MMC core by making the
regulator status a member of struct mmc_host.

I took this opportunity to also implement NULL versions of
the regulator functions so as to rid the driver code from
some ugly #ifdef CONFIG_REGULATOR clauses.

Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Reviewed-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: Liam Girdwood <lrg@slimlogic.co.uk>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Adrian Hunter <adrian.hunter@nokia.com>
Cc: Robert Jarzmik <robert.jarzmik@free.fr>
Cc: Sundar Iyer <sundar.iyer@stericsson.com>
Cc: Daniel Mack <daniel@caiaq.de>
Cc: Pierre Ossman <pierre@ossman.eu>
Cc: Matt Fleming <matt@console-pimps.org>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Eric Miao <eric.y.miao@gmail.com>
Cc: Cliff Brake <cbrake@bec-systems.com>
Cc: Jarkko Lavinen <jarkko.lavinen@nokia.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: mmc_rescan: reduce verbosity
Eric Bénard [Tue, 12 Oct 2010 07:29:37 +0000 (09:29 +0200)]
mmc: sdhci: mmc_rescan: reduce verbosity

mmc_rescan() includes a pr_info which prints 4 lines each second for
hosts configured with MMC_CAP_NEEDS_POLL. This patch enables the message
only if CONFIG_MMC_DEBUG is selected. Tested on i.MX51's sdhci-esdhc.

Signed-off-by: Eric Bénard <eric@eukrea.com>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Acked-by: Hein Tibosch <hein_tibosch@yahoo.es>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Make ID freq configurable
Hein Tibosch [Mon, 6 Sep 2010 01:37:19 +0000 (09:37 +0800)]
mmc: Make ID freq configurable

In the latest releases of the mmc driver, the freq during initialization
is set to a fixed 400 Khz.  This was reportedly too fast for several
users.  As there doesn't seem to be an ideal frequency
which-works-for-all, Pierre suggested to let the driver try several
frequencies.

This patch implements that idea. It will try mmc-initialization using
several frequencies from an array 400, 300, 200 and 100.

In case SDIO is broken, it'll still try to detect SDMEM, also at different
freqs.

Signed-off-by: Hein Tibosch <hein_tibosch@yahoo.es>
Cc: Pierre Ossman <pierre@ossman.eu>
Reviewed-by: Chris Ball <cjb@laptop.org>
Tested-by: Chris Ball <cjb@laptop.org>
Cc: Ben Nizette <bn@niasdigital.com>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Adrian Hunter <adrian.hunter@nokia.com>
Cc: Matt Fleming <matt@console-pimps.org>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: disable MMC_CAP_NEEDS_POLL in nonremovable case
Jaehoon Chung [Mon, 27 Sep 2010 08:42:20 +0000 (09:42 +0100)]
mmc: sdhci: disable MMC_CAP_NEEDS_POLL in nonremovable case

When a controller requires SDHCI_QUIRK_BROKEN_CARD_DETECTION, we poll
for card insertion/removal, and that creates interrupts.  There's no
need to be doing this if we have a non-removable card.

This patch requires cards to be removable before we're willing to set
MMC_CAP_NEEDS_POLL.

Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Acked-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Matt Fleming <matt@console-pimps.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
[cjb: modified changelog and code indentation]
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Add helper function to check if a card is removable
Matt Fleming [Mon, 27 Sep 2010 08:42:19 +0000 (09:42 +0100)]
mmc: Add helper function to check if a card is removable

There are two checks that need to be made when determining whether a
card is removable. A host controller may set MMC_CAP_NONREMOVABLE if the
controller does not support removing cards (e.g. eMMC), in which case
the card is physically non-removable. Also the 'mmc_assume_removable'
module parameter can be configured at module load time, in which case
the card may be logically non-removable.

A helper function keeps the logic in one place so that code always
checks both conditions.

Because this new function is likely to be called from modules we now
need to export the mmc_assume_removable symbol.

Signed-off-by: Matt Fleming <matt@console-pimps.org>
Acked-by: Kyungmin Park <kyungmin.park@samsung.com>
Tested-by: Jaehoon Chung <jh80.chung@samsung.com>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Makefile: Fix EXTRA_CFLAGS assignment
matt mooney [Fri, 24 Sep 2010 19:17:24 +0000 (12:17 -0700)]
mmc: Makefile: Fix EXTRA_CFLAGS assignment

The EXTRA_CFLAGS assignment in mmc/Makefile was not accomplishing
anything because this flag only has effect on sources at the same level
as the makefile (i.e., per directory). Since card/, core/, and host/
rely on MMC_DEBUG, the subdir-ccflags-y variant seems to be the
appropriate choice.

Signed-off-by: matt mooney <mfm@muteddisk.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: print out controller name for register debug
Philip Rakity [Wed, 22 Sep 2010 22:25:13 +0000 (15:25 -0700)]
mmc: sdhci: print out controller name for register debug

In a multi-controller environment it is helpful to know which controller
has problems.

Signed-off-by: Philip Rakity <prakity@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: add MMC_CAP_8_BIT_DATA in the host capabilities
Giuseppe Cavallaro [Tue, 21 Sep 2010 01:22:13 +0000 (21:22 -0400)]
mmc: sdhci: add MMC_CAP_8_BIT_DATA in the host capabilities

This patch is necessary to gain the performance boost from 8-bit data
with the sdhci-stm driver.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: Enable high-speed support for MMC cards
Zhangfei Gao [Tue, 17 Aug 2010 01:15:32 +0000 (21:15 -0400)]
mmc: sdhci: Enable high-speed support for MMC cards

MMC_CAP_MMC_HIGHSPEED allows MMC and eMMC to negotiate up to 50M
instead of the previous limit of 25M.

Signed-off-by: Zhangfei Gao <zgao6@marvell.com>
Acked-by: Matt Fleming <matt@console-pimps.org>
Acked-by: Kyungmin Park <kmpark@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: SDHC 3.0: correct f_min calculation for SD 3.0 spec
Zhangfei Gao [Mon, 20 Sep 2010 19:15:18 +0000 (15:15 -0400)]
mmc: SDHC 3.0: correct f_min calculation for SD 3.0 spec

While we're at it, add symbols for SDHCI_MAX_DIV_SPEC_{200,300}.

Signed-off-by: Zhangfei Gao <zhangfei.gao@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: SDHC 3.0: Base clock frequency change in spec 3.0
Zhangfei Gao [Fri, 20 Aug 2010 18:02:36 +0000 (14:02 -0400)]
mmc: SDHC 3.0: Base clock frequency change in spec 3.0

SDHC Spec 3.0: Capabilities Register bits[15-08] are Base Clock Frequency
      1.0/2.0: Capabilities Register bits[13-08] are Base Clock Frequency

Signed-off-by: Zhangfei Gao <zgao6@marvell.com>
Cc: David Vrabel <david.vrabel@csr.com>
Cc: Matt Fleming <matt@console-pimps.org>
Cc: Michal Miroslaw <mirqus@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: SDHC 3.0: support 10-bit divided clock mode
Zhangfei Gao [Thu, 5 Aug 2010 23:10:01 +0000 (07:10 +0800)]
mmc: SDHC 3.0: support 10-bit divided clock mode

Signed-off-by: Zhangfei Gao <zgao6@marvell.com>
Cc: Michał Mirosław <mirqus@gmail.com>
Cc: David Vrabel <david.vrabel@csr.com>
Reviewed-by: Matt Fleming <matt@console-pimps.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: sdhci: Make consistent use of CONFIG_MMC_DEBUG for "DEBUG"ging
George G. Davis [Thu, 18 Feb 2010 17:32:12 +0000 (12:32 -0500)]
mmc: sdhci: Make consistent use of CONFIG_MMC_DEBUG for "DEBUG"ging

The "6882a8c sdhci: Add better ADMA error reporting" commit added
sdhci_show_adma_error() which is built when DEBUG is defined.  Since we
already have CONFIG_MMC_DEBUG used elsewhere in this driver, may as well
make consistent use of that config knob instead.

Signed-off-by: George G. Davis <gdavis@mvista.com>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc_test: fix display of .5 KiB
Adrian Hunter [Thu, 23 Sep 2010 11:51:36 +0000 (14:51 +0300)]
mmc_test: fix display of .5 KiB

Append .5 to KiB display when there are an odd number of sectors.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc_test: fix memory allocation segment limits
Adrian Hunter [Thu, 23 Sep 2010 11:51:29 +0000 (14:51 +0300)]
mmc_test: fix memory allocation segment limits

Correctly allocate memory to meet the host controller
driver's maximum segment size and count limits.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc_test: move files from sysfs to debugfs
Andy Shevchenko [Fri, 10 Sep 2010 07:10:50 +0000 (10:10 +0300)]
mmc_test: move files from sysfs to debugfs

As proposed by Greg K-H it is more logical to keep files for the mmc_test
driver under debugfs.

Additionally this patch brings seq_file API for show() method.  It allows
to write unlimited data to the file.

Example of usage:
  # mount -t debugfs none /sys/kernel/debug
  # modprobe mmc_test
    [  581.395843] mmc_test mmc0:0001: Card claimed for testing.
  # echo 25 > /sys/kernel/debug/mmc0/mmc0\:0001/test
    [  604.568542] mmc0: Starting tests of card mmc0:0001...
    [  604.582733] mmc0: Test case 25. Best-case read performance into scattered pages...
    [  604.923553] mmc0: Transfer of 8192 sectors (4096 KiB) took 0.124664314 seconds (33644 kB/s, 32856 KiB/s)
    [  604.933227] mmc0: Result: OK
    [  604.936248] mmc0: Tests completed.
  # cat /sys/kernel/debug/mmc0/mmc0\:0001/test
    Test 25: 0
    1 8192 0.124664314 33644784

Signed-off-by: Andy Shevchenko <ext-andriy.shevchenko@nokia.com>
Cc: Greg KH <greg@kroah.com>
Cc: Adrian Hunter <adrian.hunter@nokia.com>
Cc: Chris Ball <cjb@laptop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc_test: collect data and show it via sysfs by demand
Andy Shevchenko [Wed, 1 Sep 2010 06:26:47 +0000 (09:26 +0300)]
mmc_test: collect data and show it via sysfs by demand

Make it possible to get test results via sysfs.  It helps to do tests
non-interactively.  We have the file created under sysfs already and
can use it to show test results.

Prior to this patch, the "test" file under each card's sysfs node was
write-only, and results were obtained by looking at dmesg.  This patch
improves programmatic access to the test results, making them available by
reading back from the same "test" file:

[root@host mmc0:e624]# echo 6 > test
[root@host mmc0:e624]# cat test
Test 6: 2

[cjb@laptop.org: changelog improvements]
Signed-off-by: Andy Shevchenko <ext-andriy.shevchenko@nokia.com>
Cc: Chris Ball <cjb@laptop.org>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc_test: change simple_strtol() to strict_strtol()
Andy Shevchenko [Wed, 1 Sep 2010 06:26:46 +0000 (09:26 +0300)]
mmc_test: change simple_strtol() to strict_strtol()

It's better to use strict_strtol() to convert user's input and strictly
check it. At least it forbids to interpret wrong input as a 0 and
prevents to run all tests.

Signed-off-by: Andy Shevchenko <ext-andriy.shevchenko@nokia.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc_test: use API to check card type
Andy Shevchenko [Wed, 1 Sep 2010 06:26:45 +0000 (09:26 +0300)]
mmc_test: use API to check card type

There are methods to check card type. Let's use them instead of direct checking
type bits.

Signed-off-by: Andy Shevchenko <ext-andriy.shevchenko@nokia.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc_test: fix performance tests that go over max_blk_count
Adrian Hunter [Fri, 10 Sep 2010 08:33:45 +0000 (11:33 +0300)]
mmc_test: fix performance tests that go over max_blk_count

The host controller driver limits I/O transfers to maximum
transfer size, maximum block count, maximum segment size
and maximum segment count.  The performance tests were
not obeying these limits which meant they would not work
with some drivers.  This patch fixes that.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Convert "mutex" to semaphore
Thomas Gleixner [Tue, 14 Sep 2010 11:12:35 +0000 (07:12 -0400)]
mmc: Convert "mutex" to semaphore

Get rid of init_MUTEX[_LOCKED]() and use sema_init() instead.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-mmc@vger.kernel.org
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: rename dev_to_mmc_card() to mmc_dev_to_card()
Andy Shevchenko [Sat, 18 Sep 2010 00:32:25 +0000 (20:32 -0400)]
mmc: rename dev_to_mmc_card() to mmc_dev_to_card()

Global symbols should use their subsystem name in a prefixed fashion.

Signed-off-by: Andy Shevchenko <ext-andriy.shevchenko@nokia.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: make mmc_dev_to_card() macro public
Andy Shevchenko [Fri, 20 Aug 2010 07:46:46 +0000 (10:46 +0300)]
mmc: make mmc_dev_to_card() macro public

Conversion from struct device to struct mmc_card is used more than in one
place.  Due to this it's better to have public macro for such thing.

Signed-off-by: Andy Shevchenko <ext-andriy.shevchenko@nokia.com>
Cc: Adrian Hunter <adrian.hunter@nokia.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Use snprintf, not sprintf.
JiebingLi [Thu, 17 Jun 2010 14:58:46 +0000 (15:58 +0100)]
mmc: Use snprintf, not sprintf.

Fix an issue found by klockwork. Just paranoia.

Signed-off-by: JiebingLi <jiebing.li@intel.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: USB SD Host Controller (USHC) driver
David Vrabel [Thu, 2 Sep 2010 14:15:08 +0000 (14:15 +0000)]
mmc: USB SD Host Controller (USHC) driver

Add a driver for USB SD Host Controller devices.  These devices are
Cypress Astoria chips with firmware compliant with issue 2 of CSR's USHC
specification.

[cjb: adapt to block layer deprecation of max_{hw,phys}_segs]
Signed-off-by: David Vrabel <david.vrabel@csr.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: Remove distinction between hw and phys segments
Martin K. Petersen [Fri, 10 Sep 2010 05:33:59 +0000 (01:33 -0400)]
mmc: Remove distinction between hw and phys segments

We have deprecated the distinction between hardware and physical
segments in the block layer.  Consolidate the two limits into one in
drivers/mmc/.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: au1xmmc.c: use resource_size()
H Hartley Sweeten [Mon, 14 Dec 2009 19:28:06 +0000 (14:28 -0500)]
mmc: au1xmmc.c: use resource_size()

[cjb: rebased patch against Linus]
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: <linux-mmc@vger.kernel.org>
Acked-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: atmel-mci.c: use resource_size()
H Hartley Sweeten [Mon, 14 Dec 2009 19:11:56 +0000 (14:11 -0500)]
mmc: atmel-mci.c: use resource_size()

[cjb: rebased patch against Linus]
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agommc: at91_mci.c: use resource_size()
H Hartley Sweeten [Mon, 14 Dec 2009 19:10:26 +0000 (14:10 -0500)]
mmc: at91_mci.c: use resource_size()

[cjb: rebased patch against Linus]
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: <linux-mmc@vger.kernel.org>
Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke...
Linus Torvalds [Sat, 23 Oct 2010 08:26:47 +0000 (01:26 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ryusuke/nilfs2

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2: (36 commits)
  nilfs2: eliminate sparse warning - "context imbalance"
  nilfs2: eliminate sparse warnings - "symbol not declared"
  nilfs2: get rid of bdi from nilfs object
  nilfs2: change license of exported header file
  nilfs2: add bdev freeze/thaw support
  nilfs2: accept 64-bit checkpoint numbers in cp mount option
  nilfs2: remove own inode allocator and destructor for metadata files
  nilfs2: get rid of back pointer to writable sb instance
  nilfs2: get rid of mi_nilfs back pointer to nilfs object
  nilfs2: see state of root dentry for mount check of snapshots
  nilfs2: use iget for all metadata files
  nilfs2: get rid of GCDAT inode
  nilfs2: add routines to redirect access to buffers of DAT file
  nilfs2: add routines to roll back state of DAT file
  nilfs2: add routines to save and restore bmap state
  nilfs2: do not allocate nilfs_mdt_info structure to gc-inodes
  nilfs2: allow nilfs_clear_inode to clear metadata file inodes
  nilfs2: get rid of snapshot mount flag
  nilfs2: simplify life cycle management of nilfs object
  nilfs2: do not allocate multiple super block instances for a device
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier...
Linus Torvalds [Sat, 23 Oct 2010 04:12:27 +0000 (21:12 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/vapier/blackfin

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier/blackfin: (47 commits)
  Blackfin: bfin_spi.h: add MMR peripheral layout
  Blackfin: bfin_ppi.h: start a common PPI/EPPI header
  Blackfin: bfin_can.h: add missing VERSION/VERSION2 MMRs
  Blackfin: bf538: add missing SIC_RVECT define
  Blackfin: bf561: rewrite SICA_xxx to just SIC_xxx
  Blackfin: bf54x: add missing SIC_RVECT definition
  Blackfin: H8606: move 8250 irqflags to platform resources
  Blackfin: glue XIP/ROM kernel kconfigs
  Blackfin: update sparse flags for latest upstream changes
  Blackfin: coreb: update ioctl numbers
  Blackfin: coreb: add gpl module license
  Blackfin: bf518-ezkit: add ssm2603 codec resources
  Blackfin: bf51x/bf52x: fix 16/32bit SPORT MMR helpers
  Blackfin: tll6527m: new board port
  Blackfin: bf526-ezbrd/bf527-ezkit: add NAND partition for u-boot
  Blackfin: merge kernel init memory back into main memory region
  Blackfin: gpio: add peripheral group check
  Blackfin: dma: bf54x: add missing break for SPORT1 TX IRQ
  Blackfin: add new cacheflush syscall
  Blackfin: bf548-ezkit: increase u-boot partition size
  ...

14 years agoMerge branch 'x86-trampoline-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 23 Oct 2010 03:37:50 +0000 (20:37 -0700)]
Merge branch 'x86-trampoline-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-trampoline-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86-32, mm: Add an initial page table for core bootstrapping

14 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel...
Linus Torvalds [Sat, 23 Oct 2010 03:35:12 +0000 (20:35 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jwessel/linux-2.6-kgdb

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel/linux-2.6-kgdb:
  kdb,debug_core: adjust master cpu switch logic against new debug_core locking
  debug_core: refactor locking for master/slave cpus
  x86,kgdb: remove unnecessary call to kgdb_correct_hw_break()
  debug_core: disable hw_breakpoints on all cores in kgdb_cpu_enter()
  kdb,kgdb: fix sparse fixups
  kdb: Fix oops in kdb_unregister
  kdb,ftdump: Remove reference to internal kdb include
  kdb: Allow kernel loadable modules to add kdb shell functions
  debug_core: stop rcu warnings on kernel resume
  debug_core: move all watch dog syncs to a single function
  x86,kgdb: fix debugger hw breakpoint test regression in 2.6.35

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Sat, 23 Oct 2010 03:30:48 +0000 (20:30 -0700)]
Merge git://git./linux/kernel/git/gregkh/usb-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (141 commits)
  USB: mct_u232: fix broken close
  USB: gadget: amd5536udc.c: fix error path
  USB: imx21-hcd - fix off by one resource size calculation
  usb: gadget: fix Kconfig warning
  usb: r8a66597-udc: Add processing when USB was removed.
  mxc_udc: add workaround for ENGcm09152 for i.MX35
  USB: ftdi_sio: add device ids for ScienceScope
  USB: musb: AM35x: Workaround for fifo read issue
  USB: musb: add musb support for AM35x
  USB: AM35x: Add musb support
  usb: Fix linker errors with CONFIG_PM=n
  USB: ohci-sh - use resource_size instead of defining its own resource_len macro
  USB: isp1362-hcd - use resource_size instead of defining its own resource_len macro
  USB: isp116x-hcd - use resource_size instead of defining its own resource_len macro
  USB: xhci: Fix compile error when CONFIG_PM=n
  USB: accept some invalid ep0-maxpacket values
  USB: xHCI: PCI power management implementation
  USB: xHCI: bus power management implementation
  USB: xHCI: port remote wakeup implementation
  USB: xHCI: port power management implementation
  ...

Manually fix up (non-data) conflict: the SCSI merge gad renamed the
'hw_sector_size' member to 'physical_block_size', and the USB tree
brought a new use of it.

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6
Linus Torvalds [Sat, 23 Oct 2010 02:59:04 +0000 (19:59 -0700)]
Merge git://git./linux/kernel/git/gregkh/tty-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6: (49 commits)
  serial8250: ratelimit "too much work" error
  serial: bfin_sport_uart: speed up sport RX sample rate to be 3% faster
  serial: abstraction for 8250 legacy ports
  serial/imx: check that the buffer is non-empty before sending it out
  serial: mfd: add more baud rates support
  jsm: Remove the uart port on errors
  Alchemy: Add UART PM methods.
  8250: allow platforms to override PM hook.
  altera_uart: Don't use plain integer as NULL pointer
  altera_uart: Fix missing prototype for registering an early console
  altera_uart: Fixup type usage of port flags
  altera_uart: Make it possible to use Altera UART and 8250 ports together
  altera_uart: Add support for different address strides
  altera_uart: Add support for getting mapbase and IRQ from resources
  altera_uart: Add support for polling mode (IRQ-less)
  serial: Factor out uart_poll_timeout() from 8250 driver
  serial: mark the 8250 driver as maintained
  serial: 8250: Don't delay after transmitter is ready.
  tty: MAINTAINERS: add drivers/serial/jsm/ as maintained driver
  vcs: invoke the vt update callback when /dev/vcs* is written to
  ...

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6
Linus Torvalds [Sat, 23 Oct 2010 02:36:42 +0000 (19:36 -0700)]
Merge git://git./linux/kernel/git/gregkh/driver-core-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6: (31 commits)
  driver core: Display error codes when class suspend fails
  Driver core: Add section count to memory_block struct
  Driver core: Add mutex for adding/removing memory blocks
  Driver core: Move find_memory_block routine
  hpilo: Despecificate driver from iLO generation
  driver core: Convert link_mem_sections to use find_memory_block_hinted.
  driver core: Introduce find_memory_block_hinted which utilizes kset_find_obj_hinted.
  kobject: Introduce kset_find_obj_hinted.
  driver core: fix build for CONFIG_BLOCK not enabled
  driver-core: base: change to new flag variable
  sysfs: only access bin file vm_ops with the active lock
  sysfs: Fail bin file mmap if vma close is implemented.
  FW_LOADER: fix kconfig dependency warning on HOTPLUG
  uio: Statically allocate uio_class and use class .dev_attrs.
  uio: Support 2^MINOR_BITS minors
  uio: Cleanup irq handling.
  uio: Don't clear driver data
  uio: Fix lack of locking in init_uio_class
  SYSFS: Allow boot time switching between deprecated and modern sysfs layout
  driver core: remove CONFIG_SYSFS_DEPRECATED_V2 but keep it for block devices
  ...

14 years agoocfs2: drop the BLKDEV_IFL_WAIT flag
Linus Torvalds [Sat, 23 Oct 2010 02:30:38 +0000 (19:30 -0700)]
ocfs2: drop the BLKDEV_IFL_WAIT flag

Commit dd3932eddf42 ("block: remove BLKDEV_IFL_WAIT") had removed the
flag argument to blkdev_issue_flush(), but the ocfs2 merge brought in a
new one.  It didn't cause a merge conflict, so the merges silently
worked out fine, but the result didn't actually compile.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Sat, 23 Oct 2010 00:52:29 +0000 (17:52 -0700)]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6: (56 commits)
  [CIFS] move close processing  from cifs_close to cifsFileInfo_put
  cifs: convert cifs_tcp_ses_lock from a rwlock to a spinlock
  cifs: cancel_delayed_work() + flush_scheduled_work() -> cancel_delayed_work_sync()
  Clean up two declarations of blob_len
  cifs: move cifsFileInfo_put to file.c
  cifs: convert GlobalSMBSeslock from a rwlock to regular spinlock
  [CIFS] Fix minor checkpatch warning and update cifs version
  cifs: move cifs_new_fileinfo to file.c
  cifs: eliminate pfile pointer from cifsFileInfo
  cifs: cifs_write argument change and cleanup
  cifs: clean up cifs_reopen_file
  cifs: eliminate the inode argument from cifs_new_fileinfo
  cifs: eliminate oflags option from cifs_new_fileinfo
  cifs: fix flags handling in cifs_posix_open
  cifs: eliminate cifs_posix_open_inode_helper
  cifs: handle FindFirst failure gracefully
  NTLM authentication and signing - Calculate auth response per smb session
  cifs: don't use vfsmount to pin superblock for oplock breaks
  cifs: keep dentry reference in cifsFileInfo instead of inode reference
  cifs: on multiuser mount, set ownership to current_fsuid/current_fsgid (try #7)
  ...

Fix up trivial conflict in fs/cifs/cifsfs.c due to added/removed header files

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6
Linus Torvalds [Sat, 23 Oct 2010 00:34:15 +0000 (17:34 -0700)]
Merge git://git./linux/kernel/git/jejb/scsi-misc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (84 commits)
  [SCSI] be2iscsi: SGE Len == 64K
  [SCSI] be2iscsi: Remove premature free of cid
  [SCSI] be2iscsi: More time for FW
  [SCSI] libsas: fix bug for vacant phy
  [SCSI] sd: Fix overflow with big physical blocks
  [SCSI] st: add MTWEOFI to write filemarks without flushing drive buffer
  [SCSI] libsas: Don't issue commands to devices that have been hot-removed
  [SCSI] megaraid_sas: Add Online Controller Reset to MegaRAID SAS drive
  [SCSI] lpfc 8.3.17: Update lpfc driver version to 8.3.17
  [SCSI] lpfc 8.3.17: Replace function reset methodology
  [SCSI] lpfc 8.3.17: SCSI fixes
  [SCSI] lpfc 8.3.17: BSG fixes
  [SCSI] lpfc 8.3.17: SLI Additions and Fixes
  [SCSI] lpfc 8.3.17: Code Cleanup and Locking fixes
  [SCSI] zfcp: Remove scsi_cmnd->serial_number from debug traces
  [SCSI] ipr: fix array error logging
  [SCSI] aha152x: enable PCMCIA on 64bit
  [SCSI] scsi_dh_alua: Handle all states correctly
  [SCSI] cxgb4i: connection and ddp setting update
  [SCSI] cxgb3i: fixed connection over vlan
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux...
Linus Torvalds [Sat, 23 Oct 2010 00:33:35 +0000 (17:33 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/geert/linux-m68k

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k:
  m68k: Add missing I/O macros {in,out}{w,l}_p() for !CONFIG_ISA
  m68k: Remove big kernel lock in cache flush code
  m68k: __pa(): cast arg to long
  fbdev: atafb - Remove undead ifdef ATAFB_FALCON
  zorro: Fix device_register() error handling
  fbdev/m68k: Fix section mismatches in q40fb.c
  m68k/m68knommu: merge the MMU and non-MMU traps.h
  m68k/m68knommu: merge MMU and non-MMU thread_info.h
  m68k/m68knommu: merge MMU and non-MMU atomic.h
  m68k/m68knommu: clean up page.h
  m68k/m68knommu: merge machdep.h files into a single file
  m68k/m68knommu: merge MMU and non-MMU string.h
  m68k/m68knommu: Remove dead SMP config option
  m68k: move definition of THREAD_SIZE into thread_info_mm.h
  m68k: Use asm-generic/ioctls.h (enables termiox)
  m68k: Remove dead GG2 config option

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/dlm
Linus Torvalds [Sat, 23 Oct 2010 00:33:16 +0000 (17:33 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/teigland/dlm

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/dlm:
  dlm: Fix dlm lock status block comment in dlm.h
  dlm: Don't send callback to node making lock request when "try 1cb" fails

14 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Sat, 23 Oct 2010 00:32:27 +0000 (17:32 -0700)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs: (36 commits)
  xfs: semaphore cleanup
  xfs: Extend project quotas to support 32bit project ids
  xfs: remove xfs_buf wrappers
  xfs: remove xfs_cred.h
  xfs: remove xfs_globals.h
  xfs: remove xfs_version.h
  xfs: remove xfs_refcache.h
  xfs: fix the xfs_trans_committed
  xfs: remove unused t_callback field in struct xfs_trans
  xfs: fix bogus m_maxagi check in xfs_iget
  xfs: do not use xfs_mod_incore_sb_batch for per-cpu counters
  xfs: do not use xfs_mod_incore_sb for per-cpu counters
  xfs: remove XFS_MOUNT_NO_PERCPU_SB
  xfs: pack xfs_buf structure more tightly
  xfs: convert buffer cache hash to rbtree
  xfs: serialise inode reclaim within an AG
  xfs: batch inode reclaim lookup
  xfs: implement batched inode lookups for AG walking
  xfs: split out inode walk inode grabbing
  xfs: split inode AG walking into separate code for reclaim
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu
Linus Torvalds [Sat, 23 Oct 2010 00:31:36 +0000 (17:31 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tj/percpu

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu:
  percpu: update comments to reflect that percpu allocations are always zero-filled
  percpu: Optimize __get_cpu_var()
  x86, percpu: Optimize this_cpu_ptr
  percpu: clear memory allocated with the km allocator
  percpu: fix build breakage on s390 and cleanup build configuration tests
  percpu: use percpu allocator on UP too
  percpu: reduce PCPU_MIN_UNIT_SIZE to 32k
  vmalloc: pcpu_get/free_vm_areas() aren't needed on UP

Fixed up trivial conflicts in include/linux/percpu.h

14 years agonilfs2: eliminate sparse warning - "context imbalance"
Jiro SEKIBA [Thu, 14 Oct 2010 04:52:00 +0000 (13:52 +0900)]
nilfs2: eliminate sparse warning - "context imbalance"

insert sparse annotations to fix following sparse warning.

fs/nilfs2/segment.c:2681:3: warning: context imbalance in 'nilfs_segctor_kill_thread' - unexpected unlock

nilfs_segctor_kill_thread is only called inside sc_state_lock lock.
sparse doesn't detect the context and warn "unexpected unlock".
__acquires/__releases pretend to lock/unlock the sc_state_lock for sparse.

Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: eliminate sparse warnings - "symbol not declared"
Jiro SEKIBA [Fri, 8 Oct 2010 13:37:27 +0000 (22:37 +0900)]
nilfs2: eliminate sparse warnings - "symbol not declared"

change nilfs_dat_commit_free and nilfs_inode_cachep static
to fix following warnings

fs/nilfs2/super.c:72:19: warning: symbol 'nilfs_inode_cachep' was not declared. Should it be static?
fs/nilfs2/dat.c:106:6: warning: symbol 'nilfs_dat_commit_free' was not declared. Should it be static?

Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: get rid of bdi from nilfs object
Ryusuke Konishi [Thu, 7 Oct 2010 05:19:48 +0000 (14:19 +0900)]
nilfs2: get rid of bdi from nilfs object

Nilfs now can use sb->s_bdi to get backing_dev_info, so we use it
instead of ns_bdi on the nilfs object and remove ns_bdi.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: change license of exported header file
Ryusuke Konishi [Sun, 3 Oct 2010 08:44:03 +0000 (17:44 +0900)]
nilfs2: change license of exported header file

This allows other projects to carry copies of the header file related
to ABI and disk format (i.e. "nilfs2_fs.h") without it or distributors
having to worry about effects on the project's overall license terms.
It's also desired for switching the license of nilfs library to LGPL.

Jiro SEKIBA pointed out these license issues (Message-ID:
<87tylo7msw.wl%jir@sekiba.com>), and he suggested switching license of
the library and nilfs2_fs.h to GNU Lesser General Public License.  We
take in his suggestion to avoid the license issues.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Cc: Jiro SEKIBA <jir@unicus.jp>
Cc: linux-nilfs <linux-nilfs@vger.kernel.org>
14 years agonilfs2: add bdev freeze/thaw support
Ryusuke Konishi [Mon, 20 Sep 2010 09:19:06 +0000 (18:19 +0900)]
nilfs2: add bdev freeze/thaw support

Nilfs hasn't supported the freeze/thaw feature because it didn't work
due to the peculiar design that multiple super block instances could
be allocated for a device.  This limitation was removed by the patch
"nilfs2: do not allocate multiple super block instances for a device".

So now this adds the freeze/thaw support to nilfs.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: accept 64-bit checkpoint numbers in cp mount option
Ryusuke Konishi [Wed, 15 Sep 2010 15:36:24 +0000 (00:36 +0900)]
nilfs2: accept 64-bit checkpoint numbers in cp mount option

The current implementation doesn't mount snapshots with checkpoint
numbers larger than INT_MAX since it uses match_int() for parsing
"cp=" mount option.

This uses simple_strtoull() for the conversion to resolve the issue.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: remove own inode allocator and destructor for metadata files
Ryusuke Konishi [Sun, 5 Sep 2010 04:35:53 +0000 (13:35 +0900)]
nilfs2: remove own inode allocator and destructor for metadata files

This finally removes own inode allocator and destructor functions for
metadata files.  Several routines, nilfs_mdt_new(),
nilfs_mdt_new_common(), nilfs_mdt_clear(), nilfs_mdt_destroy(), and
nilfs_alloc_inode_common() will be gone.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: get rid of back pointer to writable sb instance
Ryusuke Konishi [Sun, 5 Sep 2010 07:17:35 +0000 (16:17 +0900)]
nilfs2: get rid of back pointer to writable sb instance

Nilfs object holds a back pointer to a writable super block instance
in nilfs->ns_writer, and this became eliminable since sb is now made
per device and all inodes have a valid pointer to it.

This deletes the ns_writer pointer and a reader/writer semaphore
protecting it.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: get rid of mi_nilfs back pointer to nilfs object
Ryusuke Konishi [Sat, 4 Sep 2010 15:23:50 +0000 (00:23 +0900)]
nilfs2: get rid of mi_nilfs back pointer to nilfs object

This removes a back pointer to nilfs object from nilfs_mdt_info
structure that is attached to metadata files.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: see state of root dentry for mount check of snapshots
Ryusuke Konishi [Mon, 13 Sep 2010 02:16:34 +0000 (11:16 +0900)]
nilfs2: see state of root dentry for mount check of snapshots

After applied the patch that unified sb instances, root dentry of
snapshots can be left in dcache even after their trees are unmounted.

The orphan root dentry/inode keeps a root object, and this causes
false positive of nilfs_checkpoint_is_mounted function.

This resolves the issue by having nilfs_checkpoint_is_mounted test
whether the root dentry is busy or not.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: use iget for all metadata files
Ryusuke Konishi [Sun, 5 Sep 2010 03:20:59 +0000 (12:20 +0900)]
nilfs2: use iget for all metadata files

This makes use of iget5_locked to allocate or get inode for metadata
files to stop using own inode allocator.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: get rid of GCDAT inode
Ryusuke Konishi [Sun, 29 Aug 2010 03:44:56 +0000 (12:44 +0900)]
nilfs2: get rid of GCDAT inode

This applies prepared rollback function and redirect function of
metadata file to DAT file, and eliminates GCDAT inode.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: add routines to redirect access to buffers of DAT file
Ryusuke Konishi [Tue, 31 Aug 2010 02:40:34 +0000 (11:40 +0900)]
nilfs2: add routines to redirect access to buffers of DAT file

During garbage collection (GC), DAT file, which converts virtual block
number to real block number, may return disk block number that is not
yet written to the device.

To avoid access to unwritten blocks, the current implementation stores
changes to the caches of GCDAT during GC and atomically commit the
changes into the DAT file after they are written to the device.

This patch, instead, adds a function that makes a copy of specified
buffer and stores it in nilfs_shadow_map, and a function to get the
backup copy as needed (nilfs_mdt_freeze_buffer and
nilfs_mdt_get_frozen_buffer respectively).

Before DAT changes block number in an entry block, it makes a copy and
redirect access to the buffer so that address conversion function
(i.e. nilfs_dat_translate) refers to the old address saved in the
copy.

This patch gives requisites for such redirection.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: add routines to roll back state of DAT file
Ryusuke Konishi [Mon, 6 Sep 2010 03:05:43 +0000 (12:05 +0900)]
nilfs2: add routines to roll back state of DAT file

This adds optional function to metadata files which makes a copy of
bmap, page caches, and b-tree node cache, and rolls back to the copy
as needed.

This enhancement is intended to displace gcdat inode that provides a
similar function in a different way.

In this patch, nilfs_shadow_map structure is added to store a copy of
the foregoing states.  nilfs_mdt_setup_shadow_map relates this
structure to a metadata file.  And, nilfs_mdt_save_to_shadow_map() and
nilfs_mdt_restore_from_shadow_map() provides save and restore
functions respectively.  Finally, nilfs_mdt_clear_shadow_map() clears
states of nilfs_shadow_map.

The copy of b-tree node cache and page cache is made by duplicating
only dirty pages into corresponding caches in nilfs_shadow_map.  Their
restoration is done by clearing dirty pages from original caches and
by copying dirty pages back from nilfs_shadow_map.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: add routines to save and restore bmap state
Ryusuke Konishi [Mon, 30 Aug 2010 14:42:18 +0000 (23:42 +0900)]
nilfs2: add routines to save and restore bmap state

This adds routines to save and restore the state of bmap structure.
The bmap state is stored in a given nilfs_bmap_store object.

These routines will be used to roll back the state of dat inode
without using gcdat inode.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: do not allocate nilfs_mdt_info structure to gc-inodes
Ryusuke Konishi [Sun, 5 Sep 2010 01:14:43 +0000 (10:14 +0900)]
nilfs2: do not allocate nilfs_mdt_info structure to gc-inodes

GC-inode now doesn't need the nilfs_mdt_info structure and there is no
reason that it is a sort of metadata files.

This stops the allocation and makes them not dependent on metadata
file routines.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: allow nilfs_clear_inode to clear metadata file inodes
Ryusuke Konishi [Fri, 20 Aug 2010 14:40:54 +0000 (23:40 +0900)]
nilfs2: allow nilfs_clear_inode to clear metadata file inodes

Allows clear inode function (nilfs_clear_inode) to handle metadata
files that uses bitmap-based object alloctor.  DAT and ifile
correspond to this.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: get rid of snapshot mount flag
Ryusuke Konishi [Wed, 25 Aug 2010 14:52:46 +0000 (23:52 +0900)]
nilfs2: get rid of snapshot mount flag

This flag is a fake used to distinguish type of super block instance.
And, it got obsolete by the unification of sb.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: simplify life cycle management of nilfs object
Ryusuke Konishi [Wed, 8 Sep 2010 17:07:56 +0000 (02:07 +0900)]
nilfs2: simplify life cycle management of nilfs object

This stops pre-allocating nilfs object in nilfs_get_sb routine, and
stops managing its life cycle by reference counting.

nilfs_find_or_create_nilfs() function, nilfs->ns_mount_mutex,
nilfs_objects list, and the reference counter will be removed through
the simplification.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: do not allocate multiple super block instances for a device
Ryusuke Konishi [Sun, 15 Aug 2010 16:54:52 +0000 (01:54 +0900)]
nilfs2: do not allocate multiple super block instances for a device

This stops allocating multiple super block instances for a device.

All snapshots and a current mode mount (i.e. latest tree) will be
controlled with nilfs_root objects that are kept within an sb
instance.

nilfs_get_sb() is rewritten so that it always has a root object for
the latest tree and snapshots make additional root objects.

The root dentry of the latest tree is binded to sb->s_root even if it
isn't attached on a directory.  Root dentries of snapshots or the
latest tree are binded to mnt->mnt_root on which they are mounted.

With this patch, nilfs_find_sbinfo() function, nilfs->ns_supers list,
and nilfs->ns_current back pointer, are deleted.  In addition,
init_nilfs() and load_nilfs() are simplified since they will be called
once for a device, not repeatedly called for mount points.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: split out nilfs_attach_snapshot
Ryusuke Konishi [Wed, 25 Aug 2010 17:15:41 +0000 (02:15 +0900)]
nilfs2: split out nilfs_attach_snapshot

This splits the code to attach snapshots into a separate routine for
convenience sake.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: split out nilfs_get_root_dentry
Ryusuke Konishi [Wed, 25 Aug 2010 16:52:51 +0000 (01:52 +0900)]
nilfs2: split out nilfs_get_root_dentry

This splits the code to allocate root dentry into a separate routine
for convenience in successive changes.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: deny write access to inodes in snapshots
Ryusuke Konishi [Sun, 15 Aug 2010 14:33:57 +0000 (23:33 +0900)]
nilfs2: deny write access to inodes in snapshots

Snapshots of nilfs are read-only.

After super block instances (sb) will be unified, nilfs will need to
check write access by a way other than implicit test with
IS_RDONLY(inode).  This is because IS_RDONLY() refers to MS_RDONLY bit
of inode->i_sb->s_flags and it will become inaccurate after the
unification of sb.

To prepare for the issue, this uses i_op->permission to deny write
access to inodes in snapshots.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: use checkpoint tree for mount check of snapshots
Ryusuke Konishi [Sat, 14 Aug 2010 12:44:51 +0000 (21:44 +0900)]
nilfs2: use checkpoint tree for mount check of snapshots

This rewrites nilfs_checkpoint_is_mounted() function so that it
decides whether a checkpoint is mounted by whether the corresponding
root object is found in checkpoint tree.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: move inode count and block count into root object
Ryusuke Konishi [Sat, 14 Aug 2010 05:48:32 +0000 (14:48 +0900)]
nilfs2: move inode count and block count into root object

This moves sbi->s_inodes_count and sbi->s_blocks_count into nilfs_root
object.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: use root object to get ifile
Ryusuke Konishi [Sat, 14 Aug 2010 04:07:15 +0000 (13:07 +0900)]
nilfs2: use root object to get ifile

This rewrites functions using ifile so that they get ifile from
nilfs_root object, and will remove sbi->s_ifile.  Some functions that
don't know the root object are extended to receive it from caller.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: make snapshots in checkpoint tree exportable
Ryusuke Konishi [Thu, 26 Aug 2010 15:23:02 +0000 (00:23 +0900)]
nilfs2: make snapshots in checkpoint tree exportable

The previous export operations cannot handle multiple versions of
a filesystem if they belong to the same sb instance.

This adds a new type of file handle and extends export operations so
that they can get the inode specified by a checkpoint number as well
as an inode number and a generation number.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: set pointer to root object in inodes
Ryusuke Konishi [Wed, 25 Aug 2010 08:45:44 +0000 (17:45 +0900)]
nilfs2: set pointer to root object in inodes

This puts a pointer to nilfs_root object in the private part of
on-memory inode, and makes nilfs_iget function pick up the inode with
the same root object.

Non-root inodes inherit its nilfs_root object from parent inode.  That
of the root inode is allocated through nilfs_attach_checkpoint()
function.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: add checkpoint tree to nilfs object
Ryusuke Konishi [Sat, 14 Aug 2010 03:59:15 +0000 (12:59 +0900)]
nilfs2: add checkpoint tree to nilfs object

To hold multiple versions of a filesystem in one sb instance, a new
on-memory structure is necessary to handle one or more checkpoints.

This adds a red-black tree of checkpoints to nilfs object, and adds
lookup and create functions for them.

Each checkpoint is represented by "nilfs_root" structure, and this
structure has rb_node to configure the rb-tree.

The nilfs_root object is identified with a checkpoint number.  For
each snapshot, a nilfs_root object is allocated and the checkpoint
number of snapshot is assigned to it.  For a regular mount
(i.e. current mode mount), NILFS_CPTREE_CURRENT_CNO constant is
assigned to the corresponding nilfs_root object.

Each nilfs_root object has an ifile inode and some counters.  These
items will displace those of nilfs_sb_info structure in successive
patches.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: remove own inode hash used for GC
Ryusuke Konishi [Fri, 20 Aug 2010 10:06:11 +0000 (19:06 +0900)]
nilfs2: remove own inode hash used for GC

This uses inode hash function that vfs provides instead of the own
hash table for caching gc inodes.  This finally removes the own inode
hash from nilfs.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: separate initializer of metadata file inode
Ryusuke Konishi [Sat, 21 Aug 2010 13:01:51 +0000 (22:01 +0900)]
nilfs2: separate initializer of metadata file inode

This separates a part of initialization code of metadata file inode,
and makes it available from the nilfs iget function that a later patch
will add to.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: use iget5_locked to get inode
Ryusuke Konishi [Fri, 20 Aug 2010 12:20:29 +0000 (21:20 +0900)]
nilfs2: use iget5_locked to get inode

This uses iget5_locked instead of iget_locked so that gc cache can
look up inodes with an inode number and an optional checkpoint number.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: keep zero value in i_cno except for gc-inodes
Ryusuke Konishi [Fri, 20 Aug 2010 11:10:38 +0000 (20:10 +0900)]
nilfs2: keep zero value in i_cno except for gc-inodes

On-memory inode structures of nilfs have a member "i_cno" which stores
a checkpoint number related to the inode.  For gc-inodes, this field
indicates version of data each gc-inode caches for GC.  Log writer
temporarily uses "i_cno" to transfer the latest checkpoint number.

This stops the latter use and lets only gc-inodes use it.

The purpose of this patch is to allow the successive change use
"i_cno" for inode lookup.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: allow nilfs_dirty_inode to mark metadata file inodes dirty
Ryusuke Konishi [Fri, 20 Aug 2010 15:30:39 +0000 (00:30 +0900)]
nilfs2: allow nilfs_dirty_inode to mark metadata file inodes dirty

This allows sop->dirty_inode callback function (nilfs_dirty_inode) to
handle metadata file inodes.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: allow nilfs_destroy_inode to destroy metadata file inodes
Ryusuke Konishi [Fri, 20 Aug 2010 14:46:06 +0000 (23:46 +0900)]
nilfs2: allow nilfs_destroy_inode to destroy metadata file inodes

The current nilfs_destroy_inode() doesn't handle metadata file inodes
including gc inodes (dummy inodes used for garbage collection).

This allows nilfs_destroy_inode() to destroy inodes of metadata files.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>