The admin queue work request buffer, aq->q_wr, is allocated via kcalloc in
__ionic_create_rdma_adminq. However, it was not being freed in the
corresponding teardown function __ionic_destroy_rdma_adminq. This results
in a memory leak. Fix this leak by adding the missing kfree(aq->q_wr) in
the destruction path.
Bernard Metzler [Tue, 23 Sep 2025 14:45:36 +0000 (16:45 +0200)]
RDMA/siw: Always report immediate post SQ errors
In siw_post_send(), any immediate error encountered during processing of
the work request list must be reported to the caller, even if previous
work requests in that list were just accepted and added to the send queue.
Not reporting those errors confuses the caller, which would wait
indefinitely for the failing and potentially subsequently aborted work
requests completion.
This fixes a case where immediate errors were overwritten by subsequent
code in siw_post_send().
RDMA/bnxt_re: improve clarity in ALLOC_PAGE handler
Update uverbs_copy_to call to use sizeof(dpi) instead of sizeof(length)
when copying the device page index (DPI) back to user space. Both dpi
and length are declared as u32, so this change has no functional impact
but makes the code clearer.
Dan Carpenter [Tue, 23 Sep 2025 11:20:45 +0000 (14:20 +0300)]
RDMA/irdma: Fix positive vs negative error codes in irdma_post_send()
This code accidentally returns positive EINVAL instead of negative
-EINVAL. Some of the callers treat positive returns as success.
Add the missing '-' char.
eth header from ib_ud_header structure packs the mac
into 4B high and 2B low parts. But when 4B high is used
in memcpy, it sees it as overflow. However, this is safe
due to the 4B high and 2B low arrangement in the structure.
To avoid the memcpy warning, use ether_addr_copy to copy
the mac address.
In function ‘fortify_memcpy_chk’,
inlined from ‘ionic_set_ah_attr.isra’ at drivers/infiniband/hw/ionic/ionic_controlpath.c:609:3:
./include/linux/fortify-string.h:580:25: error: call to ‘__read_overflow2_field’ declared with attribute warning: detected read beyond size of field (2nd parameter); maybe use struct_group()? [-Werror=attribute-warning]
580 | __read_overflow2_field(q_size_field, size);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
make[6]: *** [scripts/Makefile.build:287: drivers/infiniband/hw/ionic/ionic_controlpath.o] Error 1
make[5]: *** [scripts/Makefile.build:556: drivers/infiniband/hw/ionic] Error 2
make[5]: *** Waiting for unfinished jobs....
make[4]: *** [scripts/Makefile.build:556: drivers/infiniband/hw] Error 2
make[3]: *** [scripts/Makefile.build:556: drivers/infiniband] Error 2
make[2]: *** [scripts/Makefile.build:556: drivers] Error 2
make[1]: *** [/tmp/tmp53nb1nwr/Makefile:2011: .] Error 2
make: *** [Makefile:248: __sub-make] Error 2
Fixes: e8521822c733 ("RDMA/ionic: Register device ops for control path") Reported-by: Leon Romanovsky <leon@kernel.org> Closes: https://lore.kernel.org/lkml/20250918180750.GA135135@unreal/ Signed-off-by: Abhijit Gangurde <abhijit.gangurde@amd.com> Link: https://patch.msgid.link/20250919121301.1113759-2-abhijit.gangurde@amd.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
Gui-Dong Han [Fri, 19 Sep 2025 02:52:12 +0000 (02:52 +0000)]
RDMA/rxe: Fix race in do_task() when draining
When do_task() exhausts its iteration budget (!ret), it sets the state
to TASK_STATE_IDLE to reschedule, without a secondary check on the
current task->state. This can overwrite the TASK_STATE_DRAINING state
set by a concurrent call to rxe_cleanup_task() or rxe_disable_task().
While state changes are protected by a spinlock, both rxe_cleanup_task()
and rxe_disable_task() release the lock while waiting for the task to
finish draining in the while(!is_done(task)) loop. The race occurs if
do_task() hits its iteration limit and acquires the lock in this window.
The cleanup logic may then proceed while the task incorrectly
reschedules itself, leading to a potential use-after-free.
This bug was introduced during the migration from tasklets to workqueues,
where the special handling for the draining case was lost.
Fix this by restoring the original pre-migration behavior. If the state is
TASK_STATE_DRAINING when iterations are exhausted, set cont to 1 to
force a new loop iteration. This allows the task to finish its work, so
that a subsequent iteration can reach the switch statement and correctly
transition the state to TASK_STATE_DRAINED, stopping the task as intended.
Fixes: 9b4b7c1f9f54 ("RDMA/rxe: Add workqueue support for rxe tasks") Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Gui-Dong Han <hanguidong02@gmail.com> Link: https://patch.msgid.link/20250919025212.1682087-1-hanguidong02@gmail.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
When computing the delta, the sa_local_svc_timeout_ms is read without
ib_nl_request_lock held. Though unlikely in practice, this can cause
a race condition if multiple local service threads are managing the
timeout.
Fixes: 2ca546b92a02 ("IB/sa: Route SA pathrecord query through netlink") Signed-off-by: Vlad Dumitrescu <vdumitrescu@nvidia.com> Reviewed-by: Mark Zhang <markzhang@nvidia.com> Signed-off-by: Edward Srouji <edwards@nvidia.com> Link: https://patch.msgid.link/20250916163112.98414-1-edwards@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
Currently, all master upper netdevices (e.g., bond, VRF) are treated
equally.
When a VRF netdevice is used over an IPoIB netdevice, the expected
netdev resolution is on the lower IPoIB device which has the IP address
assigned to it and not the VRF device.
The rdma_cm module (CMA) tries to match incoming requests to a
particular netdevice. When successful, it also validates that the return
path points to the same device by performing a routing table lookup.
Currently, the former would resolve to the VRF netdevice, while the
latter to the correct lower IPoIB netdevice, leading to failure in
rdma_cm.
Improve this by ignoring the VRF master netdevice, if it exists, and
instead return the lower IPoIB device.
Signed-off-by: Vlad Dumitrescu <vdumitrescu@nvidia.com> Reviewed-by: Parav Pandit <parav@nvidia.com> Signed-off-by: Edward Srouji <edwards@nvidia.com> Link: https://patch.msgid.link/20250916111103.84069-5-edwards@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
RDMA/core: Use route entry flag to decide on loopback traffic
addr_resolve() considers a destination to be local if the next-hop
device of the resolved route for the destination is the loopback
netdevice.
This fails when the source and destination IP addresses belong to
a netdev enslaved to a VRF netdev. In this case the next-hop device
is the VRF itself:
$ ip link add name myvrf up type vrf table 100
$ ip link set ens2f0np0 master myvrf up
$ ip addr add 192.168.1.1/24 dev ens2f0np0
$ ip route get 192.168.1.1 oif myvrf
local 192.168.1.1 dev myvrf table 100 src 192.168.1.1 uid 0
cache <local>
This results in packets being generated with an incorrect destination
MAC of the VRF netdevice and ib_write_bw failing with timeout.
Solve this by determining if a destination is local or not based on
the resolved route's type rather than based on its next-hop netdevice
loopback flag.
This enables to resolve loopback traffic with and without VRF
configurations in a uniform way.
Signed-off-by: Parav Pandit <parav@nvidia.com> Reviewed-by: Vlad Dumitrescu <vdumitrescu@nvidia.com> Signed-off-by: Edward Srouji <edwards@nvidia.com> Link: https://patch.msgid.link/20250916111103.84069-4-edwards@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
RDMA/core: Resolve MAC of next-hop device without ARP support
Currently, if the next-hop netdevice does not support ARP resolution,
the destination MAC address is silently set to zero without reporting
an error. This leads to incorrect behavior and may result in packet
transmission failures.
Fix this by deferring MAC resolution to the IP stack via neighbour
lookup, allowing proper resolution or error reporting as appropriate.
Fixes: 7025fcd36bd6 ("IB: address translation to map IP toIB addresses (GIDs)") Signed-off-by: Parav Pandit <parav@nvidia.com> Reviewed-by: Vlad Dumitrescu <vdumitrescu@nvidia.com> Signed-off-by: Edward Srouji <edwards@nvidia.com> Link: https://patch.msgid.link/20250916111103.84069-3-edwards@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
To reduce dependencies in IFF_LOOPBACK in route and neighbour resolution
steps, squash the static function to its single caller and simplify the
code.
Until now, network field was set even when neighbour resolution failed.
With this change, dev_addr output fields are valid only when resolution
is successful.
Signed-off-by: Parav Pandit <parav@nvidia.com> Reviewed-by: Vlad Dumitrescu <vdumitrescu@nvidia.com> Signed-off-by: Edward Srouji <edwards@nvidia.com> Link: https://patch.msgid.link/20250916111103.84069-2-edwards@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
Shiraz Saleem [Wed, 27 Aug 2025 15:25:44 +0000 (10:25 -0500)]
RDMA/irdma: Extend CQE Error and Flush Handling for GEN3 Devices
Enhance the CQE error and flush handling specific to GEN3 devices.
Unlike GEN1/2 devices, which depend on software to generate completions
in error, GEN3 devices leverage firmware to generate CQEs in error for
all WQEs posted after a QP moves to an error state.
Key changes include:
- Updating the CQ poll logic to properly advance the CQ head in the
event of a flush CQE.
- Updating the flush logic for GEN3 to pass error WQE idx
for SQ on an AE to flush out unprocessed WQEs in error.
- Isolating the decoding of AE to flush codes into a separate routine
irdma_ae_to_qp_err_code. This routine can now be leveraged to
flush error CQEs on an AE and when error CQE is received for SRQ.
Shiraz Saleem [Wed, 27 Aug 2025 15:25:42 +0000 (10:25 -0500)]
RDMA/irdma: Restrict Memory Window and CQE Timestamping to GEN3
With the deprecation of Memory Window and Timestamping support in GEN2,
move these features to be exclusive to GEN3. This iteration supports
only Type2 Memory Windows. Additionally, it includes the reporting of
the timestamp mask and Host Channel Adapter (HCA) core clock frequency
via the query device verb.
RDMA/irdma: Add support for V2 HMC resource management scheme
HMC resource initialization is updated to support V1 or V2 approach
based on the FW capability. In the V2 approach, driver receives the
assigned HMC resources count and verifies if it will fit in the given
local memory. If it doesn't fit, the driver load fails.
Signed-off-by: Vinoth Kumar Chandra Mohan <vinoth.kumar.chandra.mohan@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Link: https://patch.msgid.link/20250827152545.2056-11-tatyana.e.nikolova@intel.com Tested-by: Jacob Moroni <jmoroni@google.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
Shiraz Saleem [Wed, 27 Aug 2025 15:25:37 +0000 (10:25 -0500)]
RDMA/irdma: Add GEN3 virtual QP1 support
Add a new RDMA virtual channel op during QP1 creation that allow the
Control Plane (CP) to virtualize a regular QP as QP1 on non-default
RDMA capable vPorts. Additionally, the CP will return the Qsets to use
on the ib_device of the vPort.
Mustafa Ismail [Wed, 27 Aug 2025 15:25:36 +0000 (10:25 -0500)]
RDMA/irdma: Introduce GEN3 vPort driver support
In the IPU model, a function can host one or more logical network
endpoints called vPorts. Each vPort may be associated with either a
physical or an internal communication port, and can be RDMA capable. A
vPort features a netdev and, if RDMA capable, must have an associated
ib_dev.
This change introduces a GEN3 auxiliary vPort driver responsible for
registering a verbs device for every RDMA-capable vPort. Additionally,
the UAPI is updated to prevent the binding of GEN3 devices to older
user-space providers.
Plug into the unified HW statistics framework by adding a hardware
statistics map array for GEN3, defining the HW-specific width and
location for each counter in the statistics buffer.
Signed-off-by: Krzysztof Czurylo <krzysztof.czurylo@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Link: https://patch.msgid.link/20250827152545.2056-7-tatyana.e.nikolova@intel.com Tested-by: Jacob Moroni <jmoroni@google.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
Shiraz Saleem [Wed, 27 Aug 2025 15:25:34 +0000 (10:25 -0500)]
RDMA/irdma: Add GEN3 support for AEQ and CEQ
Extend support for GEN3 devices by programming the necessary hardware
IRQ registers and the updated descriptor fields for the Asynchronous
Event Queue (AEQ) and Completion Event Queue (CEQ). Introduce a RDMA
virtual channel operation with the Control Plane (CP) to associate
interrupt vectors appropriately with AEQ and CEQ. Add new Asynchronous
Event (AE) definitions specific to GEN3.
Additionally, refactor the AEQ and CEQ setup into the irdma_ctrl_init_hw
device control initialization routine.
This completes the PCI device level initialization for RDMA in the core
driver.
RDMA/irdma: Add GEN3 CQP support with deferred completions
GEN3 introduces asynchronous handling of Control QP (CQP) operations to
minimize head-of-line blocking. Create the CQP using the updated GEN3-
specific descriptor fields and implement the necessary support for this
deferred completion mechanism.
Signed-off-by: Krzysztof Czurylo <krzysztof.czurylo@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Link: https://patch.msgid.link/20250827152545.2056-5-tatyana.e.nikolova@intel.com Tested-by: Jacob Moroni <jmoroni@google.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
RDMA/irdma: Discover and set up GEN3 hardware register layout
Discover the hardware register layout for GEN3 devices through an RDMA
virtual channel operation with the Control Plane (CP). Set up the
corresponding hardware attributes specific to GEN3 devices.
Signed-off-by: Christopher Bednarz <christopher.n.bednarz@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Link: https://patch.msgid.link/20250827152545.2056-4-tatyana.e.nikolova@intel.com Tested-by: Jacob Moroni <jmoroni@google.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
Mustafa Ismail [Wed, 27 Aug 2025 15:25:31 +0000 (10:25 -0500)]
RDMA/irdma: Add GEN3 core driver support
Introduce support for the GEN3 auxiliary core driver, which is
responsible for initializing PCI-level RDMA resources.
Facilitate host-driver communication with the device's Control Plane (CP)
to discover capabilities and perform privileged operations through an
RDMA-specific messaging interface built atop the IDPF mailbox and virtual
channel protocol.
Establish the RDMA virtual channel message interface and incorporate
operations to retrieve the hardware version and discover capabilities
from the CP.
Additionally, set up the RDMA MMIO regions and initialize the RF structure.
Signed-off-by: Mustafa Ismail <mustafa.ismail@intel.com> Co-developed-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Link: https://patch.msgid.link/20250827152545.2056-3-tatyana.e.nikolova@intel.com Tested-by: Jacob Moroni <jmoroni@google.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
Mustafa Ismail [Wed, 27 Aug 2025 15:25:30 +0000 (10:25 -0500)]
RDMA/irdma: Refactor GEN2 auxiliary driver
Refactor the irdma auxiliary driver and associated interfaces out of main.c
and into a standalone GEN2-specific source file and rename as gen_2 driver.
This is in preparation for adding GEN3 auxiliary drivers. Each HW
generation will have its own gen-specific interface file.
Additionally, move the Address Handle hash table and associated locks
under rf struct. This will allow GEN3 code to migrate to use it easily.
Signed-off-by: Mustafa Ismail <mustafa.ismail@intel.com> Co-developed-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Link: https://patch.msgid.link/20250827152545.2056-2-tatyana.e.nikolova@intel.com Tested-by: Jacob Moroni <jmoroni@google.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
RDMA/cm: Rate limit destroy CM ID timeout error message
When the destroy CM ID timeout kicks in, you typically get a storm of
them which creates a log flooding. Hence, change pr_err() to
pr_err_ratelimited() in cm_destroy_id_wait_timeout().
RDMA/bnxt_re: Avoid GID level QoS update from the driver
The driver inserts a VLAN header into RoCE packets when the
traffic was untagged by modifying the existing GID entries.
This has caused the firmware to enforce only VLAN-based
priority mappings, ignoring other valid priority configurations
set via APP TLVs (e.g., DSCP selectors).
Driver now has support for selecting the service level (vlan id)
and traffic class (dscp) during modify_qp. So no need to override
the priority update using the update gid method. Hence removing
the code that handles the above operation.
RDMA/bnxt_re: Update sysfs entries with appropriate data
Updated the existing sysfs entries with correct data.
This change is to align the behavior with our OOB driver.
Added "board_id" sysfs entry which will provide the
VPD Part number, if exists.
net: ionic: Provide doorbell and CMB region information
The RDMA device needs information of controller memory bar and
doorbell capability to share with user context. Discover CMB regions
and express doorbell capabilities on device init.
Reviewed-by: Shannon Nelson <shannon.nelson@amd.com> Co-developed-by: Pablo Cascón <pablo.cascon@amd.com> Signed-off-by: Pablo Cascón <pablo.cascon@amd.com> Co-developed-by: Allen Hubbe <allen.hubbe@amd.com> Signed-off-by: Allen Hubbe <allen.hubbe@amd.com> Signed-off-by: Abhijit Gangurde <abhijit.gangurde@amd.com> Link: https://patch.msgid.link/20250903061606.4139957-7-abhijit.gangurde@amd.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
net: ionic: Update LIF identity with additional RDMA capabilities
Firmware sends the RDMA capability in a response for LIF_IDENTIFY
device command. Update the LIF indentify with additional RDMA
capabilities used by driver and firmware.
net: ionic: Create an auxiliary device for rdma driver
To support RDMA capable ethernet device, create an auxiliary device in
the ionic Ethernet driver. The RDMA device is modeled as an auxiliary
device to the Ethernet device.
RDMA/bnxt_re: Call strscpy() with correct size argument
In bnxt_re_register_ib(), strscpy() is called with the length of the
source string rather than the size of the destination buffer.
This is fine as long as the destination buffer is larger than the source
string, but we should still use the destination buffer size instead to
call strscpy() as intended. And since 'node_desc' has a fixed size, we
can safely omit the size argument and let strscpy() infer it using
sizeof().
Qianfeng Rong [Tue, 26 Aug 2025 15:05:56 +0000 (23:05 +0800)]
RDMA/rdmavt: Use int type to store negative error codes
Change 'ret' from u32 to int in alloc_qpn() to store -EINVAL, and remove
the 'bail' label as it simply returns 'ret'.
Storing negative error codes in an u32 causes no runtime issues, but it's
ugly as pants, Change 'ret' from u32 to int type - this change has no
runtime impact.
Edward Srouji [Sun, 24 Aug 2025 14:48:39 +0000 (17:48 +0300)]
RDMA/mlx5: Fix page size bitmap calculation for KSM mode
When using KSM (Key Scatter-gather Memory) access mode, the HW requires
the IOVA to be aligned to the selected page size.
Without this alignment, the HW may not function correctly.
Currently, mlx5_umem_mkc_find_best_pgsz() does not filter out page sizes
that would result in misaligned IOVAs for KSM mode. This can lead to
selecting page sizes that are incompatible with the given IOVA.
Fix this by filtering the page size bitmap when in KSM mode, keeping
only page sizes to which the IOVA is aligned to.
Fixes: fcfb03597b7d ("RDMA/mlx5: Align mkc page size capability check to PRM") Signed-off-by: Edward Srouji <edwards@nvidia.com> Link: https://patch.msgid.link/20250824144839.154717-1-edwards@nvidia.com Reviewed-by: Michael Guralnik <michaelgur@nvidia.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
RDMA/bnxt_re: Use firmware provided message timeout value
Before this patch, we used a hardcoded value of 500 msec as the default
value for L2 firmware message response timeout. With this commit,
the driver is using the firmware timeout value from the firmware.
As part of this change moved bnxt_re_query_hwrm_intf_version() to
bnxt_re_setup_chip_ctx() so that timeout value is queries before
sending first command.
RDMA/bnxt_re: Initialize fw with roce_mirror support
- Check FW capability for roce_mirror support.
- Initialize FW with roce_mirror support.
- When modifying QP, use unique GID for sgid incase of RawEth QP.
- Added support for create_flow and destroy_flow verbs. These
verbs are used on RawEth QP to add a specific flow action.
- To support TCP dump on RoCE, added IB_FLOW_ATTR_SNIFFER
attribute.
- In create_flow verb, driver allocates mirror_vnic and configure it
with RawEth QP. Once this is done, driver will enable mirroring.
- In destroy_flow, driver will disable mirroring and free the mirror
vnic.
- RawEth QP requires unique GID so that per function stats_ctx
is not polluted by packets mirrored to RoCE vnic.
- Added support to add unique GID when RawEth type QP is created.
- Added support to destroy unique GID when RawEth type QP is
destroyed.
- Allocated exclusive stats_ctx to use for RawEth type QP.
Moved the stats context allocation logic to a new function.
The stats context memory allocation code has been moved from
bnxt_qplib_alloc_hwctx() to the newly added bnxt_re_get_stats_ctx()
function. Also, the code to send the firmware command has been moved.
This patch is in preparation for other patches in this series.
There is no functional changes intended.
This patch is in preparation for other patches in this series.
There is no functional changes intended.
1. Rename bnxt_qplib_alloc_ctx() to bnxt_qplib_alloc_hwctx().
2. Rename bnxt_qplib_free_ctx() to bnxt_qplib_free_hwctx().
3. Reduce the number of arguments of bnxt_qplib_alloc_hwctx()
by moving a check outside of it.
When the firmware advertises that the device is capable of supporting
port mirroring on RoCE device, reserve one additional stat_ctx.
To support port mirroring feature, RDMA driver allocates one stat_ctx
for exclusive use in RawEth QP.
Signed-off-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com> Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com> Reviewed-by: Kashyap Desai <kashyap.desai@broadcom.com> Reviewed-by: Selvin Xavier <selvin.xavier@broadcom.com> Reviewed-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Link: https://patch.msgid.link/20250822040801.776196-2-kalesh-anakkur.purayil@broadcom.com Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Acked-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Leon Romanovsky <leon@kernel.org>
RDMA/bnxt_re: Report udp source port for flow_label in bnxt_re_query_qp
The firmware doesn't capture the flow_label. Therefore the value
that's always returned by qplib_qp->ah.flow_label is 0 whenever
a qp is created. And as per IB spec, udp source port can be reported
for flow_label. Hence reported udp source port for flow_label in
bnxt_re_query_qp by populating the value of qplib_qp->udp_sport
into qp_attr->ah_attr.grh.flow_label.
RDMA/bnxt_re: RoCE related hardware counters update
Support for new hardware counters added, and existing hardware
counters have been modified according to the design documents
for compatibility with open-source monitoring agents.
Signed-off-by: Vasuthevan Maheswaran <vasuthevan.maheswaran@broadcom.com> Signed-off-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com> Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Link: https://patch.msgid.link/20250814112555.221665-7-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
RDMA/bnxt_re: Optimize bnxt_qplib_get_dev_attr function
Optimize bnxt_qplib_get_dev_attr() by separating out query_version which
uses creq notification method to host. Due to serialization of cmdq by
firmware, expected latency in response to heavy multi-threaded rdma
applications might be observed.
This patch separates the version_query logic out of device attribute
query and called only during rdma driver init.
Chenna Arnoori [Thu, 14 Aug 2025 11:25:49 +0000 (16:55 +0530)]
RDMA/bnxt_re: RoCE Driver Dynamic Debug for HWRM's
Add Linux kernel dynamic debug prints to ROCE HWRM's.
Dumping request and response buffers for the ROCE HWRM's using
print_hex_dump_bytes() to be part of kernel dynmic debug.
Qianfeng Rong [Thu, 21 Aug 2025 07:22:09 +0000 (15:22 +0800)]
RDMA/erdma: Use vcalloc() instead of vzalloc()
Replace vzalloc() with vcalloc() in vmalloc_to_dma_addrs(). As noted in
the kernel documentation [1], open-coded multiplication in allocator
arguments is discouraged because it can lead to integer overflow.
Use vcalloc() to gain built-in overflow protection, making memory
allocation safer when calculating allocation size compared to explicit
multiplication.
RDMA/mlx5: Fix vport loopback forcing for MPV device
Previously loopback for MPV was supposed to be permanently enabled,
however other driver flows were able to over-ride that configuration and
disable it.
Add force_lb parameter that indicates that loopback should always be
enabled which prevents all other driver flows from disabling it.
However, this value represents the number of WQEs in units of Basic
Blocks (see MLX5_SEND_WQE_BB), not actual number of WQEs. Since the size
of a WQE can vary depending on transport type and features (e.g., atomic
operations, UMR, LSO), the actual number of WQEs can be significantly
smaller than the WQEBB count suggests.
This patch introduces a conservative estimation of the worst-case WQE size
— considering largest segments possible with 1 SGE and no inline data or
special features. It uses this to derive a more accurate max_qp_wr value.
Fixes: 938fe83c8dcb ("net/mlx5_core: New device capabilities handling") Link: https://patch.msgid.link/r/7d992c9831c997ed5c33d30973406dc2dcaf5e89.1755088725.git.leon@kernel.org Reported-by: Chuck Lever <cel@kernel.org> Closes: https://lore.kernel.org/all/20250506142202.GJ2260621@ziepe.ca/ Signed-off-by: Or Har-Toov <ohartoov@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
In rare servicing scenarios, the hardware may delete the
state of the GSI QP, preventing it from generating CQEs
for pending send WRs. Since WRs submitted to the GSI QP
hold CM resources, the device cannot be removed until
those WRs are completed. This patch marks all pending
send WRs as failed, allowing the GSI QP to release the CM
resources and enabling safe device removal.
Boshi Yu [Fri, 25 Jul 2025 05:53:54 +0000 (13:53 +0800)]
RDMA/erdma: Use dma_map_page to map scatter MTT buffer
Each high-level indirect MTT entry is assumed to point to exactly one page
of the low-level MTT buffer, but dma_map_sg may merge contiguous physical
pages when mapping. To avoid extra overhead from splitting merged regions,
use dma_map_page to map the scatter MTT buffer page by page.
Mark Zhang [Mon, 30 Jun 2025 10:52:35 +0000 (13:52 +0300)]
RDMA/ucma: Support write an event into a CM
Enable user-space to inject an event into a CM through it's event
channel. Two new events are added and supported: RDMA_CM_EVENT_USER and
RDMA_CM_EVENT_INTERNAL. With these 2 events a new event parameter "arg"
is supported, which is passed from sender to receiver transparently.
With this feature an application is able to write an event into a CM
channel with a new user-space rdmacm API. For example thread T1 could
write an event with the API:
rdma_write_cm_event(cm_id, RDMA_CM_EVENT_USER, status, arg);
and thread T2 could receive the event with rdma_get_cm_event().
Mark Zhang [Mon, 30 Jun 2025 10:52:33 +0000 (13:52 +0300)]
RDMA/cma: Support IB service record resolution
Add new UCMA command and the corresponding CMA implementation. Userspace
can send this command to request service resolution based on service
name or ID.
On a successful resolution, one or multiple service records are
returned, the first one will be used as destination address by default.
Two new CM events are added and returned to caller accordingly:
- RDMA_CM_EVENT_ADDRINFO_RESOLVED: Resolve succeeded;
- RDMA_CM_EVENT_ADDRINFO_ERROR: Resolve failed.
Internally two new CM states are added:
- RDMA_CM_ADDRINFO_QUERY: CM is in the process of IB service
resolution;
- RDMA_CM_ADDRINFO_RESOLVED: CM has finished the resolve process.
With these new states, beside existing state transfer processes, 2 new
processes are supported:
1. The default address is used:
RDMA_CM_ADDR_BOUND ->
RDMA_CM_ADDRINFO_QUERY ->
RDMA_CM_ADDRINFO_RESOLVED ->
RDMA_CM_ROUTE_QUERY
2. To use a different address:
RDMA_CM_ADDR_BOUND ->
RDMA_CM_ADDRINFO_QUERY->
RDMA_CM_ADDRINFO_RESOLVED ->
RDMA_CM_ADDR_QUERY ->
RDMA_CM_ADDR_RESOLVED ->
RDMA_CM_ROUTE_QUERY
In the 2nd case, resolve_addrinfo returns multiple records, a user
could call rdma_resolve_addr() with the one that is not the first.
Mark Zhang [Mon, 30 Jun 2025 10:52:32 +0000 (13:52 +0300)]
RDMA/sa_query: Support IB service records resolution
Add an SA query API ib_sa_service_rec_get() to support building and
sending SA query MADs that ask for service records with a specific
name or ID, and receiving and parsing responses from the SM.
Linus Torvalds [Sun, 10 Aug 2025 05:51:37 +0000 (08:51 +0300)]
Merge tag 'smp_urgent_for_v6.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull smp fixes from Borislav Petkov:
- Remove an obsolete comment and fix spelling
* tag 'smp_urgent_for_v6.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
cpu: Remove obsolete comment from takedown_cpu()
smp: Fix spelling in on_each_cpu_cond_mask()'s doc-comment
Linus Torvalds [Sun, 10 Aug 2025 05:46:47 +0000 (08:46 +0300)]
Merge tag 'irq_urgent_for_v6.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull irq fixes from Borislav Petkov:
- Fix a wrong ioremap size in mvebu-gicp
- Remove yet another compile-test case for a driver which needs an
additional dependency
- Fix a lock inversion scenario in the IRQ unit test suite
- Remove an impossible flag situation in gic-v5
- Do not iounmap resources in gic-v5 which are managed by devm
- Make sure stale, left-over interrupts in mvebu-gicp are cleared on
driver init
- Fix a reference counting mishap in msi-lib
- Fix a dereference-before-null-ptr-check case in the riscv-imsic
irqchip driver
* tag 'irq_urgent_for_v6.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
irqchip/mvebu-gicp: Use resource_size() for ioremap()
irqchip: Build IMX_MU_MSI only on ARM
genirq/test: Resolve irq lock inversion warnings
irqchip/gic-v5: Remove IRQD_RESEND_WHEN_IN_PROGRESS for ITS IRQs
irqchip/gic-v5: iwb: Fix iounmap probe failure path
irqchip/mvebu-gicp: Clear pending interrupts on init
irqchip/msi-lib: Fix fwnode refcount in msi_lib_irq_domain_select()
irqchip/riscv-imsic: Don't dereference before NULL pointer check
Linus Torvalds [Sun, 10 Aug 2025 05:15:32 +0000 (08:15 +0300)]
Merge tag 'x86_urgent_for_v6.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Borislav Petkov:
- Fix an interrupt vector setup race which leads to a non-functioning
device
- Add new Intel CPU models *and* a family: 0x12. Finally. Yippie! :-)
* tag 'x86_urgent_for_v6.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/irq: Plug vector setup race
x86/cpu: Add new Intel CPU model numbers for Wildcatlake and Novalake
Len Brown [Sun, 10 Aug 2025 01:08:26 +0000 (21:08 -0400)]
tools/power turbostat: version 2025.09.09
Probe and display L3 Cache topology
Add ability to average an added counter
(useful for pre-integrated "counters", such as Watts)
Break the limit of 64 built-in counters.
Assorted bug fixes and minor feature tweaks
/sys/devices/system/cpu/intel_uncore_frequency/package_X_die_Y/
may be readable by all, but
/sys/devices/system/cpu/intel_uncore_frequency/package_X_die_Y/current_freq_khz
may be readable only by root.
Non-root turbostat users see complaints in this scenario.
Fail probe of the interface if we can't read current_freq_khz.
Reported-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Original-patch-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
Zhang Rui [Wed, 11 Jun 2025 06:50:26 +0000 (14:50 +0800)]
tools/power turbostat: Fix DMR support
Together with the RAPL MSRs, there are more MSRs gone on DMR, including
PLR (Perf Limit Reasons), and IRTL (Package cstate Interrupt Response
Time Limit) MSRs. The configurable TDP info should also be retrieved
from TPMI based Intel Speed Select Technology feature.
Remove the access of these MSRs for DMR. Improve the DMR platform
feature table to make it more readable at the same time.
Fixes: 83075bd59de2 ("tools/power turbostat: Add initial support for DMR") Signed-off-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
tools/power turbostat: add format "average" for external attributes
External atributes with format "raw" are not printed in summary lines
for nodes/packages (or with option -S). The new format "average"
behaves like "raw" but also adds the summary data
Signed-off-by: Michael Hebenstreit <michael.hebenstreit@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
Len Brown [Sat, 12 Jul 2025 20:16:56 +0000 (16:16 -0400)]
tools/power turbostat: Support more than 64 built-in-counters
We have out-grown the ability to use a 64-bit memory location
to inventory every possible built-in counter.
Leverage the the CPU_SET(3) macros to break this barrier.
Also, break the Joules & Watts counters into two,
since we can no longer 'or' them together...
projects / thirdparty / kernel / stable.git / log
