The nvme request generation counter is intended to catch duplicate
completions. Incrementing the counter on submission means duplicates can
only be caught if the request tag is reallocated and dispatched prior to
the driver observing the corrupted CQE. Incrementing on completion
removes this window, making it possible to detect duplicate completions
in consecutive entries.
Signed-off-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Currently applications have a hard time figuring out which
nvme-over-fabrics arguments are supported for any given kernel;
the ioctl will return an error code on failure, and the application
has to guess whether this was due to an invalid argument or due
to a connection or controller error.
With this patch applications can read a list of supported
arguments by simply reading from /dev/nvme-fabrics, allowing
them to validate the connection string.
Signed-off-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
If VBT size is not a multiple of 4, the last 4-byte store will be out of
bounds of the allocated buffer. Spotted with KASAN. Round up the
allocation size.
v2: Use round_up() intead of roundup() as it's a power of 2 (Thomas)
Reported-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Fixes: a36e7dc0af ("drm/i915/dg1: Read OPROM via SPI controller")
Cc: Clint Taylor <clinton.a.taylor@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211222081654.1843211-1-jani.nikula@intel.com
When cross compiling i386_defconfig on an arm64 host with clang, there
are a few instances of '-Waddress-of-packed-member' and
'-Wgnu-variable-sized-type-not-at-end' in arch/x86/boot/compressed/,
which should both be disabled with the cc-disable-warning calls in that
directory's Makefile, which indicates that cc-disable-warning is failing
at the point of testing these flags.
The cc-disable-warning calls fail because at the point that the flags
are tested, KBUILD_CFLAGS has '-march=i386' without $(CLANG_FLAGS),
which has the '--target=' flag to tell clang what architecture it is
targeting. Without the '--target=' flag, the host architecture (arm64)
is used and i386 is not a valid value for '-march=' in that case. This
error can be seen by adding some logging to try-run:
clang-14: error: the clang compiler does not support '-march=i386'
Invoking the compiler has to succeed prior to calling cc-option or
cc-disable-warning in order to accurately test whether or not the flag
is supported; if it doesn't, the requested flag can never be added to
the compiler flags. Move $(CLANG_FLAGS) to the beginning of KBUILD_FLAGS
so that any new flags that might be added in the future can be
accurately tested.
Fixes: d5cbd80e30 ("x86/boot: Add $(CLANG_FLAGS) to compressed KBUILD_CFLAGS")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/r/20211222163040.1961481-1-nathan@kernel.org
Kernel generates mapping change message, XFRM_MSG_MAPPING,
when a source port chage is detected on a input state with UDP
encapsulation set. Kernel generates a message for each IPsec packet
with new source port. For a high speed flow per packet mapping change
message can be excessive, and can overload the user space listener.
Introduce rate limiting for XFRM_MSG_MAPPING message to the user space.
The rate limiting is configurable via netlink, when adding a new SA or
updating it. Use the new attribute XFRMA_MTIMER_THRESH in seconds.
v1->v2 change:
update xfrm_sa_len()
v2->v3 changes:
use u32 insted unsigned long to reduce size of struct xfrm_state
fix xfrm_ompat size Reported-by: kernel test robot <lkp@intel.com>
accept XFRM_MSG_MAPPING only when XFRMA_ENCAP is present
Co-developed-by: Thomas Egerer <thomas.egerer@secunet.com>
Signed-off-by: Thomas Egerer <thomas.egerer@secunet.com>
Signed-off-by: Antony Antony <antony.antony@secunet.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
SA use_time was only updated once, for the first packet.
with this fix update the use_time for every packet.
Signed-off-by: Antony Antony <antony.antony@secunet.com>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Remove the references to the old spi-nor.c file.
The old drivers/mtd/spi-nor/spi-nor.c file
is not more present and now some of its code is contained in:
drivers/mtd/spi-nor/core.c
Signed-off-by: Flavio Suligoi <f.suligoi@asem.it>
[tudor.ambarus@microchip.com:
- remove change in Documentation/driver-api/mtd/spi-nor.rst.
The documentation has to be rewritten entirely.
- update commit message]
Signed-off-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Reviewed-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Link: https://lore.kernel.org/r/20210126092516.1431913-1-f.suligoi@asem.it
Set reply queue depth of 1K for B0 and 4K for A0.
While freeing the segmented request queues use the actual queue depth that
is used while creating them.
Link: https://lore.kernel.org/r/20211220141159.16117-25-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Enhance driver to consider MPI3_IOCSTATUS_SCSI_IOC_TERMINATED as a success
for TMs issued by it and check the pending I/Os to decide the success or
failure of the task management requests instead of just considering the
MPI3_IOCSTATUS_SCSI_IOC_TERMINATED as a failure of the task management
request.
Link: https://lore.kernel.org/r/20211220141159.16117-24-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Add support for the io_uring interface in I/O-polled mode.
This feature is disabled in the driver by default. To enable the feature, a
module parameter "poll_queues" has to be set with the desired number of
polling queues.
When the feature is enabled, the driver reserves a certain number of
operational queue pairs for the poll_queues either from the available queue
pairs or creates additional queue pairs based on the operational queue
availability.
The Polling queues will have corresponding IRQ and ISR functions as similar
to default queues. However, the IRQ line is disabled by the driver for
poll_queues.
Link: https://lore.kernel.org/r/20211220141159.16117-22-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Print cable management & temperature threshold event data.
Use vendor id & device id macro definitions from MPI3 headers.
Link: https://lore.kernel.org/r/20211220141159.16117-21-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The IOC sends a Prepare for Reset Event to the host to prepare for a Soft
Reset. This event data has two reason codes:
1. Start - The host is expected to gracefully quiesce all I/O within
approximately 1 second.
2. Abort - The IOC is requesting to abort a previous Prepare for Reset
Event request. Normal I/O may be resumed.
Link: https://lore.kernel.org/r/20211220141159.16117-20-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Enhance driver to gracefully handle discrepancies in certain key data sizes
between firmware update operations as mentioned below:
- The driver displays an error message and marks the controller as
unrecoverable if the firmware reports ReplyFrameSize that is greater
than the current ReplyFrameSize.
- If the firmware reports ReplyFrameSize greater than the current
ReplyFrameSize then the driver uses the current ReplyFrameSize while
copying the reply messages.
- The driver displays an error message and marks the controller as
unrecoverable if the firmware reports MaxOperationalReplyQueues less
than the currently allocated operational reply queues count.
- If the firmware reports MaxOperationalReplyQueues that is greater than
the currently allocated operational reply queue count then the driver
ignores the new increased value and uses the previously allocated number
of operational queues only.
- If the firmware reports MaxDevHandle greater than the previously used
MaxDevHandle value after a reset then the driver re-allocates the
'device remove pending bitmap' buffer with the newer size using
krealloc().
Link: https://lore.kernel.org/r/20211220141159.16117-18-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Detect asynchronous reset that occurred in the firmware by polling for
reset history bit of IOC status register is set and if that bit is set,
then the driver waits for the controller to become ready and then
re-initializes the controller.
Also reduce the time driver is waiting for the controller to acknowledge
the reset action after issuing a specific reset action to the
controller. The wait time is reduced from 510 seconds to 30 seconds. If the
controller didn't acknowledge a specific reset action within the time
interval then the driver marks the controller as unrecoverable instead of
retrying two more times prior to giving up.
Link: https://lore.kernel.org/r/20211220141159.16117-17-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Currently the driver marks the controller as unrecoverable if there is an
asynchronous reset or fault during the initialization, reinitialization
post reset, and OS resume.
Enhance driver to retry the initialization, re-initialization, and resume
sequences for a maximum of 3 times if the controller became faulty or
asynchronously reset due to a firmware activation during the initialization
sequence.
Link: https://lore.kernel.org/r/20211220141159.16117-15-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Separate out reply and sense buffer allocation and initialization into two
routines and call only initialization routine while issuing the IOC Init
request message.
Also move out the event enable logic to a separate function.
Link: https://lore.kernel.org/r/20211220141159.16117-13-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Save snapdump and fault the controller with the given reason code if it is
already not in the fault or not in asynchronous reset. This ensures that
soft reset is issued from the watchdog thread. This will also be used to
handle initialization time faults/resets/timeout as in those cases
immediate soft reset invocation is not required.
Link: https://lore.kernel.org/r/20211220141159.16117-12-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The following special handling is needed for UNMAP commands issued to NVMe
drives:
- On B0 boards, if the parameter list length is greater than 24 and not a
16-byte multiple, then truncate the parameter list length to a 16-byte
multiple.
- On A0 boards, if the parameter list length is greater than block
descriptor data length + 8, then truncate the parameter list length to
block descriptor data length + 8 value.
Link: https://lore.kernel.org/r/20211220141159.16117-10-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Add validation for various access statuses prior to exposing attached
target device to the operating system.
Link: https://lore.kernel.org/r/20211220141159.16117-8-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The SAS4 Controller firmware exposes the SES devices in Managed PCIe Switch
as a PCIe Device Type SCSI Device
(MPI3_DEVICE0_PCIE_DEVICE_INFO_TYPE_SCSI_DEVICE).
Driver is enhanced to handle this device type by:
- Exposing the device to the upper layers and
- Not updating any hardware sectors & virtual boundary settings as these
settings are needed only for NVMe devices.
Link: https://lore.kernel.org/r/20211220141159.16117-7-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Don't issue the soft reset if internal commands are flushed out with reset
status. Soft reset needs to be issued only if commands are really timed
out.
Link: https://lore.kernel.org/r/20211220141159.16117-4-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Use spin_lock_irqsave() instead of spin_lock() while acquiring
reply_free_queue_lock & sbq_lock locks.
Link: https://lore.kernel.org/r/20211220141159.16117-3-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The driver doesn't express DMA addressing limitation under 32-bits anywhere
else, so remove the spurious GFP_DMA allocation.
Link: https://lore.kernel.org/r/20211222092247.928711-1-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The driver doesn't express DMA addressing limitation under 32-bits anywhere
else, so remove the spurious GFP_DMA allocation.
Link: https://lore.kernel.org/r/20211222092048.925829-1-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The myrs devices supports 64-bit addressing, so remove the spurious GFP_DMA
allocations.
Link: https://lore.kernel.org/r/20211222091935.925624-1-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The driver doesn't express DMA addressing limitation under 32-bits anywhere
else, so remove the spurious GFP_DMA allocation.
Link: https://lore.kernel.org/r/20211222091801.924745-1-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The driver doesn't express DMA addressing limitation under 32-bits anywhere
else, so remove the spurious GFP_DMA allocation.
Link: https://lore.kernel.org/r/20211222091630.922788-1-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The allocated buffers are used as a command payload, for which the block
layer and/or DMA API do the proper bounce buffering if needed.
Link: https://lore.kernel.org/r/20211222090842.920724-1-hch@lst.de
Reported-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The allocated buffers are used as a command payload, for which the block
layer and/or DMA API do the proper bounce buffering if needed.
Link: https://lore.kernel.org/r/20211222090311.916624-1-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
All the error handling paths of 'mlx5e_tc_add_fdb_flow()' end to 'err_out'
where 'flow_flag_set(flow, FAILED);' is called.
All but the new error handling paths added by the commits given in the
Fixes tag below.
Fix these error handling paths and branch to 'err_out'.
Fixes: 166f431ec6 ("net/mlx5e: Add indirect tc offload of ovs internal port")
Fixes: b16eb3c81f ("net/mlx5: Support internal port as decap route device")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
(cherry picked from commit 31108d142f)
When there is ct or sample action, the ct or sample rule will be deleted
and return. But if there is an extra mirror action, the forward rule can't
be deleted because of the return.
Fix it by removing the return.
Fixes: 69e2916ebc ("net/mlx5: CT: Add support for mirroring")
Fixes: f94d6389f6 ("net/mlx5e: TC, Add support to offload sample action")
Signed-off-by: Chris Mi <cmi@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
There are two ICOSQs per channel: one is needed for RX, and the other
for async operations (XSK TX, kTLS offload). Currently, the recovery
flow for both is the same, and async ICOSQ is mistakenly treated like
the regular ICOSQ.
This patch prevents running the regular ICOSQ recovery on async ICOSQ.
The purpose of async ICOSQ is to handle XSK wakeup requests and post
kTLS offload RX parameters, it has nothing to do with RQ and XSKRQ UMRs,
so the regular recovery sequence is not applicable here.
Fixes: be5323c837 ("net/mlx5e: Report and recover from CQE error on ICOSQ")
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Aya Levin <ayal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Both regular RQ and XSKRQ use the same ICOSQ for UMRs. When doing
recovery for the ICOSQ, don't forget to deactivate XSKRQ.
XSK can be opened and closed while channels are active, so a new mutex
prevents the ICOSQ recovery from running at the same time. The ICOSQ
recovery deactivates and reactivates XSKRQ, so any parallel change in
XSK state would break consistency. As the regular RQ is running, it's
not enough to just flush the recovery work, because it can be
rescheduled.
Fixes: be5323c837 ("net/mlx5e: Report and recover from CQE error on ICOSQ")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
When TC classifier action offloads are disabled (CONFIG_MLX5_CLS_ACT in
Kconfig), the mlx5e_rep_tc_receive() function which is responsible for
passing the skb to the stack (or freeing it) is defined as a nop, and
results in leaking the skb memory. Replace the nop with a call to
napi_gro_receive() to resolve the leak.
Fixes: 28e7606fa8 ("net/mlx5e: Refactor rx handler of represetor device")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Ariel Levkovich <lariel@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Only prio 1 is supported if firmware doesn't support ignore flow
level for nic mode. The offending commit removed the check wrongly.
Add it back.
Fixes: 9a99c8f125 ("net/mlx5e: E-Switch, Offload all chain 0 priorities when modify header and forward action is not supported")
Signed-off-by: Chris Mi <cmi@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
SF do not directly control the PCI device. During recovery flow SF
should not be allowed to do pci disable or pci reset, its PF will do it.
It fixes the following kernel trace:
mlx5_core.sf mlx5_core.sf.25: mlx5_health_try_recover:387:(pid 40948): starting health recovery flow
mlx5_core 0000:03:00.0: mlx5_pci_slot_reset was called
mlx5_core 0000:03:00.0: wait vital counter value 0xab175 after 1 iterations
mlx5_core.sf mlx5_core.sf.25: firmware version: 24.32.532
mlx5_core.sf mlx5_core.sf.23: mlx5_health_try_recover:387:(pid 40946): starting health recovery flow
mlx5_core 0000:03:00.0: mlx5_pci_slot_reset was called
mlx5_core 0000:03:00.0: wait vital counter value 0xab193 after 1 iterations
mlx5_core.sf mlx5_core.sf.23: firmware version: 24.32.532
mlx5_core.sf mlx5_core.sf.25: mlx5_cmd_check:813:(pid 40948): ENABLE_HCA(0x104) op_mod(0x0) failed,
status bad resource state(0x9), syndrome (0x658908)
mlx5_core.sf mlx5_core.sf.25: mlx5_function_setup:1292:(pid 40948): enable hca failed
mlx5_core.sf mlx5_core.sf.25: mlx5_health_try_recover:389:(pid 40948): health recovery failed
Fixes: 1958fc2f07 ("net/mlx5: SF, Add auxiliary device driver")
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
