IOMMU device groups are currently a rather vague associative notion
with assembly required by the user or user level driver provider to
do anything useful. This patch intends to grow the IOMMU group concept
into something a bit more consumable.
To do this, we first create an object representing the group, struct
iommu_group. This structure is allocated (iommu_group_alloc) and
filled (iommu_group_add_device) by the iommu driver. The iommu driver
is free to add devices to the group using it's own set of policies.
This allows inclusion of devices based on physical hardware or topology
limitations of the platform, as well as soft requirements, such as
multi-function trust levels or peer-to-peer protection of the
interconnects. Each device may only belong to a single iommu group,
which is linked from struct device.iommu_group. IOMMU groups are
maintained using kobject reference counting, allowing for automatic
removal of empty, unreferenced groups. It is the responsibility of
the iommu driver to remove devices from the group
(iommu_group_remove_device).
IOMMU groups also include a userspace representation in sysfs under
/sys/kernel/iommu_groups. When allocated, each group is given a
dynamically assign ID (int). The ID is managed by the core IOMMU group
code to support multiple heterogeneous iommu drivers, which could
potentially collide in group naming/numbering. This also keeps group
IDs to small, easily managed values. A directory is created under
/sys/kernel/iommu_groups for each group. A further subdirectory named
"devices" contains links to each device within the group. The iommu_group
file in the device's sysfs directory, which formerly contained a group
number when read, is now a link to the iommu group. Example:
$ ls -l /sys/kernel/iommu_groups/26/devices/
total 0
lrwxrwxrwx. 1 root root 0 Apr 17 12:57 0000:00:1e.0 ->
../../../../devices/pci0000:00/0000:00:1e.0
lrwxrwxrwx. 1 root root 0 Apr 17 12:57 0000:06:0d.0 ->
../../../../devices/pci0000:00/0000:00:1e.0/0000:06:0d.0
lrwxrwxrwx. 1 root root 0 Apr 17 12:57 0000:06:0d.1 ->
../../../../devices/pci0000:00/0000:00:1e.0/0000:06:0d.1
$ ls -l /sys/kernel/iommu_groups/26/devices/*/iommu_group
[truncating perms/owner/timestamp]
/sys/kernel/iommu_groups/26/devices/0000:00:1e.0/iommu_group ->
../../../kernel/iommu_groups/26
/sys/kernel/iommu_groups/26/devices/0000:06:0d.0/iommu_group ->
../../../../kernel/iommu_groups/26
/sys/kernel/iommu_groups/26/devices/0000:06:0d.1/iommu_group ->
../../../../kernel/iommu_groups/26
Groups also include several exported functions for use by user level
driver providers, for example VFIO. These include:
iommu_group_get(): Acquires a reference to a group from a device
iommu_group_put(): Releases reference
iommu_group_for_each_dev(): Iterates over group devices using callback
iommu_group_[un]register_notifier(): Allows notification of device add
and remove operations relevant to the group
iommu_group_id(): Return the group number
This patch also extends the IOMMU API to allow attaching groups to
domains. This is currently a simple wrapper for iterating through
devices within a group, but it's expected that the IOMMU API may
eventually make groups a more integral part of domains.
Groups intentionally do not try to manage group ownership. A user
level driver provider must independently acquire ownership for each
device within a group before making use of the group as a whole.
This may change in the future if group usage becomes more pervasive
across both DMA and IOMMU ops.
Groups intentionally do not provide a mechanism for driver locking
or otherwise manipulating driver matching/probing of devices within
the group. Such interfaces are generic to devices and beyond the
scope of IOMMU groups. If implemented, user level providers have
ready access via iommu_group_for_each_dev and group notifiers.
iommu_device_group() is removed here as it has no users. The
replacement is:
group = iommu_group_get(dev);
id = iommu_group_id(group);
iommu_group_put(group);
AMD-Vi & Intel VT-d support re-added in following patches.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
IOMMU groups allow IOMMU drivers to represent DMA visibility
and isolation of devices. Multiple devices may be grouped
together for the purposes of DMA. Placing a pointer on
struct device enable easy access for things like streaming
DMA programming and drivers like VFIO.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
The atomic notifier from twl4030/twl6030 to notifiy VBUS and ID events,
is replaced by a direct call to omap musb blue.
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Used devres API's to associate the phy with a device so that on
driver detach, release function is invoked on the devres data(usb_phy)
and devres data(usb_phy) is released.
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Add a linked list for keeping multiple PHY instances with different
types so that we can have separate USB2 and USB3 PHYs on one single
board. _get_phy_ has been changed so that the controller gets
the transceiver by type. _remove_phy_ has been added to let the phy
be removed from the phy list.
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
_transceiver() in otg.c is replaced with _phy. usb_set_transceiver is
replaced with usb_add_phy to make it similar to other usb standard
function names like usb_add_hcd.
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Request allocation is about to be made per-blkg meaning that there'll
be multiple request lists.
* Make queue full state per request_list. blk_*queue_full() functions
are renamed to blk_*rl_full() and takes @rl instead of @q.
* Rename blk_init_free_list() to blk_init_rl() and make it take @rl
instead of @q. Also add @gfp_mask parameter.
* Add blk_exit_rl() instead of destroying rl directly from
blk_release_queue().
* Add request_list->q and make request alloc/free functions -
blk_free_request(), [__]freed_request(), __get_request() - take @rl
instead of @q.
This patch doesn't introduce any functional difference.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Add q->nr_rqs[] which currently behaves the same as q->rq.count[] and
move q->rq.elvpriv to q->nr_rqs_elvpriv. blk_drain_queue() is updated
to use q->nr_rqs[] instead of q->rq.count[].
These counters separates queue-wide request statistics from the
request list and allow implementation of per-queue request allocation.
While at it, properly indent fields of struct request_list.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
iscsi_remove_host() uses bsg_remove_queue() which implements custom
queue draining. fc_bsg_remove() open-codes mostly identical logic.
The draining logic isn't correct in that blk_stop_queue() doesn't
prevent new requests from being queued - it just stops processing, so
nothing prevents new requests to be queued after the logic determines
that the queue is drained.
blk_cleanup_queue() now implements proper queue draining and these
custom draining logics aren't necessary. Drop them and use
bsg_unregister_queue() + blk_cleanup_queue() instead.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: James Smart <james.smart@emulex.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
mempool_create_node() currently assumes %GFP_KERNEL. Its only user,
blk_init_free_list(), is about to be updated to use other allocation
flags - add @gfp_mask argument to the function.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pull drm fixes from Dave Airlie:
"Nothing major in here, one radeon SI fix for tiling, and one uninit
var fix, two minor header file fixes."
* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
drm: drop comment about this header being autogenerated.
drm/edid: don't return stack garbage from supports_rb
vga_switcheroo: Add include guard
drm/radeon: SI tiling fixes for display
There are a few things that make the logging and
debugging in mac80211 less useful than it should
be right now:
* a lot of messages should be pr_info, not pr_debug
* wholesale use of pr_debug makes it require *both*
Kconfig and dynamic configuration
* there are still a lot of ifdefs
* the style is very inconsistent, sometimes the
sdata->name is printed in front
Clean up everything, introducing new macros and
separating out the station MLME debugging into
a new Kconfig symbol.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Pablo says:
====================
The following four patches provide Netfilter fixes for the cthelper
infrastructure that was recently merged mainstream, they are:
* two fixes for compilation breakage with two different configurations:
- CONFIG_NF_NAT=m and CONFIG_NF_CT_NETLINK=y
- NF_CONNTRACK_EVENTS=n and CONFIG_NETFILTER_NETLINK_QUEUE_CT=y
* two fixes for sparse warnings.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
* topic/huang-d3cold-v7:
PCI/PM: add PCIe runtime D3cold support
PCI: do not call pci_set_power_state with PCI_D3cold
PCI/PM: add runtime PM support to PCIe port
ACPI/PM: specify lowest allowed state for device sleep state
This patch adds runtime D3cold support and corresponding ACPI platform
support. This patch only enables runtime D3cold support; it does not
enable D3cold support during system suspend/hibernate.
D3cold is the deepest power saving state for a PCIe device, where its main
power is removed. While it is in D3cold, you can't access the device at
all, not even its configuration space (which is still accessible in D3hot).
Therefore the PCI PM registers can not be used to transition into/out of
the D3cold state; that must be done by platform logic such as ACPI _PR3.
To support wakeup from D3cold, a system may provide auxiliary power, which
allows a device to request wakeup using a Beacon or the sideband WAKE#
signal. WAKE# is usually connected to platform logic such as ACPI GPE.
This is quite different from other power saving states, where devices
request wakeup via a PME message on the PCIe link.
Some devices, such as those in plug-in slots, have no direct platform
logic. For example, there is usually no ACPI _PR3 for them. D3cold
support for these devices can be done via the PCIe Downstream Port leading
to the device. When the PCIe port is powered on/off, the device is powered
on/off too. Wakeup events from the device will be notified to the
corresponding PCIe port.
For more information about PCIe D3cold and corresponding ACPI support,
please refer to:
- PCI Express Base Specification Revision 2.0
- Advanced Configuration and Power Interface Specification Revision 5.0
[bhelgaas: changelog]
Reviewed-by: Rafael J. Wysocki <rjw@sisk.pl>
Originally-by: Zheng Yan <zheng.z.yan@intel.com>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Lower device sleep state can save more power, but has more exit
latency too. Sometimes, to satisfy some power QoS and other
requirement, we need to constrain the lowest device sleep state.
In this patch, a parameter to specify lowest allowed state for
acpi_pm_device_sleep_state is added. So that the caller can enforce
the constraint via the parameter.
This is needed by PCIe D3cold support, where the lowest power state
allowed may be D3_HOT instead of default D3_COLD.
CC: Len Brown <lenb@kernel.org>
CC: linux-acpi@vger.kernel.org
Reviewed-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Several forthcoming Wolfson devices are based on a common platform
known as Arizona allowing a great deal of reuse of driver code. This
patch adds core support for these devices.
In order to handle systems which do not use the generic clock API a
simple wrapper for the 32kHz clock domain in the devices is provided.
Once the generic clock API is widely available this code will be moved
over to use that.
For simplicity some WM5102 specific code is included in the core driver,
the effort involved in splitting the device out isn't worth it.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Several forthcoming Wolfson devices are based on a common platform
known as Arizona allowing a great deal of reuse of driver code. This
patch adds register definitions for these devices.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
The code handles this fine already, we just need new macros in the header
for drivers to create the controls.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Liam Girdwood <lrg@ti.com>
Several bug reports have been received recently for USB mass-storage
devices that don't handle READ CAPACITY(16) commands properly. They
report bogus sizes, in some cases becoming unusable as a result.
The bugs were triggered by commit
09b6b51b0b (SCSI & usb-storage: add
flags for VPD pages and REPORT LUNS), which caused usb-storage to stop
overriding the SCSI level reported by devices. By default, the sd
driver will try READ CAPACITY(16) first for any device whose level is
above SCSI_SPC_2.
It seems likely that any device large enough to require the use of
READ CAPACITY(16) (i.e., 2 TB or more) would be able to handle READ
CAPACITY(10) commands properly. Indeed, I don't know of any devices
that don't handle READ CAPACITY(10) properly.
Therefore this patch (as1559) adds a new flag telling the sd driver
to try READ CAPACITY(10) before READ CAPACITY(16), and sets this flag
for every USB mass-storage device. If a device really is larger than
2 TB, sd will fall back to READ CAPACITY(16) just as it used to.
This fixes Bugzilla #43391.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: Hans de Goede <hdegoede@redhat.com>
CC: "James E.J. Bottomley" <JBottomley@parallels.com>
CC: Matthew Dharm <mdharm-usb@one-eyed-alien.net>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This change is meant to add a control for disabling early socket demux.
The main motivation behind this patch is to provide an option to disable
the feature as it adds an additional cost to routing that reduces overall
throughput by up to 5%. For example one of my systems went from 12.1Mpps
to 11.6 after the early socket demux was added. It looks like the reason
for the regression is that we are now having to perform two lookups, first
the one for an established socket, and then the one for the routing table.
By adding this patch and toggling the value for ip_early_demux to 0 I am
able to get back to the 12.1Mpps I was previously seeing.
[ Move local variables in ip_rcv_finish() down into the basic
block in which they are actually used. -DaveM ]
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
* topic/jiang-mmconfig-v10:
ACPI: mark acpi_sfi_table_parse() as __init
x86/PCI: use pr_level() to replace printk(KERN_LEVEL)
x86/PCI: refine __pci_mmcfg_init() for better code readability
x86/PCI: get rid of redundant log messages
x86/PCI: simplify pci_mmcfg_late_insert_resources()
x86/PCI: update MMCONFIG information when hot-plugging PCI host bridges
PCI/ACPI: provide MMCONFIG address for PCI host bridges
x86/PCI: add pci_mmconfig_insert()/delete() for PCI root bridge hotplug
x86/PCI: prepare pci_mmcfg_check_reserved() to be called at runtime
x86/PCI: introduce pci_mmcfg_arch_map()/pci_mmcfg_arch_unmap()
x86/PCI: use RCU list to protect mmconfig list
x86/PCI: split out pci_mmconfig_alloc() for code reuse
x86/PCI: split out pci_mmcfg_check_reserved() for code reuse
Mark function acpi_sfi_table_parse() as __init to avoid warning messages:
WARNING: vmlinux.o(.text+0x4cd2d2): Section mismatch in reference from the
function acpi_sfi_table_parse.clone.0() to the function
Function acpi_sfi_table_parse() calls acpi_table_parse() and
pci_parse_mcfg(), which are both marked as __init. Currently
acpi_sfi_table_parse() is only used by MMCONFIG to scan MCFG table
at boot time only, so it's safe to mark acpi_sfi_table_parse() as __init.
Reviewed-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Jiang Liu <liuj97@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
This patch provide MMCONFIG address for PCI host bridges, which will
be used to support host bridge hotplug. It gets MMCONFIG address
by evaluating _CBA method if available.
Reviewed-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Jiang Liu <liuj97@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Pull perf updates from Ingo Molnar.
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
ftrace: Make all inline tags also include notrace
perf: Use css_tryget() to avoid propping up css refcount
perf tools: Fix synthesizing tracepoint names from the perf.data headers
perf stat: Fix default output file
perf tools: Fix endianity swapping for adds_features bitmask
These are no longer used. Every ceph_connection instance is embedded in
another structure, and refcounts manipulated via the get/put ops.
Signed-off-by: Sage Weil <sage@inktank.com>
This patch add support for the SPEAr ASoC pcm layer in ASoC
framework. The pcm layer uses common snd_dmaengine framework.
Signed-off-by: Rajeev Kumar <rajeev-dlh.kumar@st.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
LD init/built-in.o
net/built-in.o:(.data+0x4408): undefined reference to `nf_nat_tcp_seq_adjust'
make: *** [vmlinux] Error 1
This patch adds a new pointer hook (nfq_ct_nat_hook) similar to other existing
in Netfilter to solve our complicated configuration dependencies.
Reported-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Pull NFS client bugfixes from Trond Myklebust:
- Fix a write hang due to an uninitalised variable when
!defined(CONFIG_NFS_V4)
- Address upcall races in the legacy NFSv4 idmapper
- Remove an O_DIRECT refcounting issue
- Fix a pNFS refcounting bug when the file layout metadata server is
also acting as a data server
- Fix a pNFS module loading race.
* tag 'nfs-for-3.5-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
NFS: Force the legacy idmapper to be single threaded
NFS: Initialise commit_info.rpc_out when !defined(CONFIG_NFS_V4)
NFS: Fix a refcounting issue in O_DIRECT
NFSv4.1: Fix a race in set_pnfs_layoutdriver
NFSv4.1: Fix umount when filelayout DS is also the MDS
Get current rssi (in dBm) from the driver/FW.
Instead of reporting the signal received in the last
rx packet, which might be inaccurate if rx traffic is
low and beacon filtering is enabled, get the signal
from the driver/FW.
Signed-off-by: Victor Goldenshtein <victorg@ti.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
This lets us catch the USB fixes that went into 3.5-rc3 into this branch,
as we want them here as well.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Registering the platform driver before module_init allows us to log oopses
that happen during device probing.
This requires changing module_init to postcore_initcall, and switching
from platform_driver_probe to platform_driver_register because the
platform device is not registered when the platform driver is registered;
and because we use driver_register, now can't use create_bundle() (since
it will try to register the same driver once again), so we have to switch
to platform_device_register_data().
Also, some __init -> __devinit changes were needed.
Overall, the registration logic is now much clearer, since we have only
one driver registration point, and just an optional dummy device, which
is created from the module parameters.
Suggested-by: Colin Cross <ccross@android.com>
Signed-off-by: Anton Vorontsov <anton.vorontsov@linaro.org>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Pull staging tree fixes from Greg Kroah-Hartman:
"Here are a number of small fixes for the drivers/staging tree, as well
as iio and pstore drivers (which came from the staging tree in the
3.5-rc1 merge). All of these are tiny, but resolve issues that people
have been reporting.
There's also a documentation update to reflect what the iio drivers
really are doing, which is good to get straightened out.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>"
* tag 'staging-3.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
staging: r8712u: Add new USB IDs
staging: gdm72xx: Release netlink socket properly
iio: drop wrong reference from Kconfig
pstore/inode: Make pstore_fill_super() static
pstore/ram: Should zap persistent zone on unlink
pstore/ram_core: Factor persistent_ram_zap() out of post_init()
pstore/ram_core: Do not reset restored zone's position and size
pstore/ram: Should update old dmesg buffer before reading
staging:iio:ad7298: Fix linker error due to missing IIO kfifo buffer
Revert "staging: usbip: bugfix for stack corruption on 64-bit architectures"
staging: usbip: bugfix for stack corruption on 64-bit architectures
staging/comedi: fix build for USB not enabled
staging: omapdrm: fix crash when freeing bad fb
staging:iio:ad7606: Re-add missing scale attribute
iio: Fix potential use after free
staging:iio: remove num_interrupt_lines from documentation
iio: documentation: Add out_altvoltage and friends
Pull driver core and printk fixes from Greg Kroah-Hartman:
"Here are some fixes for 3.5-rc4 that resolve the kmsg problems that
people have reported showing up after the printk and kmsg changes went
into 3.5-rc1. There are also a smattering of other tiny fixes for the
extcon and hyper-v drivers that people have reported.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>"
* tag 'driver-core-3.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core:
extcon: max8997: Add missing kfree for info->edev in max8997_muic_remove()
extcon: Set platform drvdata in gpio_extcon_probe() and fix irq leak
extcon: Fix wrong index in max8997_extcon_cable[]
kmsg - kmsg_dump() fix CONFIG_PRINTK=n compilation
printk: return -EINVAL if the message len is bigger than the buf size
printk: use mutex lock to stop syslog_seq from going wild
kmsg - kmsg_dump() use iterator to receive log buffer content
vme: change maintainer e-mail address
Extcon: Don't try to create duplicate link names
driver core: fixup reversed deferred probe order
printk: Fix alignment of buf causing crash on ARM EABI
Tools: hv: verify origin of netlink connector message
* emailed from Andrew Morton <akpm@linux-foundation.org>: (21 patches)
mm/memblock: fix overlapping allocation when doubling reserved array
c/r: prctl: Move PR_GET_TID_ADDRESS to a proper place
pidns: find_new_reaper() can no longer switch to init_pid_ns.child_reaper
pidns: guarantee that the pidns init will be the last pidns process reaped
fault-inject: avoid call to random32() if fault injection is disabled
Viresh has moved
get_maintainer: Fix --help warning
mm/memory.c: fix kernel-doc warnings
mm: fix kernel-doc warnings
mm: correctly synchronize rss-counters at exit/exec
mm, thp: print useful information when mmap_sem is unlocked in zap_pmd_range
h8300: use the declarations provided by <asm/sections.h>
h8300: fix use of extinct _sbss and _ebss
xtensa: use the declarations provided by <asm/sections.h>
xtensa: use "test -e" instead of bashism "test -a"
xtensa: replace xtensa-specific _f{data,text} by _s{data,text}
memcg: fix use_hierarchy css_is_ancestor oops regression
mm, oom: fix and cleanup oom score calculations
nilfs2: ensure proper cache clearing for gc-inodes
thp: avoid atomic64_read in pmd_read_atomic for 32bit PAE
...
In the x86 32bit PAE CONFIG_TRANSPARENT_HUGEPAGE=y case while holding the
mmap_sem for reading, cmpxchg8b cannot be used to read pmd contents under
Xen.
So instead of dealing only with "consistent" pmdvals in
pmd_none_or_trans_huge_or_clear_bad() (which would be conceptually
simpler) we let pmd_none_or_trans_huge_or_clear_bad() deal with pmdvals
where the low 32bit and high 32bit could be inconsistent (to avoid having
to use cmpxchg8b).
The only guarantee we get from pmd_read_atomic is that if the low part of
the pmd was found null, the high part will be null too (so the pmd will be
considered unstable). And if the low part of the pmd is found "stable"
later, then it means the whole pmd was read atomically (because after a
pmd is stable, neither MADV_DONTNEED nor page faults can alter it anymore,
and we read the high part after the low part).
In the 32bit PAE x86 case, it is enough to read the low part of the pmdval
atomically to declare the pmd as "stable" and that's true for THP and no
THP, furthermore in the THP case we also have a barrier() that will
prevent any inconsistent pmdvals to be cached by a later re-read of the
*pmd.
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Jonathan Nieder <jrnieder@gmail.com>
Cc: Ulrich Obergfell <uobergfe@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Petr Matousek <pmatouse@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jan Beulich <jbeulich@suse.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@gmail.com>
Tested-by: Andrew Jones <drjones@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
On arches that do not support this_cpu_cmpxchg_double() slab_lock is used
to do atomic cmpxchg() on double word which contains page->_count. The
page count can be changed from get_page() or put_page() without taking
slab_lock. That corrupts page counter.
Fix it by moving page->_count out of cmpxchg_double data. So that slub
does no change it while updating slub meta-data in struct page.
[akpm@linux-foundation.org: use standard comment layout, tweak comment text]
Reported-by: Amey Bhide <abhide@nicira.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
