| Age | Commit message (Collapse) | Author |
|---|
|
The USB core hub thread (khubd) is designed with external USB hubs in
mind. It expects that if a port status change bit is set, the hub will
continue to send a notification through the hub status data transfer.
Basically, it expects hub notifications to be level-triggered.
The xHCI host controller is designed to be edge-triggered on the logical
'OR' of all the port status change bits. When all port status change
bits are clear, and a new change bit is set, the xHC will generate a
Port Status Change Event. If another change bit is set in the same port
status register before the first bit is cleared, it will not send
another event.
This means that the hub code may lose port status changes because of
race conditions between clearing change bits. The user sees this as a
"dead port" that doesn't react to device connects.
The fix is to turn on port polling whenever a new change bit is set.
Once the USB core issues a hub status request that shows that no change
bits are set in any USB ports, turn off port polling.
We can't allow the USB core to poll the roothub for port events during
host suspend because if the PCI host is in D3cold, the port registers
will be all f's. Instead, stop the port polling timer, and
unconditionally restart it when the host resumes. If there are no port
change bits set after the resume, the first call to hub_status_data will
disable polling.
This patch should be backported to stable kernels with the first xHCI
support, 2.6.31 and newer, that include the commit
0f2a79300a1471cf92ab43af165ea13555c8b0a5 "USB: xhci: Root hub support."
There will be merge conflicts because the check for HC_STATE_SUSPENDED
was moved into xhci_suspend in 3.8.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable@vger.kernel.org
|
|
An empty port can transition to either Inactive or Compliance Mode if a
newly connected USB 3.0 device fails to link train. In that case, we
issue a warm reset. Some devices, such as John's Roseweil eusb3
enclosure, slip back into Compliance Mode after the warm reset.
The current warm reset code does not check for device connect status on
warm reset completion, and it incorrectly reports the warm reset
succeeded. This causes the USB core to attempt to send a Set Address
control transfer to a port in Compliance Mode, which will always fail.
Make hub_port_wait_reset check the current connect status and link state
after the warm reset completes. Return a failure status if the device
is disconnected or the link state is Compliance Mode or SS.Inactive.
Make hub_events disable the port if warm reset fails. This will disable
the port, and then bring it back into the RxDetect state. Make the USB
core ignore the connect change until the device reconnects.
Note that this patch does NOT handle connected devices slipping into the
Inactive state very well. This is a concern, because devices can go
into the Inactive state on U1/U2 exit failure. However, the fix for
that case is too large for stable, so it will be submitted in a separate
patch.
This patch should be backported to kernels as old as 3.2, contain the
commit ID 75d7cf72ab9fa01dc70877aa5c68e8ef477229dc "usbcore: refine warm
reset logic"
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: John Covici <covici@ccs.covici.com>
Cc: stable@vger.kernel.org
|
|
The port reset code bails out early if the current connect status is
cleared (device disconnected). If we're issuing a hot reset, it may
also look at the link state before the reset is finished.
Section 10.14.2.6 of the USB 3.0 spec says that when a port enters the
Error state or Resetting state, the port connection bit retains the
value from the previous state. Therefore we can't trust it until the
reset finishes. Also, the xHCI spec section 4.19.1.2.5 says software
shall ignore the link state while the port is resetting, as it can be in
an unknown state.
The port state during reset is also unknown for USB 2.0 hubs. The hub
sends a reset signal by driving the bus into an SE0 state. This
overwhelms the "connect" signal from the device, so the port can't tell
whether anything is connected or not.
Fix the port reset code to ignore the port link state and current
connect bit until the reset finishes, and USB_PORT_STAT_RESET is
cleared.
Remove the check for USB_PORT_STAT_C_BH_RESET in the warm reset case,
because it's redundant. When the warm reset finishes, the port reset
bit will be cleared at the same time USB_PORT_STAT_C_BH_RESET is set.
Remove the now-redundant check for a cleared USB_PORT_STAT_RESET bit
in the code to deal with the finished reset.
This patch should be backported to all stable kernels.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable@vger.kernel.org
|
|
John's NEC 0.96 xHCI host controller needs a longer timeout for a warm
reset to complete. The logs show it takes 650ms to complete the warm
reset, so extend the hub reset timeout to 800ms to be on the safe side.
This commit should be backported to kernels as old as 3.2, that contain
the commit 75d7cf72ab9fa01dc70877aa5c68e8ef477229dc "usbcore: refine
warm reset logic".
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: John Covici <covici@ccs.covici.com>
Cc: stable@vger.kernel.org
|
|
If hot and warm reset fails, or a port remains in the Compliance Mode,
the USB core needs to be able to disable a USB 3.0 port. Unlike USB 2.0
ports, once the port is placed into the Disabled link state, it will not
report any new device connects. To get device connect notifications, we
need to put the link into the Disabled state, and then the RxDetect
state.
The xHCI driver needs to atomically clear all change bits on USB 3.0
port disable, so that we get Port Status Change Events for future port
changes. We could technically do this in the USB core instead of in the
xHCI roothub code, since the port state machine can't advance out of the
disabled state until we set the link state to RxDetect. However,
external USB 3.0 hubs don't need this code. They are level-triggered,
not edge-triggered like xHCI, so they will continue to send interrupt
events when any change bit is set. Therefore it doesn't make sense to
put this code in the USB core.
This patch is part of a series to fix several reports of infinite loops
on device enumeration failure. This includes John, when he boots with
a USB 3.0 device (Roseweil eusb3 enclosure) attached to his NEC 0.96
host controller. The fix requires warm reset support, so it does not
make sense to backport this patch to stable kernels without warm reset
support.
This patch should be backported to kernels as old as 3.2, contain the
commit ID 75d7cf72ab9fa01dc70877aa5c68e8ef477229dc "usbcore: refine warm
reset logic"
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Reported-by: John Covici <covici@ccs.covici.com>
Cc: stable@vger.kernel.org
|
|
When the USB core finishes reseting a USB device, the xHCI driver sends
a Reset Device command to the host. The xHC then updates its internal
representation of the USB device to the 'Default' device state. If the
device was already in the Default state, the xHC will complete the
command with an error status.
If a device needs to be reset several times during enumeration, the
second reset will always fail because of the xHCI Reset Device command.
This can cause issues during enumeration.
For example, usb_reset_and_verify_device calls into hub_port_init in a
loop. Say that on the first call into hub_port_init, the device is
successfully reset, but doesn't respond to several set address control
transfers. Then the port will be disabled, but the udev will remain in
tact. usb_reset_and_verify_device will call into hub_port_init again.
On the second call into hub_port_init, the device will be reset, and the
xHCI driver will issue a Reset Device command. This command will fail
(because the device is already in the Default state), and
usb_reset_and_verify_device will fail. The port will be disabled, and
the device won't be able to enumerate.
Fix this by ignoring the return value of the HCD reset_device callback.
This commit should be backported to kernels as old as 3.2, that contain
the commit 75d7cf72ab9fa01dc70877aa5c68e8ef477229dc "usbcore: refine
warm reset logic".
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable@vger.kernel.org
|
|
USB 3.0 hubs and roothubs will automatically transition a failed hot
reset to a warm (BH) reset. In that case, the warm reset change bit
will be set, and the link state change bit may also be set. Change
hub_port_finish_reset to unconditionally clear those change bits for USB
3.0 hubs. If these bits are not cleared, we may lose port change events
from the roothub.
This commit should be backported to kernels as old as 3.2, that contain
the commit 75d7cf72ab9fa01dc70877aa5c68e8ef477229dc "usbcore: refine
warm reset logic".
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable@vger.kernel.org
|
|
A high speed control or bulk endpoint may have bInterval set to zero,
which means it does not NAK. If bInterval is non-zero, it means the
endpoint NAKs at a rate of 2^(bInterval - 1).
The xHCI code to compute the NAK interval does not handle the special
case of zero properly. The current code unconditionally subtracts one
from bInterval and uses it as an exponent. This causes a very large
bInterval to be used, and warning messages like these will be printed:
usb 1-1: ep 0x1 - rounding interval to 32768 microframes, ep desc says 0 microframes
This may cause the xHCI host hardware to reject the Configure Endpoint
command, which means the HS device will be unusable under xHCI ports.
This patch should be backported to kernels as old as 2.6.31, that contain
commit dfa49c4ad120a784ef1ff0717168aa79f55a483a "USB: xhci - fix math in
xhci_get_endpoint_interval()".
Reported-by: Vincent Pelletier <plr.vincent@gmail.com>
Suggested-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org
|
|
Pull watchdog updates from Wim Van Sebroeck:
"This includes some fixes and code improvements (like
clk_prepare_enable and clk_disable_unprepare), conversion from the
omap_wdt and twl4030_wdt drivers to the watchdog framework, addition
of the SB8x0 chipset support and the DA9055 Watchdog driver and some
OF support for the davinci_wdt driver."
* git://www.linux-watchdog.org/linux-watchdog: (22 commits)
watchdog: mei: avoid oops in watchdog unregister code path
watchdog: Orion: Fix possible null-deference in orion_wdt_probe
watchdog: sp5100_tco: Add SB8x0 chipset support
watchdog: davinci_wdt: add OF support
watchdog: da9052: Fix invalid free of devm_ allocated data
watchdog: twl4030_wdt: Change TWL4030_MODULE_PM_RECEIVER to TWL_MODULE_PM_RECEIVER
watchdog: remove depends on CONFIG_EXPERIMENTAL
watchdog: Convert dev_printk(KERN_<LEVEL> to dev_<level>(
watchdog: DA9055 Watchdog driver
watchdog: omap_wdt: eliminate goto
watchdog: omap_wdt: delete redundant platform_set_drvdata() calls
watchdog: omap_wdt: convert to devm_ functions
watchdog: omap_wdt: convert to new watchdog core
watchdog: WatchDog Timer Driver Core: fix comment
watchdog: s3c2410_wdt: use clk_prepare_enable and clk_disable_unprepare
watchdog: imx2_wdt: Select the driver via ARCH_MXC
watchdog: cpu5wdt.c: add missing del_timer call
watchdog: hpwdt.c: Increase version string
watchdog: Convert twl4030_wdt to watchdog core
davinci_wdt: preparation for switch to common clock framework
...
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm
Pull dm update from Alasdair G Kergon:
"Miscellaneous device-mapper fixes, cleanups and performance
improvements.
Of particular note:
- Disable broken WRITE SAME support in all targets except linear and
striped. Use it when kcopyd is zeroing blocks.
- Remove several mempools from targets by moving the data into the
bio's new front_pad area(which dm calls 'per_bio_data').
- Fix a race in thin provisioning if discards are misused.
- Prevent userspace from interfering with the ioctl parameters and
use kmalloc for the data buffer if it's small instead of vmalloc.
- Throttle some annoying error messages when I/O fails."
* tag 'dm-3.8-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-dm: (36 commits)
dm stripe: add WRITE SAME support
dm: remove map_info
dm snapshot: do not use map_context
dm thin: dont use map_context
dm raid1: dont use map_context
dm flakey: dont use map_context
dm raid1: rename read_record to bio_record
dm: move target request nr to dm_target_io
dm snapshot: use per_bio_data
dm verity: use per_bio_data
dm raid1: use per_bio_data
dm: introduce per_bio_data
dm kcopyd: add WRITE SAME support to dm_kcopyd_zero
dm linear: add WRITE SAME support
dm: add WRITE SAME support
dm: prepare to support WRITE SAME
dm ioctl: use kmalloc if possible
dm ioctl: remove PF_MEMALLOC
dm persistent data: improve improve space map block alloc failure message
dm thin: use DMERR_LIMIT for errors
...
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband
Pull more infiniband changes from Roland Dreier:
"Second batch of InfiniBand/RDMA changes for 3.8:
- cxgb4 changes to fix lookup engine hash collisions
- mlx4 changes to make flow steering usable
- fix to IPoIB to avoid pinning dst reference for too long"
* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
RDMA/cxgb4: Fix bug for active and passive LE hash collision path
RDMA/cxgb4: Fix LE hash collision bug for passive open connection
RDMA/cxgb4: Fix LE hash collision bug for active open connection
mlx4_core: Allow choosing flow steering mode
mlx4_core: Adjustments to Flow Steering activation logic for SR-IOV
mlx4_core: Fix error flow in the flow steering wrapper
mlx4_core: Add QPN enforcement for flow steering rules set by VFs
cxgb4: Add LE hash collision bug fix path in LLD driver
cxgb4: Add T4 filter support
IPoIB: Call skb_dst_drop() once skb is enqueued for sending
|
|
Rename stripe_map_discard to stripe_map_range and reuse it for WRITE
SAME bio processing.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
This patch removes map_info from bio-based device mapper targets.
map_info is still used for request-based targets.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Eliminate struct map_info from dm-snap.
map_info->ptr was used in dm-snap to indicate if the bio was tracked.
If map_info->ptr was non-NULL, the bio was linked in tracked_chunk_hash.
This patch removes the use of map_info->ptr. We determine if the bio was
tracked based on hlist_unhashed(&c->node). If hlist_unhashed is true,
the bio is not tracked, if it is false, the bio is tracked.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
This patch removes endio_hook_pool from dm-thin and uses per-bio data instead.
This patch removes any use of map_info in preparation for the next patch
that removes map_info from bio-based device mapper.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Don't use map_info any more in dm-raid1.
map_info was used for writes to hold the region number. For this purpose
we add a new field dm_bio_details to dm_raid1_bio_record.
map_info was used for reads to hold a pointer to dm_raid1_bio_record (if
the pointer was non-NULL, bio details were saved; if the pointer was
NULL, bio details were not saved). We use
dm_raid1_bio_record.details->bi_bdev for this purpose. If bi_bdev is
NULL, details were not saved, if bi_bdev is non-NULL, details were
saved.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Replace map_info with a per-bio structure "struct per_bio_data" in dm-flakey.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Rename struct read_record to bio_record in dm-raid1.
In the following patch, the structure will be used for both read and
write bios, so rename it.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
This patch moves target_request_nr from map_info to dm_target_io and
makes it accessible with dm_bio_get_target_request_nr.
This patch is a preparation for the next patch that removes map_info.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Replace tracked_chunk_pool with per_bio_data in dm-snap.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Replace io_mempool with per_bio_data in dm-verity.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Replace read_record_pool with per_bio_data in dm-raid1.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Introduce a field per_bio_data_size in struct dm_target.
Targets can set this field in the constructor. If a target sets this
field to a non-zero value, "per_bio_data_size" bytes of auxiliary data
are allocated for each bio submitted to the target. These data can be
used for any purpose by the target and help us improve performance by
removing some per-target mempools.
Per-bio data is accessed with dm_per_bio_data. The
argument data_size must be the same as the value per_bio_data_size in
dm_target.
If the target has a pointer to per_bio_data, it can get a pointer to
the bio with dm_bio_from_per_bio_data() function (data_size must be the
same as the value passed to dm_per_bio_data).
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Add WRITE SAME support to dm-io and make it accessible to
dm_kcopyd_zero(). dm_kcopyd_zero() provides an asynchronous interface
whereas the blkdev_issue_write_same() interface is synchronous.
WRITE SAME is a SCSI command that can be leveraged for more efficient
zeroing of a specified logical extent of a device which supports it.
Only a single zeroed logical block is transfered to the target for each
WRITE SAME and the target then writes that same block across the
specified extent.
The dm thin target uses this.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
The linear target can already support WRITE SAME requests so signal
this by setting num_write_same_requests to 1.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
WRITE SAME bios have a payload that contain a single page. When
cloning WRITE SAME bios DM has no need to modify the bi_io_vec
attributes (and doing so would be detrimental). DM need only alter the
start and end of the WRITE SAME bio accordingly.
Rather than duplicate __clone_and_map_discard, factor out a common
function that is also used by __clone_and_map_write_same.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Allow targets to opt in to WRITE SAME support by setting
'num_write_same_requests' in the dm_target structure.
A dm device will only advertise WRITE SAME support if all its
targets and all its underlying devices support it.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
If the parameter buffer is small enough, try to allocate it with kmalloc()
rather than vmalloc().
vmalloc is noticeably slower than kmalloc because it has to manipulate
page tables.
In my tests, on PA-RISC this patch speeds up activation 13 times.
On Opteron this patch speeds up activation by 5%.
This patch introduces a new function free_params() to free the
parameters and this uses new flags that record whether or not vmalloc()
was used and whether or not the input buffer must be wiped after use.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
When allocating memory for the userspace ioctl data, set some
appropriate GPF flags directly instead of using PF_MEMALLOC.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Improve space map error message when unable to allocate a new
metadata block.
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Throttle all errors logged from the IO path by dm thin.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Nearly all of persistent-data is in the IO path so throttle error
messages with DMERR_LIMIT to limit the amount logged when
something has gone wrong.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Reinstate a useful error message when the block manager buffer validator fails.
This was mistakenly eliminated when the block manager was converted to use
dm-bufio.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
If the user does not supply a bitmap region_size to the dm raid target,
a reasonable size is computed automatically. If this is not a power of 2,
the md code will report an error later.
This patch catches the problem early and rounds the region_size to the
next power of two.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Remove unused @data_block parameter from cell_defer.
Change thin_bio_map to use many returns rather than setting a variable.
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Rename cell_defer_except() to cell_defer_no_holder() which describes
its function more clearly.
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
track_chunk is always called with interrupts enabled. Consequently, we
do not need to save and restore interrupt state in "flags" variable.
This patch changes spin_lock_irqsave to spin_lock_irq and
spin_unlock_irqrestore to spin_unlock_irq.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Use a defined macro DM_ENDIO_INCOMPLETE instead of a numeric constant.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
mempool_alloc can't fail if __GFP_WAIT is specified, so the condition
that tests if read_record is non-NULL is always true.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
If "ignore_discard" is specified when creating the thin pool device then
discard support is disabled for that device. The pool device's status
should reflect this fact rather than stating "no_discard_passdown"
(which implies discards are enabled but passdown is disabled).
Reported-by: Zdenek Kabelac <zkabelac@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
When deleting nested btrees, the code forgets to delete the innermost
btree. The thin-metadata code serendipitously compensates for this by
claiming there is one extra layer in the tree.
This patch corrects both problems.
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
When discards are prepared it is best to directly wake the worker that
will process them. The worker will be woken anyway, via periodic
commit, but there is no reason to not wake_worker here.
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
There is a race when discard bios and non-discard bios are issued
simultaneously to the same block.
Discard support is expensive for all thin devices precisely because you
have to be careful to quiesce the area you're discarding. DM thin must
handle this conflicting IO pattern (simultaneous non-discard vs discard)
even though a sane application shouldn't be issuing such IO.
The race manifests as follows:
1. A non-discard bio is mapped in thin_bio_map.
This doesn't lock out parallel activity to the same block.
2. A discard bio is issued to the same block as the non-discard bio.
3. The discard bio is locked in a dm_bio_prison_cell in process_discard
to lock out parallel activity against the same block.
4. The non-discard bio's mapping continues and its all_io_entry is
incremented so the bio is accounted for in the thin pool's all_io_ds
which is a dm_deferred_set used to track time locality of non-discard IO.
5. The non-discard bio is finally locked in a dm_bio_prison_cell in
process_bio.
The race can result in deadlock, leaving the block layer hanging waiting
for completion of a discard bio that never completes, e.g.:
INFO: task ruby:15354 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
ruby D ffffffff8160f0e0 0 15354 15314 0x00000000
ffff8802fb08bc58 0000000000000082 ffff8802fb08bfd8 0000000000012900
ffff8802fb08a010 0000000000012900 0000000000012900 0000000000012900
ffff8802fb08bfd8 0000000000012900 ffff8803324b9480 ffff88032c6f14c0
Call Trace:
[<ffffffff814e5a19>] schedule+0x29/0x70
[<ffffffff814e3d85>] schedule_timeout+0x195/0x220
[<ffffffffa06b9bc1>] ? _dm_request+0x111/0x160 [dm_mod]
[<ffffffff814e589e>] wait_for_common+0x11e/0x190
[<ffffffff8107a170>] ? try_to_wake_up+0x2b0/0x2b0
[<ffffffff814e59ed>] wait_for_completion+0x1d/0x20
[<ffffffff81233289>] blkdev_issue_discard+0x219/0x260
[<ffffffff81233e79>] blkdev_ioctl+0x6e9/0x7b0
[<ffffffff8119a65c>] block_ioctl+0x3c/0x40
[<ffffffff8117539c>] do_vfs_ioctl+0x8c/0x340
[<ffffffff8119a547>] ? block_llseek+0x67/0xb0
[<ffffffff811756f1>] sys_ioctl+0xa1/0xb0
[<ffffffff810561f6>] ? sys_rt_sigprocmask+0x86/0xd0
[<ffffffff814ef099>] system_call_fastpath+0x16/0x1b
The thinp-test-suite's test_discard_random_sectors reliably hits this
deadlock on fast SSD storage.
The fix for this race is that the all_io_entry for a bio must be
incremented whilst the dm_bio_prison_cell is held for the bio's
associated virtual and physical blocks. That cell locking wasn't
occurring early enough in thin_bio_map. This patch fixes this.
Care is taken to always call the new function inc_all_io_entry() with
the relevant cells locked, but they are generally unlocked before
calling issue() to try to avoid holding the cells locked across
generic_submit_request.
Also, now that thin_bio_map may lock bios in a cell, process_bio() is no
longer the only thread that will do so. Because of this we must be sure
to use cell_defer_except() to release all non-holder entries, that
were added by the other thread, because they must be deferred.
This patch depends on "dm thin: replace dm_cell_release_singleton with
cell_defer_except".
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: stable@vger.kernel.org
|
|
Change existing users of the function dm_cell_release_singleton to share
cell_defer_except instead, and then remove the now-unused function.
Everywhere that calls dm_cell_release_singleton, the bio in question
is the holder of the cell.
If there are no non-holder entries in the cell then cell_defer_except
behaves exactly like dm_cell_release_singleton. Conversely, if there
*are* non-holder entries then dm_cell_release_singleton must not be used
because those entries would need to be deferred.
Consequently, it is safe to replace use of dm_cell_release_singleton
with cell_defer_except.
This patch is a pre-requisite for "dm thin: fix race between
simultaneous io and discards to same block".
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
WRITE SAME bios are not yet handled correctly by device-mapper so
disable their use on device-mapper devices by setting
max_write_same_sectors to zero.
As an example, a ciphertext device is incompatible because the data
gets changed according to the location at which it written and so the
dm crypt target cannot support it.
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org
Cc: Milan Broz <mbroz@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Abort dm ioctl processing if userspace changes the data_size parameter
after we validated it but before we finished copying the data buffer
from userspace.
The dm ioctl parameters are processed in the following sequence:
1. ctl_ioctl() calls copy_params();
2. copy_params() makes a first copy of the fixed-sized portion of the
userspace parameters into the local variable "tmp";
3. copy_params() then validates tmp.data_size and allocates a new
structure big enough to hold the complete data and copies the whole
userspace buffer there;
4. ctl_ioctl() reads userspace data the second time and copies the whole
buffer into the pointer "param";
5. ctl_ioctl() reads param->data_size without any validation and stores it
in the variable "input_param_size";
6. "input_param_size" is further used as the authoritative size of the
kernel buffer.
The problem is that userspace code could change the contents of user
memory between steps 2 and 4. In particular, the data_size parameter
can be changed to an invalid value after the kernel has validated it.
This lets userspace force the kernel to access invalid kernel memory.
The fix is to ensure that the size has not changed at step 4.
This patch shouldn't have a security impact because CAP_SYS_ADMIN is
required to run this code, but it should be fixed anyway.
Reported-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
Cc: stable@kernel.org
|
|
This patch fixes a compilation failure on sparc32 by renaming struct node.
struct node is already defined in include/linux/node.h. On sparc32, it
happens to be included through other dependencies and persistent-data
doesn't compile because of conflicting declarations.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alasdair G Kergon <agk@redhat.com>
|
|
Pull vfio update from Alex Williamson.
* tag 'vfio-for-v3.8-v2' of git://github.com/awilliam/linux-vfio:
vfio-pci: Enable device before attempting reset
VFIO: fix out of order labels for error recovery in vfio_pci_init()
VFIO: use ACCESS_ONCE() to guard access to dev->driver
VFIO: unregister IOMMU notifier on error recovery path
vfio-pci: Re-order device reset
vfio: simplify kmalloc+copy_from_user to memdup_user
|
|
Merge the rest of Andrew's patches for -rc1:
"A bunch of fixes and misc missed-out-on things.
That'll do for -rc1. I still have a batch of IPC patches which still
have a possible bug report which I'm chasing down."
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (25 commits)
keys: use keyring_alloc() to create module signing keyring
keys: fix unreachable code
sendfile: allows bypassing of notifier events
SGI-XP: handle non-fatal traps
fat: fix incorrect function comment
Documentation: ABI: remove testing/sysfs-devices-node
proc: fix inconsistent lock state
linux/kernel.h: fix DIV_ROUND_CLOSEST with unsigned divisors
memcg: don't register hotcpu notifier from ->css_alloc()
checkpatch: warn on uapi #includes that #include <uapi/...
revert "rtc: recycle id when unloading a rtc driver"
mm: clean up transparent hugepage sysfs error messages
hfsplus: add error message for the case of failure of sync fs in delayed_sync_fs() method
hfsplus: rework processing of hfs_btree_write() returned error
hfsplus: rework processing errors in hfsplus_free_extents()
hfsplus: avoid crash on failed block map free
kcmp: include linux/ptrace.h
drivers/rtc/rtc-imxdi.c: must include <linux/spinlock.h>
mm: cma: WARN if freed memory is still in use
exec: do not leave bprm->interp on stack
...
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull VFS update from Al Viro:
"fscache fixes, ESTALE patchset, vmtruncate removal series, assorted
misc stuff."
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (79 commits)
vfs: make lremovexattr retry once on ESTALE error
vfs: make removexattr retry once on ESTALE
vfs: make llistxattr retry once on ESTALE error
vfs: make listxattr retry once on ESTALE error
vfs: make lgetxattr retry once on ESTALE
vfs: make getxattr retry once on an ESTALE error
vfs: allow lsetxattr() to retry once on ESTALE errors
vfs: allow setxattr to retry once on ESTALE errors
vfs: allow utimensat() calls to retry once on an ESTALE error
vfs: fix user_statfs to retry once on ESTALE errors
vfs: make fchownat retry once on ESTALE errors
vfs: make fchmodat retry once on ESTALE errors
vfs: have chroot retry once on ESTALE error
vfs: have chdir retry lookup and call once on ESTALE error
vfs: have faccessat retry once on an ESTALE error
vfs: have do_sys_truncate retry once on an ESTALE error
vfs: fix renameat to retry on ESTALE errors
vfs: make do_unlinkat retry once on ESTALE errors
vfs: make do_rmdir retry once on ESTALE errors
vfs: add a flags argument to user_path_parent
...
|
