Implement IRQ handlers for immediate and delayed replies and
notifications. Communication is synchronous and allows for serialization
of maximum one message at a time.
DSP may respond with ADSP_PENDING status for a request - known as
delayed reply - and when situation occurs, framework keeps the lock and
awaits upcoming response through IPCD channel which is handled in
bottom-half. Immediate replies spawn no BH at all as their processing is
very short.
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20200929141247.8058-3-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Declare base structures, registers and extension routines for the catpt
solution.
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20200929141247.8058-2-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Replace commas with semicolons. What is done is essentially described by
the following Coccinelle semantic patch (http://coccinelle.lip6.fr/):
// <smpl>
@@ expression e1,e2; @@
e1
-,
+;
e2
... when any
// </smpl>
Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://lore.kernel.org/r/1601233948-11629-17-git-send-email-Julia.Lawall@inria.fr
The region size reported by the firmware for mc and software
portals was less than allocated by the hardware. This may be
problematic when mmapping the region in user space because the
region size is less than page size. However the size as reserved
by the hardware is 64K.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-14-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In virtual machines the device-id range is defined
between 0x10000-0x20000. The reason for using such a
large range is to avoid overlapping with the PCI range.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Bharat Bhushan <Bharat.Bhushan@nxp.com>
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-13-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The IRQ pool handling functions can be used by both DPRC
driver and VFIO. Adapt and export those functions.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-12-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Before destroying the mc_io, check first that it was
allocated.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-11-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Both DPRC driver and VFIO driver use the same initialization code
for the DPRC. Introduced a new function which groups this
initialization code. The function is exported and may be
used by VFIO as well.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-10-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Create and export a cleanup function for DPRC. The function
is used by the DPRC driver, but it will be used by the VFIO
driver as well.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-9-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently the DPRC scan function is used only by the bus driver.
But the same functionality will be needed by the VFIO driver.
To support this, the dprc scan function was exported and a little
bit adjusted to fit both scenarios. Also the scan mutex initialization
is done when the bus object is created, not in dprc_probe in order
to be used by both VFIO and bus driver.
Similarily dprc_remove_devices is exported to be used by VFIO.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-8-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are already firmware API commands that have multiple
versions. For each multiple version command, another command
to retrieve the API version is issued. This may introduce an important
overhead. The version does not change while the system is running,
so the DPRC API version can be safely cached.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-6-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The QMAN region is memory mapped, so it should be of type
IORESOURCE_MEM. The region flags bits were wrongly used to
pass additional information. Use the bus specific bits for
this purpose.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-5-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch is required for vfio-fsl-mc meta driver to successfully bind
layerscape container devices for device passthrough. This patch adds
a mechanism to allow a layerscape device to specify a driver rather than
a layerscape driver provide a device match.
Example to allow a device (dprc.1) to specifically bind
with driver (vfio-fsl-mc):-
- echo vfio-fsl-mc > /sys/bus/fsl-mc/devices/dprc.1/driver_override
- echo dprc.1 > /sys/bus/fsl-mc/drivers/fsl_mc_dprc/unbind
- echo dprc.1 > /sys/bus/fsl-mc/drivers/vfio-fsl-mc/bind
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Bharat Bhushan <Bharat.Bhushan@nxp.com>
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-4-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Prepare the dprc_scan_objects function to be used by
the VFIO mc driver code. The function is used to scan the mc
objects by the bus driver. The same functionality is
needed by the VFIO mc driver, but in this case the
interrupt configuration is delayed until the userspace
configures the interrupts. In order to use the same function
in both drivers add a new parameter.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-3-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The total number of interrupts is only used for some checks
outside the dprc_scan_objects function. Furthermore, in some
situations the check is made twice. Move the bounds check inside
the function for all situations.
Reviewed-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Link: https://lore.kernel.org/r/20200929085441.17448-2-diana.craciun@oss.nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Read and write io memory should address align on ARCH ARM. Change to use
memcpy_toio to avoid kernel panic caused by the address un-align issue.
Signed-off-by: Sherry Sun <sherry.sun@nxp.com>
Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com>
Link: https://lore.kernel.org/r/20200929091106.24624-5-sherry.sun@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since struct _mic_vring_info and vring are allocated together and follow
vring, if the vring_size() is not four bytes aligned, which will cause
the start address of struct _mic_vring_info is not four byte aligned.
For example, when vring entries is 128, the vring_size() will be 5126
bytes. The _mic_vring_info struct layout in ddr looks like:
0x90002400: 00000000 00390000 EE010000 0000C0FF
Here 0x39 is the avail_idx member, and 0xC0FFEE01 is the magic member.
When EP use ioread32(magic) to reads the magic in RC's share memory, it
will cause kernel panic on ARM64 platform due to the cross-byte io read.
Here read magic in user space use le32toh(vr0->info->magic) will meet
the same issue.
So add round_up(x,4) for vring_size, then the struct _mic_vring_info
will store in this way:
0x90002400: 00000000 00000000 00000039 C0FFEE01
Which will avoid kernel panic when read magic in struct _mic_vring_info.
Signed-off-by: Sherry Sun <sherry.sun@nxp.com>
Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com>
Link: https://lore.kernel.org/r/20200929091106.24624-4-sherry.sun@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On my platform (i.MX53) bus access sometimes fails with
w1_search: max_slave_count 64 reached, will continue next search.
The reason is the use of jiffies to implement a 200us timeout in
mxc_w1_ds2_touch_bit().
On some platforms the jiffies timer resolution is insufficient for this.
Fix by replacing jiffies by ktime_get().
For consistency apply the same change to the other use of jiffies in
mxc_w1_ds2_reset_bus().
Fixes: f80b2581a7 ("w1: mxc_w1: Optimize mxc_w1_ds2_touch_bit()")
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Martin Fuzzey <martin.fuzzey@flowbird.group>
Link: https://lore.kernel.org/r/1601455030-6607-1-git-send-email-martin.fuzzey@flowbird.group
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Set VIRTIO_F_ACCESS_PLATFORM feature for vop driver, as the DMA mapping
details shouldn't decide on the virtio implementation, but the host PCIe
implementation.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sherry Sun <sherry.sun@nxp.com>
Signed-off-by: Joakim Zhang <qiangqing.zhang@nxp.com>
Link: https://lore.kernel.org/r/20200929084944.24146-1-sherry.sun@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A struct device is a dynamic structure, with reference counting.
"Tricking" the kernel to make a dynamic structure static, by working
around the driver core release detection logic, is not nice.
Because of this, this code has been used as an example for others on
"how to do things", which is just about the worst thing possible to have
happen.
Fix this all up by making the platform device dynamic and providing a
real release function.
Fixes: b02f6a2ef0 ("platform/x86: intel_pmc_core: Attach using APCI HID "INT33A1"")
Cc: Rajneesh Bhardwaj <rajneesh.bhardwaj@intel.com>
Cc: Vishwanath Somayaji <vishwanath.somayaji@intel.com>
Cc: Darren Hart <dvhart@infradead.org>
Cc: Andy Shevchenko <andy@infradead.org>
Reported-by: Maximilian Luz <luzmaximilian@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Rajat Jain <rajatja@google.com>
Link: https://lore.kernel.org/r/20200923184803.192265-1-gregkh@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Refactor: Handle this NFS version-specific mapping in the only
place where nfserr_wrongsec is generated.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Refactor: I'm about to change the return value from .pc_func. Clear
the way by replacing the RETURN_STATUS() macro with logic that
plants the status code directly into the response structure.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Remove special dispatcher logic for NFSv2 error responses. These are
rare to the point of becoming extinct, but all NFS responses have to
pay the cost of the extra conditional branches.
With this change, the NFSv2 error cases now get proper
xdr_ressize_check() calls.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
nfsd_release_fhandle() assumes that rqstp->rq_resp always points to
an nfsd_fhandle struct. In fact, no NFSv2 procedure uses struct
nfsd_fhandle as its response structure.
So far that has been "safe" to do because the res structs put the
resp->fh field at that same offset as struct nfsd_fhandle. I don't
think that's a guarantee, though, and there is certainly nothing
preventing a developer from altering the fields in those structures.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
nfsd_dispatch() is a hot path. Ensure the compiler takes the
processing of rare error cases out of line.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
For consistency and code legibility, use a similar organization of
variables as svc_generic_dispatch().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Add a documenting comment for the function. Remove comments that
simply describe obvious aspects of the code, but leave comments
that explain the differences in processing of each NFS version.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Reorder the arms so the compiler places checks for the most frequent
case first.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
nfsd_dispatch() is a hot path. Let's optimize the XDR method calls
for the by-far common case, which is that the XDR methods are indeed
present.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Clean up: Follow-up on ten-year-old commit b9081d90f5 ("NFS: kill
off complicated macro 'PROC'") by performing the same conversion in
the NFSACL code. To reduce the chance of error, I copied the original
C preprocessor output and then made some minor edits.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Clean up: Follow-up on ten-year-old commit b9081d90f5 ("NFS: kill
off complicated macro 'PROC'") by performing the same conversion in
the lockd code. To reduce the chance of error, I copied the original
C preprocessor output and then made some minor edits.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
There's no protection in nfsd_dispatch() against a NULL .pc_func
helpers. A malicious NFS client can trigger a crash by invoking the
unused/unsupported NFSv2 ROOT or WRITECACHE procedures.
The current NFSD dispatcher does not support returning a void reply
to a non-NULL procedure, so the reply to both of these is wrong, for
the moment.
Cc: <stable@vger.kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
The latest reference to usbfs_conn_disc_event() removed in
commit fb28d58b72 ("USB: remove CONFIG_USB_DEVICEFS")
in 2012 and now a user poll() waits infinitely for content changes.
Signed-off-by: Sergey Korolev <s.korolev@ndmsystems.com>
Link: https://lore.kernel.org/r/20200809161233.13135-1-s.korolev@ndmsystems.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
STMicroelectronics USB Type-C port controllers use I2C interface to
configure, control and read the operation status of the device. All ST USB
Type-C port controllers are based on the same I2C register map. That's why
this driver can be used with all ST USB Type-C ICs.
Some ST USB Type-C port controllers are Dual Role Port (DRP), only Sink or
Source, some supports USB Power Delivery. This can be configured through
connector device tree bindings.
This driver is a basic Type-C port controller driver, with no power
delivery support. It allows to configure ST USB Type-C port controller.
Interrupt is supported and enables CC connection events, to detect
attach and detach and update Type-C subsystem accordingly as well as usb
role switch.
ST USB Type-C port controller can be supplied in three different ways
depending on the target application:
- through VDD pin only (so VDD is the main supply)
- through VSYS pin only (so VSYS is the main supply)
- through VDD and VSYS pins.
When both VDD and VSYS power supplies are present, the low power supply
VSYS is selected as main supply when VSYS voltage is above 3.1V, else
VDD is selected as main supply.
In case of Source or Dual port type, if VDD supply is present, it has to be
enabled in case of Source power role to provide Vbus. When interrupt
support is available, VDD supply is dynamically managed upon attach/detach
interrupt. When there is no interrupt support, VDD supply is enabled by
default.
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Amelie Delaunay <amelie.delaunay@st.com>
Link: https://lore.kernel.org/r/20200924090049.9041-5-amelie.delaunay@st.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch adds a function that converts power operation mode string into
power operation mode value.
It is useful to configure power operation mode through device tree
property, as power capabilities may be linked to hardware design.
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Amelie Delaunay <amelie.delaunay@st.com>
Link: https://lore.kernel.org/r/20200924090049.9041-3-amelie.delaunay@st.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Driver ->power_on and ->power_off callbacks leaks internal SMCC firmware
return codes to phy caller. This patch converts SMCC error codes to
standard linux errno codes. Include file linux/arm-smccc.h already provides
defines for SMCC error codes, so use them instead of custom driver defines.
Note that return value is signed 32bit, but stored in unsigned long type
with zero padding.
Tested-by: Tomasz Maciej Nowak <tmn505@gmail.com>
Link: https://lore.kernel.org/r/20200902144344.16684-2-pali@kernel.org
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Reviewed-by: Rob Herring <robh@kernel.org>
After installing archlinux, the mute led and micmute led are not working
at all. This patch fix this issue by applying a fixup from similar
model. These mute leds are confirmed working on HP Elitebook 845 G7.
Signed-off-by: Qiu Wenbo <qiuwenbo@kylinos.com.cn>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20201002124454.7240-1-qiuwenbo@kylinos.com.cn
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Some platforms have only super speed data bus connected to this device
and high speed data bus directly connected to the SoC. In such platforms
modelling connector as a child of this device is making it non compliant
with usb connector bindings. By modelling connector node as standalone
device node along with this device and the SoC data bus will make it
compliant with usb connector bindings.
Update the driver to handle this model by using OF graph API to get the
connector fwnode and usb role switch class API to get role switch handle.
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20200920134905.4370-5-biju.das.jz@bp.renesas.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Document HS and SS data bus for the "usb-role-switch" enabled case.
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20200920134905.4370-4-biju.das.jz@bp.renesas.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Convert ti,hd3ss3220.txt to YAML. Updated the binding documentation
as graph bindings of this device model Super Speed (SS) data bus to
the Super Speed (SS) capable connector.
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Link: https://lore.kernel.org/r/20200920134905.4370-3-biju.das.jz@bp.renesas.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All slots in sysrq_key_table[] are either used, reserved or at least
commented with their intended use. This patch adds capital letter versions
available, which means adding 26 more entries.
For already existing SysRq operations the user presses Alt-SysRq-<key>, and
for the newly added ones Alt-Shift-SysRq-<key>.
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://lore.kernel.org/r/20200818112825.6445-2-andrzej.p@collabora.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The list_lru_count() returns the pre node count, but the new xattr
shrinkers are memcg aware, so the shrinkers should return per memcg
count by calling list_lru_shrink_count() instead. Otherwise over-shrink
might be experienced. The problem was spotted by visual code
inspection.
Cc: Trond Myklebust <trond.myklebust@hammerspace.com>
Cc: Anna Schumaker <anna.schumaker@netapp.com>
Cc: Frank van der Linden <fllinden@amazon.com>
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Frank van der Linden <fllinden@amazon.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Since commit 0e0cb35b41 ("NFSv4: Handle NFS4ERR_OLD_STATEID in
CLOSE/OPEN_DOWNGRADE") the following livelock may occur if a CLOSE races
with the update of the nfs_state:
Process 1 Process 2 Server
========= ========= ========
OPEN file
OPEN file
Reply OPEN (1)
Reply OPEN (2)
Update state (1)
CLOSE file (1)
Reply OLD_STATEID (1)
CLOSE file (2)
Reply CLOSE (-1)
Update state (2)
wait for state change
OPEN file
wake
CLOSE file
OPEN file
wake
CLOSE file
...
...
We can avoid this situation by not issuing an immediate retry with a bumped
seqid when CLOSE/OPEN_DOWNGRADE receives NFS4ERR_OLD_STATEID. Instead,
take the same approach used by OPEN and wait at least 5 seconds for
outstanding stateid updates to complete if we can detect that we're out of
sequence.
Note that after this change it is still possible (though unlikely) that
CLOSE waits a full 5 seconds, bumps the seqid, and retries -- and that
attempt races with another OPEN at the same time. In order to avoid this
race (which would result in the livelock), update
nfs_need_update_open_stateid() to handle the case where:
- the state is NFS_OPEN_STATE, and
- the stateid doesn't match the current open stateid
Finally, nfs_need_update_open_stateid() is modified to be idempotent and
renamed to better suit the purpose of signaling that the stateid passed
is the next stateid in sequence.
Fixes: 0e0cb35b41 ("NFSv4: Handle NFS4ERR_OLD_STATEID in CLOSE/OPEN_DOWNGRADE")
Cc: stable@vger.kernel.org # v5.4+
Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
