Merge tag 'caps-pr-20251204' of git://git.kernel.org/pub/scm/linux/kernel/git/sergeh/linux

Pull capabilities update from Serge Hallyn:
 "Ryan Foster had sent a patch to add testing of the
  rootid_owns_currentns() function. That patch pointed out
  that this function was not as clear as it should be. Fix it:

   - Clarify the intent of the function in the name

   - Split the function so that the base functionality is easier to test
     from a kunit test"

* tag 'caps-pr-20251204' of git://git.kernel.org/pub/scm/linux/kernel/git/sergeh/linux:
  Clarify the rootid_owns_currentns

.clang-format

@@ -415,6 +415,7 @@ ForEachMacros:
   - 'for_each_prop_dlc_cpus'
   - 'for_each_prop_dlc_platforms'
   - 'for_each_property_of_node'
+  - 'for_each_pt_level_entry'
   - 'for_each_rdt_resource'
   - 'for_each_reg'
   - 'for_each_reg_filtered'

.mailmap

@@ -345,7 +345,8 @@ Jayachandran C <c.jayachandran@gmail.com> <jayachandranc@netlogicmicro.com>
 Jayachandran C <c.jayachandran@gmail.com> <jchandra@broadcom.com>
 Jayachandran C <c.jayachandran@gmail.com> <jchandra@digeo.com>
 Jayachandran C <c.jayachandran@gmail.com> <jnair@caviumnetworks.com>
-<jean-philippe@linaro.org> <jean-philippe.brucker@arm.com>
+Jean-Philippe Brucker <jpb@kernel.org> <jean-philippe.brucker@arm.com>
+Jean-Philippe Brucker <jpb@kernel.org> <jean-philippe@linaro.org>
 Jean-Michel Hautbois <jeanmichel.hautbois@yoseli.org> <jeanmichel.hautbois@ideasonboard.com>
 Jean Tourrilhes <jt@hpl.hp.com>
 Jeevan Shriram <quic_jshriram@quicinc.com> <jshriram@codeaurora.org>

Documentation/ABI/testing/sysfs-bus-cxl

@@ -496,8 +496,17 @@ Description:
 		changed, only freed by writing 0. The kernel makes no guarantees
 		that data is maintained over an address space freeing event, and
 		there is no guarantee that a free followed by an allocate
-		results in the same address being allocated.
+		results in the same address being allocated. If extended linear
+		cache is present, the size indicates extended linear cache size
+		plus the CXL region size.
 
+What:		/sys/bus/cxl/devices/regionZ/extended_linear_cache_size
+Date:		October, 2025
+KernelVersion:	v6.19
+Contact:	linux-cxl@vger.kernel.org
+Description:
+		(RO) The size of extended linear cache, if there is an extended
+		linear cache. Otherwise the attribute will not be visible.
 
 What:		/sys/bus/cxl/devices/regionZ/mode
 Date:		January, 2023

Documentation/devicetree/bindings/ata/eswin,eic7700-ahci.yaml

@@ -0,0 +1,79 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/ata/eswin,eic7700-ahci.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Eswin EIC7700 SoC SATA Controller
+
+maintainers:
+  - Yulin Lu <luyulin@eswincomputing.com>
+  - Huan He <hehuan1@eswincomputing.com>
+
+description:
+  AHCI SATA controller embedded into the EIC7700 SoC is based on the DWC AHCI
+  SATA v5.00a IP core.
+
+select:
+  properties:
+    compatible:
+      const: eswin,eic7700-ahci
+  required:
+    - compatible
+
+allOf:
+  - $ref: snps,dwc-ahci-common.yaml#
+
+properties:
+  compatible:
+    items:
+      - const: eswin,eic7700-ahci
+      - const: snps,dwc-ahci
+
+  clocks:
+    minItems: 2
+    maxItems: 2
+
+  clock-names:
+    items:
+      - const: pclk
+      - const: aclk
+
+  resets:
+    maxItems: 1
+
+  reset-names:
+    const: arst
+
+  ports-implemented:
+    const: 1
+
+required:
+  - compatible
+  - reg
+  - interrupts
+  - clocks
+  - clock-names
+  - resets
+  - reset-names
+  - phys
+  - phy-names
+  - ports-implemented
+
+unevaluatedProperties: false
+
+examples:
+  - |
+    sata@50420000 {
+        compatible = "eswin,eic7700-ahci", "snps,dwc-ahci";
+        reg = <0x50420000 0x10000>;
+        interrupt-parent = <&plic>;
+        interrupts = <58>;
+        clocks = <&clock 171>, <&clock 186>;
+        clock-names = "pclk", "aclk";
+        phys = <&sata_phy>;
+        phy-names = "sata-phy";
+        ports-implemented = <0x1>;
+        resets = <&reset 96>;
+        reset-names = "arst";
+    };

Documentation/devicetree/bindings/ata/snps,dwc-ahci.yaml

@@ -33,6 +33,10 @@ properties:
       - description: SPEAr1340 AHCI SATA device
         const: snps,spear-ahci
 
+  iommus:
+    minItems: 1
+    maxItems: 3
+
 patternProperties:
   "^sata-port@[0-9a-e]$":
     $ref: /schemas/ata/snps,dwc-ahci-common.yaml#/$defs/dwc-ahci-port

Documentation/devicetree/bindings/iommu/arm,smmu.yaml

@@ -35,6 +35,8 @@ properties:
       - description: Qcom SoCs implementing "qcom,smmu-500" and "arm,mmu-500"
         items:
           - enum:
+              - qcom,glymur-smmu-500
+              - qcom,kaanapali-smmu-500
               - qcom,milos-smmu-500
               - qcom,qcm2290-smmu-500
               - qcom,qcs615-smmu-500

Documentation/devicetree/bindings/iommu/mediatek,iommu.yaml

@@ -82,6 +82,9 @@ properties:
           - mediatek,mt8188-iommu-vdo        # generation two
           - mediatek,mt8188-iommu-vpp        # generation two
           - mediatek,mt8188-iommu-infra      # generation two
+          - mediatek,mt8189-iommu-apu        # generation two
+          - mediatek,mt8189-iommu-infra      # generation two
+          - mediatek,mt8189-iommu-mm         # generation two
           - mediatek,mt8192-m4u  # generation two
           - mediatek,mt8195-iommu-vdo        # generation two
           - mediatek,mt8195-iommu-vpp        # generation two
@@ -128,6 +131,7 @@ properties:
       This is the mtk_m4u_id according to the HW. Specifies the mtk_m4u_id as
       defined in
       dt-binding/memory/mediatek,mt8188-memory-port.h for mt8188,
+      dt-binding/memory/mediatek,mt8189-memory-port.h for mt8189,
       dt-binding/memory/mt2701-larb-port.h for mt2701 and mt7623,
       dt-binding/memory/mt2712-larb-port.h for mt2712,
       dt-binding/memory/mt6779-larb-port.h for mt6779,
@@ -164,6 +168,7 @@ allOf:
               - mediatek,mt8186-iommu-mm
               - mediatek,mt8188-iommu-vdo
               - mediatek,mt8188-iommu-vpp
+              - mediatek,mt8189-iommu-mm
               - mediatek,mt8192-m4u
               - mediatek,mt8195-iommu-vdo
               - mediatek,mt8195-iommu-vpp
@@ -180,6 +185,7 @@ allOf:
             - mediatek,mt8186-iommu-mm
             - mediatek,mt8188-iommu-vdo
             - mediatek,mt8188-iommu-vpp
+            - mediatek,mt8189-iommu-mm
             - mediatek,mt8192-m4u
             - mediatek,mt8195-iommu-vdo
             - mediatek,mt8195-iommu-vpp
@@ -208,6 +214,8 @@ allOf:
             contains:
               enum:
                 - mediatek,mt8188-iommu-infra
+                - mediatek,mt8189-iommu-apu
+                - mediatek,mt8189-iommu-infra
                 - mediatek,mt8195-iommu-infra
 
     then:

Documentation/devicetree/bindings/iommu/qcom,iommu.yaml

@@ -32,14 +32,18 @@ properties:
           - const: qcom,msm-iommu-v2
 
   clocks:
+    minItems: 2
     items:
       - description: Clock required for IOMMU register group access
       - description: Clock required for underlying bus access
+      - description: Clock required for Translation Buffer Unit access
 
   clock-names:
+    minItems: 2
     items:
       - const: iface
       - const: bus
+      - const: tbu
 
   power-domains:
     maxItems: 1

Documentation/driver-api/cxl/allocation/page-allocator.rst

@@ -41,37 +41,6 @@ To simplify this, the page allocator will prefer :code:`ZONE_MOVABLE` over
 will fallback to allocate from :code:`ZONE_NORMAL`.
 
 
-Zone and Node Quirks
-====================
-Let's consider a configuration where the local DRAM capacity is largely onlined
-into :code:`ZONE_NORMAL`, with no :code:`ZONE_MOVABLE` capacity present. The
-CXL capacity has the opposite configuration - all onlined in
-:code:`ZONE_MOVABLE`.
-
-Under the default allocation policy, the page allocator will completely skip
-:code:`ZONE_MOVABLE` as a valid allocation target.  This is because, as of
-Linux v6.15, the page allocator does (approximately) the following: ::
-
-  for (each zone in local_node):
-
-    for (each node in fallback_order):
-
-      attempt_allocation(gfp_flags);
-
-Because the local node does not have :code:`ZONE_MOVABLE`, the CXL node is
-functionally unreachable for direct allocation.  As a result, the only way
-for CXL capacity to be used is via `demotion` in the reclaim path.
-
-This configuration also means that if the DRAM ndoe has :code:`ZONE_MOVABLE`
-capacity - when that capacity is depleted, the page allocator will actually
-prefer CXL :code:`ZONE_MOVABLE` pages over DRAM :code:`ZONE_NORMAL` pages.
-
-We may wish to invert this priority in future Linux versions.
-
-If `demotion` and `swap` are disabled, Linux will begin to cause OOM crashes
-when the DRAM nodes are depleted. See the reclaim section for more details.
-
-
 CGroups and CPUSets
 ===================
 Finally, assuming CXL memory is reachable via the page allocation (i.e. onlined

Documentation/driver-api/generic_pt.rst

@@ -0,0 +1,137 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+========================
+Generic Radix Page Table
+========================
+
+.. kernel-doc:: include/linux/generic_pt/common.h
+	:doc: Generic Radix Page Table
+
+.. kernel-doc:: drivers/iommu/generic_pt/pt_defs.h
+	:doc: Generic Page Table Language
+
+Usage
+=====
+
+Generic PT is structured as a multi-compilation system. Since each format
+provides an API using a common set of names there can be only one format active
+within a compilation unit. This design avoids function pointers around the low
+level API.
+
+Instead the function pointers can end up at the higher level API (i.e.
+map/unmap, etc.) and the per-format code can be directly inlined into the
+per-format compilation unit. For something like IOMMU each format will be
+compiled into a per-format IOMMU operations kernel module.
+
+For this to work the .c file for each compilation unit will include both the
+format headers and the generic code for the implementation. For instance in an
+implementation compilation unit the headers would normally be included as
+follows:
+
+generic_pt/fmt/iommu_amdv1.c::
+
+	#include <linux/generic_pt/common.h>
+	#include "defs_amdv1.h"
+	#include "../pt_defs.h"
+	#include "amdv1.h"
+	#include "../pt_common.h"
+	#include "../pt_iter.h"
+	#include "../iommu_pt.h"  /* The IOMMU implementation */
+
+iommu_pt.h includes definitions that will generate the operations functions for
+map/unmap/etc. using the definitions provided by AMDv1. The resulting module
+will have exported symbols named like pt_iommu_amdv1_init().
+
+Refer to drivers/iommu/generic_pt/fmt/iommu_template.h for an example of how the
+IOMMU implementation uses multi-compilation to generate per-format ops structs
+pointers.
+
+The format code is written so that the common names arise from #defines to
+distinct format specific names. This is intended to aid debuggability by
+avoiding symbol clashes across all the different formats.
+
+Exported symbols and other global names are mangled using a per-format string
+via the NS() helper macro.
+
+The format uses struct pt_common as the top-level struct for the table,
+and each format will have its own struct pt_xxx which embeds it to store
+format-specific information.
+
+The implementation will further wrap struct pt_common in its own top-level
+struct, such as struct pt_iommu_amdv1.
+
+Format functions at the struct pt_common level
+----------------------------------------------
+
+.. kernel-doc:: include/linux/generic_pt/common.h
+	:identifiers:
+.. kernel-doc:: drivers/iommu/generic_pt/pt_common.h
+
+Iteration Helpers
+-----------------
+
+.. kernel-doc:: drivers/iommu/generic_pt/pt_iter.h
+
+Writing a Format
+----------------
+
+It is best to start from a simple format that is similar to the target. x86_64
+is usually a good reference for something simple, and AMDv1 is something fairly
+complete.
+
+The required inline functions need to be implemented in the format header.
+These should all follow the standard pattern of::
+
+ static inline pt_oaddr_t amdv1pt_entry_oa(const struct pt_state *pts)
+ {
+	[..]
+ }
+ #define pt_entry_oa amdv1pt_entry_oa
+
+where a uniquely named per-format inline function provides the implementation
+and a define maps it to the generic name. This is intended to make debug symbols
+work better. inline functions should always be used as the prototypes in
+pt_common.h will cause the compiler to validate the function signature to
+prevent errors.
+
+Review pt_fmt_defaults.h to understand some of the optional inlines.
+
+Once the format compiles then it should be run through the generic page table
+kunit test in kunit_generic_pt.h using kunit. For example::
+
+   $ tools/testing/kunit/kunit.py run --build_dir build_kunit_x86_64 --arch x86_64 --kunitconfig ./drivers/iommu/generic_pt/.kunitconfig amdv1_fmt_test.*
+   [...]
+   [11:15:08] Testing complete. Ran 9 tests: passed: 9
+   [11:15:09] Elapsed time: 3.137s total, 0.001s configuring, 2.368s building, 0.311s running
+
+The generic tests are intended to prove out the format functions and give
+clearer failures to speed up finding the problems. Once those pass then the
+entire kunit suite should be run.
+
+IOMMU Invalidation Features
+---------------------------
+
+Invalidation is how the page table algorithms synchronize with a HW cache of the
+page table memory, typically called the TLB (or IOTLB for IOMMU cases).
+
+The TLB can store present PTEs, non-present PTEs and table pointers, depending
+on its design. Every HW has its own approach on how to describe what has changed
+to have changed items removed from the TLB.
+
+PT_FEAT_FLUSH_RANGE
+~~~~~~~~~~~~~~~~~~~
+
+PT_FEAT_FLUSH_RANGE is the easiest scheme to understand. It tries to generate a
+single range invalidation for each operation, over-invalidating if there are
+gaps of VA that don't need invalidation. This trades off impacted VA for number
+of invalidation operations. It does not keep track of what is being invalidated;
+however, if pages have to be freed then page table pointers have to be cleaned
+from the walk cache. The range can start/end at any page boundary.
+
+PT_FEAT_FLUSH_RANGE_NO_GAPS
+~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+PT_FEAT_FLUSH_RANGE_NO_GAPS is similar to PT_FEAT_FLUSH_RANGE; however, it tries
+to minimize the amount of impacted VA by issuing extra flush operations. This is
+useful if the cost of processing VA is very high, for instance because a
+hypervisor is processing the page table with a shadowing algorithm.

Documentation/driver-api/index.rst

@@ -93,6 +93,7 @@ Subsystem-specific APIs
    frame-buffer
    aperture
    generic-counter
+   generic_pt
    gpio/index
    hsi
    hte/index

Documentation/driver-api/pci/p2pdma.rst

@@ -9,22 +9,48 @@ between two devices on the bus. This type of transaction is henceforth
 called Peer-to-Peer (or P2P). However, there are a number of issues that
 make P2P transactions tricky to do in a perfectly safe way.
 
-One of the biggest issues is that PCI doesn't require forwarding
-transactions between hierarchy domains, and in PCIe, each Root Port
-defines a separate hierarchy domain. To make things worse, there is no
-simple way to determine if a given Root Complex supports this or not.
-(See PCIe r4.0, sec 1.3.1). Therefore, as of this writing, the kernel
-only supports doing P2P when the endpoints involved are all behind the
-same PCI bridge, as such devices are all in the same PCI hierarchy
-domain, and the spec guarantees that all transactions within the
-hierarchy will be routable, but it does not require routing
-between hierarchies.
+For PCIe the routing of Transaction Layer Packets (TLPs) is well-defined up
+until they reach a host bridge or root port. If the path includes PCIe switches
+then based on the ACS settings the transaction can route entirely within
+the PCIe hierarchy and never reach the root port. The kernel will evaluate
+the PCIe topology and always permit P2P in these well-defined cases.
 
-The second issue is that to make use of existing interfaces in Linux,
-memory that is used for P2P transactions needs to be backed by struct
-pages. However, PCI BARs are not typically cache coherent so there are
-a few corner case gotchas with these pages so developers need to
-be careful about what they do with them.
+However, if the P2P transaction reaches the host bridge then it might have to
+hairpin back out the same root port, be routed inside the CPU SOC to another
+PCIe root port, or routed internally to the SOC.
+
+The PCIe specification doesn't define the forwarding of transactions between
+hierarchy domains and kernel defaults to blocking such routing. There is an
+allow list to allow detecting known-good HW, in which case P2P between any
+two PCIe devices will be permitted.
+
+Since P2P inherently is doing transactions between two devices it requires two
+drivers to be co-operating inside the kernel. The providing driver has to convey
+its MMIO to the consuming driver. To meet the driver model lifecycle rules the
+MMIO must have all DMA mapping removed, all CPU accesses prevented, all page
+table mappings undone before the providing driver completes remove().
+
+This requires the providing and consuming driver to actively work together to
+guarantee that the consuming driver has stopped using the MMIO during a removal
+cycle. This is done by either a synchronous invalidation shutdown or waiting
+for all usage refcounts to reach zero.
+
+At the lowest level the P2P subsystem offers a naked struct p2p_provider that
+delegates lifecycle management to the providing driver. It is expected that
+drivers using this option will wrap their MMIO memory in DMABUF and use DMABUF
+to provide an invalidation shutdown. These MMIO addresess have no struct page, and
+if used with mmap() must create special PTEs. As such there are very few
+kernel uAPIs that can accept pointers to them; in particular they cannot be used
+with read()/write(), including O_DIRECT.
+
+Building on this, the subsystem offers a layer to wrap the MMIO in a ZONE_DEVICE
+pgmap of MEMORY_DEVICE_PCI_P2PDMA to create struct pages. The lifecycle of
+pgmap ensures that when the pgmap is destroyed all other drivers have stopped
+using the MMIO. This option works with O_DIRECT flows, in some cases, if the
+underlying subsystem supports handling MEMORY_DEVICE_PCI_P2PDMA through
+FOLL_PCI_P2PDMA. The use of FOLL_LONGTERM is prevented. As this relies on pgmap
+it also relies on architecture support along with alignment and minimum size
+limitations.
 
 
 Driver Writer's Guide
@@ -114,14 +140,39 @@ allocating scatter-gather lists with P2P memory.
 Struct Page Caveats
 -------------------
 
-Driver writers should be very careful about not passing these special
-struct pages to code that isn't prepared for it. At this time, the kernel
-interfaces do not have any checks for ensuring this. This obviously
-precludes passing these pages to userspace.
+While the MEMORY_DEVICE_PCI_P2PDMA pages can be installed in VMAs,
+pin_user_pages() and related will not return them unless FOLL_PCI_P2PDMA is set.
 
-P2P memory is also technically IO memory but should never have any side
-effects behind it. Thus, the order of loads and stores should not be important
-and ioreadX(), iowriteX() and friends should not be necessary.
+The MEMORY_DEVICE_PCI_P2PDMA pages require care to support in the kernel. The
+KVA is still MMIO and must still be accessed through the normal
+readX()/writeX()/etc helpers. Direct CPU access (e.g. memcpy) is forbidden, just
+like any other MMIO mapping. While this will actually work on some
+architectures, others will experience corruption or just crash in the kernel.
+Supporting FOLL_PCI_P2PDMA in a subsystem requires scrubbing it to ensure no CPU
+access happens.
+
+
+Usage With DMABUF
+=================
+
+DMABUF provides an alternative to the above struct page-based
+client/provider/orchestrator system and should be used when struct page
+doesn't exist. In this mode the exporting driver will wrap
+some of its MMIO in a DMABUF and give the DMABUF FD to userspace.
+
+Userspace can then pass the FD to an importing driver which will ask the
+exporting driver to map it to the importer.
+
+In this case the initiator and target pci_devices are known and the P2P subsystem
+is used to determine the mapping type. The phys_addr_t-based DMA API is used to
+establish the dma_addr_t.
+
+Lifecycle is controlled by DMABUF move_notify(). When the exporting driver wants
+to remove() it must deliver an invalidation shutdown to all DMABUF importing
+drivers through move_notify() and synchronously DMA unmap all the MMIO.
+
+No importing driver can continue to have a DMA map to the MMIO after the
+exporting driver has destroyed its p2p_provider.
 
 
 P2P DMA Support Library

MAINTAINERS

@@ -388,7 +388,7 @@ B:	https://bugzilla.kernel.org
 F:	drivers/acpi/*thermal*
 
 ACPI VIOT DRIVER
-M:	Jean-Philippe Brucker <jean-philippe@linaro.org>
+M:	Jean-Philippe Brucker <jpb@kernel.org>
 L:	linux-acpi@vger.kernel.org
 L:	iommu@lists.linux.dev
 S:	Maintained
@@ -2269,7 +2269,7 @@ F:	drivers/iommu/arm/
 F:	drivers/iommu/io-pgtable-arm*
 
 ARM SMMU SVA SUPPORT
-R:	Jean-Philippe Brucker <jean-philippe@linaro.org>
+R:	Jean-Philippe Brucker <jpb@kernel.org>
 F:	drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c
 
 ARM SUB-ARCHITECTURES
@@ -5243,6 +5243,13 @@ W:	http://www.broadcom.com
 F:	drivers/infiniband/hw/bnxt_re/
 F:	include/uapi/rdma/bnxt_re-abi.h
 
+BROADCOM 800 GIGABIT ROCE DRIVER
+M:	Siva Reddy Kallam <siva.kallam@broadcom.com>
+L:	linux-rdma@vger.kernel.org
+S:	Supported
+W:	http://www.broadcom.com
+F:	drivers/infiniband/hw/bng_re/
+
 BROADCOM NVRAM DRIVER
 M:	Rafał Miłecki <zajec5@gmail.com>
 L:	linux-mips@vger.kernel.org
@@ -27214,6 +27221,13 @@ L:	virtualization@lists.linux.dev
 S:	Maintained
 F:	drivers/vfio/pci/virtio
 
+VFIO XE PCI DRIVER
+M:	Michał Winiarski <michal.winiarski@intel.com>
+L:	kvm@vger.kernel.org
+L:	intel-xe@lists.freedesktop.org
+S:	Supported
+F:	drivers/vfio/pci/xe
+
 VGA_SWITCHEROO
 R:	Lukas Wunner <lukas@wunner.de>
 S:	Maintained
@@ -27455,7 +27469,7 @@ F:	drivers/virtio/virtio_input.c
 F:	include/uapi/linux/virtio_input.h
 
 VIRTIO IOMMU DRIVER
-M:	Jean-Philippe Brucker <jean-philippe@linaro.org>
+M:	Jean-Philippe Brucker <jpb@kernel.org>
 L:	virtualization@lists.linux.dev
 S:	Maintained
 F:	drivers/iommu/virtio-iommu.c

arch/powerpc/include/asm/mem_encrypt.h

@@ -9,6 +9,9 @@
 #define _ASM_POWERPC_MEM_ENCRYPT_H
 
 #include <asm/svm.h>
+#include <linux/types.h>
+
+struct device;
 
 static inline bool force_dma_unencrypted(struct device *dev)
 {

arch/powerpc/kernel/iommu.c

@@ -1156,7 +1156,8 @@ EXPORT_SYMBOL_GPL(iommu_add_device);
  */
 static int
 spapr_tce_platform_iommu_attach_dev(struct iommu_domain *platform_domain,
-				    struct device *dev)
+				    struct device *dev,
+				    struct iommu_domain *old)
 {
 	struct iommu_domain *domain = iommu_get_domain_for_dev(dev);
 	struct iommu_table_group *table_group;
@@ -1189,7 +1190,7 @@ static struct iommu_domain spapr_tce_platform_domain = {
 
 static int
 spapr_tce_blocked_iommu_attach_dev(struct iommu_domain *platform_domain,
-				     struct device *dev)
+				   struct device *dev, struct iommu_domain *old)
 {
 	struct iommu_group *grp = iommu_group_get(dev);
 	struct iommu_table_group *table_group;

block/blk-mq-dma.c

@@ -84,7 +84,7 @@ static inline bool blk_can_dma_map_iova(struct request *req,
 
 static bool blk_dma_map_bus(struct blk_dma_iter *iter, struct phys_vec *vec)
 {
-	iter->addr = pci_p2pdma_bus_addr_map(&iter->p2pdma, vec->paddr);
+	iter->addr = pci_p2pdma_bus_addr_map(iter->p2pdma.mem, vec->paddr);
 	iter->len = vec->len;
 	return true;
 }

drivers/acpi/numa/hmat.c

@@ -910,12 +910,13 @@ static void hmat_register_target(struct memory_target *target)
 	 * Register generic port perf numbers. The nid may not be
 	 * initialized and is still NUMA_NO_NODE.
 	 */
-	mutex_lock(&target_lock);
-	if (*(u16 *)target->gen_port_device_handle) {
-		hmat_update_generic_target(target);
-		target->registered = true;
+	scoped_guard(mutex, &target_lock) {
+		if (*(u16 *)target->gen_port_device_handle) {
+			hmat_update_generic_target(target);
+			target->registered = true;
+			return;
+		}
 	}
-	mutex_unlock(&target_lock);
 
 	hmat_hotplug_target(target);
 }

drivers/amba/Kconfig

@@ -5,7 +5,7 @@ config ARM_AMBA
 if ARM_AMBA
 
 config TEGRA_AHB
-	bool
+	bool "Enable AHB driver for NVIDIA Tegra SoCs" if COMPILE_TEST
 	default y if ARCH_TEGRA
 	help
 	  Adds AHB configuration functionality for NVIDIA Tegra SoCs,

drivers/ata/libata-core.c

@@ -4216,6 +4216,10 @@ static const struct ata_dev_quirks_entry __ata_dev_quirks[] = {
 	/* Apacer models with LPM issues */
 	{ "Apacer AS340*",		NULL,	ATA_QUIRK_NOLPM },
 
+	/* Silicon Motion models with LPM issues */
+	{ "MD619HXCLDE3TC",		"TCVAID", ATA_QUIRK_NOLPM },
+	{ "MD619GXCLDE3TC",		"TCV35D", ATA_QUIRK_NOLPM },
+
 	/* These specific Samsung models/firmware-revs do not handle LPM well */
 	{ "SAMSUNG MZMPC128HBFU-000MV", "CXM14M1Q", ATA_QUIRK_NOLPM },
 	{ "SAMSUNG SSD PM830 mSATA *",  "CXM13D1Q", ATA_QUIRK_NOLPM },

drivers/ata/libata-sff.c

@@ -3191,7 +3191,8 @@ void ata_sff_port_init(struct ata_port *ap)
 
 int __init ata_sff_init(void)
 {
-	ata_sff_wq = alloc_workqueue("ata_sff", WQ_MEM_RECLAIM, WQ_MAX_ACTIVE);
+	ata_sff_wq = alloc_workqueue("ata_sff", WQ_MEM_RECLAIM | WQ_PERCPU,
+				     WQ_MAX_ACTIVE);
 	if (!ata_sff_wq)
 		return -ENOMEM;
 

drivers/ata/pata_it821x.c

@@ -75,6 +75,7 @@
 #include <linux/blkdev.h>
 #include <linux/delay.h>
 #include <linux/slab.h>
+#include <linux/string.h>
 #include <scsi/scsi_host.h>
 #include <linux/libata.h>
 
@@ -632,9 +633,9 @@ static void it821x_display_disk(struct ata_port *ap, int n, u8 *buf)
 		cbl = "";
 
 	if (mode)
-		snprintf(mbuf, 8, "%5s%d", mtype, mode - 1);
+		snprintf(mbuf, sizeof(mbuf), "%5s%d", mtype, mode - 1);
 	else
-		strcpy(mbuf, "PIO");
+		strscpy(mbuf, "PIO");
 	if (buf[52] == 4)
 		ata_port_info(ap, "%d: %-6s %-8s          %s %s\n",
 				n, mbuf, types[buf[52]], id, cbl);

drivers/ata/pata_pcmcia.c

@@ -344,6 +344,7 @@ static const struct pcmcia_device_id pcmcia_devices[] = {
 	PCMCIA_DEVICE_PROD_ID2("NinjaATA-", 0xebe0bd79),
 	PCMCIA_DEVICE_PROD_ID12("PCMCIA", "CD-ROM", 0x281f1c5d, 0x66536591),
 	PCMCIA_DEVICE_PROD_ID12("PCMCIA", "PnPIDE", 0x281f1c5d, 0x0c694728),
+	PCMCIA_DEVICE_PROD_ID2("PCMCIA ATA/ATAPI Adapter", 0x888d7b73),
 	PCMCIA_DEVICE_PROD_ID12("SHUTTLE TECHNOLOGY LTD.", "PCCARD-IDE/ATAPI Adapter", 0x4a3f0ba0, 0x322560e1),
 	PCMCIA_DEVICE_PROD_ID12("SEAGATE", "ST1", 0x87c1b330, 0xe1f30883),
 	PCMCIA_DEVICE_PROD_ID12("SAMSUNG", "04/05/06", 0x43d74cb4, 0x6a22777d),

drivers/char/tpm/tpm-chip.c

@@ -230,42 +230,6 @@ struct tpm_chip *tpm_default_chip(void)
 }
 EXPORT_SYMBOL_GPL(tpm_default_chip);
 
-/**
- * tpm_find_get_ops() - find and reserve a TPM chip
- * @chip:	a &struct tpm_chip instance, %NULL for the default chip
- *
- * Finds a TPM chip and reserves its class device and operations. The chip must
- * be released with tpm_put_ops() after use.
- * This function is for internal use only. It supports existing TPM callers
- * by accepting NULL, but those callers should be converted to pass in a chip
- * directly.
- *
- * Return:
- * A reserved &struct tpm_chip instance.
- * %NULL if a chip is not found.
- * %NULL if the chip is not available.
- */
-struct tpm_chip *tpm_find_get_ops(struct tpm_chip *chip)
-{
-	int rc;
-
-	if (chip) {
-		if (!tpm_try_get_ops(chip))
-			return chip;
-		return NULL;
-	}
-
-	chip = tpm_default_chip();
-	if (!chip)
-		return NULL;
-	rc = tpm_try_get_ops(chip);
-	/* release additional reference we got from tpm_default_chip() */
-	put_device(&chip->dev);
-	if (rc)
-		return NULL;
-	return chip;
-}
-
 /**
  * tpm_dev_release() - free chip memory and the device number
  * @dev: the character device for the TPM chip
@@ -282,7 +246,6 @@ static void tpm_dev_release(struct device *dev)
 
 	kfree(chip->work_space.context_buf);
 	kfree(chip->work_space.session_buf);
-	kfree(chip->allocated_banks);
 #ifdef CONFIG_TCG_TPM2_HMAC
 	kfree(chip->auth);
 #endif

drivers/char/tpm/tpm-dev-common.c

@@ -275,7 +275,8 @@ void tpm_common_release(struct file *file, struct file_priv *priv)
 
 int __init tpm_dev_common_init(void)
 {
-	tpm_dev_wq = alloc_workqueue("tpm_dev_wq", WQ_MEM_RECLAIM, 0);
+	tpm_dev_wq = alloc_workqueue("tpm_dev_wq", WQ_MEM_RECLAIM | WQ_PERCPU,
+				     0);
 
 	return !tpm_dev_wq ? -ENOMEM : 0;
 }

drivers/char/tpm/tpm-interface.c

@@ -313,10 +313,13 @@ int tpm_is_tpm2(struct tpm_chip *chip)
 {
 	int rc;
 
-	chip = tpm_find_get_ops(chip);
 	if (!chip)
 		return -ENODEV;
 
+	rc = tpm_try_get_ops(chip);
+	if (rc)
+		return rc;
+
 	rc = (chip->flags & TPM_CHIP_FLAG_TPM2) != 0;
 
 	tpm_put_ops(chip);
@@ -338,10 +341,13 @@ int tpm_pcr_read(struct tpm_chip *chip, u32 pcr_idx,
 {
 	int rc;
 
-	chip = tpm_find_get_ops(chip);
 	if (!chip)
 		return -ENODEV;
 
+	rc = tpm_try_get_ops(chip);
+	if (rc)
+		return rc;
+
 	if (chip->flags & TPM_CHIP_FLAG_TPM2)
 		rc = tpm2_pcr_read(chip, pcr_idx, digest, NULL);
 	else
@@ -369,10 +375,13 @@ int tpm_pcr_extend(struct tpm_chip *chip, u32 pcr_idx,
 	int rc;
 	int i;
 
-	chip = tpm_find_get_ops(chip);
 	if (!chip)
 		return -ENODEV;
 
+	rc = tpm_try_get_ops(chip);
+	if (rc)
+		return rc;
+
 	for (i = 0; i < chip->nr_allocated_banks; i++) {
 		if (digests[i].alg_id != chip->allocated_banks[i].alg_id) {
 			rc = -EINVAL;
@@ -492,10 +501,13 @@ int tpm_get_random(struct tpm_chip *chip, u8 *out, size_t max)
 	if (!out || max > TPM_MAX_RNG_DATA)
 		return -EINVAL;
 
-	chip = tpm_find_get_ops(chip);
 	if (!chip)
 		return -ENODEV;
 
+	rc = tpm_try_get_ops(chip);
+	if (rc)
+		return rc;
+
 	if (chip->flags & TPM_CHIP_FLAG_TPM2)
 		rc = tpm2_get_random(chip, out, max);
 	else

drivers/char/tpm/tpm.h

@@ -267,7 +267,6 @@ static inline void tpm_msleep(unsigned int delay_msec)
 int tpm_chip_bootstrap(struct tpm_chip *chip);
 int tpm_chip_start(struct tpm_chip *chip);
 void tpm_chip_stop(struct tpm_chip *chip);
-struct tpm_chip *tpm_find_get_ops(struct tpm_chip *chip);
 
 struct tpm_chip *tpm_chip_alloc(struct device *dev,
 				const struct tpm_class_ops *ops);

drivers/char/tpm/tpm1-cmd.c

@@ -799,11 +799,6 @@ int tpm1_pm_suspend(struct tpm_chip *chip, u32 tpm_suspend_pcr)
  */
 int tpm1_get_pcr_allocation(struct tpm_chip *chip)
 {
-	chip->allocated_banks = kcalloc(1, sizeof(*chip->allocated_banks),
-					GFP_KERNEL);
-	if (!chip->allocated_banks)
-		return -ENOMEM;
-
 	chip->allocated_banks[0].alg_id = TPM_ALG_SHA1;
 	chip->allocated_banks[0].digest_size = hash_digest_size[HASH_ALGO_SHA1];
 	chip->allocated_banks[0].crypto_id = HASH_ALGO_SHA1;

drivers/char/tpm/tpm2-cmd.c

@@ -550,11 +550,9 @@ ssize_t tpm2_get_pcr_allocation(struct tpm_chip *chip)
 
 	nr_possible_banks = be32_to_cpup(
 		(__be32 *)&buf.data[TPM_HEADER_SIZE + 5]);
-
-	chip->allocated_banks = kcalloc(nr_possible_banks,
-					sizeof(*chip->allocated_banks),
-					GFP_KERNEL);
-	if (!chip->allocated_banks) {
+	if (nr_possible_banks > TPM2_MAX_PCR_BANKS) {
+		pr_err("tpm: out of bank capacity: %u > %u\n",
+		       nr_possible_banks, TPM2_MAX_PCR_BANKS);
 		rc = -ENOMEM;
 		goto out;
 	}

drivers/char/tpm/tpm_crb.c

@@ -179,6 +179,7 @@ static int crb_try_pluton_doorbell(struct crb_priv *priv, bool wait_for_complete
  *
  * @dev:  crb device
  * @priv: crb private data
+ * @loc:  locality
  *
  * Write CRB_CTRL_REQ_GO_IDLE to TPM_CRB_CTRL_REQ
  * The device should respond within TIMEOUT_C by clearing the bit.
@@ -233,6 +234,7 @@ static int crb_go_idle(struct tpm_chip *chip)
  *
  * @dev:  crb device
  * @priv: crb private data
+ * @loc:  locality
  *
  * Write CRB_CTRL_REQ_CMD_READY to TPM_CRB_CTRL_REQ
  * and poll till the device acknowledge it by clearing the bit.
@@ -412,7 +414,7 @@ static int crb_do_acpi_start(struct tpm_chip *chip)
 #ifdef CONFIG_ARM64
 /*
  * This is a TPM Command Response Buffer start method that invokes a
- * Secure Monitor Call to requrest the firmware to execute or cancel
+ * Secure Monitor Call to request the firmware to execute or cancel
  * a TPM 2.0 command.
  */
 static int tpm_crb_smc_start(struct device *dev, unsigned long func_id)

drivers/char/tpm/tpm_tis_core.c

@@ -265,8 +265,7 @@ static u8 tpm_tis_status(struct tpm_chip *chip)
 
 			/*
 			 * Dump stack for forensics, as invalid TPM_STS.x could be
-			 * potentially triggered by impaired tpm_try_get_ops() or
-			 * tpm_find_get_ops().
+			 * potentially triggered by impaired tpm_try_get_ops().
 			 */
 			dump_stack();
 		}

drivers/crypto/hisilicon/qm.c

@@ -3032,11 +3032,36 @@ static void qm_put_pci_res(struct hisi_qm *qm)
 	pci_release_mem_regions(pdev);
 }
 
+static void hisi_mig_region_clear(struct hisi_qm *qm)
+{
+	u32 val;
+
+	/* Clear migration region set of PF */
+	if (qm->fun_type == QM_HW_PF && qm->ver > QM_HW_V3) {
+		val = readl(qm->io_base + QM_MIG_REGION_SEL);
+		val &= ~QM_MIG_REGION_EN;
+		writel(val, qm->io_base + QM_MIG_REGION_SEL);
+	}
+}
+
+static void hisi_mig_region_enable(struct hisi_qm *qm)
+{
+	u32 val;
+
+	/* Select migration region of PF */
+	if (qm->fun_type == QM_HW_PF && qm->ver > QM_HW_V3) {
+		val = readl(qm->io_base + QM_MIG_REGION_SEL);
+		val |= QM_MIG_REGION_EN;
+		writel(val, qm->io_base + QM_MIG_REGION_SEL);
+	}
+}
+
 static void hisi_qm_pci_uninit(struct hisi_qm *qm)
 {
 	struct pci_dev *pdev = qm->pdev;
 
 	pci_free_irq_vectors(pdev);
+	hisi_mig_region_clear(qm);
 	qm_put_pci_res(qm);
 	pci_disable_device(pdev);
 }
@@ -5752,6 +5777,7 @@ int hisi_qm_init(struct hisi_qm *qm)
 		goto err_free_qm_memory;
 
 	qm_cmd_init(qm);
+	hisi_mig_region_enable(qm);
 
 	return 0;
 
@@ -5890,6 +5916,7 @@ static int qm_rebuild_for_resume(struct hisi_qm *qm)
 	}
 
 	qm_cmd_init(qm);
+	hisi_mig_region_enable(qm);
 	hisi_qm_dev_err_init(qm);
 	/* Set the doorbell timeout to QM_DB_TIMEOUT_CFG ns. */
 	writel(QM_DB_TIMEOUT_SET, qm->io_base + QM_DB_TIMEOUT_CFG);

drivers/cxl/acpi.c

@@ -11,25 +11,36 @@
 #include "cxlpci.h"
 #include "cxl.h"
 
-struct cxl_cxims_data {
-	int nr_maps;
-	u64 xormaps[] __counted_by(nr_maps);
-};
-
 static const guid_t acpi_cxl_qtg_id_guid =
 	GUID_INIT(0xF365F9A6, 0xA7DE, 0x4071,
 		  0xA6, 0x6A, 0xB4, 0x0C, 0x0B, 0x4F, 0x8E, 0x52);
 
-static u64 cxl_apply_xor_maps(struct cxl_root_decoder *cxlrd, u64 addr)
+#define HBIW_TO_NR_MAPS_SIZE (CXL_DECODER_MAX_INTERLEAVE + 1)
+static const int hbiw_to_nr_maps[HBIW_TO_NR_MAPS_SIZE] = {
+	[1] = 0, [2] = 1, [3] = 0, [4] = 2, [6] = 1, [8] = 3, [12] = 2, [16] = 4
+};
+
+static const int valid_hbiw[] = { 1, 2, 3, 4, 6, 8, 12, 16 };
+
+u64 cxl_do_xormap_calc(struct cxl_cxims_data *cximsd, u64 addr, int hbiw)
 {
-	struct cxl_cxims_data *cximsd = cxlrd->platform_data;
-	int hbiw = cxlrd->cxlsd.nr_targets;
+	int nr_maps_to_apply = -1;
 	u64 val;
 	int pos;
 
-	/* No xormaps for host bridge interleave ways of 1 or 3 */
-	if (hbiw == 1 || hbiw == 3)
-		return addr;
+	/*
+	 * Strictly validate hbiw since this function is used for testing and
+	 * that nullifies any expectation of trusted parameters from the CXL
+	 * Region Driver.
+	 */
+	for (int i = 0; i < ARRAY_SIZE(valid_hbiw); i++) {
+		if (valid_hbiw[i] == hbiw) {
+			nr_maps_to_apply = hbiw_to_nr_maps[hbiw];
+			break;
+		}
+	}
+	if (nr_maps_to_apply == -1 || nr_maps_to_apply > cximsd->nr_maps)
+		return ULLONG_MAX;
 
 	/*
 	 * In regions using XOR interleave arithmetic the CXL HPA may not
@@ -60,6 +71,14 @@ static u64 cxl_apply_xor_maps(struct cxl_root_decoder *cxlrd, u64 addr)
 
 	return addr;
 }
+EXPORT_SYMBOL_FOR_MODULES(cxl_do_xormap_calc, "cxl_translate");
+
+static u64 cxl_apply_xor_maps(struct cxl_root_decoder *cxlrd, u64 addr)
+{
+	struct cxl_cxims_data *cximsd = cxlrd->platform_data;
+
+	return cxl_do_xormap_calc(cximsd, addr, cxlrd->cxlsd.nr_targets);
+}
 
 struct cxl_cxims_context {
 	struct device *dev;
@@ -353,7 +372,7 @@ static int cxl_acpi_set_cache_size(struct cxl_root_decoder *cxlrd)
 
 	rc = hmat_get_extended_linear_cache_size(&res, nid, &cache_size);
 	if (rc)
-		return rc;
+		return 0;
 
 	/*
 	 * The cache range is expected to be within the CFMWS.
@@ -378,21 +397,18 @@ static void cxl_setup_extended_linear_cache(struct cxl_root_decoder *cxlrd)
 	int rc;
 
 	rc = cxl_acpi_set_cache_size(cxlrd);
-	if (!rc)
-		return;
-
-	if (rc != -EOPNOTSUPP) {
+	if (rc) {
 		/*
-		 * Failing to support extended linear cache region resize does not
+		 * Failing to retrieve extended linear cache region resize does not
 		 * prevent the region from functioning. Only causes cxl list showing
 		 * incorrect region size.
 		 */
 		dev_warn(cxlrd->cxlsd.cxld.dev.parent,
-			 "Extended linear cache calculation failed rc:%d\n", rc);
-	}
+			 "Extended linear cache retrieval failed rc:%d\n", rc);
 
-	/* Ignoring return code */
-	cxlrd->cache_size = 0;
+		/* Ignoring return code */
+		cxlrd->cache_size = 0;
+	}
 }
 
 DEFINE_FREE(put_cxlrd, struct cxl_root_decoder *,
@@ -453,8 +469,6 @@ static int __cxl_parse_cfmws(struct acpi_cedt_cfmws *cfmws,
 		ig = CXL_DECODER_MIN_GRANULARITY;
 	cxld->interleave_granularity = ig;
 
-	cxl_setup_extended_linear_cache(cxlrd);
-
 	if (cfmws->interleave_arithmetic == ACPI_CEDT_CFMWS_ARITHMETIC_XOR) {
 		if (ways != 1 && ways != 3) {
 			cxims_ctx = (struct cxl_cxims_context) {
@@ -470,19 +484,14 @@ static int __cxl_parse_cfmws(struct acpi_cedt_cfmws *cfmws,
 				return -EINVAL;
 			}
 		}
+		cxlrd->ops.hpa_to_spa = cxl_apply_xor_maps;
+		cxlrd->ops.spa_to_hpa = cxl_apply_xor_maps;
 	}
 
+	cxl_setup_extended_linear_cache(cxlrd);
+
 	cxlrd->qos_class = cfmws->qtg_id;
 
-	if (cfmws->interleave_arithmetic == ACPI_CEDT_CFMWS_ARITHMETIC_XOR) {
-		cxlrd->ops = kzalloc(sizeof(*cxlrd->ops), GFP_KERNEL);
-		if (!cxlrd->ops)
-			return -ENOMEM;
-
-		cxlrd->ops->hpa_to_spa = cxl_apply_xor_maps;
-		cxlrd->ops->spa_to_hpa = cxl_apply_xor_maps;
-	}
-
 	rc = cxl_decoder_add(cxld);
 	if (rc)
 		return rc;

drivers/cxl/core/cdat.c

@@ -826,7 +826,7 @@ static struct xarray *cxl_switch_gather_bandwidth(struct cxl_region *cxlr,
 		cxl_coordinates_combine(coords, coords, ctx->coord);
 
 		/*
-		 * Take the min of the calculated bandwdith and the upstream
+		 * Take the min of the calculated bandwidth and the upstream
 		 * switch SSLBIS bandwidth if there's a parent switch
 		 */
 		if (!is_root)
@@ -949,7 +949,7 @@ static struct xarray *cxl_hb_gather_bandwidth(struct xarray *xa)
 /**
  * cxl_region_update_bandwidth - Update the bandwidth access coordinates of a region
  * @cxlr: The region being operated on
- * @input_xa: xarray holds cxl_perf_ctx wht calculated bandwidth per ACPI0017 instance
+ * @input_xa: xarray holds cxl_perf_ctx with calculated bandwidth per ACPI0017 instance
  */
 static void cxl_region_update_bandwidth(struct cxl_region *cxlr,
 					struct xarray *input_xa)

drivers/cxl/core/hdm.c

@@ -905,6 +905,9 @@ static void cxl_decoder_reset(struct cxl_decoder *cxld)
 	if ((cxld->flags & CXL_DECODER_F_ENABLE) == 0)
 		return;
 
+	if (test_bit(CXL_DECODER_F_LOCK, &cxld->flags))
+		return;
+
 	if (port->commit_end == id)
 		cxl_port_commit_reap(cxld);
 	else

drivers/cxl/core/pci.c

@@ -71,85 +71,6 @@ struct cxl_dport *__devm_cxl_add_dport_by_dev(struct cxl_port *port,
 }
 EXPORT_SYMBOL_NS_GPL(__devm_cxl_add_dport_by_dev, "CXL");
 
-struct cxl_walk_context {
-	struct pci_bus *bus;
-	struct cxl_port *port;
-	int type;
-	int error;
-	int count;
-};
-
-static int match_add_dports(struct pci_dev *pdev, void *data)
-{
-	struct cxl_walk_context *ctx = data;
-	struct cxl_port *port = ctx->port;
-	int type = pci_pcie_type(pdev);
-	struct cxl_register_map map;
-	struct cxl_dport *dport;
-	u32 lnkcap, port_num;
-	int rc;
-
-	if (pdev->bus != ctx->bus)
-		return 0;
-	if (!pci_is_pcie(pdev))
-		return 0;
-	if (type != ctx->type)
-		return 0;
-	if (pci_read_config_dword(pdev, pci_pcie_cap(pdev) + PCI_EXP_LNKCAP,
-				  &lnkcap))
-		return 0;
-
-	rc = cxl_find_regblock(pdev, CXL_REGLOC_RBI_COMPONENT, &map);
-	if (rc)
-		dev_dbg(&port->dev, "failed to find component registers\n");
-
-	port_num = FIELD_GET(PCI_EXP_LNKCAP_PN, lnkcap);
-	dport = devm_cxl_add_dport(port, &pdev->dev, port_num, map.resource);
-	if (IS_ERR(dport)) {
-		ctx->error = PTR_ERR(dport);
-		return PTR_ERR(dport);
-	}
-	ctx->count++;
-
-	return 0;
-}
-
-/**
- * devm_cxl_port_enumerate_dports - enumerate downstream ports of the upstream port
- * @port: cxl_port whose ->uport_dev is the upstream of dports to be enumerated
- *
- * Returns a positive number of dports enumerated or a negative error
- * code.
- */
-int devm_cxl_port_enumerate_dports(struct cxl_port *port)
-{
-	struct pci_bus *bus = cxl_port_to_pci_bus(port);
-	struct cxl_walk_context ctx;
-	int type;
-
-	if (!bus)
-		return -ENXIO;
-
-	if (pci_is_root_bus(bus))
-		type = PCI_EXP_TYPE_ROOT_PORT;
-	else
-		type = PCI_EXP_TYPE_DOWNSTREAM;
-
-	ctx = (struct cxl_walk_context) {
-		.port = port,
-		.bus = bus,
-		.type = type,
-	};
-	pci_walk_bus(bus, match_add_dports, &ctx);
-
-	if (ctx.count == 0)
-		return -ENODEV;
-	if (ctx.error)
-		return ctx.error;
-	return ctx.count;
-}
-EXPORT_SYMBOL_NS_GPL(devm_cxl_port_enumerate_dports, "CXL");
-
 static int cxl_dvsec_mem_range_valid(struct cxl_dev_state *cxlds, int id)
 {
 	struct pci_dev *pdev = to_pci_dev(cxlds->dev);
@@ -1217,6 +1138,14 @@ int cxl_gpf_port_setup(struct cxl_dport *dport)
 	return 0;
 }
 
+struct cxl_walk_context {
+	struct pci_bus *bus;
+	struct cxl_port *port;
+	int type;
+	int error;
+	int count;
+};
+
 static int count_dports(struct pci_dev *pdev, void *data)
 {
 	struct cxl_walk_context *ctx = data;

drivers/cxl/core/port.c

@@ -459,7 +459,6 @@ static void cxl_root_decoder_release(struct device *dev)
 	if (atomic_read(&cxlrd->region_id) >= 0)
 		memregion_free(atomic_read(&cxlrd->region_id));
 	__cxl_decoder_release(&cxlrd->cxlsd.cxld);
-	kfree(cxlrd->ops);
 	kfree(cxlrd);
 }
 

drivers/cxl/core/region.c

@@ -245,6 +245,9 @@ static void cxl_region_decode_reset(struct cxl_region *cxlr, int count)
 	struct cxl_region_params *p = &cxlr->params;
 	int i;
 
+	if (test_bit(CXL_REGION_F_LOCK, &cxlr->flags))
+		return;
+
 	/*
 	 * Before region teardown attempt to flush, evict any data cached for
 	 * this region, or scream loudly about missing arch / platform support
@@ -419,6 +422,9 @@ static ssize_t commit_store(struct device *dev, struct device_attribute *attr,
 		return len;
 	}
 
+	if (test_bit(CXL_REGION_F_LOCK, &cxlr->flags))
+		return -EPERM;
+
 	rc = queue_reset(cxlr);
 	if (rc)
 		return rc;
@@ -461,21 +467,6 @@ static ssize_t commit_show(struct device *dev, struct device_attribute *attr,
 }
 static DEVICE_ATTR_RW(commit);
 
-static umode_t cxl_region_visible(struct kobject *kobj, struct attribute *a,
-				  int n)
-{
-	struct device *dev = kobj_to_dev(kobj);
-	struct cxl_region *cxlr = to_cxl_region(dev);
-
-	/*
-	 * Support tooling that expects to find a 'uuid' attribute for all
-	 * regions regardless of mode.
-	 */
-	if (a == &dev_attr_uuid.attr && cxlr->mode != CXL_PARTMODE_PMEM)
-		return 0444;
-	return a->mode;
-}
-
 static ssize_t interleave_ways_show(struct device *dev,
 				    struct device_attribute *attr, char *buf)
 {
@@ -754,6 +745,21 @@ static ssize_t size_show(struct device *dev, struct device_attribute *attr,
 }
 static DEVICE_ATTR_RW(size);
 
+static ssize_t extended_linear_cache_size_show(struct device *dev,
+					       struct device_attribute *attr,
+					       char *buf)
+{
+	struct cxl_region *cxlr = to_cxl_region(dev);
+	struct cxl_region_params *p = &cxlr->params;
+	ssize_t rc;
+
+	ACQUIRE(rwsem_read_intr, rwsem)(&cxl_rwsem.region);
+	if ((rc = ACQUIRE_ERR(rwsem_read_intr, &rwsem)))
+		return rc;
+	return sysfs_emit(buf, "%#llx\n", p->cache_size);
+}
+static DEVICE_ATTR_RO(extended_linear_cache_size);
+
 static struct attribute *cxl_region_attrs[] = {
 	&dev_attr_uuid.attr,
 	&dev_attr_commit.attr,
@@ -762,9 +768,34 @@ static struct attribute *cxl_region_attrs[] = {
 	&dev_attr_resource.attr,
 	&dev_attr_size.attr,
 	&dev_attr_mode.attr,
+	&dev_attr_extended_linear_cache_size.attr,
 	NULL,
 };
 
+static umode_t cxl_region_visible(struct kobject *kobj, struct attribute *a,
+				  int n)
+{
+	struct device *dev = kobj_to_dev(kobj);
+	struct cxl_region *cxlr = to_cxl_region(dev);
+
+	/*
+	 * Support tooling that expects to find a 'uuid' attribute for all
+	 * regions regardless of mode.
+	 */
+	if (a == &dev_attr_uuid.attr && cxlr->mode != CXL_PARTMODE_PMEM)
+		return 0444;
+
+	/*
+	 * Don't display extended linear cache attribute if there is no
+	 * extended linear cache.
+	 */
+	if (a == &dev_attr_extended_linear_cache_size.attr &&
+	    cxlr->params.cache_size == 0)
+		return 0;
+
+	return a->mode;
+}
+
 static const struct attribute_group cxl_region_group = {
 	.attrs = cxl_region_attrs,
 	.is_visible = cxl_region_visible,
@@ -838,16 +869,16 @@ static int match_free_decoder(struct device *dev, const void *data)
 	return 1;
 }
 
-static bool region_res_match_cxl_range(const struct cxl_region_params *p,
-				       const struct range *range)
+static bool spa_maps_hpa(const struct cxl_region_params *p,
+			 const struct range *range)
 {
 	if (!p->res)
 		return false;
 
 	/*
-	 * If an extended linear cache region then the CXL range is assumed
-	 * to be fronted by the DRAM range in current known implementation.
-	 * This assumption will be made until a variant implementation exists.
+	 * The extended linear cache region is constructed by a 1:1 ratio
+	 * where the SPA maps equal amounts of DRAM and CXL HPA capacity with
+	 * CXL decoders at the high end of the SPA range.
 	 */
 	return p->res->start + p->cache_size == range->start &&
 		p->res->end == range->end;
@@ -865,7 +896,7 @@ static int match_auto_decoder(struct device *dev, const void *data)
 	cxld = to_cxl_decoder(dev);
 	r = &cxld->hpa_range;
 
-	if (region_res_match_cxl_range(p, r))
+	if (spa_maps_hpa(p, r))
 		return 1;
 
 	return 0;
@@ -1059,6 +1090,16 @@ static int cxl_rr_assign_decoder(struct cxl_port *port, struct cxl_region *cxlr,
 	return 0;
 }
 
+static void cxl_region_set_lock(struct cxl_region *cxlr,
+				struct cxl_decoder *cxld)
+{
+	if (!test_bit(CXL_DECODER_F_LOCK, &cxld->flags))
+		return;
+
+	set_bit(CXL_REGION_F_LOCK, &cxlr->flags);
+	clear_bit(CXL_REGION_F_NEEDS_RESET, &cxlr->flags);
+}
+
 /**
  * cxl_port_attach_region() - track a region's interest in a port by endpoint
  * @port: port to add a new region reference 'struct cxl_region_ref'
@@ -1170,6 +1211,8 @@ static int cxl_port_attach_region(struct cxl_port *port,
 		}
 	}
 
+	cxl_region_set_lock(cxlr, cxld);
+
 	rc = cxl_rr_ep_add(cxl_rr, cxled);
 	if (rc) {
 		dev_dbg(&cxlr->dev,
@@ -1328,7 +1371,7 @@ static int cxl_port_setup_targets(struct cxl_port *port,
 				  struct cxl_endpoint_decoder *cxled)
 {
 	struct cxl_root_decoder *cxlrd = to_cxl_root_decoder(cxlr->dev.parent);
-	int parent_iw, parent_ig, ig, iw, rc, inc = 0, pos = cxled->pos;
+	int parent_iw, parent_ig, ig, iw, rc, pos = cxled->pos;
 	struct cxl_port *parent_port = to_cxl_port(port->dev.parent);
 	struct cxl_region_ref *cxl_rr = cxl_rr_load(port, cxlr);
 	struct cxl_memdev *cxlmd = cxled_to_memdev(cxled);
@@ -1465,7 +1508,7 @@ static int cxl_port_setup_targets(struct cxl_port *port,
 	if (test_bit(CXL_REGION_F_AUTO, &cxlr->flags)) {
 		if (cxld->interleave_ways != iw ||
 		    (iw > 1 && cxld->interleave_granularity != ig) ||
-		    !region_res_match_cxl_range(p, &cxld->hpa_range) ||
+		    !spa_maps_hpa(p, &cxld->hpa_range) ||
 		    ((cxld->flags & CXL_DECODER_F_ENABLE) == 0)) {
 			dev_err(&cxlr->dev,
 				"%s:%s %s expected iw: %d ig: %d %pr\n",
@@ -1520,9 +1563,8 @@ add_target:
 		cxlsd->target[cxl_rr->nr_targets_set] = ep->dport;
 		cxlsd->cxld.target_map[cxl_rr->nr_targets_set] = ep->dport->port_id;
 	}
-	inc = 1;
+	cxl_rr->nr_targets_set++;
 out_target_set:
-	cxl_rr->nr_targets_set += inc;
 	dev_dbg(&cxlr->dev, "%s:%s target[%d] = %s for %s:%s @ %d\n",
 		dev_name(port->uport_dev), dev_name(&port->dev),
 		cxl_rr->nr_targets_set - 1, dev_name(ep->dport->dport_dev),
@@ -2439,6 +2481,7 @@ static struct cxl_region *cxl_region_alloc(struct cxl_root_decoder *cxlrd, int i
 	dev->bus = &cxl_bus_type;
 	dev->type = &cxl_region_type;
 	cxlr->id = id;
+	cxl_region_set_lock(cxlr, &cxlrd->cxlsd.cxld);
 
 	return cxlr;
 }
@@ -2924,38 +2967,119 @@ static bool cxl_is_hpa_in_chunk(u64 hpa, struct cxl_region *cxlr, int pos)
 	return false;
 }
 
-static bool has_hpa_to_spa(struct cxl_root_decoder *cxlrd)
+#define CXL_POS_ZERO 0
+/**
+ * cxl_validate_translation_params
+ * @eiw: encoded interleave ways
+ * @eig: encoded interleave granularity
+ * @pos: position in interleave
+ *
+ * Callers pass CXL_POS_ZERO when no position parameter needs validating.
+ *
+ * Returns: 0 on success, -EINVAL on first invalid parameter
+ */
+int cxl_validate_translation_params(u8 eiw, u16 eig, int pos)
 {
-	return cxlrd->ops && cxlrd->ops->hpa_to_spa;
-}
+	int ways, gran;
 
-static bool has_spa_to_hpa(struct cxl_root_decoder *cxlrd)
-{
-	return cxlrd->ops && cxlrd->ops->spa_to_hpa;
-}
-
-u64 cxl_dpa_to_hpa(struct cxl_region *cxlr, const struct cxl_memdev *cxlmd,
-		   u64 dpa)
-{
-	struct cxl_root_decoder *cxlrd = to_cxl_root_decoder(cxlr->dev.parent);
-	u64 dpa_offset, hpa_offset, bits_upper, mask_upper, hpa;
-	struct cxl_region_params *p = &cxlr->params;
-	struct cxl_endpoint_decoder *cxled = NULL;
-	u16 eig = 0;
-	u8 eiw = 0;
-	int pos;
-
-	for (int i = 0; i < p->nr_targets; i++) {
-		cxled = p->targets[i];
-		if (cxlmd == cxled_to_memdev(cxled))
-			break;
+	if (eiw_to_ways(eiw, &ways)) {
+		pr_debug("%s: invalid eiw=%u\n", __func__, eiw);
+		return -EINVAL;
 	}
-	if (!cxled || cxlmd != cxled_to_memdev(cxled))
+	if (eig_to_granularity(eig, &gran)) {
+		pr_debug("%s: invalid eig=%u\n", __func__, eig);
+		return -EINVAL;
+	}
+	if (pos < 0 || pos >= ways) {
+		pr_debug("%s: invalid pos=%d for ways=%u\n", __func__, pos,
+			 ways);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_FOR_MODULES(cxl_validate_translation_params, "cxl_translate");
+
+u64 cxl_calculate_dpa_offset(u64 hpa_offset, u8 eiw, u16 eig)
+{
+	u64 dpa_offset, bits_lower, bits_upper, temp;
+	int ret;
+
+	ret = cxl_validate_translation_params(eiw, eig, CXL_POS_ZERO);
+	if (ret)
 		return ULLONG_MAX;
 
-	pos = cxled->pos;
-	ways_to_eiw(p->interleave_ways, &eiw);
-	granularity_to_eig(p->interleave_granularity, &eig);
+	/*
+	 * DPA offset: CXL Spec 3.2 Section 8.2.4.20.13
+	 * Lower bits [IG+7:0] pass through unchanged
+	 * (eiw < 8)
+	 *	Per spec: DPAOffset[51:IG+8] = (HPAOffset[51:IG+IW+8] >> IW)
+	 *	Clear the position bits to isolate upper section, then
+	 *	reverse the left shift by eiw that occurred during DPA->HPA
+	 * (eiw >= 8)
+	 *	Per spec: DPAOffset[51:IG+8] = HPAOffset[51:IG+IW] / 3
+	 *	Extract upper bits from the correct bit range and divide by 3
+	 *	to recover the original DPA upper bits
+	 */
+	bits_lower = hpa_offset & GENMASK_ULL(eig + 7, 0);
+	if (eiw < 8) {
+		temp = hpa_offset &= ~GENMASK_ULL(eig + eiw + 8 - 1, 0);
+		dpa_offset = temp >> eiw;
+	} else {
+		bits_upper = div64_u64(hpa_offset >> (eig + eiw), 3);
+		dpa_offset = bits_upper << (eig + 8);
+	}
+	dpa_offset |= bits_lower;
+
+	return dpa_offset;
+}
+EXPORT_SYMBOL_FOR_MODULES(cxl_calculate_dpa_offset, "cxl_translate");
+
+int cxl_calculate_position(u64 hpa_offset, u8 eiw, u16 eig)
+{
+	unsigned int ways = 0;
+	u64 shifted, rem;
+	int pos, ret;
+
+	ret = cxl_validate_translation_params(eiw, eig, CXL_POS_ZERO);
+	if (ret)
+		return ret;
+
+	if (!eiw)
+		/* position is 0 if no interleaving */
+		return 0;
+
+	/*
+	 * Interleave position: CXL Spec 3.2 Section 8.2.4.20.13
+	 * eiw < 8
+	 *	Position is in the IW bits at HPA_OFFSET[IG+8+IW-1:IG+8].
+	 *	Per spec "remove IW bits starting with bit position IG+8"
+	 * eiw >= 8
+	 *	Position is not explicitly stored in HPA_OFFSET bits. It is
+	 *	derived from the modulo operation of the upper bits using
+	 *	the total number of interleave ways.
+	 */
+	if (eiw < 8) {
+		pos = (hpa_offset >> (eig + 8)) & GENMASK(eiw - 1, 0);
+	} else {
+		shifted = hpa_offset >> (eig + 8);
+		eiw_to_ways(eiw, &ways);
+		div64_u64_rem(shifted, ways, &rem);
+		pos = rem;
+	}
+
+	return pos;
+}
+EXPORT_SYMBOL_FOR_MODULES(cxl_calculate_position, "cxl_translate");
+
+u64 cxl_calculate_hpa_offset(u64 dpa_offset, int pos, u8 eiw, u16 eig)
+{
+	u64 mask_upper, hpa_offset, bits_upper;
+	int ret;
+
+	ret = cxl_validate_translation_params(eiw, eig, pos);
+	if (ret)
+		return ULLONG_MAX;
 
 	/*
 	 * The device position in the region interleave set was removed
@@ -2967,9 +3091,6 @@ u64 cxl_dpa_to_hpa(struct cxl_region *cxlr, const struct cxl_memdev *cxlmd,
 	 * 8.2.4.19.13 Implementation Note: Device Decode Logic
 	 */
 
-	/* Remove the dpa base */
-	dpa_offset = dpa - cxl_dpa_resource_start(cxled);
-
 	mask_upper = GENMASK_ULL(51, eig + 8);
 
 	if (eiw < 8) {
@@ -2984,12 +3105,43 @@ u64 cxl_dpa_to_hpa(struct cxl_region *cxlr, const struct cxl_memdev *cxlmd,
 	/* The lower bits remain unchanged */
 	hpa_offset |= dpa_offset & GENMASK_ULL(eig + 7, 0);
 
+	return hpa_offset;
+}
+EXPORT_SYMBOL_FOR_MODULES(cxl_calculate_hpa_offset, "cxl_translate");
+
+u64 cxl_dpa_to_hpa(struct cxl_region *cxlr, const struct cxl_memdev *cxlmd,
+		   u64 dpa)
+{
+	struct cxl_root_decoder *cxlrd = to_cxl_root_decoder(cxlr->dev.parent);
+	struct cxl_region_params *p = &cxlr->params;
+	struct cxl_endpoint_decoder *cxled = NULL;
+	u64 dpa_offset, hpa_offset, hpa;
+	u16 eig = 0;
+	u8 eiw = 0;
+	int pos;
+
+	for (int i = 0; i < p->nr_targets; i++) {
+		if (cxlmd == cxled_to_memdev(p->targets[i])) {
+			cxled = p->targets[i];
+			break;
+		}
+	}
+	if (!cxled)
+		return ULLONG_MAX;
+
+	pos = cxled->pos;
+	ways_to_eiw(p->interleave_ways, &eiw);
+	granularity_to_eig(p->interleave_granularity, &eig);
+
+	dpa_offset = dpa - cxl_dpa_resource_start(cxled);
+	hpa_offset = cxl_calculate_hpa_offset(dpa_offset, pos, eiw, eig);
+
 	/* Apply the hpa_offset to the region base address */
 	hpa = hpa_offset + p->res->start + p->cache_size;
 
 	/* Root decoder translation overrides typical modulo decode */
-	if (has_hpa_to_spa(cxlrd))
-		hpa = cxlrd->ops->hpa_to_spa(cxlrd, hpa);
+	if (cxlrd->ops.hpa_to_spa)
+		hpa = cxlrd->ops.hpa_to_spa(cxlrd, hpa);
 
 	if (!cxl_resource_contains_addr(p->res, hpa)) {
 		dev_dbg(&cxlr->dev,
@@ -2998,7 +3150,7 @@ u64 cxl_dpa_to_hpa(struct cxl_region *cxlr, const struct cxl_memdev *cxlmd,
 	}
 
 	/* Simple chunk check, by pos & gran, only applies to modulo decodes */
-	if (!has_hpa_to_spa(cxlrd) && (!cxl_is_hpa_in_chunk(hpa, cxlr, pos)))
+	if (!cxlrd->ops.hpa_to_spa && !cxl_is_hpa_in_chunk(hpa, cxlr, pos))
 		return ULLONG_MAX;
 
 	return hpa;
@@ -3016,8 +3168,6 @@ static int region_offset_to_dpa_result(struct cxl_region *cxlr, u64 offset,
 	struct cxl_root_decoder *cxlrd = to_cxl_root_decoder(cxlr->dev.parent);
 	struct cxl_endpoint_decoder *cxled;
 	u64 hpa, hpa_offset, dpa_offset;
-	u64 bits_upper, bits_lower;
-	u64 shifted, rem, temp;
 	u16 eig = 0;
 	u8 eiw = 0;
 	int pos;
@@ -3033,56 +3183,21 @@ static int region_offset_to_dpa_result(struct cxl_region *cxlr, u64 offset,
 	 * If the root decoder has SPA to CXL HPA callback, use it. Otherwise
 	 * CXL HPA is assumed to equal SPA.
 	 */
-	if (has_spa_to_hpa(cxlrd)) {
-		hpa = cxlrd->ops->spa_to_hpa(cxlrd, p->res->start + offset);
+	if (cxlrd->ops.spa_to_hpa) {
+		hpa = cxlrd->ops.spa_to_hpa(cxlrd, p->res->start + offset);
 		hpa_offset = hpa - p->res->start;
 	} else {
 		hpa_offset = offset;
 	}
-	/*
-	 * Interleave position: CXL Spec 3.2 Section 8.2.4.20.13
-	 * eiw < 8
-	 *	Position is in the IW bits at HPA_OFFSET[IG+8+IW-1:IG+8].
-	 *	Per spec "remove IW bits starting with bit position IG+8"
-	 * eiw >= 8
-	 *	Position is not explicitly stored in HPA_OFFSET bits. It is
-	 *	derived from the modulo operation of the upper bits using
-	 *	the total number of interleave ways.
-	 */
-	if (eiw < 8) {
-		pos = (hpa_offset >> (eig + 8)) & GENMASK(eiw - 1, 0);
-	} else {
-		shifted = hpa_offset >> (eig + 8);
-		div64_u64_rem(shifted, p->interleave_ways, &rem);
-		pos = rem;
-	}
+
+	pos = cxl_calculate_position(hpa_offset, eiw, eig);
 	if (pos < 0 || pos >= p->nr_targets) {
 		dev_dbg(&cxlr->dev, "Invalid position %d for %d targets\n",
 			pos, p->nr_targets);
 		return -ENXIO;
 	}
 
-	/*
-	 * DPA offset: CXL Spec 3.2 Section 8.2.4.20.13
-	 * Lower bits [IG+7:0] pass through unchanged
-	 * (eiw < 8)
-	 *	Per spec: DPAOffset[51:IG+8] = (HPAOffset[51:IG+IW+8] >> IW)
-	 *	Clear the position bits to isolate upper section, then
-	 *	reverse the left shift by eiw that occurred during DPA->HPA
-	 * (eiw >= 8)
-	 *	Per spec: DPAOffset[51:IG+8] = HPAOffset[51:IG+IW] / 3
-	 *	Extract upper bits from the correct bit range and divide by 3
-	 *	to recover the original DPA upper bits
-	 */
-	bits_lower = hpa_offset & GENMASK_ULL(eig + 7, 0);
-	if (eiw < 8) {
-		temp = hpa_offset &= ~((u64)GENMASK(eig + eiw + 8 - 1, 0));
-		dpa_offset = temp >> eiw;
-	} else {
-		bits_upper = div64_u64(hpa_offset >> (eig + eiw), 3);
-		dpa_offset = bits_upper << (eig + 8);
-	}
-	dpa_offset |= bits_lower;
+	dpa_offset = cxl_calculate_dpa_offset(hpa_offset, eiw, eig);
 
 	/* Look-up and return the result: a memdev and a DPA */
 	for (int i = 0; i < p->nr_targets; i++) {
@@ -3398,7 +3513,7 @@ static int match_region_by_range(struct device *dev, const void *data)
 	p = &cxlr->params;
 
 	guard(rwsem_read)(&cxl_rwsem.region);
-	return region_res_match_cxl_range(p, r);
+	return spa_maps_hpa(p, r);
 }
 
 static int cxl_extended_linear_cache_resize(struct cxl_region *cxlr,
@@ -3479,6 +3594,10 @@ static int __construct_region(struct cxl_region *cxlr,
 			 "Extended linear cache calculation failed rc:%d\n", rc);
 	}
 
+	rc = sysfs_update_group(&cxlr->dev.kobj, &cxl_region_group);
+	if (rc)
+		return rc;
+
 	rc = insert_resource(cxlrd->res, res);
 	if (rc) {
 		/*

drivers/cxl/cxl.h

@@ -451,7 +451,7 @@ struct cxl_root_decoder {
 	void *platform_data;
 	struct mutex range_lock;
 	int qos_class;
-	struct cxl_rd_ops *ops;
+	struct cxl_rd_ops ops;
 	struct cxl_switch_decoder cxlsd;
 };
 
@@ -517,6 +517,14 @@ enum cxl_partition_mode {
  */
 #define CXL_REGION_F_NEEDS_RESET 1
 
+/*
+ * Indicate whether this region is locked due to 1 or more decoders that have
+ * been locked. The approach of all or nothing is taken with regard to the
+ * locked attribute. CXL_REGION_F_NEEDS_RESET should not be set if this flag is
+ * set.
+ */
+#define CXL_REGION_F_LOCK 2
+
 /**
  * struct cxl_region - CXL region
  * @dev: This region's device
@@ -738,6 +746,25 @@ static inline bool is_cxl_root(struct cxl_port *port)
 	return port->uport_dev == port->dev.parent;
 }
 
+/* Address translation functions exported to cxl_translate test module only */
+int cxl_validate_translation_params(u8 eiw, u16 eig, int pos);
+u64 cxl_calculate_hpa_offset(u64 dpa_offset, int pos, u8 eiw, u16 eig);
+u64 cxl_calculate_dpa_offset(u64 hpa_offset, u8 eiw, u16 eig);
+int cxl_calculate_position(u64 hpa_offset, u8 eiw, u16 eig);
+struct cxl_cxims_data {
+	int nr_maps;
+	u64 xormaps[] __counted_by(nr_maps);
+};
+
+#if IS_ENABLED(CONFIG_CXL_ACPI)
+u64 cxl_do_xormap_calc(struct cxl_cxims_data *cximsd, u64 addr, int hbiw);
+#else
+static inline u64 cxl_do_xormap_calc(struct cxl_cxims_data *cximsd, u64 addr, int hbiw)
+{
+	return ULLONG_MAX;
+}
+#endif
+
 int cxl_num_decoders_committed(struct cxl_port *port);
 bool is_cxl_port(const struct device *dev);
 struct cxl_port *to_cxl_port(const struct device *dev);

drivers/cxl/cxlpci.h

@@ -127,7 +127,6 @@ static inline bool cxl_pci_flit_256(struct pci_dev *pdev)
 	return lnksta2 & PCI_EXP_LNKSTA2_FLIT;
 }
 
-int devm_cxl_port_enumerate_dports(struct cxl_port *port);
 struct cxl_dev_state;
 void read_cdat_data(struct cxl_port *port);
 void cxl_cor_error_detected(struct pci_dev *pdev);

drivers/cxl/pci.c

@@ -136,7 +136,7 @@ static irqreturn_t cxl_pci_mbox_irq(int irq, void *id)
 	if (opcode == CXL_MBOX_OP_SANITIZE) {
 		mutex_lock(&cxl_mbox->mbox_mutex);
 		if (mds->security.sanitize_node)
-			mod_delayed_work(system_wq, &mds->security.poll_dwork, 0);
+			mod_delayed_work(system_percpu_wq, &mds->security.poll_dwork, 0);
 		mutex_unlock(&cxl_mbox->mbox_mutex);
 	} else {
 		/* short-circuit the wait in __cxl_pci_mbox_send_cmd() */

drivers/dma-buf/Makefile

@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0-only
 obj-y := dma-buf.o dma-fence.o dma-fence-array.o dma-fence-chain.o \
-	 dma-fence-unwrap.o dma-resv.o
+	 dma-fence-unwrap.o dma-resv.o dma-buf-mapping.o
 obj-$(CONFIG_DMABUF_HEAPS)	+= dma-heap.o
 obj-$(CONFIG_DMABUF_HEAPS)	+= heaps/
 obj-$(CONFIG_SYNC_FILE)		+= sync_file.o

drivers/dma-buf/dma-buf-mapping.c

@@ -0,0 +1,248 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * DMA BUF Mapping Helpers
+ *
+ */
+#include <linux/dma-buf-mapping.h>
+#include <linux/dma-resv.h>
+
+static struct scatterlist *fill_sg_entry(struct scatterlist *sgl, size_t length,
+					 dma_addr_t addr)
+{
+	unsigned int len, nents;
+	int i;
+
+	nents = DIV_ROUND_UP(length, UINT_MAX);
+	for (i = 0; i < nents; i++) {
+		len = min_t(size_t, length, UINT_MAX);
+		length -= len;
+		/*
+		 * DMABUF abuses scatterlist to create a scatterlist
+		 * that does not have any CPU list, only the DMA list.
+		 * Always set the page related values to NULL to ensure
+		 * importers can't use it. The phys_addr based DMA API
+		 * does not require the CPU list for mapping or unmapping.
+		 */
+		sg_set_page(sgl, NULL, 0, 0);
+		sg_dma_address(sgl) = addr + (dma_addr_t)i * UINT_MAX;
+		sg_dma_len(sgl) = len;
+		sgl = sg_next(sgl);
+	}
+
+	return sgl;
+}
+
+static unsigned int calc_sg_nents(struct dma_iova_state *state,
+				  struct dma_buf_phys_vec *phys_vec,
+				  size_t nr_ranges, size_t size)
+{
+	unsigned int nents = 0;
+	size_t i;
+
+	if (!state || !dma_use_iova(state)) {
+		for (i = 0; i < nr_ranges; i++)
+			nents += DIV_ROUND_UP(phys_vec[i].len, UINT_MAX);
+	} else {
+		/*
+		 * In IOVA case, there is only one SG entry which spans
+		 * for whole IOVA address space, but we need to make sure
+		 * that it fits sg->length, maybe we need more.
+		 */
+		nents = DIV_ROUND_UP(size, UINT_MAX);
+	}
+
+	return nents;
+}
+
+/**
+ * struct dma_buf_dma - holds DMA mapping information
+ * @sgt:    Scatter-gather table
+ * @state:  DMA IOVA state relevant in IOMMU-based DMA
+ * @size:   Total size of DMA transfer
+ */
+struct dma_buf_dma {
+	struct sg_table sgt;
+	struct dma_iova_state *state;
+	size_t size;
+};
+
+/**
+ * dma_buf_phys_vec_to_sgt - Returns the scatterlist table of the attachment
+ * from arrays of physical vectors. This funciton is intended for MMIO memory
+ * only.
+ * @attach:	[in]	attachment whose scatterlist is to be returned
+ * @provider:	[in]	p2pdma provider
+ * @phys_vec:	[in]	array of physical vectors
+ * @nr_ranges:	[in]	number of entries in phys_vec array
+ * @size:	[in]	total size of phys_vec
+ * @dir:	[in]	direction of DMA transfer
+ *
+ * Returns sg_table containing the scatterlist to be returned; returns ERR_PTR
+ * on error. May return -EINTR if it is interrupted by a signal.
+ *
+ * On success, the DMA addresses and lengths in the returned scatterlist are
+ * PAGE_SIZE aligned.
+ *
+ * A mapping must be unmapped by using dma_buf_free_sgt().
+ *
+ * NOTE: This function is intended for exporters. If direct traffic routing is
+ * mandatory exporter should call routing pci_p2pdma_map_type() before calling
+ * this function.
+ */
+struct sg_table *dma_buf_phys_vec_to_sgt(struct dma_buf_attachment *attach,
+					 struct p2pdma_provider *provider,
+					 struct dma_buf_phys_vec *phys_vec,
+					 size_t nr_ranges, size_t size,
+					 enum dma_data_direction dir)
+{
+	unsigned int nents, mapped_len = 0;
+	struct dma_buf_dma *dma;
+	struct scatterlist *sgl;
+	dma_addr_t addr;
+	size_t i;
+	int ret;
+
+	dma_resv_assert_held(attach->dmabuf->resv);
+
+	if (WARN_ON(!attach || !attach->dmabuf || !provider))
+		/* This function is supposed to work on MMIO memory only */
+		return ERR_PTR(-EINVAL);
+
+	dma = kzalloc(sizeof(*dma), GFP_KERNEL);
+	if (!dma)
+		return ERR_PTR(-ENOMEM);
+
+	switch (pci_p2pdma_map_type(provider, attach->dev)) {
+	case PCI_P2PDMA_MAP_BUS_ADDR:
+		/*
+		 * There is no need in IOVA at all for this flow.
+		 */
+		break;
+	case PCI_P2PDMA_MAP_THRU_HOST_BRIDGE:
+		dma->state = kzalloc(sizeof(*dma->state), GFP_KERNEL);
+		if (!dma->state) {
+			ret = -ENOMEM;
+			goto err_free_dma;
+		}
+
+		dma_iova_try_alloc(attach->dev, dma->state, 0, size);
+		break;
+	default:
+		ret = -EINVAL;
+		goto err_free_dma;
+	}
+
+	nents = calc_sg_nents(dma->state, phys_vec, nr_ranges, size);
+	ret = sg_alloc_table(&dma->sgt, nents, GFP_KERNEL | __GFP_ZERO);
+	if (ret)
+		goto err_free_state;
+
+	sgl = dma->sgt.sgl;
+
+	for (i = 0; i < nr_ranges; i++) {
+		if (!dma->state) {
+			addr = pci_p2pdma_bus_addr_map(provider,
+						       phys_vec[i].paddr);
+		} else if (dma_use_iova(dma->state)) {
+			ret = dma_iova_link(attach->dev, dma->state,
+					    phys_vec[i].paddr, 0,
+					    phys_vec[i].len, dir,
+					    DMA_ATTR_MMIO);
+			if (ret)
+				goto err_unmap_dma;
+
+			mapped_len += phys_vec[i].len;
+		} else {
+			addr = dma_map_phys(attach->dev, phys_vec[i].paddr,
+					    phys_vec[i].len, dir,
+					    DMA_ATTR_MMIO);
+			ret = dma_mapping_error(attach->dev, addr);
+			if (ret)
+				goto err_unmap_dma;
+		}
+
+		if (!dma->state || !dma_use_iova(dma->state))
+			sgl = fill_sg_entry(sgl, phys_vec[i].len, addr);
+	}
+
+	if (dma->state && dma_use_iova(dma->state)) {
+		WARN_ON_ONCE(mapped_len != size);
+		ret = dma_iova_sync(attach->dev, dma->state, 0, mapped_len);
+		if (ret)
+			goto err_unmap_dma;
+
+		sgl = fill_sg_entry(sgl, mapped_len, dma->state->addr);
+	}
+
+	dma->size = size;
+
+	/*
+	 * No CPU list included — set orig_nents = 0 so others can detect
+	 * this via SG table (use nents only).
+	 */
+	dma->sgt.orig_nents = 0;
+
+
+	/*
+	 * SGL must be NULL to indicate that SGL is the last one
+	 * and we allocated correct number of entries in sg_alloc_table()
+	 */
+	WARN_ON_ONCE(sgl);
+	return &dma->sgt;
+
+err_unmap_dma:
+	if (!i || !dma->state) {
+		; /* Do nothing */
+	} else if (dma_use_iova(dma->state)) {
+		dma_iova_destroy(attach->dev, dma->state, mapped_len, dir,
+				 DMA_ATTR_MMIO);
+	} else {
+		for_each_sgtable_dma_sg(&dma->sgt, sgl, i)
+			dma_unmap_phys(attach->dev, sg_dma_address(sgl),
+				       sg_dma_len(sgl), dir, DMA_ATTR_MMIO);
+	}
+	sg_free_table(&dma->sgt);
+err_free_state:
+	kfree(dma->state);
+err_free_dma:
+	kfree(dma);
+	return ERR_PTR(ret);
+}
+EXPORT_SYMBOL_NS_GPL(dma_buf_phys_vec_to_sgt, "DMA_BUF");
+
+/**
+ * dma_buf_free_sgt- unmaps the buffer
+ * @attach:	[in]	attachment to unmap buffer from
+ * @sgt:	[in]	scatterlist info of the buffer to unmap
+ * @dir:	[in]	direction of DMA transfer
+ *
+ * This unmaps a DMA mapping for @attached obtained
+ * by dma_buf_phys_vec_to_sgt().
+ */
+void dma_buf_free_sgt(struct dma_buf_attachment *attach, struct sg_table *sgt,
+		      enum dma_data_direction dir)
+{
+	struct dma_buf_dma *dma = container_of(sgt, struct dma_buf_dma, sgt);
+	int i;
+
+	dma_resv_assert_held(attach->dmabuf->resv);
+
+	if (!dma->state) {
+		; /* Do nothing */
+	} else if (dma_use_iova(dma->state)) {
+		dma_iova_destroy(attach->dev, dma->state, dma->size, dir,
+				 DMA_ATTR_MMIO);
+	} else {
+		struct scatterlist *sgl;
+
+		for_each_sgtable_dma_sg(sgt, sgl, i)
+			dma_unmap_phys(attach->dev, sg_dma_address(sgl),
+				       sg_dma_len(sgl), dir, DMA_ATTR_MMIO);
+	}
+
+	sg_free_table(sgt);
+	kfree(dma->state);
+	kfree(dma);
+
+}
+EXPORT_SYMBOL_NS_GPL(dma_buf_free_sgt, "DMA_BUF");

drivers/gpu/drm/i915/Makefile

@@ -239,6 +239,8 @@ i915-y += \
 	display/intel_cdclk.o \
 	display/intel_cmtg.o \
 	display/intel_color.o \
+	display/intel_colorop.o \
+	display/intel_color_pipeline.o \
 	display/intel_combo_phy.o \
 	display/intel_connector.o \
 	display/intel_crtc.o \

drivers/gpu/drm/i915/display/intel_color.c

@@ -32,6 +32,8 @@
 #include "intel_display_utils.h"
 #include "intel_dsb.h"
 #include "intel_vrr.h"
+#include "skl_universal_plane.h"
+#include "skl_universal_plane_regs.h"
 
 struct intel_color_funcs {
 	int (*color_check)(struct intel_atomic_state *state,
@@ -87,6 +89,14 @@ struct intel_color_funcs {
 	 * Read config other than LUTs and CSCs, before them. Optional.
 	 */
 	void (*get_config)(struct intel_crtc_state *crtc_state);
+
+	/* Plane CSC*/
+	void (*load_plane_csc_matrix)(struct intel_dsb *dsb,
+				      const struct intel_plane_state *plane_state);
+
+	/* Plane Pre/Post CSC */
+	void (*load_plane_luts)(struct intel_dsb *dsb,
+				const struct intel_plane_state *plane_state);
 };
 
 #define CTM_COEFF_SIGN	(1ULL << 63)
@@ -609,6 +619,8 @@ static u16 ctm_to_twos_complement(u64 coeff, int int_bits, int frac_bits)
 	if (CTM_COEFF_NEGATIVE(coeff))
 		c = -c;
 
+	int_bits = max(int_bits, 1);
+
 	c = clamp(c, -(s64)BIT(int_bits + frac_bits - 1),
 		  (s64)(BIT(int_bits + frac_bits - 1) - 1));
 
@@ -3836,6 +3848,266 @@ static void icl_read_luts(struct intel_crtc_state *crtc_state)
 	}
 }
 
+static void
+xelpd_load_plane_csc_matrix(struct intel_dsb *dsb,
+			    const struct intel_plane_state *plane_state)
+{
+	struct intel_display *display = to_intel_display(plane_state);
+	const struct drm_plane_state *state = &plane_state->uapi;
+	enum pipe pipe = to_intel_plane(state->plane)->pipe;
+	enum plane_id plane = to_intel_plane(state->plane)->id;
+	const struct drm_property_blob *blob = plane_state->hw.ctm;
+	struct drm_color_ctm_3x4 *ctm;
+	const u64 *input;
+	u16 coeffs[9] = {};
+	int i, j;
+
+	if (!icl_is_hdr_plane(display, plane) || !blob)
+		return;
+
+	ctm = blob->data;
+	input = ctm->matrix;
+
+	/*
+	 * Convert fixed point S31.32 input to format supported by the
+	 * hardware.
+	 */
+	for (i = 0, j = 0; i < ARRAY_SIZE(coeffs); i++) {
+		u64 abs_coeff = ((1ULL << 63) - 1) & input[j];
+
+		/*
+		 * Clamp input value to min/max supported by
+		 * hardware.
+		 */
+		abs_coeff = clamp_val(abs_coeff, 0, CTM_COEFF_4_0 - 1);
+
+		/* sign bit */
+		if (CTM_COEFF_NEGATIVE(input[j]))
+			coeffs[i] |= 1 << 15;
+
+		if (abs_coeff < CTM_COEFF_0_125)
+			coeffs[i] |= (3 << 12) |
+				      ILK_CSC_COEFF_FP(abs_coeff, 12);
+		else if (abs_coeff < CTM_COEFF_0_25)
+			coeffs[i] |= (2 << 12) |
+				      ILK_CSC_COEFF_FP(abs_coeff, 11);
+		else if (abs_coeff < CTM_COEFF_0_5)
+			coeffs[i] |= (1 << 12) |
+				      ILK_CSC_COEFF_FP(abs_coeff, 10);
+		else if (abs_coeff < CTM_COEFF_1_0)
+			coeffs[i] |= ILK_CSC_COEFF_FP(abs_coeff, 9);
+		else if (abs_coeff < CTM_COEFF_2_0)
+			coeffs[i] |= (7 << 12) |
+				      ILK_CSC_COEFF_FP(abs_coeff, 8);
+		else
+			coeffs[i] |= (6 << 12) |
+				      ILK_CSC_COEFF_FP(abs_coeff, 7);
+
+		/* Skip postoffs */
+		if (!((j + 2) % 4))
+			j += 2;
+		else
+			j++;
+	}
+
+	intel_de_write_dsb(display, dsb, PLANE_CSC_COEFF(pipe, plane, 0),
+			   coeffs[0] << 16 | coeffs[1]);
+	intel_de_write_dsb(display, dsb, PLANE_CSC_COEFF(pipe, plane, 1),
+			   coeffs[2] << 16);
+
+	intel_de_write_dsb(display, dsb, PLANE_CSC_COEFF(pipe, plane, 2),
+			   coeffs[3] << 16 | coeffs[4]);
+	intel_de_write_dsb(display, dsb, PLANE_CSC_COEFF(pipe, plane, 3),
+			   coeffs[5] << 16);
+
+	intel_de_write_dsb(display, dsb, PLANE_CSC_COEFF(pipe, plane, 4),
+			   coeffs[6] << 16 | coeffs[7]);
+	intel_de_write_dsb(display, dsb, PLANE_CSC_COEFF(pipe, plane, 5),
+			   coeffs[8] << 16);
+
+	intel_de_write_dsb(display, dsb, PLANE_CSC_PREOFF(pipe, plane, 0), 0);
+	intel_de_write_dsb(display, dsb, PLANE_CSC_PREOFF(pipe, plane, 1), 0);
+	intel_de_write_dsb(display, dsb, PLANE_CSC_PREOFF(pipe, plane, 2), 0);
+
+	/*
+	 * Conversion from S31.32 to S0.12. BIT[12] is the signed bit
+	 */
+	intel_de_write_dsb(display, dsb,
+			   PLANE_CSC_POSTOFF(pipe, plane, 0),
+			   ctm_to_twos_complement(input[3], 0, 12));
+	intel_de_write_dsb(display, dsb,
+			   PLANE_CSC_POSTOFF(pipe, plane, 1),
+			   ctm_to_twos_complement(input[7], 0, 12));
+	intel_de_write_dsb(display, dsb,
+			   PLANE_CSC_POSTOFF(pipe, plane, 2),
+			   ctm_to_twos_complement(input[11], 0, 12));
+}
+
+static void
+xelpd_program_plane_pre_csc_lut(struct intel_dsb *dsb,
+				const struct intel_plane_state *plane_state)
+{
+	struct intel_display *display = to_intel_display(plane_state);
+	const struct drm_plane_state *state = &plane_state->uapi;
+	enum pipe pipe = to_intel_plane(state->plane)->pipe;
+	enum plane_id plane = to_intel_plane(state->plane)->id;
+	const struct drm_color_lut32 *pre_csc_lut = plane_state->hw.degamma_lut->data;
+	u32 i, lut_size;
+
+	if (icl_is_hdr_plane(display, plane)) {
+		lut_size = 128;
+
+		intel_de_write_dsb(display, dsb,
+				   PLANE_PRE_CSC_GAMC_INDEX_ENH(pipe, plane, 0),
+				   PLANE_PAL_PREC_AUTO_INCREMENT);
+
+		if (pre_csc_lut) {
+			for (i = 0; i < lut_size; i++) {
+				u32 lut_val = drm_color_lut32_extract(pre_csc_lut[i].green, 24);
+
+				intel_de_write_dsb(display, dsb,
+						   PLANE_PRE_CSC_GAMC_DATA_ENH(pipe, plane, 0),
+						   lut_val);
+			}
+
+			/* Program the max register to clamp values > 1.0. */
+			/* TODO: Restrict to 0x7ffffff */
+			do {
+				intel_de_write_dsb(display, dsb,
+						   PLANE_PRE_CSC_GAMC_DATA_ENH(pipe, plane, 0),
+						   (1 << 24));
+			} while (i++ > 130);
+		} else {
+			for (i = 0; i < lut_size; i++) {
+				u32 v = (i * ((1 << 24) - 1)) / (lut_size - 1);
+
+				intel_de_write_dsb(display, dsb,
+						   PLANE_PRE_CSC_GAMC_DATA_ENH(pipe, plane, 0), v);
+			}
+
+			do {
+				intel_de_write_dsb(display, dsb,
+						   PLANE_PRE_CSC_GAMC_DATA_ENH(pipe, plane, 0),
+						   1 << 24);
+			} while (i++ < 130);
+		}
+
+		intel_de_write_dsb(display, dsb, PLANE_PRE_CSC_GAMC_INDEX_ENH(pipe, plane, 0), 0);
+	}
+}
+
+static void
+xelpd_program_plane_post_csc_lut(struct intel_dsb *dsb,
+				 const struct intel_plane_state *plane_state)
+{
+	struct intel_display *display = to_intel_display(plane_state);
+	const struct drm_plane_state *state = &plane_state->uapi;
+	enum pipe pipe = to_intel_plane(state->plane)->pipe;
+	enum plane_id plane = to_intel_plane(state->plane)->id;
+	const struct drm_color_lut32 *post_csc_lut = plane_state->hw.gamma_lut->data;
+	u32 i, lut_size, lut_val;
+
+	if (icl_is_hdr_plane(display, plane)) {
+		intel_de_write_dsb(display, dsb, PLANE_POST_CSC_GAMC_INDEX_ENH(pipe, plane, 0),
+				   PLANE_PAL_PREC_AUTO_INCREMENT);
+		/* TODO: Add macro */
+		intel_de_write_dsb(display, dsb, PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH(pipe, plane, 0),
+				   PLANE_PAL_PREC_AUTO_INCREMENT);
+		if (post_csc_lut) {
+			lut_size = 32;
+			for (i = 0; i < lut_size; i++) {
+				lut_val = drm_color_lut32_extract(post_csc_lut[i].green, 24);
+
+				intel_de_write_dsb(display, dsb,
+						   PLANE_POST_CSC_GAMC_DATA_ENH(pipe, plane, 0),
+						   lut_val);
+			}
+
+			/* Segment 2 */
+			do {
+				intel_de_write_dsb(display, dsb,
+						   PLANE_POST_CSC_GAMC_DATA_ENH(pipe, plane, 0),
+						   (1 << 24));
+			} while (i++ < 34);
+		} else {
+			/*TODO: Add for segment 0 */
+			lut_size = 32;
+			for (i = 0; i < lut_size; i++) {
+				u32 v = (i * ((1 << 24) - 1)) / (lut_size - 1);
+
+				intel_de_write_dsb(display, dsb,
+						   PLANE_POST_CSC_GAMC_DATA_ENH(pipe, plane, 0), v);
+			}
+
+			do {
+				intel_de_write_dsb(display, dsb,
+						   PLANE_POST_CSC_GAMC_DATA_ENH(pipe, plane, 0),
+						   1 << 24);
+			} while (i++ < 34);
+		}
+
+		intel_de_write_dsb(display, dsb, PLANE_POST_CSC_GAMC_INDEX_ENH(pipe, plane, 0), 0);
+		intel_de_write_dsb(display, dsb,
+				   PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH(pipe, plane, 0), 0);
+	}
+}
+
+static void
+xelpd_plane_load_luts(struct intel_dsb *dsb, const struct intel_plane_state *plane_state)
+{
+	if (plane_state->hw.degamma_lut)
+		xelpd_program_plane_pre_csc_lut(dsb, plane_state);
+
+	if (plane_state->hw.gamma_lut)
+		xelpd_program_plane_post_csc_lut(dsb, plane_state);
+}
+
+static u32 glk_3dlut_10(const struct drm_color_lut32 *color)
+{
+	return REG_FIELD_PREP(LUT_3D_DATA_RED_MASK, drm_color_lut32_extract(color->red, 10)) |
+		REG_FIELD_PREP(LUT_3D_DATA_GREEN_MASK, drm_color_lut32_extract(color->green, 10)) |
+		REG_FIELD_PREP(LUT_3D_DATA_BLUE_MASK, drm_color_lut32_extract(color->blue, 10));
+}
+
+static void glk_load_lut_3d(struct intel_dsb *dsb,
+			    struct intel_crtc *crtc,
+			    const struct drm_property_blob *blob)
+{
+	struct intel_display *display = to_intel_display(crtc->base.dev);
+	const struct drm_color_lut32 *lut = blob->data;
+	int i, lut_size = drm_color_lut32_size(blob);
+	enum pipe pipe = crtc->pipe;
+
+	if (!dsb && intel_de_read(display, LUT_3D_CTL(pipe)) & LUT_3D_READY) {
+		drm_err(display->drm, "[CRTC:%d:%s] 3D LUT not ready, not loading LUTs\n",
+			crtc->base.base.id, crtc->base.name);
+		return;
+	}
+
+	intel_de_write_dsb(display, dsb, LUT_3D_INDEX(pipe), LUT_3D_AUTO_INCREMENT);
+	for (i = 0; i < lut_size; i++)
+		intel_de_write_dsb(display, dsb, LUT_3D_DATA(pipe), glk_3dlut_10(&lut[i]));
+	intel_de_write_dsb(display, dsb, LUT_3D_INDEX(pipe), 0);
+}
+
+static void glk_lut_3d_commit(struct intel_dsb *dsb, struct intel_crtc *crtc, bool enable)
+{
+	struct intel_display *display = to_intel_display(crtc);
+	enum pipe pipe = crtc->pipe;
+	u32 val = 0;
+
+	if (!dsb && intel_de_read(display, LUT_3D_CTL(pipe)) & LUT_3D_READY) {
+		drm_err(display->drm, "[CRTC:%d:%s] 3D LUT not ready, not committing change\n",
+			crtc->base.base.id, crtc->base.name);
+		return;
+	}
+
+	if (enable)
+		val = LUT_3D_ENABLE | LUT_3D_READY | LUT_3D_BIND_PLANE_1;
+
+	intel_de_write_dsb(display, dsb, LUT_3D_CTL(pipe), val);
+}
+
 static const struct intel_color_funcs chv_color_funcs = {
 	.color_check = chv_color_check,
 	.color_commit_arm = i9xx_color_commit_arm,
@@ -3883,6 +4155,8 @@ static const struct intel_color_funcs tgl_color_funcs = {
 	.lut_equal = icl_lut_equal,
 	.read_csc = icl_read_csc,
 	.get_config = skl_get_config,
+	.load_plane_csc_matrix = xelpd_load_plane_csc_matrix,
+	.load_plane_luts = xelpd_plane_load_luts,
 };
 
 static const struct intel_color_funcs icl_color_funcs = {
@@ -3963,6 +4237,67 @@ static const struct intel_color_funcs ilk_color_funcs = {
 	.get_config = ilk_get_config,
 };
 
+void intel_color_plane_commit_arm(struct intel_dsb *dsb,
+				  const struct intel_plane_state *plane_state)
+{
+	struct intel_display *display = to_intel_display(plane_state);
+	struct intel_crtc *crtc = to_intel_crtc(plane_state->uapi.crtc);
+
+	if (crtc && intel_color_crtc_has_3dlut(display, crtc->pipe))
+		glk_lut_3d_commit(dsb, crtc, !!plane_state->hw.lut_3d);
+}
+
+static void
+intel_color_load_plane_csc_matrix(struct intel_dsb *dsb,
+				  const struct intel_plane_state *plane_state)
+{
+	struct intel_display *display = to_intel_display(plane_state);
+
+	if (display->funcs.color->load_plane_csc_matrix)
+		display->funcs.color->load_plane_csc_matrix(dsb, plane_state);
+}
+
+static void
+intel_color_load_plane_luts(struct intel_dsb *dsb,
+			    const struct intel_plane_state *plane_state)
+{
+	struct intel_display *display = to_intel_display(plane_state);
+
+	if (display->funcs.color->load_plane_luts)
+		display->funcs.color->load_plane_luts(dsb, plane_state);
+}
+
+bool
+intel_color_crtc_has_3dlut(struct intel_display *display, enum pipe pipe)
+{
+	if (DISPLAY_VER(display) >= 12)
+		return pipe == PIPE_A || pipe == PIPE_B;
+	else
+		return false;
+}
+
+static void
+intel_color_load_3dlut(struct intel_dsb *dsb,
+		       const struct intel_plane_state *plane_state)
+{
+	struct intel_display *display = to_intel_display(plane_state);
+	struct intel_crtc *crtc = to_intel_crtc(plane_state->uapi.crtc);
+
+	if (crtc && intel_color_crtc_has_3dlut(display, crtc->pipe))
+		glk_load_lut_3d(dsb, crtc, plane_state->hw.lut_3d);
+}
+
+void intel_color_plane_program_pipeline(struct intel_dsb *dsb,
+					const struct intel_plane_state *plane_state)
+{
+	if (plane_state->hw.ctm)
+		intel_color_load_plane_csc_matrix(dsb, plane_state);
+	if (plane_state->hw.degamma_lut || plane_state->hw.gamma_lut)
+		intel_color_load_plane_luts(dsb, plane_state);
+	if (plane_state->hw.lut_3d)
+		intel_color_load_3dlut(dsb, plane_state);
+}
+
 void intel_color_crtc_init(struct intel_crtc *crtc)
 {
 	struct intel_display *display = to_intel_display(crtc);

drivers/gpu/drm/i915/display/intel_color.h

@@ -13,7 +13,9 @@ struct intel_crtc_state;
 struct intel_crtc;
 struct intel_display;
 struct intel_dsb;
+struct intel_plane_state;
 struct drm_property_blob;
+enum pipe;
 
 void intel_color_init_hooks(struct intel_display *display);
 int intel_color_init(struct intel_display *display);
@@ -40,5 +42,9 @@ bool intel_color_lut_equal(const struct intel_crtc_state *crtc_state,
 			   const struct drm_property_blob *blob2,
 			   bool is_pre_csc_lut);
 void intel_color_assert_luts(const struct intel_crtc_state *crtc_state);
-
+void intel_color_plane_program_pipeline(struct intel_dsb *dsb,
+					const struct intel_plane_state *plane_state);
+void intel_color_plane_commit_arm(struct intel_dsb *dsb,
+				  const struct intel_plane_state *plane_state);
+bool intel_color_crtc_has_3dlut(struct intel_display *display, enum pipe pipe);
 #endif /* __INTEL_COLOR_H__ */

drivers/gpu/drm/i915/display/intel_color_pipeline.c

@@ -0,0 +1,99 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright © 2025 Intel Corporation
+ */
+#include "intel_color.h"
+#include "intel_colorop.h"
+#include "intel_color_pipeline.h"
+#include "intel_de.h"
+#include "intel_display_types.h"
+#include "skl_universal_plane.h"
+
+#define MAX_COLOR_PIPELINES 1
+#define PLANE_DEGAMMA_SIZE 128
+#define PLANE_GAMMA_SIZE 32
+
+static
+int _intel_color_pipeline_plane_init(struct drm_plane *plane, struct drm_prop_enum_list *list,
+				     enum pipe pipe)
+{
+	struct drm_device *dev = plane->dev;
+	struct intel_display *display = to_intel_display(dev);
+	struct drm_colorop *prev_op;
+	struct intel_colorop *colorop;
+	int ret;
+
+	colorop = intel_colorop_create(INTEL_PLANE_CB_PRE_CSC_LUT);
+
+	ret = drm_plane_colorop_curve_1d_lut_init(dev, &colorop->base, plane,
+						  PLANE_DEGAMMA_SIZE,
+						  DRM_COLOROP_LUT1D_INTERPOLATION_LINEAR,
+						  DRM_COLOROP_FLAG_ALLOW_BYPASS);
+
+	if (ret)
+		return ret;
+
+	list->type = colorop->base.base.id;
+	list->name = kasprintf(GFP_KERNEL, "Color Pipeline %d", colorop->base.base.id);
+
+	/* TODO: handle failures and clean up */
+	prev_op = &colorop->base;
+
+	if (DISPLAY_VER(display) >= 35 &&
+	    intel_color_crtc_has_3dlut(display, pipe) &&
+	    plane->type == DRM_PLANE_TYPE_PRIMARY) {
+		colorop = intel_colorop_create(INTEL_PLANE_CB_3DLUT);
+
+		ret = drm_plane_colorop_3dlut_init(dev, &colorop->base, plane, 17,
+						   DRM_COLOROP_LUT3D_INTERPOLATION_TETRAHEDRAL,
+						   true);
+		if (ret)
+			return ret;
+
+		drm_colorop_set_next_property(prev_op, &colorop->base);
+
+		prev_op = &colorop->base;
+	}
+
+	colorop = intel_colorop_create(INTEL_PLANE_CB_CSC);
+	ret = drm_plane_colorop_ctm_3x4_init(dev, &colorop->base, plane,
+					     DRM_COLOROP_FLAG_ALLOW_BYPASS);
+	if (ret)
+		return ret;
+
+	drm_colorop_set_next_property(prev_op, &colorop->base);
+	prev_op = &colorop->base;
+
+	colorop = intel_colorop_create(INTEL_PLANE_CB_POST_CSC_LUT);
+	ret = drm_plane_colorop_curve_1d_lut_init(dev, &colorop->base, plane,
+						  PLANE_GAMMA_SIZE,
+						  DRM_COLOROP_LUT1D_INTERPOLATION_LINEAR,
+						  DRM_COLOROP_FLAG_ALLOW_BYPASS);
+	if (ret)
+		return ret;
+
+	drm_colorop_set_next_property(prev_op, &colorop->base);
+
+	return 0;
+}
+
+int intel_color_pipeline_plane_init(struct drm_plane *plane, enum pipe pipe)
+{
+	struct drm_device *dev = plane->dev;
+	struct intel_display *display = to_intel_display(dev);
+	struct drm_prop_enum_list pipelines[MAX_COLOR_PIPELINES];
+	int len = 0;
+	int ret;
+
+	/* Currently expose pipeline only for HDR planes */
+	if (!icl_is_hdr_plane(display, to_intel_plane(plane)->id))
+		return 0;
+
+	/* Add pipeline consisting of transfer functions */
+	ret = _intel_color_pipeline_plane_init(plane, &pipelines[len], pipe);
+	if (ret)
+		return ret;
+	len++;
+
+	return drm_plane_create_color_pipeline_property(plane, pipelines, len);
+}

drivers/gpu/drm/i915/display/intel_color_pipeline.h

@@ -0,0 +1,14 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2025 Intel Corporation
+ */
+
+#ifndef __INTEL_COLOR_PIPELINE_H__
+#define __INTEL_COLOR_PIPELINE_H__
+
+struct drm_plane;
+enum pipe;
+
+int intel_color_pipeline_plane_init(struct drm_plane *plane, enum pipe pipe);
+
+#endif /* __INTEL_COLOR_PIPELINE_H__ */

drivers/gpu/drm/i915/display/intel_color_regs.h

@@ -316,4 +316,33 @@
 #define   SKL_BOTTOM_COLOR_CSC_ENABLE		REG_BIT(30)
 #define SKL_BOTTOM_COLOR(pipe)		_MMIO_PIPE(pipe, _SKL_BOTTOM_COLOR_A, _SKL_BOTTOM_COLOR_B)
 
+/* 3D LUT */
+#define _LUT_3D_CTL_A		0x490A4
+#define _LUT_3D_CTL_B		0x491A4
+#define LUT_3D_CTL(pipe)	_MMIO_PIPE(pipe, _LUT_3D_CTL_A, _LUT_3D_CTL_B)
+#define   LUT_3D_ENABLE			REG_BIT(31)
+#define   LUT_3D_READY			REG_BIT(30)
+#define   LUT_3D_BINDING_MASK		REG_GENMASK(23, 22)
+#define   LUT_3D_BIND_PIPE		REG_FIELD_PREP(LUT_3D_BINDING_MASK, 0)
+#define   LUT_3D_BIND_PLANE_1		REG_FIELD_PREP(LUT_3D_BINDING_MASK, 1)
+#define   LUT_3D_BIND_PLANE_2		REG_FIELD_PREP(LUT_3D_BINDING_MASK, 2)
+#define   LUT_3D_BIND_PLANE_3		REG_FIELD_PREP(LUT_3D_BINDING_MASK, 3)
+
+#define _LUT_3D_INDEX_A		0x490A8
+#define _LUT_3D_INDEX_B		0x491A8
+#define LUT_3D_INDEX(pipe)	_MMIO_PIPE(pipe, _LUT_3D_INDEX_A, _LUT_3D_INDEX_B)
+#define   LUT_3D_AUTO_INCREMENT		REG_BIT(13)
+#define   LUT_3D_INDEX_VALUE_MASK	REG_GENMASK(12, 0)
+#define   LUT_3D_INDEX_VALUE(x)		REG_FIELD_PREP(LUT_3D_INDEX_VALUE_MASK, (x))
+
+#define _LUT_3D_DATA_A		0x490AC
+#define _LUT_3D_DATA_B		0x491AC
+#define LUT_3D_DATA(pipe)	_MMIO_PIPE(pipe, _LUT_3D_DATA_A, _LUT_3D_DATA_B)
+#define   LUT_3D_DATA_RED_MASK		REG_GENMASK(29, 20)
+#define   LUT_3D_DATA_GREEN_MASK	REG_GENMASK(19, 10)
+#define   LUT_3D_DATA_BLUE_MASK		REG_GENMASK(9, 0)
+#define   LUT_3D_DATA_RED(x)		REG_FIELD_PREP(LUT_3D_DATA_RED_MASK, (x))
+#define   LUT_3D_DATA_GREEN(x)		REG_FIELD_PREP(LUT_3D_DATA_GREEN_MASK, (x))
+#define   LUT_3D_DATA_BLUE(x)		REG_FIELD_PREP(LUT_3D_DATA_BLUE_MASK, (x))
+
 #endif /* __INTEL_COLOR_REGS_H__ */

drivers/gpu/drm/i915/display/intel_colorop.c

@@ -0,0 +1,35 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright © 2025 Intel Corporation
+ */
+#include "intel_colorop.h"
+
+struct intel_colorop *to_intel_colorop(struct drm_colorop *colorop)
+{
+	return container_of(colorop, struct intel_colorop, base);
+}
+
+struct intel_colorop *intel_colorop_alloc(void)
+{
+	struct intel_colorop *colorop;
+
+	colorop = kzalloc(sizeof(*colorop), GFP_KERNEL);
+	if (!colorop)
+		return ERR_PTR(-ENOMEM);
+
+	return colorop;
+}
+
+struct intel_colorop *intel_colorop_create(enum intel_color_block id)
+{
+	struct intel_colorop *colorop;
+
+	colorop = intel_colorop_alloc();
+
+	if (IS_ERR(colorop))
+		return colorop;
+
+	colorop->id = id;
+
+	return colorop;
+}

drivers/gpu/drm/i915/display/intel_colorop.h

@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2025 Intel Corporation
+ */
+
+#ifndef __INTEL_COLOROP_H__
+#define __INTEL_COLOROP_H__
+
+#include "intel_display_types.h"
+
+struct intel_colorop *to_intel_colorop(struct drm_colorop *colorop);
+struct intel_colorop *intel_colorop_alloc(void);
+struct intel_colorop *intel_colorop_create(enum intel_color_block id);
+
+#endif /* __INTEL_COLOROP_H__ */

drivers/gpu/drm/i915/display/intel_display.c

@@ -7304,6 +7304,7 @@ static void intel_atomic_dsb_finish(struct intel_atomic_state *state,
 	struct intel_display *display = to_intel_display(state);
 	struct intel_crtc_state *new_crtc_state =
 		intel_atomic_get_new_crtc_state(state, crtc);
+	unsigned int size = new_crtc_state->plane_color_changed ? 8192 : 1024;
 
 	if (!new_crtc_state->use_flipq &&
 	    !new_crtc_state->use_dsb &&
@@ -7314,10 +7315,12 @@ static void intel_atomic_dsb_finish(struct intel_atomic_state *state,
 	 * Rough estimate:
 	 * ~64 registers per each plane * 8 planes = 512
 	 * Double that for pipe stuff and other overhead.
+	 * ~4913 registers for 3DLUT
+	 * ~200 color registers * 3 HDR planes
 	 */
 	new_crtc_state->dsb_commit = intel_dsb_prepare(state, crtc, INTEL_DSB_0,
 						       new_crtc_state->use_dsb ||
-						       new_crtc_state->use_flipq ? 1024 : 16);
+						       new_crtc_state->use_flipq ? size : 16);
 	if (!new_crtc_state->dsb_commit) {
 		new_crtc_state->use_flipq = false;
 		new_crtc_state->use_dsb = false;

drivers/gpu/drm/i915/display/intel_display_limits.h

@@ -138,4 +138,13 @@ enum hpd_pin {
 	HPD_NUM_PINS
 };
 
+enum intel_color_block {
+	INTEL_PLANE_CB_PRE_CSC_LUT,
+	INTEL_PLANE_CB_CSC,
+	INTEL_PLANE_CB_POST_CSC_LUT,
+	INTEL_PLANE_CB_3DLUT,
+
+	INTEL_CB_MAX
+};
+
 #endif /* __INTEL_DISPLAY_LIMITS_H__ */

drivers/gpu/drm/i915/display/intel_display_types.h

@@ -646,6 +646,7 @@ struct intel_plane_state {
 		enum drm_color_encoding color_encoding;
 		enum drm_color_range color_range;
 		enum drm_scaling_filter scaling_filter;
+		struct drm_property_blob *ctm, *degamma_lut, *gamma_lut, *lut_3d;
 	} hw;
 
 	struct i915_vma *ggtt_vma;
@@ -1391,6 +1392,9 @@ struct intel_crtc_state {
 		u8 silence_period_sym_clocks;
 		u8 lfps_half_cycle_num_of_syms;
 	} alpm_state;
+
+	/* to track changes in plane color blocks */
+	bool plane_color_changed;
 };
 
 enum intel_pipe_crc_source {
@@ -1985,6 +1989,11 @@ struct intel_dp_mst_encoder {
 	struct intel_connector *connector;
 };
 
+struct intel_colorop {
+	struct drm_colorop base;
+	enum intel_color_block id;
+};
+
 static inline struct intel_encoder *
 intel_attached_encoder(struct intel_connector *connector)
 {

drivers/gpu/drm/i915/display/intel_plane.c

@@ -49,6 +49,7 @@
 #include "i9xx_plane_regs.h"
 #include "intel_cdclk.h"
 #include "intel_cursor.h"
+#include "intel_colorop.h"
 #include "intel_display_rps.h"
 #include "intel_display_trace.h"
 #include "intel_display_types.h"
@@ -336,6 +337,58 @@ intel_plane_copy_uapi_plane_damage(struct intel_plane_state *new_plane_state,
 		*damage = drm_plane_state_src(&new_uapi_plane_state->uapi);
 }
 
+static bool
+intel_plane_colorop_replace_blob(struct intel_plane_state *plane_state,
+				 struct intel_colorop *intel_colorop,
+				 struct drm_property_blob *blob)
+{
+	if (intel_colorop->id == INTEL_PLANE_CB_CSC)
+		return drm_property_replace_blob(&plane_state->hw.ctm, blob);
+	else if (intel_colorop->id == INTEL_PLANE_CB_PRE_CSC_LUT)
+		return	drm_property_replace_blob(&plane_state->hw.degamma_lut, blob);
+	else if (intel_colorop->id == INTEL_PLANE_CB_POST_CSC_LUT)
+		return drm_property_replace_blob(&plane_state->hw.gamma_lut, blob);
+	else if (intel_colorop->id == INTEL_PLANE_CB_3DLUT)
+		return	drm_property_replace_blob(&plane_state->hw.lut_3d, blob);
+
+	return false;
+}
+
+static void
+intel_plane_color_copy_uapi_to_hw_state(struct intel_plane_state *plane_state,
+					const struct intel_plane_state *from_plane_state,
+					struct intel_crtc *crtc)
+{
+	struct drm_colorop *iter_colorop, *colorop;
+	struct drm_colorop_state *new_colorop_state;
+	struct drm_atomic_state *state = plane_state->uapi.state;
+	struct intel_colorop *intel_colorop;
+	struct drm_property_blob *blob;
+	struct intel_atomic_state *intel_atomic_state = to_intel_atomic_state(state);
+	struct intel_crtc_state *new_crtc_state = intel_atomic_state ?
+		intel_atomic_get_new_crtc_state(intel_atomic_state, crtc) : NULL;
+	bool changed = false;
+	int i = 0;
+
+	iter_colorop = plane_state->uapi.color_pipeline;
+
+	while (iter_colorop) {
+		for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) {
+			if (new_colorop_state->colorop == iter_colorop) {
+				blob = new_colorop_state->bypass ? NULL : new_colorop_state->data;
+				intel_colorop = to_intel_colorop(colorop);
+				changed |= intel_plane_colorop_replace_blob(plane_state,
+									    intel_colorop,
+									    blob);
+			}
+		}
+		iter_colorop = iter_colorop->next;
+	}
+
+	if (new_crtc_state && changed)
+		new_crtc_state->plane_color_changed = true;
+}
+
 void intel_plane_copy_uapi_to_hw_state(struct intel_plane_state *plane_state,
 				       const struct intel_plane_state *from_plane_state,
 				       struct intel_crtc *crtc)
@@ -364,6 +417,8 @@ void intel_plane_copy_uapi_to_hw_state(struct intel_plane_state *plane_state,
 
 	plane_state->uapi.src = drm_plane_state_src(&from_plane_state->uapi);
 	plane_state->uapi.dst = drm_plane_state_dest(&from_plane_state->uapi);
+
+	intel_plane_color_copy_uapi_to_hw_state(plane_state, from_plane_state, crtc);
 }
 
 void intel_plane_copy_hw_state(struct intel_plane_state *plane_state,

drivers/gpu/drm/i915/display/skl_universal_plane.c

@@ -11,6 +11,8 @@
 
 #include "pxp/intel_pxp.h"
 #include "intel_bo.h"
+#include "intel_color.h"
+#include "intel_color_pipeline.h"
 #include "intel_de.h"
 #include "intel_display_irq.h"
 #include "intel_display_regs.h"
@@ -1275,6 +1277,18 @@ static u32 glk_plane_color_ctl(const struct intel_plane_state *plane_state)
 	if (plane_state->force_black)
 		plane_color_ctl |= PLANE_COLOR_PLANE_CSC_ENABLE;
 
+	if (plane_state->hw.degamma_lut)
+		plane_color_ctl |= PLANE_COLOR_PRE_CSC_GAMMA_ENABLE;
+
+	if (plane_state->hw.ctm)
+		plane_color_ctl |= PLANE_COLOR_PLANE_CSC_ENABLE;
+
+	if (plane_state->hw.gamma_lut) {
+		plane_color_ctl &= ~PLANE_COLOR_PLANE_GAMMA_DISABLE;
+		if (drm_color_lut32_size(plane_state->hw.gamma_lut) != 32)
+			plane_color_ctl |= PLANE_COLOR_POST_CSC_GAMMA_MULTSEG_ENABLE;
+	}
+
 	return plane_color_ctl;
 }
 
@@ -1556,6 +1570,8 @@ icl_plane_update_noarm(struct intel_dsb *dsb,
 	plane_color_ctl = plane_state->color_ctl |
 		glk_plane_color_ctl_crtc(crtc_state);
 
+	intel_color_plane_program_pipeline(dsb, plane_state);
+
 	/* The scaler will handle the output position */
 	if (plane_state->scaler_id >= 0) {
 		crtc_x = 0;
@@ -1657,6 +1673,8 @@ icl_plane_update_arm(struct intel_dsb *dsb,
 
 	icl_plane_update_sel_fetch_arm(dsb, plane, crtc_state, plane_state);
 
+	intel_color_plane_commit_arm(dsb, plane_state);
+
 	/*
 	 * In order to have FBC for fp16 formats pixel normalizer block must be
 	 * active. Check if pixel normalizer block need to be enabled for FBC.
@@ -3001,6 +3019,9 @@ skl_universal_plane_create(struct intel_display *display,
 					  DRM_COLOR_YCBCR_BT709,
 					  DRM_COLOR_YCBCR_LIMITED_RANGE);
 
+	if (DISPLAY_VER(display) >= 12)
+		intel_color_pipeline_plane_init(&plane->base, pipe);
+
 	drm_plane_create_alpha_property(&plane->base);
 	drm_plane_create_blend_mode_property(&plane->base,
 					     BIT(DRM_MODE_BLEND_PIXEL_NONE) |

drivers/gpu/drm/i915/display/skl_universal_plane_regs.h

@@ -254,6 +254,8 @@
 #define   PLANE_COLOR_PIPE_CSC_ENABLE			REG_BIT(23) /* Pre-ICL */
 #define   PLANE_COLOR_PLANE_CSC_ENABLE			REG_BIT(21) /* ICL+ */
 #define   PLANE_COLOR_INPUT_CSC_ENABLE			REG_BIT(20) /* ICL+ */
+#define   PLANE_COLOR_POST_CSC_GAMMA_MULTSEG_ENABLE	REG_BIT(15) /* TGL+ */
+#define   PLANE_COLOR_PRE_CSC_GAMMA_ENABLE		REG_BIT(14)
 #define   PLANE_COLOR_CSC_MODE_MASK			REG_GENMASK(19, 17)
 #define   PLANE_COLOR_CSC_MODE_BYPASS			REG_FIELD_PREP(PLANE_COLOR_CSC_MODE_MASK, 0)
 #define   PLANE_COLOR_CSC_MODE_YUV601_TO_RGB601		REG_FIELD_PREP(PLANE_COLOR_CSC_MODE_MASK, 1)
@@ -290,6 +292,119 @@
 									   _PLANE_INPUT_CSC_POSTOFF_HI_1_A, _PLANE_INPUT_CSC_POSTOFF_HI_1_B, \
 									   _PLANE_INPUT_CSC_POSTOFF_HI_2_A, _PLANE_INPUT_CSC_POSTOFF_HI_2_B)
 
+#define _MMIO_PLANE_GAMC(plane, i, a, b)  _MMIO(_PIPE(plane, a, b) + (i) * 4)
+
+#define _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_1_A	0x70160
+#define _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_1_B	0x71160
+#define _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_2_A	0x70260
+#define _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_2_B	0x71260
+#define _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_1(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_1_A, \
+								      _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_1_B)
+#define _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_2(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_2_A, \
+								      _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_2_B)
+#define PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_1(pipe), \
+										 _PLANE_POST_CSC_GAMC_SEG0_INDEX_ENH_2(pipe))
+
+#define _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_1_A	0x70164
+#define _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_1_B	0x71164
+#define _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_2_A	0x70264
+#define _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_2_B	0x71264
+#define _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_1(pipe)	_PIPE(pipe, _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_1_A, \
+							      _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_1_B)
+#define _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_2(pipe)	_PIPE(pipe, _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_2_A, \
+							      _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_2_B)
+#define PLANE_POST_CSC_GAMC_SEG0_DATA_ENH(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_1(pipe), \
+										 _PLANE_POST_CSC_GAMC_SEG0_DATA_ENH_2(pipe))
+
+#define _PLANE_POST_CSC_GAMC_INDEX_ENH_1_A	0x701d8
+#define _PLANE_POST_CSC_GAMC_INDEX_ENH_1_B	0x711d8
+#define _PLANE_POST_CSC_GAMC_INDEX_ENH_2_A	0x702d8
+#define _PLANE_POST_CSC_GAMC_INDEX_ENH_2_B	0x712d8
+#define _PLANE_POST_CSC_GAMC_INDEX_ENH_1(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_INDEX_ENH_1_A, \
+							      _PLANE_POST_CSC_GAMC_INDEX_ENH_1_B)
+#define _PLANE_POST_CSC_GAMC_INDEX_ENH_2(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_INDEX_ENH_2_A, \
+							      _PLANE_POST_CSC_GAMC_INDEX_ENH_2_B)
+#define PLANE_POST_CSC_GAMC_INDEX_ENH(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_POST_CSC_GAMC_INDEX_ENH_1(pipe), \
+									 _PLANE_POST_CSC_GAMC_INDEX_ENH_2(pipe))
+
+#define _PLANE_POST_CSC_GAMC_DATA_ENH_1_A	0x701dc
+#define _PLANE_POST_CSC_GAMC_DATA_ENH_1_B	0x711dc
+#define _PLANE_POST_CSC_GAMC_DATA_ENH_2_A	0x702dc
+#define _PLANE_POST_CSC_GAMC_DATA_ENH_2_B	0x712dc
+#define _PLANE_POST_CSC_GAMC_DATA_ENH_1(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_DATA_ENH_1_A, \
+							      _PLANE_POST_CSC_GAMC_DATA_ENH_1_B)
+#define _PLANE_POST_CSC_GAMC_DATA_ENH_2(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_DATA_ENH_2_A, \
+							      _PLANE_POST_CSC_GAMC_DATA_ENH_2_B)
+#define PLANE_POST_CSC_GAMC_DATA_ENH(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_POST_CSC_GAMC_DATA_ENH_1(pipe), \
+									 _PLANE_POST_CSC_GAMC_DATA_ENH_2(pipe))
+
+#define _PLANE_POST_CSC_GAMC_INDEX_1_A		0x704d8
+#define _PLANE_POST_CSC_GAMC_INDEX_1_B		0x714d8
+#define _PLANE_POST_CSC_GAMC_INDEX_2_A		0x705d8
+#define _PLANE_POST_CSC_GAMC_INDEX_2_B		0x715d8
+#define _PLANE_POST_CSC_GAMC_INDEX_1(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_INDEX_1_A, \
+							      _PLANE_POST_CSC_GAMC_INDEX_1_B)
+#define _PLANE_POST_CSC_GAMC_INDEX_2(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_INDEX_2_A, \
+							      _PLANE_POST_CSC_GAMC_INDEX_2_B)
+#define PLANE_POST_CSC_GAMC_INDEX(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_POST_CSC_GAMC_INDEX_1(pipe), \
+									 _PLANE_POST_CSC_GAMC_INDEX_2(pipe))
+
+#define _PLANE_POST_CSC_GAMC_DATA_1_A		0x704dc
+#define _PLANE_POST_CSC_GAMC_DATA_1_B		0x714dc
+#define _PLANE_POST_CSC_GAMC_DATA_2_A		0x705dc
+#define _PLANE_POST_CSC_GAMC_DATA_2_B		0x715dc
+#define _PLANE_POST_CSC_GAMC_DATA_1(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_DATA_1_A, \
+							      _PLANE_POST_CSC_GAMC_DATA_1_B)
+#define _PLANE_POST_CSC_GAMC_DATA_2(pipe)		_PIPE(pipe, _PLANE_POST_CSC_GAMC_DATA_2_A, \
+							      _PLANE_POST_CSC_GAMC_DATA_2_B)
+#define PLANE_POST_CSC_GAMC_DATA(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_POST_CSC_GAMC_DATA_1(pipe), \
+									 _PLANE_POST_CSC_GAMC_DATA_2(pipe))
+
+#define _PLANE_PRE_CSC_GAMC_INDEX_ENH_1_A	0x701d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_ENH_1_B	0x711d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_ENH_2_A	0x702d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_ENH_2_B	0x712d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_ENH_1(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_INDEX_ENH_1_A, \
+							      _PLANE_PRE_CSC_GAMC_INDEX_ENH_1_B)
+#define _PLANE_PRE_CSC_GAMC_INDEX_ENH_2(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_INDEX_ENH_2_A, \
+							      _PLANE_PRE_CSC_GAMC_INDEX_ENH_2_B)
+#define PLANE_PRE_CSC_GAMC_INDEX_ENH(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_PRE_CSC_GAMC_INDEX_ENH_1(pipe), \
+									 _PLANE_PRE_CSC_GAMC_INDEX_ENH_2(pipe))
+#define  PLANE_PAL_PREC_AUTO_INCREMENT          REG_BIT(10)
+
+#define _PLANE_PRE_CSC_GAMC_DATA_ENH_1_A	0x701d4
+#define _PLANE_PRE_CSC_GAMC_DATA_ENH_1_B	0x711d4
+#define _PLANE_PRE_CSC_GAMC_DATA_ENH_2_A	0x702d4
+#define _PLANE_PRE_CSC_GAMC_DATA_ENH_2_B	0x712d4
+#define _PLANE_PRE_CSC_GAMC_DATA_ENH_1(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_DATA_ENH_1_A, \
+							      _PLANE_PRE_CSC_GAMC_DATA_ENH_1_B)
+#define _PLANE_PRE_CSC_GAMC_DATA_ENH_2(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_DATA_ENH_2_A, \
+							      _PLANE_PRE_CSC_GAMC_DATA_ENH_2_B)
+#define PLANE_PRE_CSC_GAMC_DATA_ENH(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_PRE_CSC_GAMC_DATA_ENH_1(pipe), \
+									 _PLANE_PRE_CSC_GAMC_DATA_ENH_2(pipe))
+
+#define _PLANE_PRE_CSC_GAMC_INDEX_1_A		0x704d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_1_B		0x714d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_2_A		0x705d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_2_B		0x715d0
+#define _PLANE_PRE_CSC_GAMC_INDEX_1(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_INDEX_1_A, \
+							      _PLANE_PRE_CSC_GAMC_INDEX_1_B)
+#define _PLANE_PRE_CSC_GAMC_INDEX_2(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_INDEX_2_A, \
+							      _PLANE_PRE_CSC_GAMC_INDEX_2_B)
+#define PLANE_PRE_CSC_GAMC_INDEX(pipe, plane, i)	_MMIO_PLANE_GAMC(plane, i, _PLANE_PRE_CSC_GAMC_INDEX_1(pipe), \
+									 _PLANE_PRE_CSC_GAMC_INDEX_2(pipe))
+
+#define _PLANE_PRE_CSC_GAMC_DATA_1_A		0x704d4
+#define _PLANE_PRE_CSC_GAMC_DATA_1_B		0x714d4
+#define _PLANE_PRE_CSC_GAMC_DATA_2_A		0x705d4
+#define _PLANE_PRE_CSC_GAMC_DATA_2_B		0x715d4
+#define _PLANE_PRE_CSC_GAMC_DATA_1(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_DATA_1_A, \
+							      _PLANE_PRE_CSC_GAMC_DATA_1_B)
+#define _PLANE_PRE_CSC_GAMC_DATA_2(pipe)		_PIPE(pipe, _PLANE_PRE_CSC_GAMC_DATA_2_A, \
+							      _PLANE_PRE_CSC_GAMC_DATA_2_B)
+#define PLANE_PRE_CSC_GAMC_DATA(pipe, plane, i)		_MMIO_PLANE_GAMC(plane, i, _PLANE_PRE_CSC_GAMC_DATA_1(pipe), \
+									 _PLANE_PRE_CSC_GAMC_DATA_2(pipe))
+
 #define _PLANE_CSC_RY_GY_1_A			0x70210
 #define _PLANE_CSC_RY_GY_2_A			0x70310
 #define _PLANE_CSC_RY_GY_1_B			0x71210

drivers/gpu/drm/i915/gvt/kvmgt.c

@@ -1141,6 +1141,122 @@ static int intel_vgpu_set_irqs(struct intel_vgpu *vgpu, u32 flags,
 	return func(vgpu, index, start, count, flags, data);
 }
 
+static int intel_vgpu_ioctl_get_region_info(struct vfio_device *vfio_dev,
+					    struct vfio_region_info *info,
+					    struct vfio_info_cap *caps)
+{
+	struct vfio_region_info_cap_sparse_mmap *sparse = NULL;
+	struct intel_vgpu *vgpu = vfio_dev_to_vgpu(vfio_dev);
+	int nr_areas = 1;
+	int cap_type_id;
+	unsigned int i;
+	int ret;
+
+	switch (info->index) {
+	case VFIO_PCI_CONFIG_REGION_INDEX:
+		info->offset = VFIO_PCI_INDEX_TO_OFFSET(info->index);
+		info->size = vgpu->gvt->device_info.cfg_space_size;
+		info->flags = VFIO_REGION_INFO_FLAG_READ |
+			      VFIO_REGION_INFO_FLAG_WRITE;
+		break;
+	case VFIO_PCI_BAR0_REGION_INDEX:
+		info->offset = VFIO_PCI_INDEX_TO_OFFSET(info->index);
+		info->size = vgpu->cfg_space.bar[info->index].size;
+		if (!info->size) {
+			info->flags = 0;
+			break;
+		}
+
+		info->flags = VFIO_REGION_INFO_FLAG_READ |
+			      VFIO_REGION_INFO_FLAG_WRITE;
+		break;
+	case VFIO_PCI_BAR1_REGION_INDEX:
+		info->offset = VFIO_PCI_INDEX_TO_OFFSET(info->index);
+		info->size = 0;
+		info->flags = 0;
+		break;
+	case VFIO_PCI_BAR2_REGION_INDEX:
+		info->offset = VFIO_PCI_INDEX_TO_OFFSET(info->index);
+		info->flags = VFIO_REGION_INFO_FLAG_CAPS |
+			      VFIO_REGION_INFO_FLAG_MMAP |
+			      VFIO_REGION_INFO_FLAG_READ |
+			      VFIO_REGION_INFO_FLAG_WRITE;
+		info->size = gvt_aperture_sz(vgpu->gvt);
+
+		sparse = kzalloc(struct_size(sparse, areas, nr_areas),
+				 GFP_KERNEL);
+		if (!sparse)
+			return -ENOMEM;
+
+		sparse->header.id = VFIO_REGION_INFO_CAP_SPARSE_MMAP;
+		sparse->header.version = 1;
+		sparse->nr_areas = nr_areas;
+		cap_type_id = VFIO_REGION_INFO_CAP_SPARSE_MMAP;
+		sparse->areas[0].offset =
+			PAGE_ALIGN(vgpu_aperture_offset(vgpu));
+		sparse->areas[0].size = vgpu_aperture_sz(vgpu);
+		break;
+
+	case VFIO_PCI_BAR3_REGION_INDEX ... VFIO_PCI_BAR5_REGION_INDEX:
+		info->offset = VFIO_PCI_INDEX_TO_OFFSET(info->index);
+		info->size = 0;
+		info->flags = 0;
+
+		gvt_dbg_core("get region info bar:%d\n", info->index);
+		break;
+
+	case VFIO_PCI_ROM_REGION_INDEX:
+	case VFIO_PCI_VGA_REGION_INDEX:
+		info->offset = VFIO_PCI_INDEX_TO_OFFSET(info->index);
+		info->size = 0;
+		info->flags = 0;
+
+		gvt_dbg_core("get region info index:%d\n", info->index);
+		break;
+	default: {
+		struct vfio_region_info_cap_type cap_type = {
+			.header.id = VFIO_REGION_INFO_CAP_TYPE,
+			.header.version = 1
+		};
+
+		if (info->index >= VFIO_PCI_NUM_REGIONS + vgpu->num_regions)
+			return -EINVAL;
+		info->index = array_index_nospec(
+			info->index, VFIO_PCI_NUM_REGIONS + vgpu->num_regions);
+
+		i = info->index - VFIO_PCI_NUM_REGIONS;
+
+		info->offset = VFIO_PCI_INDEX_TO_OFFSET(info->index);
+		info->size = vgpu->region[i].size;
+		info->flags = vgpu->region[i].flags;
+
+		cap_type.type = vgpu->region[i].type;
+		cap_type.subtype = vgpu->region[i].subtype;
+
+		ret = vfio_info_add_capability(caps, &cap_type.header,
+					       sizeof(cap_type));
+		if (ret)
+			return ret;
+	}
+	}
+
+	if ((info->flags & VFIO_REGION_INFO_FLAG_CAPS) && sparse) {
+		ret = -EINVAL;
+		if (cap_type_id == VFIO_REGION_INFO_CAP_SPARSE_MMAP) {
+			ret = vfio_info_add_capability(
+				caps, &sparse->header,
+				struct_size(sparse, areas, sparse->nr_areas));
+		}
+		if (ret) {
+			kfree(sparse);
+			return ret;
+		}
+	}
+
+	kfree(sparse);
+	return 0;
+}
+
 static long intel_vgpu_ioctl(struct vfio_device *vfio_dev, unsigned int cmd,
 			     unsigned long arg)
 {
@@ -1169,152 +1285,6 @@ static long intel_vgpu_ioctl(struct vfio_device *vfio_dev, unsigned int cmd,
 		return copy_to_user((void __user *)arg, &info, minsz) ?
 			-EFAULT : 0;
 
-	} else if (cmd == VFIO_DEVICE_GET_REGION_INFO) {
-		struct vfio_region_info info;
-		struct vfio_info_cap caps = { .buf = NULL, .size = 0 };
-		unsigned int i;
-		int ret;
-		struct vfio_region_info_cap_sparse_mmap *sparse = NULL;
-		int nr_areas = 1;
-		int cap_type_id;
-
-		minsz = offsetofend(struct vfio_region_info, offset);
-
-		if (copy_from_user(&info, (void __user *)arg, minsz))
-			return -EFAULT;
-
-		if (info.argsz < minsz)
-			return -EINVAL;
-
-		switch (info.index) {
-		case VFIO_PCI_CONFIG_REGION_INDEX:
-			info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
-			info.size = vgpu->gvt->device_info.cfg_space_size;
-			info.flags = VFIO_REGION_INFO_FLAG_READ |
-				     VFIO_REGION_INFO_FLAG_WRITE;
-			break;
-		case VFIO_PCI_BAR0_REGION_INDEX:
-			info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
-			info.size = vgpu->cfg_space.bar[info.index].size;
-			if (!info.size) {
-				info.flags = 0;
-				break;
-			}
-
-			info.flags = VFIO_REGION_INFO_FLAG_READ |
-				     VFIO_REGION_INFO_FLAG_WRITE;
-			break;
-		case VFIO_PCI_BAR1_REGION_INDEX:
-			info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
-			info.size = 0;
-			info.flags = 0;
-			break;
-		case VFIO_PCI_BAR2_REGION_INDEX:
-			info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
-			info.flags = VFIO_REGION_INFO_FLAG_CAPS |
-					VFIO_REGION_INFO_FLAG_MMAP |
-					VFIO_REGION_INFO_FLAG_READ |
-					VFIO_REGION_INFO_FLAG_WRITE;
-			info.size = gvt_aperture_sz(vgpu->gvt);
-
-			sparse = kzalloc(struct_size(sparse, areas, nr_areas),
-					 GFP_KERNEL);
-			if (!sparse)
-				return -ENOMEM;
-
-			sparse->header.id = VFIO_REGION_INFO_CAP_SPARSE_MMAP;
-			sparse->header.version = 1;
-			sparse->nr_areas = nr_areas;
-			cap_type_id = VFIO_REGION_INFO_CAP_SPARSE_MMAP;
-			sparse->areas[0].offset =
-					PAGE_ALIGN(vgpu_aperture_offset(vgpu));
-			sparse->areas[0].size = vgpu_aperture_sz(vgpu);
-			break;
-
-		case VFIO_PCI_BAR3_REGION_INDEX ... VFIO_PCI_BAR5_REGION_INDEX:
-			info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
-			info.size = 0;
-			info.flags = 0;
-
-			gvt_dbg_core("get region info bar:%d\n", info.index);
-			break;
-
-		case VFIO_PCI_ROM_REGION_INDEX:
-		case VFIO_PCI_VGA_REGION_INDEX:
-			info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
-			info.size = 0;
-			info.flags = 0;
-
-			gvt_dbg_core("get region info index:%d\n", info.index);
-			break;
-		default:
-			{
-				struct vfio_region_info_cap_type cap_type = {
-					.header.id = VFIO_REGION_INFO_CAP_TYPE,
-					.header.version = 1 };
-
-				if (info.index >= VFIO_PCI_NUM_REGIONS +
-						vgpu->num_regions)
-					return -EINVAL;
-				info.index =
-					array_index_nospec(info.index,
-							VFIO_PCI_NUM_REGIONS +
-							vgpu->num_regions);
-
-				i = info.index - VFIO_PCI_NUM_REGIONS;
-
-				info.offset =
-					VFIO_PCI_INDEX_TO_OFFSET(info.index);
-				info.size = vgpu->region[i].size;
-				info.flags = vgpu->region[i].flags;
-
-				cap_type.type = vgpu->region[i].type;
-				cap_type.subtype = vgpu->region[i].subtype;
-
-				ret = vfio_info_add_capability(&caps,
-							&cap_type.header,
-							sizeof(cap_type));
-				if (ret)
-					return ret;
-			}
-		}
-
-		if ((info.flags & VFIO_REGION_INFO_FLAG_CAPS) && sparse) {
-			ret = -EINVAL;
-			if (cap_type_id == VFIO_REGION_INFO_CAP_SPARSE_MMAP)
-				ret = vfio_info_add_capability(&caps,
-					&sparse->header,
-					struct_size(sparse, areas,
-						    sparse->nr_areas));
-			if (ret) {
-				kfree(sparse);
-				return ret;
-			}
-		}
-
-		if (caps.size) {
-			info.flags |= VFIO_REGION_INFO_FLAG_CAPS;
-			if (info.argsz < sizeof(info) + caps.size) {
-				info.argsz = sizeof(info) + caps.size;
-				info.cap_offset = 0;
-			} else {
-				vfio_info_cap_shift(&caps, sizeof(info));
-				if (copy_to_user((void __user *)arg +
-						  sizeof(info), caps.buf,
-						  caps.size)) {
-					kfree(caps.buf);
-					kfree(sparse);
-					return -EFAULT;
-				}
-				info.cap_offset = sizeof(info);
-			}
-
-			kfree(caps.buf);
-		}
-
-		kfree(sparse);
-		return copy_to_user((void __user *)arg, &info, minsz) ?
-			-EFAULT : 0;
 	} else if (cmd == VFIO_DEVICE_GET_IRQ_INFO) {
 		struct vfio_irq_info info;
 
@@ -1477,6 +1447,7 @@ static const struct vfio_device_ops intel_vgpu_dev_ops = {
 	.write		= intel_vgpu_write,
 	.mmap		= intel_vgpu_mmap,
 	.ioctl		= intel_vgpu_ioctl,
+	.get_region_info_caps = intel_vgpu_ioctl_get_region_info,
 	.dma_unmap	= intel_vgpu_dma_unmap,
 	.bind_iommufd	= vfio_iommufd_emulated_bind,
 	.unbind_iommufd = vfio_iommufd_emulated_unbind,

drivers/gpu/drm/xe/Makefile

@@ -184,6 +184,10 @@ xe-$(CONFIG_PCI_IOV) += \
 	xe_sriov_pf_sysfs.o \
 	xe_tile_sriov_pf_debugfs.o
 
+ifdef CONFIG_XE_VFIO_PCI
+	xe-$(CONFIG_PCI_IOV) += xe_sriov_vfio.o
+endif
+
 # include helpers for tests even when XE is built-in
 ifdef CONFIG_DRM_XE_KUNIT_TEST
 xe-y += tests/xe_kunit_helpers.o
@@ -242,6 +246,8 @@ xe-$(CONFIG_DRM_XE_DISPLAY) += \
 	i915-display/intel_cdclk.o \
 	i915-display/intel_cmtg.o \
 	i915-display/intel_color.o \
+	i915-display/intel_colorop.o \
+	i915-display/intel_color_pipeline.o \
 	i915-display/intel_combo_phy.o \
 	i915-display/intel_connector.o \
 	i915-display/intel_crtc.o \

drivers/gpu/drm/xe/xe_gpu_scheduler.h

@@ -54,13 +54,14 @@ static inline void xe_sched_tdr_queue_imm(struct xe_gpu_scheduler *sched)
 static inline void xe_sched_resubmit_jobs(struct xe_gpu_scheduler *sched)
 {
 	struct drm_sched_job *s_job;
+	bool restore_replay = false;
 
 	list_for_each_entry(s_job, &sched->base.pending_list, list) {
 		struct drm_sched_fence *s_fence = s_job->s_fence;
 		struct dma_fence *hw_fence = s_fence->parent;
 
-		if (to_xe_sched_job(s_job)->skip_emit ||
-		    (hw_fence && !dma_fence_is_signaled(hw_fence)))
+		restore_replay |= to_xe_sched_job(s_job)->restore_replay;
+		if (restore_replay || (hw_fence && !dma_fence_is_signaled(hw_fence)))
 			sched->base.ops->run_job(s_job);
 	}
 }

drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c

@@ -711,7 +711,7 @@ static u64 pf_profile_fair_ggtt(struct xe_gt *gt, unsigned int num_vfs)
 	if (num_vfs > 56)
 		return SZ_64M - SZ_8M;
 
-	return rounddown_pow_of_two(shareable / num_vfs);
+	return rounddown_pow_of_two(div_u64(shareable, num_vfs));
 }
 
 /**

drivers/gpu/drm/xe/xe_gt_sriov_pf_migration.c

@@ -17,6 +17,7 @@
 #include "xe_gt_sriov_pf_helpers.h"
 #include "xe_gt_sriov_pf_migration.h"
 #include "xe_gt_sriov_printk.h"
+#include "xe_guc.h"
 #include "xe_guc_buf.h"
 #include "xe_guc_ct.h"
 #include "xe_migrate.h"
@@ -1023,6 +1024,12 @@ static void action_ring_cleanup(void *arg)
 	ptr_ring_cleanup(r, destroy_pf_packet);
 }
 
+static void pf_gt_migration_check_support(struct xe_gt *gt)
+{
+	if (GUC_FIRMWARE_VER(&gt->uc.guc) < MAKE_GUC_VER(70, 54, 0))
+		xe_sriov_pf_migration_disable(gt_to_xe(gt), "requires GuC version >= 70.54.0");
+}
+
 /**
  * xe_gt_sriov_pf_migration_init() - Initialize support for VF migration.
  * @gt: the &xe_gt
@@ -1039,6 +1046,8 @@ int xe_gt_sriov_pf_migration_init(struct xe_gt *gt)
 
 	xe_gt_assert(gt, IS_SRIOV_PF(xe));
 
+	pf_gt_migration_check_support(gt);
+
 	if (!pf_migration_supported(gt))
 		return 0;
 

drivers/gpu/drm/xe/xe_guc_submit.c

@@ -822,7 +822,7 @@ static void submit_exec_queue(struct xe_exec_queue *q, struct xe_sched_job *job)
 
 	xe_gt_assert(guc_to_gt(guc), exec_queue_registered(q));
 
-	if (!job->skip_emit || job->last_replay) {
+	if (!job->restore_replay || job->last_replay) {
 		if (xe_exec_queue_is_parallel(q))
 			wq_item_append(q);
 		else
@@ -881,10 +881,10 @@ guc_exec_queue_run_job(struct drm_sched_job *drm_job)
 	if (!killed_or_banned_or_wedged && !xe_sched_job_is_error(job)) {
 		if (!exec_queue_registered(q))
 			register_exec_queue(q, GUC_CONTEXT_NORMAL);
-		if (!job->skip_emit)
+		if (!job->restore_replay)
 			q->ring_ops->emit_job(job);
 		submit_exec_queue(q, job);
-		job->skip_emit = false;
+		job->restore_replay = false;
 	}
 
 	/*
@@ -2112,6 +2112,18 @@ static void guc_exec_queue_revert_pending_state_change(struct xe_guc *guc,
 	q->guc->resume_time = 0;
 }
 
+static void lrc_parallel_clear(struct xe_lrc *lrc)
+{
+	struct xe_device *xe = gt_to_xe(lrc->gt);
+	struct iosys_map map = xe_lrc_parallel_map(lrc);
+	int i;
+
+	for (i = 0; i < WQ_SIZE / sizeof(u32); ++i)
+		parallel_write(xe, map, wq[i],
+			       FIELD_PREP(WQ_TYPE_MASK, WQ_TYPE_NOOP) |
+			       FIELD_PREP(WQ_LEN_MASK, 0));
+}
+
 /*
  * This function is quite complex but only real way to ensure no state is lost
  * during VF resume flows. The function scans the queue state, make adjustments
@@ -2135,8 +2147,8 @@ static void guc_exec_queue_pause(struct xe_guc *guc, struct xe_exec_queue *q)
 	guc_exec_queue_revert_pending_state_change(guc, q);
 
 	if (xe_exec_queue_is_parallel(q)) {
-		struct xe_device *xe = guc_to_xe(guc);
-		struct iosys_map map = xe_lrc_parallel_map(q->lrc[0]);
+		/* Pairs with WRITE_ONCE in __xe_exec_queue_init  */
+		struct xe_lrc *lrc = READ_ONCE(q->lrc[0]);
 
 		/*
 		 * NOP existing WQ commands that may contain stale GGTT
@@ -2144,14 +2156,14 @@ static void guc_exec_queue_pause(struct xe_guc *guc, struct xe_exec_queue *q)
 		 * seems to get confused if the WQ head/tail pointers are
 		 * adjusted.
 		 */
-		for (i = 0; i < WQ_SIZE / sizeof(u32); ++i)
-			parallel_write(xe, map, wq[i],
-				       FIELD_PREP(WQ_TYPE_MASK, WQ_TYPE_NOOP) |
-				       FIELD_PREP(WQ_LEN_MASK, 0));
+		if (lrc)
+			lrc_parallel_clear(lrc);
 	}
 
 	job = xe_sched_first_pending_job(sched);
 	if (job) {
+		job->restore_replay = true;
+
 		/*
 		 * Adjust software tail so jobs submitted overwrite previous
 		 * position in ring buffer with new GGTT addresses.
@@ -2241,17 +2253,18 @@ static void guc_exec_queue_unpause_prepare(struct xe_guc *guc,
 					   struct xe_exec_queue *q)
 {
 	struct xe_gpu_scheduler *sched = &q->guc->sched;
-	struct drm_sched_job *s_job;
 	struct xe_sched_job *job = NULL;
+	bool restore_replay = false;
 
-	list_for_each_entry(s_job, &sched->base.pending_list, list) {
-		job = to_xe_sched_job(s_job);
+	list_for_each_entry(job, &sched->base.pending_list, drm.list) {
+		restore_replay |= job->restore_replay;
+		if (restore_replay) {
+			xe_gt_dbg(guc_to_gt(guc), "Replay JOB - guc_id=%d, seqno=%d",
+				  q->guc->id, xe_sched_job_seqno(job));
 
-		xe_gt_dbg(guc_to_gt(guc), "Replay JOB - guc_id=%d, seqno=%d",
-			  q->guc->id, xe_sched_job_seqno(job));
-
-		q->ring_ops->emit_job(job);
-		job->skip_emit = true;
+			q->ring_ops->emit_job(job);
+			job->restore_replay = true;
+		}
 	}
 
 	if (job)

drivers/gpu/drm/xe/xe_pagefault.c

@@ -102,7 +102,6 @@ retry_userptr:
 
 	/* Lock VM and BOs dma-resv */
 	xe_validation_ctx_init(&ctx, &vm->xe->val, &exec, (struct xe_val_flags) {});
-	drm_exec_init(&exec, 0, 0);
 	drm_exec_until_all_locked(&exec) {
 		err = xe_pagefault_begin(&exec, vma, tile->mem.vram,
 					 needs_vram == 1);

drivers/gpu/drm/xe/xe_pci.c

@@ -1223,6 +1223,23 @@ static struct pci_driver xe_pci_driver = {
 #endif
 };
 
+/**
+ * xe_pci_to_pf_device() - Get PF &xe_device.
+ * @pdev: the VF &pci_dev device
+ *
+ * Return: pointer to PF &xe_device, NULL otherwise.
+ */
+struct xe_device *xe_pci_to_pf_device(struct pci_dev *pdev)
+{
+	struct drm_device *drm;
+
+	drm = pci_iov_get_pf_drvdata(pdev, &xe_pci_driver);
+	if (IS_ERR(drm))
+		return NULL;
+
+	return to_xe_device(drm);
+}
+
 int xe_register_pci_driver(void)
 {
 	return pci_register_driver(&xe_pci_driver);

drivers/gpu/drm/xe/xe_pci.h

@@ -6,7 +6,10 @@
 #ifndef _XE_PCI_H_
 #define _XE_PCI_H_
 
+struct pci_dev;
+
 int xe_register_pci_driver(void);
 void xe_unregister_pci_driver(void);
+struct xe_device *xe_pci_to_pf_device(struct pci_dev *pdev);
 
 #endif

drivers/gpu/drm/xe/xe_pm.c

@@ -726,6 +726,13 @@ static void xe_pm_runtime_lockdep_prime(void)
 /**
  * xe_pm_runtime_get - Get a runtime_pm reference and resume synchronously
  * @xe: xe device instance
+ *
+ * When possible, scope-based runtime PM (through guard(xe_pm_runtime)) is
+ * be preferred over direct usage of this function.  Manual get/put handling
+ * should only be used when the function contains goto-based logic which
+ * can break scope-based handling, or when the lifetime of the runtime PM
+ * reference does not match a specific scope (e.g., runtime PM obtained in one
+ * function and released in a different one).
  */
 void xe_pm_runtime_get(struct xe_device *xe)
 {
@@ -758,6 +765,13 @@ void xe_pm_runtime_put(struct xe_device *xe)
  * xe_pm_runtime_get_ioctl - Get a runtime_pm reference before ioctl
  * @xe: xe device instance
  *
+ * When possible, scope-based runtime PM (through
+ * ACQUIRE(xe_pm_runtime_ioctl, ...)) is be preferred over direct usage of this
+ * function.  Manual get/put handling should only be used when the function
+ * contains goto-based logic which can break scope-based handling, or when the
+ * lifetime of the runtime PM reference does not match a specific scope (e.g.,
+ * runtime PM obtained in one function and released in a different one).
+ *
  * Returns: Any number greater than or equal to 0 for success, negative error
  * code otherwise.
  */
@@ -827,6 +841,13 @@ static bool xe_pm_suspending_or_resuming(struct xe_device *xe)
  * It will warn if not protected.
  * The reference should be put back after this function regardless, since it
  * will always bump the usage counter, regardless.
+ *
+ * When possible, scope-based runtime PM (through guard(xe_pm_runtime_noresume))
+ * is be preferred over direct usage of this function.  Manual get/put handling
+ * should only be used when the function contains goto-based logic which can
+ * break scope-based handling, or when the lifetime of the runtime PM reference
+ * does not match a specific scope (e.g., runtime PM obtained in one function
+ * and released in a different one).
  */
 void xe_pm_runtime_get_noresume(struct xe_device *xe)
 {

drivers/gpu/drm/xe/xe_pm.h

@@ -6,6 +6,7 @@
 #ifndef _XE_PM_H_
 #define _XE_PM_H_
 
+#include <linux/cleanup.h>
 #include <linux/pm_runtime.h>
 
 #define DEFAULT_VRAM_THRESHOLD 300 /* in MB */
@@ -37,4 +38,20 @@ int xe_pm_block_on_suspend(struct xe_device *xe);
 void xe_pm_might_block_on_suspend(void);
 int xe_pm_module_init(void);
 
+static inline void __xe_pm_runtime_noop(struct xe_device *xe) {}
+
+DEFINE_GUARD(xe_pm_runtime, struct xe_device *,
+	     xe_pm_runtime_get(_T), xe_pm_runtime_put(_T))
+DEFINE_GUARD(xe_pm_runtime_noresume, struct xe_device *,
+	     xe_pm_runtime_get_noresume(_T), xe_pm_runtime_put(_T))
+DEFINE_GUARD_COND(xe_pm_runtime, _ioctl, xe_pm_runtime_get_ioctl(_T), _RET >= 0)
+
+/*
+ * Used when a function needs to release runtime PM in all possible cases
+ * and error paths, but the wakeref was already acquired by a different
+ * function (i.e., get() has already happened so only a put() is needed).
+ */
+DEFINE_GUARD(xe_pm_runtime_release_only, struct xe_device *,
+	     __xe_pm_runtime_noop(_T), xe_pm_runtime_put(_T));
+
 #endif

drivers/gpu/drm/xe/xe_sched_job_types.h

@@ -63,8 +63,8 @@ struct xe_sched_job {
 	bool ring_ops_flush_tlb;
 	/** @ggtt: mapped in ggtt. */
 	bool ggtt;
-	/** @skip_emit: skip emitting the job */
-	bool skip_emit;
+	/** @restore_replay: job being replayed for restore */
+	bool restore_replay;
 	/** @last_replay: last job being replayed */
 	bool last_replay;
 	/** @ptrs: per instance pointers. */

drivers/gpu/drm/xe/xe_sriov_pf_migration.c

@@ -46,13 +46,37 @@ bool xe_sriov_pf_migration_supported(struct xe_device *xe)
 {
 	xe_assert(xe, IS_SRIOV_PF(xe));
 
-	return xe->sriov.pf.migration.supported;
+	return IS_ENABLED(CONFIG_DRM_XE_DEBUG) || !xe->sriov.pf.migration.disabled;
 }
 
-static bool pf_check_migration_support(struct xe_device *xe)
+/**
+ * xe_sriov_pf_migration_disable() - Turn off SR-IOV VF migration support on PF.
+ * @xe: the &xe_device instance.
+ * @fmt: format string for the log message, to be combined with following VAs.
+ */
+void xe_sriov_pf_migration_disable(struct xe_device *xe, const char *fmt, ...)
 {
-	/* XXX: for now this is for feature enabling only */
-	return IS_ENABLED(CONFIG_DRM_XE_DEBUG);
+	struct va_format vaf;
+	va_list va_args;
+
+	xe_assert(xe, IS_SRIOV_PF(xe));
+
+	va_start(va_args, fmt);
+	vaf.fmt = fmt;
+	vaf.va  = &va_args;
+	xe_sriov_notice(xe, "migration %s: %pV\n",
+			IS_ENABLED(CONFIG_DRM_XE_DEBUG) ?
+			"missing prerequisite" : "disabled",
+			&vaf);
+	va_end(va_args);
+
+	xe->sriov.pf.migration.disabled = true;
+}
+
+static void pf_migration_check_support(struct xe_device *xe)
+{
+	if (!xe_device_has_memirq(xe))
+		xe_sriov_pf_migration_disable(xe, "requires memory-based IRQ support");
 }
 
 static void pf_migration_cleanup(void *arg)
@@ -77,7 +101,8 @@ int xe_sriov_pf_migration_init(struct xe_device *xe)
 
 	xe_assert(xe, IS_SRIOV_PF(xe));
 
-	xe->sriov.pf.migration.supported = pf_check_migration_support(xe);
+	pf_migration_check_support(xe);
+
 	if (!xe_sriov_pf_migration_supported(xe))
 		return 0;
 

drivers/gpu/drm/xe/xe_sriov_pf_migration.h

@@ -14,6 +14,7 @@ struct xe_sriov_packet;
 
 int xe_sriov_pf_migration_init(struct xe_device *xe);
 bool xe_sriov_pf_migration_supported(struct xe_device *xe);
+void xe_sriov_pf_migration_disable(struct xe_device *xe, const char *fmt, ...);
 int xe_sriov_pf_migration_restore_produce(struct xe_device *xe, unsigned int vfid,
 					  struct xe_sriov_packet *data);
 struct xe_sriov_packet *

drivers/gpu/drm/xe/xe_sriov_pf_migration_types.h

@@ -14,8 +14,8 @@
  * struct xe_sriov_pf_migration - Xe device level VF migration data
  */
 struct xe_sriov_pf_migration {
-	/** @supported: indicates whether VF migration feature is supported */
-	bool supported;
+	/** @disabled: indicates whether VF migration feature is disabled */
+	bool disabled;
 };
 
 /**

drivers/gpu/drm/xe/xe_sriov_vfio.c

@@ -0,0 +1,80 @@
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright © 2025 Intel Corporation
+ */
+
+#include <drm/intel/xe_sriov_vfio.h>
+#include <linux/cleanup.h>
+
+#include "xe_pci.h"
+#include "xe_pm.h"
+#include "xe_sriov_pf_control.h"
+#include "xe_sriov_pf_helpers.h"
+#include "xe_sriov_pf_migration.h"
+
+struct xe_device *xe_sriov_vfio_get_pf(struct pci_dev *pdev)
+{
+	return xe_pci_to_pf_device(pdev);
+}
+EXPORT_SYMBOL_FOR_MODULES(xe_sriov_vfio_get_pf, "xe-vfio-pci");
+
+bool xe_sriov_vfio_migration_supported(struct xe_device *xe)
+{
+	if (!IS_SRIOV_PF(xe))
+		return -EPERM;
+
+	return xe_sriov_pf_migration_supported(xe);
+}
+EXPORT_SYMBOL_FOR_MODULES(xe_sriov_vfio_migration_supported, "xe-vfio-pci");
+
+#define DEFINE_XE_SRIOV_VFIO_FUNCTION(_type, _func, _impl)			\
+_type xe_sriov_vfio_##_func(struct xe_device *xe, unsigned int vfid)		\
+{										\
+	if (!IS_SRIOV_PF(xe))							\
+		return -EPERM;							\
+	if (vfid == PFID || vfid > xe_sriov_pf_num_vfs(xe))			\
+		return -EINVAL;							\
+										\
+	guard(xe_pm_runtime_noresume)(xe);					\
+										\
+	return xe_sriov_pf_##_impl(xe, vfid);					\
+}										\
+EXPORT_SYMBOL_FOR_MODULES(xe_sriov_vfio_##_func, "xe-vfio-pci")
+
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, wait_flr_done, control_wait_flr);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, suspend_device, control_pause_vf);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, resume_device, control_resume_vf);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, stop_copy_enter, control_trigger_save_vf);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, stop_copy_exit, control_finish_save_vf);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, resume_data_enter, control_trigger_restore_vf);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, resume_data_exit, control_finish_restore_vf);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(int, error, control_stop_vf);
+DEFINE_XE_SRIOV_VFIO_FUNCTION(ssize_t, stop_copy_size, migration_size);
+
+ssize_t xe_sriov_vfio_data_read(struct xe_device *xe, unsigned int vfid,
+				char __user *buf, size_t len)
+{
+	if (!IS_SRIOV_PF(xe))
+		return -EPERM;
+	if (vfid == PFID || vfid > xe_sriov_pf_num_vfs(xe))
+		return -EINVAL;
+
+	guard(xe_pm_runtime_noresume)(xe);
+
+	return xe_sriov_pf_migration_read(xe, vfid, buf, len);
+}
+EXPORT_SYMBOL_FOR_MODULES(xe_sriov_vfio_data_read, "xe-vfio-pci");
+
+ssize_t xe_sriov_vfio_data_write(struct xe_device *xe, unsigned int vfid,
+				 const char __user *buf, size_t len)
+{
+	if (!IS_SRIOV_PF(xe))
+		return -EPERM;
+	if (vfid == PFID || vfid > xe_sriov_pf_num_vfs(xe))
+		return -EINVAL;
+
+	guard(xe_pm_runtime_noresume)(xe);
+
+	return xe_sriov_pf_migration_write(xe, vfid, buf, len);
+}
+EXPORT_SYMBOL_FOR_MODULES(xe_sriov_vfio_data_write, "xe-vfio-pci");

drivers/infiniband/Kconfig

@@ -80,6 +80,7 @@ config INFINIBAND_VIRT_DMA
 if INFINIBAND_USER_ACCESS || !INFINIBAND_USER_ACCESS
 if !UML
 source "drivers/infiniband/hw/bnxt_re/Kconfig"
+source "drivers/infiniband/hw/bng_re/Kconfig"
 source "drivers/infiniband/hw/cxgb4/Kconfig"
 source "drivers/infiniband/hw/efa/Kconfig"
 source "drivers/infiniband/hw/erdma/Kconfig"

drivers/infiniband/core/cm.c

@@ -34,7 +34,6 @@ MODULE_AUTHOR("Sean Hefty");
 MODULE_DESCRIPTION("InfiniBand CM");
 MODULE_LICENSE("Dual BSD/GPL");
 
-#define CM_DESTROY_ID_WAIT_TIMEOUT 10000 /* msecs */
 #define CM_DIRECT_RETRY_CTX ((void *) 1UL)
 #define CM_MRA_SETTING 24 /* 4.096us * 2^24 = ~68.7 seconds */
 
@@ -1057,6 +1056,7 @@ static void cm_destroy_id(struct ib_cm_id *cm_id, int err)
 {
 	struct cm_id_private *cm_id_priv;
 	enum ib_cm_state old_state;
+	unsigned long timeout;
 	struct cm_work *work;
 	int ret;
 
@@ -1167,10 +1167,9 @@ retest:
 
 	xa_erase(&cm.local_id_table, cm_local_id(cm_id->local_id));
 	cm_deref_id(cm_id_priv);
+	timeout = msecs_to_jiffies((cm_id_priv->max_cm_retries * cm_id_priv->timeout_ms * 5) / 4);
 	do {
-		ret = wait_for_completion_timeout(&cm_id_priv->comp,
-						  msecs_to_jiffies(
-						  CM_DESTROY_ID_WAIT_TIMEOUT));
+		ret = wait_for_completion_timeout(&cm_id_priv->comp, timeout);
 		if (!ret) /* timeout happened */
 			cm_destroy_id_wait_timeout(cm_id, old_state);
 	} while (!ret);
@@ -4518,7 +4517,7 @@ static int __init ib_cm_init(void)
 	get_random_bytes(&cm.random_id_operand, sizeof cm.random_id_operand);
 	INIT_LIST_HEAD(&cm.timewait_list);
 
-	cm.wq = alloc_workqueue("ib_cm", 0, 1);
+	cm.wq = alloc_workqueue("ib_cm", WQ_PERCPU, 1);
 	if (!cm.wq) {
 		ret = -ENOMEM;
 		goto error2;

drivers/infiniband/core/cma.c

@@ -4475,6 +4475,8 @@ int rdma_connect_locked(struct rdma_cm_id *id,
 		container_of(id, struct rdma_id_private, id);
 	int ret;
 
+	lockdep_assert_held(&id_priv->handler_mutex);
+
 	if (!cma_comp_exch(id_priv, RDMA_CM_ROUTE_RESOLVED, RDMA_CM_CONNECT))
 		return -EINVAL;
 

drivers/infiniband/core/device.c

@@ -3021,7 +3021,7 @@ static int __init ib_core_init(void)
 {
 	int ret = -ENOMEM;
 
-	ib_wq = alloc_workqueue("infiniband", 0, 0);
+	ib_wq = alloc_workqueue("infiniband", WQ_PERCPU, 0);
 	if (!ib_wq)
 		return -ENOMEM;
 
@@ -3031,7 +3031,7 @@ static int __init ib_core_init(void)
 		goto err;
 
 	ib_comp_wq = alloc_workqueue("ib-comp-wq",
-			WQ_HIGHPRI | WQ_MEM_RECLAIM | WQ_SYSFS, 0);
+			WQ_HIGHPRI | WQ_MEM_RECLAIM | WQ_SYSFS | WQ_PERCPU, 0);
 	if (!ib_comp_wq)
 		goto err_unbound;
 

drivers/infiniband/core/restrack.c

@@ -175,7 +175,7 @@ void rdma_restrack_new(struct rdma_restrack_entry *res,
 EXPORT_SYMBOL(rdma_restrack_new);
 
 /**
- * rdma_restrack_add() - add object to the reource tracking database
+ * rdma_restrack_add() - add object to the resource tracking database
  * @res:  resource entry
  */
 void rdma_restrack_add(struct rdma_restrack_entry *res)
@@ -277,7 +277,7 @@ int rdma_restrack_put(struct rdma_restrack_entry *res)
 EXPORT_SYMBOL(rdma_restrack_put);
 
 /**
- * rdma_restrack_del() - delete object from the reource tracking database
+ * rdma_restrack_del() - delete object from the resource tracking database
  * @res:  resource entry
  */
 void rdma_restrack_del(struct rdma_restrack_entry *res)

drivers/infiniband/core/ucma.c

@@ -366,7 +366,7 @@ static int ucma_event_handler(struct rdma_cm_id *cm_id,
 	if (event->event == RDMA_CM_EVENT_DEVICE_REMOVAL) {
 		xa_lock(&ctx_table);
 		if (xa_load(&ctx_table, ctx->id) == ctx)
-			queue_work(system_unbound_wq, &ctx->close_work);
+			queue_work(system_dfl_wq, &ctx->close_work);
 		xa_unlock(&ctx_table);
 	}
 	return 0;

drivers/infiniband/core/umem.c

@@ -45,6 +45,8 @@
 
 #include "uverbs.h"
 
+#define RESCHED_LOOP_CNT_THRESHOLD 0x1000
+
 static void __ib_umem_release(struct ib_device *dev, struct ib_umem *umem, int dirty)
 {
 	bool make_dirty = umem->writable && dirty;
@@ -55,10 +57,14 @@ static void __ib_umem_release(struct ib_device *dev, struct ib_umem *umem, int d
 		ib_dma_unmap_sgtable_attrs(dev, &umem->sgt_append.sgt,
 					   DMA_BIDIRECTIONAL, 0);
 
-	for_each_sgtable_sg(&umem->sgt_append.sgt, sg, i)
+	for_each_sgtable_sg(&umem->sgt_append.sgt, sg, i) {
 		unpin_user_page_range_dirty_lock(sg_page(sg),
 			DIV_ROUND_UP(sg->length, PAGE_SIZE), make_dirty);
 
+		if (i && !(i % RESCHED_LOOP_CNT_THRESHOLD))
+			cond_resched();
+	}
+
 	sg_free_append_table(&umem->sgt_append);
 }
 

drivers/infiniband/core/verbs.c

@@ -148,6 +148,7 @@ __attribute_const__ int ib_rate_to_mult(enum ib_rate rate)
 	case IB_RATE_400_GBPS: return 160;
 	case IB_RATE_600_GBPS: return 240;
 	case IB_RATE_800_GBPS: return 320;
+	case IB_RATE_1600_GBPS: return 640;
 	default:	       return  -1;
 	}
 }
@@ -178,6 +179,7 @@ __attribute_const__ enum ib_rate mult_to_ib_rate(int mult)
 	case 160: return IB_RATE_400_GBPS;
 	case 240: return IB_RATE_600_GBPS;
 	case 320: return IB_RATE_800_GBPS;
+	case 640: return IB_RATE_1600_GBPS;
 	default:  return IB_RATE_PORT_CURRENT;
 	}
 }
@@ -208,6 +210,7 @@ __attribute_const__ int ib_rate_to_mbps(enum ib_rate rate)
 	case IB_RATE_400_GBPS: return 425000;
 	case IB_RATE_600_GBPS: return 637500;
 	case IB_RATE_800_GBPS: return 850000;
+	case IB_RATE_1600_GBPS: return 1700000;
 	default:	       return -1;
 	}
 }

drivers/infiniband/hw/Makefile

@@ -13,5 +13,6 @@ obj-$(CONFIG_INFINIBAND_HFI1)		+= hfi1/
 obj-$(CONFIG_INFINIBAND_HNS_HIP08)	+= hns/
 obj-$(CONFIG_INFINIBAND_QEDR)		+= qedr/
 obj-$(CONFIG_INFINIBAND_BNXT_RE)	+= bnxt_re/
+obj-$(CONFIG_INFINIBAND_BNG_RE)		+= bng_re/
 obj-$(CONFIG_INFINIBAND_ERDMA)		+= erdma/
 obj-$(CONFIG_INFINIBAND_IONIC)		+= ionic/

drivers/infiniband/hw/bng_re/Kconfig

@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0-only
+config INFINIBAND_BNG_RE
+	tristate "Broadcom Next generation RoCE HCA support"
+	depends on 64BIT
+	depends on INET && DCB && BNGE
+	help
+	  This driver supports Broadcom Next generation
+	  50/100/200/400/800 gigabit RoCE HCAs. The module
+	  will be called bng_re. To compile this driver
+	  as a module, choose M here.

drivers/infiniband/hw/bng_re/Makefile

@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: GPL-2.0
+ccflags-y := -I $(srctree)/drivers/net/ethernet/broadcom/bnge -I $(srctree)/drivers/infiniband/hw/bnxt_re
+
+obj-$(CONFIG_INFINIBAND_BNG_RE) += bng_re.o
+
+bng_re-y := bng_dev.o bng_fw.o \
+	    bng_res.o bng_sp.o \
+	    bng_debugfs.o

drivers/infiniband/hw/bng_re/bng_debugfs.c

@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2025 Broadcom.
+#include <linux/debugfs.h>
+#include <linux/pci.h>
+
+#include <rdma/ib_verbs.h>
+
+#include "bng_res.h"
+#include "bng_fw.h"
+#include "bnge.h"
+#include "bnge_auxr.h"
+#include "bng_re.h"
+#include "bng_debugfs.h"
+
+static struct dentry *bng_re_debugfs_root;
+
+void bng_re_debugfs_add_pdev(struct bng_re_dev *rdev)
+{
+	struct pci_dev *pdev = rdev->aux_dev->pdev;
+
+	rdev->dbg_root =
+		debugfs_create_dir(dev_name(&pdev->dev), bng_re_debugfs_root);
+}
+
+void bng_re_debugfs_rem_pdev(struct bng_re_dev *rdev)
+{
+	debugfs_remove_recursive(rdev->dbg_root);
+	rdev->dbg_root = NULL;
+}
+
+void bng_re_register_debugfs(void)
+{
+	bng_re_debugfs_root = debugfs_create_dir("bng_re", NULL);
+}
+
+void bng_re_unregister_debugfs(void)
+{
+	debugfs_remove(bng_re_debugfs_root);
+}

drivers/infiniband/hw/bng_re/bng_debugfs.h

@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+// Copyright (c) 2025 Broadcom.
+
+#ifndef __BNG_RE_DEBUGFS__
+#define __BNG_RE_DEBUGFS__
+
+void bng_re_debugfs_add_pdev(struct bng_re_dev *rdev);
+void bng_re_debugfs_rem_pdev(struct bng_re_dev *rdev);
+
+void bng_re_register_debugfs(void);
+void bng_re_unregister_debugfs(void);
+#endif

drivers/infiniband/hw/bng_re/bng_dev.c

@@ -0,0 +1,534 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2025 Broadcom.
+
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/auxiliary_bus.h>
+
+#include <rdma/ib_verbs.h>
+
+#include "bng_res.h"
+#include "bng_sp.h"
+#include "bng_fw.h"
+#include "bnge.h"
+#include "bnge_auxr.h"
+#include "bng_re.h"
+#include "bnge_hwrm.h"
+#include "bng_debugfs.h"
+
+MODULE_AUTHOR("Siva Reddy Kallam <siva.kallam@broadcom.com>");
+MODULE_DESCRIPTION(BNG_RE_DESC);
+MODULE_LICENSE("Dual BSD/GPL");
+
+static struct bng_re_dev *bng_re_dev_add(struct auxiliary_device *adev,
+					 struct bnge_auxr_dev *aux_dev)
+{
+	struct bng_re_dev *rdev;
+
+	/* Allocate bng_re_dev instance */
+	rdev = ib_alloc_device(bng_re_dev, ibdev);
+	if (!rdev) {
+		pr_err("%s: bng_re_dev allocation failure!", KBUILD_MODNAME);
+		return NULL;
+	}
+
+	/* Assign auxiliary device specific data */
+	rdev->netdev = aux_dev->net;
+	rdev->aux_dev = aux_dev;
+	rdev->adev = adev;
+	rdev->fn_id = rdev->aux_dev->pdev->devfn;
+
+	return rdev;
+}
+
+
+static int bng_re_register_netdev(struct bng_re_dev *rdev)
+{
+	struct bnge_auxr_dev *aux_dev;
+
+	aux_dev = rdev->aux_dev;
+	return bnge_register_dev(aux_dev, rdev->adev);
+}
+
+static void bng_re_destroy_chip_ctx(struct bng_re_dev *rdev)
+{
+	struct bng_re_chip_ctx *chip_ctx;
+
+	if (!rdev->chip_ctx)
+		return;
+
+	kfree(rdev->dev_attr);
+	rdev->dev_attr = NULL;
+
+	chip_ctx = rdev->chip_ctx;
+	rdev->chip_ctx = NULL;
+	rdev->rcfw.res = NULL;
+	rdev->bng_res.cctx = NULL;
+	rdev->bng_res.pdev = NULL;
+	kfree(chip_ctx);
+}
+
+static int bng_re_setup_chip_ctx(struct bng_re_dev *rdev)
+{
+	struct bng_re_chip_ctx *chip_ctx;
+	struct bnge_auxr_dev *aux_dev;
+	int rc = -ENOMEM;
+
+	aux_dev = rdev->aux_dev;
+	rdev->bng_res.pdev = aux_dev->pdev;
+	rdev->rcfw.res = &rdev->bng_res;
+	chip_ctx = kzalloc(sizeof(*chip_ctx), GFP_KERNEL);
+	if (!chip_ctx)
+		return -ENOMEM;
+	chip_ctx->chip_num = aux_dev->chip_num;
+	chip_ctx->hw_stats_size = aux_dev->hw_ring_stats_size;
+
+	rdev->chip_ctx = chip_ctx;
+	rdev->bng_res.cctx = rdev->chip_ctx;
+	rdev->dev_attr = kzalloc(sizeof(*rdev->dev_attr), GFP_KERNEL);
+	if (!rdev->dev_attr)
+		goto free_chip_ctx;
+	rdev->bng_res.dattr = rdev->dev_attr;
+
+	return 0;
+free_chip_ctx:
+	kfree(rdev->chip_ctx);
+	rdev->chip_ctx = NULL;
+	return rc;
+}
+
+static void bng_re_init_hwrm_hdr(struct input *hdr, u16 opcd)
+{
+	hdr->req_type = cpu_to_le16(opcd);
+	hdr->cmpl_ring = cpu_to_le16(-1);
+	hdr->target_id = cpu_to_le16(-1);
+}
+
+static void bng_re_fill_fw_msg(struct bnge_fw_msg *fw_msg, void *msg,
+			       int msg_len, void *resp, int resp_max_len,
+			       int timeout)
+{
+	fw_msg->msg = msg;
+	fw_msg->msg_len = msg_len;
+	fw_msg->resp = resp;
+	fw_msg->resp_max_len = resp_max_len;
+	fw_msg->timeout = timeout;
+}
+
+static int bng_re_net_ring_free(struct bng_re_dev *rdev,
+				u16 fw_ring_id, int type)
+{
+	struct bnge_auxr_dev *aux_dev = rdev->aux_dev;
+	struct hwrm_ring_free_input req = {};
+	struct hwrm_ring_free_output resp;
+	struct bnge_fw_msg fw_msg = {};
+	int rc = -EINVAL;
+
+	if (!rdev)
+		return rc;
+
+	if (!aux_dev)
+		return rc;
+
+	bng_re_init_hwrm_hdr((void *)&req, HWRM_RING_FREE);
+	req.ring_type = type;
+	req.ring_id = cpu_to_le16(fw_ring_id);
+	bng_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp,
+			    sizeof(resp), BNGE_DFLT_HWRM_CMD_TIMEOUT);
+	rc = bnge_send_msg(aux_dev, &fw_msg);
+	if (rc)
+		ibdev_err(&rdev->ibdev, "Failed to free HW ring:%d :%#x",
+			  req.ring_id, rc);
+	return rc;
+}
+
+static int bng_re_net_ring_alloc(struct bng_re_dev *rdev,
+				 struct bng_re_ring_attr *ring_attr,
+				 u16 *fw_ring_id)
+{
+	struct bnge_auxr_dev *aux_dev = rdev->aux_dev;
+	struct hwrm_ring_alloc_input req = {};
+	struct hwrm_ring_alloc_output resp;
+	struct bnge_fw_msg fw_msg = {};
+	int rc = -EINVAL;
+
+	if (!aux_dev)
+		return rc;
+
+	bng_re_init_hwrm_hdr((void *)&req, HWRM_RING_ALLOC);
+	req.enables = 0;
+	req.page_tbl_addr =  cpu_to_le64(ring_attr->dma_arr[0]);
+	if (ring_attr->pages > 1) {
+		/* Page size is in log2 units */
+		req.page_size = BNGE_PAGE_SHIFT;
+		req.page_tbl_depth = 1;
+	}
+	req.fbo = 0;
+	/* Association of ring index with doorbell index and MSIX number */
+	req.logical_id = cpu_to_le16(ring_attr->lrid);
+	req.length = cpu_to_le32(ring_attr->depth + 1);
+	req.ring_type = ring_attr->type;
+	req.int_mode = ring_attr->mode;
+	bng_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp,
+			   sizeof(resp), BNGE_DFLT_HWRM_CMD_TIMEOUT);
+	rc = bnge_send_msg(aux_dev, &fw_msg);
+	if (!rc)
+		*fw_ring_id = le16_to_cpu(resp.ring_id);
+
+	return rc;
+}
+
+static int bng_re_stats_ctx_free(struct bng_re_dev *rdev)
+{
+	struct bnge_auxr_dev *aux_dev = rdev->aux_dev;
+	struct hwrm_stat_ctx_free_input req = {};
+	struct hwrm_stat_ctx_free_output resp = {};
+	struct bnge_fw_msg fw_msg = {};
+	int rc = -EINVAL;
+
+	if (!aux_dev)
+		return rc;
+
+	bng_re_init_hwrm_hdr((void *)&req, HWRM_STAT_CTX_FREE);
+	req.stat_ctx_id = cpu_to_le32(rdev->stats_ctx.fw_id);
+	bng_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp,
+			   sizeof(resp), BNGE_DFLT_HWRM_CMD_TIMEOUT);
+	rc = bnge_send_msg(aux_dev, &fw_msg);
+	if (rc)
+		ibdev_err(&rdev->ibdev, "Failed to free HW stats context %#x",
+			  rc);
+
+	return rc;
+}
+
+static int bng_re_stats_ctx_alloc(struct bng_re_dev *rdev)
+{
+	struct bnge_auxr_dev *aux_dev = rdev->aux_dev;
+	struct bng_re_stats *stats = &rdev->stats_ctx;
+	struct hwrm_stat_ctx_alloc_output resp = {};
+	struct hwrm_stat_ctx_alloc_input req = {};
+	struct bnge_fw_msg fw_msg = {};
+	int rc = -EINVAL;
+
+	stats->fw_id = BNGE_INVALID_STATS_CTX_ID;
+
+	if (!aux_dev)
+		return rc;
+
+	bng_re_init_hwrm_hdr((void *)&req, HWRM_STAT_CTX_ALLOC);
+	req.update_period_ms = cpu_to_le32(1000);
+	req.stats_dma_addr = cpu_to_le64(stats->dma_map);
+	req.stats_dma_length = cpu_to_le16(rdev->chip_ctx->hw_stats_size);
+	req.stat_ctx_flags = STAT_CTX_ALLOC_REQ_STAT_CTX_FLAGS_ROCE;
+	bng_re_fill_fw_msg(&fw_msg, (void *)&req, sizeof(req), (void *)&resp,
+			   sizeof(resp), BNGE_DFLT_HWRM_CMD_TIMEOUT);
+	rc = bnge_send_msg(aux_dev, &fw_msg);
+	if (!rc)
+		stats->fw_id = le32_to_cpu(resp.stat_ctx_id);
+	return rc;
+}
+
+static void bng_re_query_hwrm_version(struct bng_re_dev *rdev)
+{
+	struct bnge_auxr_dev *aux_dev = rdev->aux_dev;
+	struct hwrm_ver_get_output ver_get_resp = {};
+	struct hwrm_ver_get_input ver_get_req = {};
+	struct bng_re_chip_ctx *cctx;
+	struct bnge_fw_msg fw_msg = {};
+	int rc;
+
+	bng_re_init_hwrm_hdr((void *)&ver_get_req, HWRM_VER_GET);
+	ver_get_req.hwrm_intf_maj = HWRM_VERSION_MAJOR;
+	ver_get_req.hwrm_intf_min = HWRM_VERSION_MINOR;
+	ver_get_req.hwrm_intf_upd = HWRM_VERSION_UPDATE;
+	bng_re_fill_fw_msg(&fw_msg, (void *)&ver_get_req, sizeof(ver_get_req),
+			    (void *)&ver_get_resp, sizeof(ver_get_resp),
+			    BNGE_DFLT_HWRM_CMD_TIMEOUT);
+	rc = bnge_send_msg(aux_dev, &fw_msg);
+	if (rc) {
+		ibdev_err(&rdev->ibdev, "Failed to query HW version, rc = 0x%x",
+			  rc);
+		return;
+	}
+
+	cctx = rdev->chip_ctx;
+	cctx->hwrm_intf_ver =
+		(u64)le16_to_cpu(ver_get_resp.hwrm_intf_major) << 48 |
+		(u64)le16_to_cpu(ver_get_resp.hwrm_intf_minor) << 32 |
+		(u64)le16_to_cpu(ver_get_resp.hwrm_intf_build) << 16 |
+		le16_to_cpu(ver_get_resp.hwrm_intf_patch);
+
+	cctx->hwrm_cmd_max_timeout = le16_to_cpu(ver_get_resp.max_req_timeout);
+
+	if (!cctx->hwrm_cmd_max_timeout)
+		cctx->hwrm_cmd_max_timeout = BNG_ROCE_FW_MAX_TIMEOUT;
+}
+
+static void bng_re_dev_uninit(struct bng_re_dev *rdev)
+{
+	int rc;
+	bng_re_debugfs_rem_pdev(rdev);
+
+	if (test_and_clear_bit(BNG_RE_FLAG_RCFW_CHANNEL_EN, &rdev->flags)) {
+		rc = bng_re_deinit_rcfw(&rdev->rcfw);
+		if (rc)
+			ibdev_warn(&rdev->ibdev,
+				   "Failed to deinitialize RCFW: %#x", rc);
+		bng_re_stats_ctx_free(rdev);
+		bng_re_free_stats_ctx_mem(rdev->bng_res.pdev, &rdev->stats_ctx);
+		bng_re_disable_rcfw_channel(&rdev->rcfw);
+		bng_re_net_ring_free(rdev, rdev->rcfw.creq.ring_id,
+			     RING_ALLOC_REQ_RING_TYPE_NQ);
+		bng_re_free_rcfw_channel(&rdev->rcfw);
+	}
+
+	kfree(rdev->nqr);
+	rdev->nqr = NULL;
+	bng_re_destroy_chip_ctx(rdev);
+	if (test_and_clear_bit(BNG_RE_FLAG_NETDEV_REGISTERED, &rdev->flags))
+		bnge_unregister_dev(rdev->aux_dev);
+}
+
+static int bng_re_dev_init(struct bng_re_dev *rdev)
+{
+	struct bng_re_ring_attr rattr = {};
+	struct bng_re_creq_ctx *creq;
+	u32 db_offt;
+	int vid;
+	u8 type;
+	int rc;
+
+	/* Registered a new RoCE device instance to netdev */
+	rc = bng_re_register_netdev(rdev);
+	if (rc) {
+		ibdev_err(&rdev->ibdev,
+				"Failed to register with netedev: %#x\n", rc);
+		return -EINVAL;
+	}
+
+	set_bit(BNG_RE_FLAG_NETDEV_REGISTERED, &rdev->flags);
+
+	if (rdev->aux_dev->auxr_info->msix_requested < BNG_RE_MIN_MSIX) {
+		ibdev_err(&rdev->ibdev,
+			  "RoCE requires minimum 2 MSI-X vectors, but only %d reserved\n",
+			  rdev->aux_dev->auxr_info->msix_requested);
+		bnge_unregister_dev(rdev->aux_dev);
+		clear_bit(BNG_RE_FLAG_NETDEV_REGISTERED, &rdev->flags);
+		return -EINVAL;
+	}
+	ibdev_dbg(&rdev->ibdev, "Got %d MSI-X vectors\n",
+		  rdev->aux_dev->auxr_info->msix_requested);
+
+	rc = bng_re_setup_chip_ctx(rdev);
+	if (rc) {
+		bnge_unregister_dev(rdev->aux_dev);
+		clear_bit(BNG_RE_FLAG_NETDEV_REGISTERED, &rdev->flags);
+		ibdev_err(&rdev->ibdev, "Failed to get chip context\n");
+		return -EINVAL;
+	}
+
+	bng_re_query_hwrm_version(rdev);
+
+	rc = bng_re_alloc_fw_channel(&rdev->bng_res, &rdev->rcfw);
+	if (rc) {
+		ibdev_err(&rdev->ibdev,
+			  "Failed to allocate RCFW Channel: %#x\n", rc);
+		goto fail;
+	}
+
+	/* Allocate nq record memory */
+	rdev->nqr = kzalloc(sizeof(*rdev->nqr), GFP_KERNEL);
+	if (!rdev->nqr) {
+		bng_re_destroy_chip_ctx(rdev);
+		bnge_unregister_dev(rdev->aux_dev);
+		clear_bit(BNG_RE_FLAG_NETDEV_REGISTERED, &rdev->flags);
+		return -ENOMEM;
+	}
+
+	rdev->nqr->num_msix = rdev->aux_dev->auxr_info->msix_requested;
+	memcpy(rdev->nqr->msix_entries, rdev->aux_dev->msix_info,
+	       sizeof(struct bnge_msix_info) * rdev->nqr->num_msix);
+
+	type = RING_ALLOC_REQ_RING_TYPE_NQ;
+	creq = &rdev->rcfw.creq;
+	rattr.dma_arr = creq->hwq.pbl[BNG_PBL_LVL_0].pg_map_arr;
+	rattr.pages = creq->hwq.pbl[creq->hwq.level].pg_count;
+	rattr.type = type;
+	rattr.mode = RING_ALLOC_REQ_INT_MODE_MSIX;
+	rattr.depth = BNG_FW_CREQE_MAX_CNT - 1;
+	rattr.lrid = rdev->nqr->msix_entries[BNG_RE_CREQ_NQ_IDX].ring_idx;
+	rc = bng_re_net_ring_alloc(rdev, &rattr, &creq->ring_id);
+	if (rc) {
+		ibdev_err(&rdev->ibdev, "Failed to allocate CREQ: %#x\n", rc);
+		goto free_rcfw;
+	}
+	db_offt = rdev->nqr->msix_entries[BNG_RE_CREQ_NQ_IDX].db_offset;
+	vid = rdev->nqr->msix_entries[BNG_RE_CREQ_NQ_IDX].vector;
+
+	rc = bng_re_enable_fw_channel(&rdev->rcfw,
+					vid, db_offt);
+	if (rc) {
+		ibdev_err(&rdev->ibdev, "Failed to enable RCFW channel: %#x\n",
+			  rc);
+		goto free_ring;
+	}
+
+	rc = bng_re_get_dev_attr(&rdev->rcfw);
+	if (rc)
+		goto disable_rcfw;
+
+	bng_re_debugfs_add_pdev(rdev);
+	rc = bng_re_alloc_stats_ctx_mem(rdev->bng_res.pdev, rdev->chip_ctx,
+					&rdev->stats_ctx);
+	if (rc) {
+		ibdev_err(&rdev->ibdev,
+			  "Failed to allocate stats context: %#x\n", rc);
+		goto disable_rcfw;
+	}
+
+	rc = bng_re_stats_ctx_alloc(rdev);
+	if (rc) {
+		ibdev_err(&rdev->ibdev,
+			  "Failed to allocate QPLIB context: %#x\n", rc);
+		goto free_stats_ctx;
+	}
+
+	rc = bng_re_init_rcfw(&rdev->rcfw, &rdev->stats_ctx);
+	if (rc) {
+		ibdev_err(&rdev->ibdev,
+			  "Failed to initialize RCFW: %#x\n", rc);
+		goto free_sctx;
+	}
+	set_bit(BNG_RE_FLAG_RCFW_CHANNEL_EN, &rdev->flags);
+
+	return 0;
+free_sctx:
+	bng_re_stats_ctx_free(rdev);
+free_stats_ctx:
+	bng_re_free_stats_ctx_mem(rdev->bng_res.pdev, &rdev->stats_ctx);
+disable_rcfw:
+	bng_re_disable_rcfw_channel(&rdev->rcfw);
+free_ring:
+	bng_re_net_ring_free(rdev, rdev->rcfw.creq.ring_id, type);
+free_rcfw:
+	bng_re_free_rcfw_channel(&rdev->rcfw);
+fail:
+	bng_re_dev_uninit(rdev);
+	return rc;
+}
+
+static int bng_re_add_device(struct auxiliary_device *adev)
+{
+	struct bnge_auxr_priv *auxr_priv =
+		container_of(adev, struct bnge_auxr_priv, aux_dev);
+	struct bng_re_en_dev_info *dev_info;
+	struct bng_re_dev *rdev;
+	int rc;
+
+	dev_info = auxiliary_get_drvdata(adev);
+
+	rdev = bng_re_dev_add(adev, auxr_priv->auxr_dev);
+	if (!rdev) {
+		rc = -ENOMEM;
+		goto exit;
+	}
+
+	dev_info->rdev = rdev;
+
+	rc = bng_re_dev_init(rdev);
+	if (rc)
+		goto re_dev_dealloc;
+
+	return 0;
+
+re_dev_dealloc:
+	ib_dealloc_device(&rdev->ibdev);
+exit:
+	return rc;
+}
+
+
+static void bng_re_remove_device(struct bng_re_dev *rdev,
+				 struct auxiliary_device *aux_dev)
+{
+	bng_re_dev_uninit(rdev);
+	ib_dealloc_device(&rdev->ibdev);
+}
+
+
+static int bng_re_probe(struct auxiliary_device *adev,
+			const struct auxiliary_device_id *id)
+{
+	struct bnge_auxr_priv *aux_priv =
+		container_of(adev, struct bnge_auxr_priv, aux_dev);
+	struct bng_re_en_dev_info *en_info;
+	int rc;
+
+	en_info = kzalloc(sizeof(*en_info), GFP_KERNEL);
+	if (!en_info)
+		return -ENOMEM;
+
+	en_info->auxr_dev = aux_priv->auxr_dev;
+
+	auxiliary_set_drvdata(adev, en_info);
+
+	rc = bng_re_add_device(adev);
+	if (rc)
+		kfree(en_info);
+
+	return rc;
+}
+
+static void bng_re_remove(struct auxiliary_device *adev)
+{
+	struct bng_re_en_dev_info *dev_info = auxiliary_get_drvdata(adev);
+	struct bng_re_dev *rdev;
+
+	rdev = dev_info->rdev;
+
+	if (rdev)
+		bng_re_remove_device(rdev, adev);
+	kfree(dev_info);
+}
+
+static const struct auxiliary_device_id bng_re_id_table[] = {
+	{ .name = BNG_RE_ADEV_NAME ".rdma", },
+	{},
+};
+
+MODULE_DEVICE_TABLE(auxiliary, bng_re_id_table);
+
+static struct auxiliary_driver bng_re_driver = {
+	.name = "rdma",
+	.probe = bng_re_probe,
+	.remove = bng_re_remove,
+	.id_table = bng_re_id_table,
+};
+
+static int __init bng_re_mod_init(void)
+{
+	int rc;
+
+
+	bng_re_register_debugfs();
+
+	rc = auxiliary_driver_register(&bng_re_driver);
+	if (rc) {
+		pr_err("%s: Failed to register auxiliary driver\n",
+		       KBUILD_MODNAME);
+		goto unreg_debugfs;
+	}
+	return 0;
+unreg_debugfs:
+	bng_re_unregister_debugfs();
+	return rc;
+}
+
+static void __exit bng_re_mod_exit(void)
+{
+	auxiliary_driver_unregister(&bng_re_driver);
+	bng_re_unregister_debugfs();
+}
+
+module_init(bng_re_mod_init);
+module_exit(bng_re_mod_exit);

drivers/infiniband/hw/bng_re/bng_fw.c

@@ -0,0 +1,767 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2025 Broadcom.
+#include <linux/pci.h>
+
+#include "roce_hsi.h"
+#include "bng_res.h"
+#include "bng_fw.h"
+#include "bng_sp.h"
+
+/**
+ * bng_re_map_rc  -  map return type based on opcode
+ * @opcode:  roce slow path opcode
+ *
+ * case #1
+ * Firmware initiated error recovery is a safe state machine and
+ * driver can consider all the underlying rdma resources are free.
+ * In this state, it is safe to return success for opcodes related to
+ * destroying rdma resources (like destroy qp, destroy cq etc.).
+ *
+ * case #2
+ * If driver detect potential firmware stall, it is not safe state machine
+ * and the driver can not consider all the underlying rdma resources are
+ * freed.
+ * In this state, it is not safe to return success for opcodes related to
+ * destroying rdma resources (like destroy qp, destroy cq etc.).
+ *
+ * Scope of this helper function is only for case #1.
+ *
+ * Returns:
+ * 0 to communicate success to caller.
+ * Non zero error code to communicate failure to caller.
+ */
+static int bng_re_map_rc(u8 opcode)
+{
+	switch (opcode) {
+	case CMDQ_BASE_OPCODE_DESTROY_QP:
+	case CMDQ_BASE_OPCODE_DESTROY_SRQ:
+	case CMDQ_BASE_OPCODE_DESTROY_CQ:
+	case CMDQ_BASE_OPCODE_DEALLOCATE_KEY:
+	case CMDQ_BASE_OPCODE_DEREGISTER_MR:
+	case CMDQ_BASE_OPCODE_DELETE_GID:
+	case CMDQ_BASE_OPCODE_DESTROY_QP1:
+	case CMDQ_BASE_OPCODE_DESTROY_AH:
+	case CMDQ_BASE_OPCODE_DEINITIALIZE_FW:
+	case CMDQ_BASE_OPCODE_MODIFY_ROCE_CC:
+	case CMDQ_BASE_OPCODE_SET_LINK_AGGR_MODE:
+		return 0;
+	default:
+		return -ETIMEDOUT;
+	}
+}
+
+void bng_re_free_rcfw_channel(struct bng_re_rcfw *rcfw)
+{
+	kfree(rcfw->crsqe_tbl);
+	bng_re_free_hwq(rcfw->res, &rcfw->cmdq.hwq);
+	bng_re_free_hwq(rcfw->res, &rcfw->creq.hwq);
+	rcfw->pdev = NULL;
+}
+
+int bng_re_alloc_fw_channel(struct bng_re_res *res,
+			    struct bng_re_rcfw *rcfw)
+{
+	struct bng_re_hwq_attr hwq_attr = {};
+	struct bng_re_sg_info sginfo = {};
+	struct bng_re_cmdq_ctx *cmdq;
+	struct bng_re_creq_ctx *creq;
+
+	rcfw->pdev = res->pdev;
+	cmdq = &rcfw->cmdq;
+	creq = &rcfw->creq;
+	rcfw->res = res;
+
+	sginfo.pgsize = PAGE_SIZE;
+	sginfo.pgshft = PAGE_SHIFT;
+
+	hwq_attr.sginfo = &sginfo;
+	hwq_attr.res = rcfw->res;
+	hwq_attr.depth = BNG_FW_CREQE_MAX_CNT;
+	hwq_attr.stride = BNG_FW_CREQE_UNITS;
+	hwq_attr.type = BNG_HWQ_TYPE_QUEUE;
+
+	if (bng_re_alloc_init_hwq(&creq->hwq, &hwq_attr)) {
+		dev_err(&rcfw->pdev->dev,
+			"HW channel CREQ allocation failed\n");
+		goto fail;
+	}
+
+	rcfw->cmdq_depth = BNG_FW_CMDQE_MAX_CNT;
+
+	sginfo.pgsize = bng_fw_cmdqe_page_size(rcfw->cmdq_depth);
+	hwq_attr.depth = rcfw->cmdq_depth & 0x7FFFFFFF;
+	hwq_attr.stride = BNG_FW_CMDQE_UNITS;
+	hwq_attr.type = BNG_HWQ_TYPE_CTX;
+	if (bng_re_alloc_init_hwq(&cmdq->hwq, &hwq_attr)) {
+		dev_err(&rcfw->pdev->dev,
+			"HW channel CMDQ allocation failed\n");
+		goto fail;
+	}
+
+	rcfw->crsqe_tbl = kcalloc(cmdq->hwq.max_elements,
+				  sizeof(*rcfw->crsqe_tbl), GFP_KERNEL);
+	if (!rcfw->crsqe_tbl)
+		goto fail;
+
+	spin_lock_init(&rcfw->tbl_lock);
+
+	rcfw->max_timeout = res->cctx->hwrm_cmd_max_timeout;
+	return 0;
+
+fail:
+	bng_re_free_rcfw_channel(rcfw);
+	return -ENOMEM;
+}
+
+static int bng_re_process_qp_event(struct bng_re_rcfw *rcfw,
+				   struct creq_qp_event *qp_event,
+				   u32 *num_wait)
+{
+	struct bng_re_hwq *hwq = &rcfw->cmdq.hwq;
+	struct bng_re_crsqe *crsqe;
+	u32 req_size;
+	u16 cookie;
+	bool is_waiter_alive;
+	struct pci_dev *pdev;
+	u32 wait_cmds = 0;
+	int rc = 0;
+
+	pdev = rcfw->pdev;
+	switch (qp_event->event) {
+	case CREQ_QP_EVENT_EVENT_QP_ERROR_NOTIFICATION:
+		dev_err(&pdev->dev, "Received QP error notification\n");
+		break;
+	default:
+		/*
+		 * Command Response
+		 * cmdq->lock needs to be acquired to synchronie
+		 * the command send and completion reaping. This function
+		 * is always called with creq->lock held. Using
+		 * the nested variant of spin_lock.
+		 *
+		 */
+
+		spin_lock_nested(&hwq->lock, SINGLE_DEPTH_NESTING);
+		cookie = le16_to_cpu(qp_event->cookie);
+		cookie &= BNG_FW_MAX_COOKIE_VALUE;
+		crsqe = &rcfw->crsqe_tbl[cookie];
+
+		if (WARN_ONCE(test_bit(FIRMWARE_STALL_DETECTED,
+				       &rcfw->cmdq.flags),
+		    "Unreponsive rcfw channel detected.!!")) {
+			dev_info(&pdev->dev,
+				 "rcfw timedout: cookie = %#x, free_slots = %d",
+				 cookie, crsqe->free_slots);
+			spin_unlock(&hwq->lock);
+			return rc;
+		}
+
+		if (crsqe->is_waiter_alive) {
+			if (crsqe->resp) {
+				memcpy(crsqe->resp, qp_event, sizeof(*qp_event));
+				/* Insert write memory barrier to ensure that
+				 * response data is copied before clearing the
+				 * flags
+				 */
+				smp_wmb();
+			}
+		}
+
+		wait_cmds++;
+
+		req_size = crsqe->req_size;
+		is_waiter_alive = crsqe->is_waiter_alive;
+
+		crsqe->req_size = 0;
+		if (!is_waiter_alive)
+			crsqe->resp = NULL;
+
+		crsqe->is_in_used = false;
+
+		hwq->cons += req_size;
+
+		spin_unlock(&hwq->lock);
+	}
+	*num_wait += wait_cmds;
+	return rc;
+}
+
+/* function events */
+static int bng_re_process_func_event(struct bng_re_rcfw *rcfw,
+				     struct creq_func_event *func_event)
+{
+	switch (func_event->event) {
+	case CREQ_FUNC_EVENT_EVENT_TX_WQE_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_TX_DATA_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_RX_WQE_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_RX_DATA_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_CQ_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_TQM_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_CFCQ_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_CFCS_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_CFCC_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_CFCM_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_TIM_ERROR:
+	case CREQ_FUNC_EVENT_EVENT_VF_COMM_REQUEST:
+	case CREQ_FUNC_EVENT_EVENT_RESOURCE_EXHAUSTED:
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+/* CREQ Completion handlers */
+static void bng_re_service_creq(struct tasklet_struct *t)
+{
+	struct bng_re_rcfw *rcfw = from_tasklet(rcfw, t, creq.creq_tasklet);
+	struct bng_re_creq_ctx *creq = &rcfw->creq;
+	u32 type, budget = BNG_FW_CREQ_ENTRY_POLL_BUDGET;
+	struct bng_re_hwq *hwq = &creq->hwq;
+	struct creq_base *creqe;
+	u32 num_wakeup = 0;
+	u32 hw_polled = 0;
+
+	/* Service the CREQ until budget is over */
+	spin_lock_bh(&hwq->lock);
+	while (budget > 0) {
+		creqe = bng_re_get_qe(hwq, hwq->cons, NULL);
+		if (!BNG_FW_CREQ_CMP_VALID(creqe, creq->creq_db.dbinfo.flags))
+			break;
+		/* The valid test of the entry must be done first before
+		 * reading any further.
+		 */
+		dma_rmb();
+
+		type = creqe->type & CREQ_BASE_TYPE_MASK;
+		switch (type) {
+		case CREQ_BASE_TYPE_QP_EVENT:
+			bng_re_process_qp_event
+				(rcfw, (struct creq_qp_event *)creqe,
+				 &num_wakeup);
+			creq->stats.creq_qp_event_processed++;
+			break;
+		case CREQ_BASE_TYPE_FUNC_EVENT:
+			if (!bng_re_process_func_event
+			    (rcfw, (struct creq_func_event *)creqe))
+				creq->stats.creq_func_event_processed++;
+			else
+				dev_warn(&rcfw->pdev->dev,
+					 "aeqe:%#x Not handled\n", type);
+			break;
+		default:
+			if (type != ASYNC_EVENT_CMPL_TYPE_HWRM_ASYNC_EVENT)
+				dev_warn(&rcfw->pdev->dev,
+					 "creqe with event 0x%x not handled\n",
+					 type);
+			break;
+		}
+		budget--;
+		hw_polled++;
+		bng_re_hwq_incr_cons(hwq->max_elements, &hwq->cons,
+				     1, &creq->creq_db.dbinfo.flags);
+	}
+
+	if (hw_polled)
+		bng_re_ring_nq_db(&creq->creq_db.dbinfo,
+				  rcfw->res->cctx, true);
+	spin_unlock_bh(&hwq->lock);
+	if (num_wakeup)
+		wake_up_nr(&rcfw->cmdq.waitq, num_wakeup);
+}
+
+static int __send_message_basic_sanity(struct bng_re_rcfw *rcfw,
+				       struct bng_re_cmdqmsg *msg,
+				       u8 opcode)
+{
+	struct bng_re_cmdq_ctx *cmdq;
+
+	cmdq = &rcfw->cmdq;
+
+	if (test_bit(FIRMWARE_STALL_DETECTED, &cmdq->flags))
+		return -ETIMEDOUT;
+
+	if (test_bit(FIRMWARE_INITIALIZED_FLAG, &cmdq->flags) &&
+	    opcode == CMDQ_BASE_OPCODE_INITIALIZE_FW) {
+		dev_err(&rcfw->pdev->dev, "RCFW already initialized!");
+		return -EINVAL;
+	}
+
+	if (!test_bit(FIRMWARE_INITIALIZED_FLAG, &cmdq->flags) &&
+	    (opcode != CMDQ_BASE_OPCODE_QUERY_FUNC &&
+	     opcode != CMDQ_BASE_OPCODE_INITIALIZE_FW &&
+	     opcode != CMDQ_BASE_OPCODE_QUERY_VERSION)) {
+		dev_err(&rcfw->pdev->dev,
+			"RCFW not initialized, reject opcode 0x%x",
+			opcode);
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static int __send_message(struct bng_re_rcfw *rcfw,
+			  struct bng_re_cmdqmsg *msg, u8 opcode)
+{
+	u32 bsize, free_slots, required_slots;
+	struct bng_re_cmdq_ctx *cmdq;
+	struct bng_re_crsqe *crsqe;
+	struct bng_fw_cmdqe *cmdqe;
+	struct bng_re_hwq *hwq;
+	u32 sw_prod, cmdq_prod;
+	struct pci_dev *pdev;
+	u16 cookie;
+	u8 *preq;
+
+	cmdq = &rcfw->cmdq;
+	hwq = &cmdq->hwq;
+	pdev = rcfw->pdev;
+
+	/* Cmdq are in 16-byte units, each request can consume 1 or more
+	 * cmdqe
+	 */
+	spin_lock_bh(&hwq->lock);
+	required_slots = bng_re_get_cmd_slots(msg->req);
+	free_slots = HWQ_FREE_SLOTS(hwq);
+	cookie = cmdq->seq_num & BNG_FW_MAX_COOKIE_VALUE;
+	crsqe = &rcfw->crsqe_tbl[cookie];
+
+	if (required_slots >= free_slots) {
+		dev_info_ratelimited(&pdev->dev,
+				     "CMDQ is full req/free %d/%d!",
+				     required_slots, free_slots);
+		spin_unlock_bh(&hwq->lock);
+		return -EAGAIN;
+	}
+	__set_cmdq_base_cookie(msg->req, msg->req_sz, cpu_to_le16(cookie));
+
+	bsize = bng_re_set_cmd_slots(msg->req);
+	crsqe->free_slots = free_slots;
+	crsqe->resp = (struct creq_qp_event *)msg->resp;
+	crsqe->is_waiter_alive = true;
+	crsqe->is_in_used = true;
+	crsqe->opcode = opcode;
+
+	crsqe->req_size = __get_cmdq_base_cmd_size(msg->req, msg->req_sz);
+	if (__get_cmdq_base_resp_size(msg->req, msg->req_sz) && msg->sb) {
+		struct bng_re_rcfw_sbuf *sbuf = msg->sb;
+
+		__set_cmdq_base_resp_addr(msg->req, msg->req_sz,
+					  cpu_to_le64(sbuf->dma_addr));
+		__set_cmdq_base_resp_size(msg->req, msg->req_sz,
+					  ALIGN(sbuf->size,
+						BNG_FW_CMDQE_UNITS) /
+						BNG_FW_CMDQE_UNITS);
+	}
+
+	preq = (u8 *)msg->req;
+	do {
+		/* Locate the next cmdq slot */
+		sw_prod = HWQ_CMP(hwq->prod, hwq);
+		cmdqe = bng_re_get_qe(hwq, sw_prod, NULL);
+		/* Copy a segment of the req cmd to the cmdq */
+		memset(cmdqe, 0, sizeof(*cmdqe));
+		memcpy(cmdqe, preq, min_t(u32, bsize, sizeof(*cmdqe)));
+		preq += min_t(u32, bsize, sizeof(*cmdqe));
+		bsize -= min_t(u32, bsize, sizeof(*cmdqe));
+		hwq->prod++;
+	} while (bsize > 0);
+	cmdq->seq_num++;
+
+	cmdq_prod = hwq->prod & 0xFFFF;
+	if (test_bit(FIRMWARE_FIRST_FLAG, &cmdq->flags)) {
+		/* The very first doorbell write
+		 * is required to set this flag
+		 * which prompts the FW to reset
+		 * its internal pointers
+		 */
+		cmdq_prod |= BIT(FIRMWARE_FIRST_FLAG);
+		clear_bit(FIRMWARE_FIRST_FLAG, &cmdq->flags);
+	}
+	/* ring CMDQ DB */
+	wmb();
+	writel(cmdq_prod, cmdq->cmdq_mbox.prod);
+	writel(BNG_FW_CMDQ_TRIG_VAL, cmdq->cmdq_mbox.db);
+	spin_unlock_bh(&hwq->lock);
+	/* Return the CREQ response pointer */
+	return 0;
+}
+
+/**
+ * __wait_for_resp   -	Don't hold the cpu context and wait for response
+ * @rcfw:    rcfw channel instance of rdev
+ * @cookie:  cookie to track the command
+ *
+ * Wait for command completion in sleepable context.
+ *
+ * Returns:
+ * 0 if command is completed by firmware.
+ * Non zero error code for rest of the case.
+ */
+static int __wait_for_resp(struct bng_re_rcfw *rcfw, u16 cookie)
+{
+	struct bng_re_cmdq_ctx *cmdq;
+	struct bng_re_crsqe *crsqe;
+
+	cmdq = &rcfw->cmdq;
+	crsqe = &rcfw->crsqe_tbl[cookie];
+
+	do {
+		wait_event_timeout(cmdq->waitq,
+				   !crsqe->is_in_used,
+				   secs_to_jiffies(rcfw->max_timeout));
+
+		if (!crsqe->is_in_used)
+			return 0;
+
+		bng_re_service_creq(&rcfw->creq.creq_tasklet);
+
+		if (!crsqe->is_in_used)
+			return 0;
+	} while (true);
+};
+
+/**
+ * bng_re_rcfw_send_message   -	interface to send
+ * and complete rcfw command.
+ * @rcfw:   rcfw channel instance of rdev
+ * @msg:    message to send
+ *
+ * This function does not account shadow queue depth. It will send
+ * all the command unconditionally as long as send queue is not full.
+ *
+ * Returns:
+ * 0 if command completed by firmware.
+ * Non zero if the command is not completed by firmware.
+ */
+int bng_re_rcfw_send_message(struct bng_re_rcfw *rcfw,
+			     struct bng_re_cmdqmsg *msg)
+{
+	struct creq_qp_event *evnt = (struct creq_qp_event *)msg->resp;
+	struct bng_re_crsqe *crsqe;
+	u16 cookie;
+	int rc;
+	u8 opcode;
+
+	opcode = __get_cmdq_base_opcode(msg->req, msg->req_sz);
+
+	rc = __send_message_basic_sanity(rcfw, msg, opcode);
+	if (rc)
+		return rc == -ENXIO ? bng_re_map_rc(opcode) : rc;
+
+	rc = __send_message(rcfw, msg, opcode);
+	if (rc)
+		return rc;
+
+	cookie = le16_to_cpu(__get_cmdq_base_cookie(msg->req, msg->req_sz))
+				& BNG_FW_MAX_COOKIE_VALUE;
+
+	rc = __wait_for_resp(rcfw, cookie);
+
+	if (rc) {
+		spin_lock_bh(&rcfw->cmdq.hwq.lock);
+		crsqe = &rcfw->crsqe_tbl[cookie];
+		crsqe->is_waiter_alive = false;
+		if (rc == -ENODEV)
+			set_bit(FIRMWARE_STALL_DETECTED, &rcfw->cmdq.flags);
+		spin_unlock_bh(&rcfw->cmdq.hwq.lock);
+		return -ETIMEDOUT;
+	}
+
+	if (evnt->status) {
+		/* failed with status */
+		dev_err(&rcfw->pdev->dev, "cmdq[%#x]=%#x status %#x\n",
+			cookie, opcode, evnt->status);
+		rc = -EIO;
+	}
+
+	return rc;
+}
+
+static int bng_re_map_cmdq_mbox(struct bng_re_rcfw *rcfw)
+{
+	struct bng_re_cmdq_mbox *mbox;
+	resource_size_t bar_reg;
+	struct pci_dev *pdev;
+
+	pdev = rcfw->pdev;
+	mbox = &rcfw->cmdq.cmdq_mbox;
+
+	mbox->reg.bar_id = BNG_FW_COMM_PCI_BAR_REGION;
+	mbox->reg.len = BNG_FW_COMM_SIZE;
+	mbox->reg.bar_base = pci_resource_start(pdev, mbox->reg.bar_id);
+	if (!mbox->reg.bar_base) {
+		dev_err(&pdev->dev,
+			"CMDQ BAR region %d resc start is 0!\n",
+			mbox->reg.bar_id);
+		return -ENOMEM;
+	}
+
+	bar_reg = mbox->reg.bar_base + BNG_FW_COMM_BASE_OFFSET;
+	mbox->reg.len = BNG_FW_COMM_SIZE;
+	mbox->reg.bar_reg = ioremap(bar_reg, mbox->reg.len);
+	if (!mbox->reg.bar_reg) {
+		dev_err(&pdev->dev,
+			"CMDQ BAR region %d mapping failed\n",
+			mbox->reg.bar_id);
+		return -ENOMEM;
+	}
+
+	mbox->prod = (void  __iomem *)(mbox->reg.bar_reg +
+			BNG_FW_PF_VF_COMM_PROD_OFFSET);
+	mbox->db = (void __iomem *)(mbox->reg.bar_reg + BNG_FW_COMM_TRIG_OFFSET);
+	return 0;
+}
+
+static irqreturn_t bng_re_creq_irq(int irq, void *dev_instance)
+{
+	struct bng_re_rcfw *rcfw = dev_instance;
+	struct bng_re_creq_ctx *creq;
+	struct bng_re_hwq *hwq;
+	u32 sw_cons;
+
+	creq = &rcfw->creq;
+	hwq = &creq->hwq;
+	/* Prefetch the CREQ element */
+	sw_cons = HWQ_CMP(hwq->cons, hwq);
+	bng_re_get_qe(hwq, sw_cons, NULL);
+
+	tasklet_schedule(&creq->creq_tasklet);
+	return IRQ_HANDLED;
+}
+
+int bng_re_rcfw_start_irq(struct bng_re_rcfw *rcfw, int msix_vector,
+			  bool need_init)
+{
+	struct bng_re_creq_ctx *creq;
+	struct bng_re_res *res;
+	int rc;
+
+	creq = &rcfw->creq;
+	res = rcfw->res;
+
+	if (creq->irq_handler_avail)
+		return -EFAULT;
+
+	creq->msix_vec = msix_vector;
+	if (need_init)
+		tasklet_setup(&creq->creq_tasklet, bng_re_service_creq);
+	else
+		tasklet_enable(&creq->creq_tasklet);
+
+	creq->irq_name = kasprintf(GFP_KERNEL, "bng_re-creq@pci:%s",
+				   pci_name(res->pdev));
+	if (!creq->irq_name)
+		return -ENOMEM;
+	rc = request_irq(creq->msix_vec, bng_re_creq_irq, 0,
+			 creq->irq_name, rcfw);
+	if (rc) {
+		kfree(creq->irq_name);
+		creq->irq_name = NULL;
+		tasklet_disable(&creq->creq_tasklet);
+		return rc;
+	}
+	creq->irq_handler_avail = true;
+
+	bng_re_ring_nq_db(&creq->creq_db.dbinfo, res->cctx, true);
+	atomic_inc(&rcfw->rcfw_intr_enabled);
+
+	return 0;
+}
+
+static int bng_re_map_creq_db(struct bng_re_rcfw *rcfw, u32 reg_offt)
+{
+	struct bng_re_creq_db *creq_db;
+	resource_size_t bar_reg;
+	struct pci_dev *pdev;
+
+	pdev = rcfw->pdev;
+	creq_db = &rcfw->creq.creq_db;
+
+	creq_db->dbinfo.flags = 0;
+	creq_db->reg.bar_id = BNG_FW_COMM_CONS_PCI_BAR_REGION;
+	creq_db->reg.bar_base = pci_resource_start(pdev, creq_db->reg.bar_id);
+	if (!creq_db->reg.bar_id)
+		dev_err(&pdev->dev,
+			"CREQ BAR region %d resc start is 0!",
+			creq_db->reg.bar_id);
+
+	bar_reg = creq_db->reg.bar_base + reg_offt;
+
+	creq_db->reg.len = BNG_FW_CREQ_DB_LEN;
+	creq_db->reg.bar_reg = ioremap(bar_reg, creq_db->reg.len);
+	if (!creq_db->reg.bar_reg) {
+		dev_err(&pdev->dev,
+			"CREQ BAR region %d mapping failed",
+			creq_db->reg.bar_id);
+		return -ENOMEM;
+	}
+	creq_db->dbinfo.db = creq_db->reg.bar_reg;
+	creq_db->dbinfo.hwq = &rcfw->creq.hwq;
+	creq_db->dbinfo.xid = rcfw->creq.ring_id;
+	return 0;
+}
+
+void bng_re_rcfw_stop_irq(struct bng_re_rcfw *rcfw, bool kill)
+{
+	struct bng_re_creq_ctx *creq;
+
+	creq = &rcfw->creq;
+
+	if (!creq->irq_handler_avail)
+		return;
+
+	creq->irq_handler_avail = false;
+	/* Mask h/w interrupts */
+	bng_re_ring_nq_db(&creq->creq_db.dbinfo, rcfw->res->cctx, false);
+	/* Sync with last running IRQ-handler */
+	synchronize_irq(creq->msix_vec);
+	free_irq(creq->msix_vec, rcfw);
+	kfree(creq->irq_name);
+	creq->irq_name = NULL;
+	atomic_set(&rcfw->rcfw_intr_enabled, 0);
+	if (kill)
+		tasklet_kill(&creq->creq_tasklet);
+	tasklet_disable(&creq->creq_tasklet);
+}
+
+void bng_re_disable_rcfw_channel(struct bng_re_rcfw *rcfw)
+{
+	struct bng_re_creq_ctx *creq;
+	struct bng_re_cmdq_ctx *cmdq;
+
+	creq = &rcfw->creq;
+	cmdq = &rcfw->cmdq;
+	/* Make sure the HW channel is stopped! */
+	bng_re_rcfw_stop_irq(rcfw, true);
+
+	iounmap(cmdq->cmdq_mbox.reg.bar_reg);
+	iounmap(creq->creq_db.reg.bar_reg);
+
+	cmdq->cmdq_mbox.reg.bar_reg = NULL;
+	creq->creq_db.reg.bar_reg = NULL;
+	creq->msix_vec = 0;
+}
+
+static void bng_re_start_rcfw(struct bng_re_rcfw *rcfw)
+{
+	struct bng_re_cmdq_ctx *cmdq;
+	struct bng_re_creq_ctx *creq;
+	struct bng_re_cmdq_mbox *mbox;
+	struct cmdq_init init = {0};
+
+	cmdq = &rcfw->cmdq;
+	creq = &rcfw->creq;
+	mbox = &cmdq->cmdq_mbox;
+
+	init.cmdq_pbl = cpu_to_le64(cmdq->hwq.pbl[BNG_PBL_LVL_0].pg_map_arr[0]);
+	init.cmdq_size_cmdq_lvl =
+			cpu_to_le16(((rcfw->cmdq_depth <<
+				      CMDQ_INIT_CMDQ_SIZE_SFT) &
+				    CMDQ_INIT_CMDQ_SIZE_MASK) |
+				    ((cmdq->hwq.level <<
+				      CMDQ_INIT_CMDQ_LVL_SFT) &
+				    CMDQ_INIT_CMDQ_LVL_MASK));
+	init.creq_ring_id = cpu_to_le16(creq->ring_id);
+	/* Write to the mailbox register */
+	__iowrite32_copy(mbox->reg.bar_reg, &init, sizeof(init) / 4);
+}
+
+int bng_re_enable_fw_channel(struct bng_re_rcfw *rcfw,
+			     int msix_vector,
+			     int cp_bar_reg_off)
+{
+	struct bng_re_cmdq_ctx *cmdq;
+	int rc;
+
+	cmdq = &rcfw->cmdq;
+
+	/* Assign defaults */
+	cmdq->seq_num = 0;
+	set_bit(FIRMWARE_FIRST_FLAG, &cmdq->flags);
+	init_waitqueue_head(&cmdq->waitq);
+
+	rc = bng_re_map_cmdq_mbox(rcfw);
+	if (rc)
+		return rc;
+
+	rc = bng_re_map_creq_db(rcfw, cp_bar_reg_off);
+	if (rc)
+		return rc;
+
+	rc = bng_re_rcfw_start_irq(rcfw, msix_vector, true);
+	if (rc) {
+		dev_err(&rcfw->pdev->dev,
+			"Failed to request IRQ for CREQ rc = 0x%x\n", rc);
+		bng_re_disable_rcfw_channel(rcfw);
+		return rc;
+	}
+
+	bng_re_start_rcfw(rcfw);
+	return 0;
+}
+
+int bng_re_deinit_rcfw(struct bng_re_rcfw *rcfw)
+{
+	struct creq_deinitialize_fw_resp resp = {};
+	struct cmdq_deinitialize_fw req = {};
+	struct bng_re_cmdqmsg msg = {};
+	int rc;
+
+	bng_re_rcfw_cmd_prep((struct cmdq_base *)&req,
+			     CMDQ_BASE_OPCODE_DEINITIALIZE_FW,
+			     sizeof(req));
+	bng_re_fill_cmdqmsg(&msg, &req, &resp, NULL,
+			    sizeof(req), sizeof(resp), 0);
+	rc = bng_re_rcfw_send_message(rcfw, &msg);
+	if (rc)
+		return rc;
+
+	clear_bit(FIRMWARE_INITIALIZED_FLAG, &rcfw->cmdq.flags);
+	return 0;
+}
+static inline bool _is_hw_retx_supported(u16 dev_cap_flags)
+{
+	return dev_cap_flags &
+		(CREQ_QUERY_FUNC_RESP_SB_HW_REQUESTER_RETX_ENABLED |
+		 CREQ_QUERY_FUNC_RESP_SB_HW_RESPONDER_RETX_ENABLED);
+}
+
+#define BNG_RE_HW_RETX(a) _is_hw_retx_supported((a))
+static inline bool _is_optimize_modify_qp_supported(u16 dev_cap_ext_flags2)
+{
+	return dev_cap_ext_flags2 &
+	       CREQ_QUERY_FUNC_RESP_SB_OPTIMIZE_MODIFY_QP_SUPPORTED;
+}
+
+int bng_re_init_rcfw(struct bng_re_rcfw *rcfw,
+		     struct bng_re_stats *stats_ctx)
+{
+	struct creq_initialize_fw_resp resp = {};
+	struct cmdq_initialize_fw req = {};
+	struct bng_re_cmdqmsg msg = {};
+	int rc;
+	u16 flags = 0;
+
+	bng_re_rcfw_cmd_prep((struct cmdq_base *)&req,
+			     CMDQ_BASE_OPCODE_INITIALIZE_FW,
+			     sizeof(req));
+	/* Supply (log-base-2-of-host-page-size - base-page-shift)
+	 * to bono to adjust the doorbell page sizes.
+	 */
+	req.log2_dbr_pg_size = cpu_to_le16(PAGE_SHIFT -
+					   BNG_FW_DBR_BASE_PAGE_SHIFT);
+	if (BNG_RE_HW_RETX(rcfw->res->dattr->dev_cap_flags))
+		flags |= CMDQ_INITIALIZE_FW_FLAGS_HW_REQUESTER_RETX_SUPPORTED;
+	if (_is_optimize_modify_qp_supported(rcfw->res->dattr->dev_cap_flags2))
+		flags |= CMDQ_INITIALIZE_FW_FLAGS_OPTIMIZE_MODIFY_QP_SUPPORTED;
+	req.flags |= cpu_to_le16(flags);
+	req.stat_ctx_id = cpu_to_le32(stats_ctx->fw_id);
+	bng_re_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
+	rc = bng_re_rcfw_send_message(rcfw, &msg);
+	if (rc)
+		return rc;
+	set_bit(FIRMWARE_INITIALIZED_FLAG, &rcfw->cmdq.flags);
+	return 0;
+}

drivers/infiniband/hw/bng_re/bng_fw.h

@@ -0,0 +1,211 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+// Copyright (c) 2025 Broadcom.
+
+#ifndef __BNG_FW_H__
+#define __BNG_FW_H__
+
+#include "bng_tlv.h"
+
+/* FW DB related */
+#define BNG_FW_CMDQ_TRIG_VAL		1
+#define BNG_FW_COMM_PCI_BAR_REGION	0
+#define BNG_FW_COMM_CONS_PCI_BAR_REGION	2
+#define BNG_FW_DBR_BASE_PAGE_SHIFT	12
+#define BNG_FW_COMM_SIZE		0x104
+#define BNG_FW_COMM_BASE_OFFSET		0x600
+#define BNG_FW_COMM_TRIG_OFFSET		0x100
+#define BNG_FW_PF_VF_COMM_PROD_OFFSET	0xc
+#define BNG_FW_CREQ_DB_LEN		8
+
+/* CREQ */
+#define BNG_FW_CREQE_MAX_CNT		(64 * 1024)
+#define BNG_FW_CREQE_UNITS		16
+#define BNG_FW_CREQ_ENTRY_POLL_BUDGET	0x100
+#define BNG_FW_CREQ_CMP_VALID(hdr, pass)			\
+	(!!((hdr)->v & CREQ_BASE_V) ==				\
+	   !((pass) & BNG_RE_FLAG_EPOCH_CONS_MASK))
+#define BNG_FW_CREQ_ENTRY_POLL_BUDGET	0x100
+
+/* CMDQ */
+struct bng_fw_cmdqe {
+	u8	data[16];
+};
+
+#define BNG_FW_CMDQE_MAX_CNT		8192
+#define BNG_FW_CMDQE_UNITS		sizeof(struct bng_fw_cmdqe)
+#define BNG_FW_CMDQE_BYTES(depth)	((depth) * BNG_FW_CMDQE_UNITS)
+
+#define BNG_FW_MAX_COOKIE_VALUE		(BNG_FW_CMDQE_MAX_CNT - 1)
+#define BNG_FW_CMD_IS_BLOCKING		0x8000
+
+/* Crsq buf is 1024-Byte */
+struct bng_re_crsbe {
+	u8			data[1024];
+};
+
+
+static inline u32 bng_fw_cmdqe_npages(u32 depth)
+{
+	u32 npages;
+
+	npages = BNG_FW_CMDQE_BYTES(depth) / PAGE_SIZE;
+	if (BNG_FW_CMDQE_BYTES(depth) % PAGE_SIZE)
+		npages++;
+	return npages;
+}
+
+static inline u32 bng_fw_cmdqe_page_size(u32 depth)
+{
+	return (bng_fw_cmdqe_npages(depth) * PAGE_SIZE);
+}
+struct bng_re_cmdq_mbox {
+	struct bng_re_reg_desc		reg;
+	void __iomem			*prod;
+	void __iomem			*db;
+};
+
+/* HWQ */
+struct bng_re_cmdq_ctx {
+	struct bng_re_hwq		hwq;
+	struct bng_re_cmdq_mbox		cmdq_mbox;
+	unsigned long			flags;
+#define FIRMWARE_INITIALIZED_FLAG	(0)
+#define FIRMWARE_STALL_DETECTED		(3)
+#define FIRMWARE_FIRST_FLAG		(31)
+	wait_queue_head_t		waitq;
+	u32				seq_num;
+};
+
+struct bng_re_creq_db {
+	struct bng_re_reg_desc	reg;
+	struct bng_re_db_info	dbinfo;
+};
+
+struct bng_re_creq_stat {
+	u64	creq_qp_event_processed;
+	u64	creq_func_event_processed;
+};
+
+struct bng_re_creq_ctx {
+	struct bng_re_hwq		hwq;
+	struct bng_re_creq_db		creq_db;
+	struct bng_re_creq_stat		stats;
+	struct tasklet_struct		creq_tasklet;
+	u16				ring_id;
+	int				msix_vec;
+	bool				irq_handler_avail;
+	char				*irq_name;
+};
+
+struct bng_re_crsqe {
+	struct creq_qp_event	*resp;
+	u32			req_size;
+	/* Free slots at the time of submission */
+	u32			free_slots;
+	u8			opcode;
+	bool			is_waiter_alive;
+	bool			is_in_used;
+};
+
+struct bng_re_rcfw_sbuf {
+	void *sb;
+	dma_addr_t dma_addr;
+	u32 size;
+};
+
+/* RoCE FW Communication Channels */
+struct bng_re_rcfw {
+	struct pci_dev		*pdev;
+	struct bng_re_res	*res;
+	struct bng_re_cmdq_ctx	cmdq;
+	struct bng_re_creq_ctx	creq;
+	struct bng_re_crsqe	*crsqe_tbl;
+	/* To synchronize the qp-handle hash table */
+	spinlock_t		tbl_lock;
+	u32			cmdq_depth;
+	/* cached from chip cctx for quick reference in slow path */
+	u16			max_timeout;
+	atomic_t		rcfw_intr_enabled;
+};
+
+struct bng_re_cmdqmsg {
+	struct cmdq_base	*req;
+	struct creq_base	*resp;
+	void			*sb;
+	u32			req_sz;
+	u32			res_sz;
+	u8			block;
+};
+
+static inline void bng_re_rcfw_cmd_prep(struct cmdq_base *req,
+					u8 opcode, u8 cmd_size)
+{
+	req->opcode = opcode;
+	req->cmd_size = cmd_size;
+}
+
+static inline void bng_re_fill_cmdqmsg(struct bng_re_cmdqmsg *msg,
+				       void *req, void *resp, void *sb,
+				       u32 req_sz, u32 res_sz, u8 block)
+{
+	msg->req = req;
+	msg->resp = resp;
+	msg->sb = sb;
+	msg->req_sz = req_sz;
+	msg->res_sz = res_sz;
+	msg->block = block;
+}
+
+/* Get the number of command units required for the req. The
+ * function returns correct value only if called before
+ * setting using bng_re_set_cmd_slots
+ */
+static inline u32 bng_re_get_cmd_slots(struct cmdq_base *req)
+{
+	u32 cmd_units = 0;
+
+	if (HAS_TLV_HEADER(req)) {
+		struct roce_tlv *tlv_req = (struct roce_tlv *)req;
+
+		cmd_units = tlv_req->total_size;
+	} else {
+		cmd_units = (req->cmd_size + BNG_FW_CMDQE_UNITS - 1) /
+			    BNG_FW_CMDQE_UNITS;
+	}
+
+	return cmd_units;
+}
+
+static inline u32 bng_re_set_cmd_slots(struct cmdq_base *req)
+{
+	u32 cmd_byte = 0;
+
+	if (HAS_TLV_HEADER(req)) {
+		struct roce_tlv *tlv_req = (struct roce_tlv *)req;
+
+		cmd_byte = tlv_req->total_size * BNG_FW_CMDQE_UNITS;
+	} else {
+		cmd_byte = req->cmd_size;
+		req->cmd_size = (req->cmd_size + BNG_FW_CMDQE_UNITS - 1) /
+				 BNG_FW_CMDQE_UNITS;
+	}
+
+	return cmd_byte;
+}
+
+void bng_re_free_rcfw_channel(struct bng_re_rcfw *rcfw);
+int bng_re_alloc_fw_channel(struct bng_re_res *res,
+			    struct bng_re_rcfw *rcfw);
+int bng_re_enable_fw_channel(struct bng_re_rcfw *rcfw,
+			     int msix_vector,
+			     int cp_bar_reg_off);
+void bng_re_disable_rcfw_channel(struct bng_re_rcfw *rcfw);
+int bng_re_rcfw_start_irq(struct bng_re_rcfw *rcfw, int msix_vector,
+			  bool need_init);
+void bng_re_rcfw_stop_irq(struct bng_re_rcfw *rcfw, bool kill);
+int bng_re_rcfw_send_message(struct bng_re_rcfw *rcfw,
+			     struct bng_re_cmdqmsg *msg);
+int bng_re_init_rcfw(struct bng_re_rcfw *rcfw,
+		     struct bng_re_stats *stats_ctx);
+int bng_re_deinit_rcfw(struct bng_re_rcfw *rcfw);
+#endif

drivers/infiniband/hw/bng_re/bng_re.h

@@ -0,0 +1,85 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+// Copyright (c) 2025 Broadcom.
+
+#ifndef __BNG_RE_H__
+#define __BNG_RE_H__
+
+#include "bng_res.h"
+
+#define BNG_RE_ADEV_NAME		"bng_en"
+
+#define BNG_RE_DESC	"Broadcom 800G RoCE Driver"
+
+#define	rdev_to_dev(rdev)	((rdev) ? (&(rdev)->ibdev.dev) : NULL)
+
+#define BNG_RE_MIN_MSIX		2
+#define BNG_RE_MAX_MSIX		BNGE_MAX_ROCE_MSIX
+
+#define BNG_RE_CREQ_NQ_IDX	0
+
+#define BNGE_INVALID_STATS_CTX_ID	-1
+/* NQ specific structures  */
+struct bng_re_nq_db {
+	struct bng_re_reg_desc	reg;
+	struct bng_re_db_info	dbinfo;
+};
+
+struct bng_re_nq {
+	struct pci_dev			*pdev;
+	struct bng_re_res		*res;
+	char				*name;
+	struct bng_re_hwq		hwq;
+	struct bng_re_nq_db		nq_db;
+	u16				ring_id;
+	int				msix_vec;
+	cpumask_t			mask;
+	struct tasklet_struct		nq_tasklet;
+	bool				requested;
+	int				budget;
+	u32				load;
+
+	struct workqueue_struct		*cqn_wq;
+};
+
+struct bng_re_nq_record {
+	struct bnge_msix_info	msix_entries[BNG_RE_MAX_MSIX];
+	struct bng_re_nq	nq[BNG_RE_MAX_MSIX];
+	int			num_msix;
+	/* serialize NQ access */
+	struct mutex		load_lock;
+};
+
+struct bng_re_en_dev_info {
+	struct bng_re_dev *rdev;
+	struct bnge_auxr_dev *auxr_dev;
+};
+
+struct bng_re_ring_attr {
+	dma_addr_t	*dma_arr;
+	int		pages;
+	int		type;
+	u32		depth;
+	u32		lrid; /* Logical ring id */
+	u8		mode;
+};
+
+struct bng_re_dev {
+	struct ib_device		ibdev;
+	unsigned long			flags;
+#define BNG_RE_FLAG_NETDEV_REGISTERED		0
+#define BNG_RE_FLAG_RCFW_CHANNEL_EN		1
+	struct net_device		*netdev;
+	struct auxiliary_device         *adev;
+	struct bnge_auxr_dev		*aux_dev;
+	struct bng_re_chip_ctx		*chip_ctx;
+	int				fn_id;
+	struct bng_re_res		bng_res;
+	struct bng_re_rcfw		rcfw;
+	struct bng_re_nq_record		*nqr;
+	/* Device Resources */
+	struct bng_re_dev_attr		*dev_attr;
+	struct dentry			*dbg_root;
+	struct bng_re_stats		stats_ctx;
+};
+
+#endif

drivers/infiniband/hw/bng_re/bng_res.c

@@ -0,0 +1,279 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2025 Broadcom.
+
+#include <linux/pci.h>
+#include <linux/vmalloc.h>
+#include <rdma/ib_umem.h>
+
+#include <linux/bnxt/hsi.h>
+#include "bng_res.h"
+#include "roce_hsi.h"
+
+/* Stats */
+void bng_re_free_stats_ctx_mem(struct pci_dev *pdev,
+			       struct bng_re_stats *stats)
+{
+	if (stats->dma) {
+		dma_free_coherent(&pdev->dev, stats->size,
+				  stats->dma, stats->dma_map);
+	}
+	memset(stats, 0, sizeof(*stats));
+	stats->fw_id = -1;
+}
+
+int bng_re_alloc_stats_ctx_mem(struct pci_dev *pdev,
+			       struct bng_re_chip_ctx *cctx,
+			       struct bng_re_stats *stats)
+{
+	memset(stats, 0, sizeof(*stats));
+	stats->fw_id = -1;
+	stats->size = cctx->hw_stats_size;
+	stats->dma = dma_alloc_coherent(&pdev->dev, stats->size,
+					&stats->dma_map, GFP_KERNEL);
+	if (!stats->dma)
+		return -ENOMEM;
+
+	return 0;
+}
+
+static void bng_free_pbl(struct bng_re_res  *res, struct bng_re_pbl *pbl)
+{
+	struct pci_dev *pdev = res->pdev;
+	int i;
+
+	for (i = 0; i < pbl->pg_count; i++) {
+		if (pbl->pg_arr[i])
+			dma_free_coherent(&pdev->dev, pbl->pg_size,
+					  (void *)((unsigned long)
+					     pbl->pg_arr[i] &
+						PAGE_MASK),
+					  pbl->pg_map_arr[i]);
+		else
+			dev_warn(&pdev->dev,
+					"PBL free pg_arr[%d] empty?!\n", i);
+		pbl->pg_arr[i] = NULL;
+	}
+
+	vfree(pbl->pg_arr);
+	pbl->pg_arr = NULL;
+	vfree(pbl->pg_map_arr);
+	pbl->pg_map_arr = NULL;
+	pbl->pg_count = 0;
+	pbl->pg_size = 0;
+}
+
+static int bng_alloc_pbl(struct bng_re_res  *res,
+			 struct bng_re_pbl *pbl,
+			 struct bng_re_sg_info *sginfo)
+{
+	struct pci_dev *pdev = res->pdev;
+	u32 pages;
+	int i;
+
+	if (sginfo->nopte)
+		return 0;
+	pages = sginfo->npages;
+
+	/* page ptr arrays */
+	pbl->pg_arr = vmalloc_array(pages, sizeof(void *));
+	if (!pbl->pg_arr)
+		return -ENOMEM;
+
+	pbl->pg_map_arr = vmalloc_array(pages, sizeof(dma_addr_t));
+	if (!pbl->pg_map_arr) {
+		vfree(pbl->pg_arr);
+		pbl->pg_arr = NULL;
+		return -ENOMEM;
+	}
+	pbl->pg_count = 0;
+	pbl->pg_size = sginfo->pgsize;
+
+	for (i = 0; i < pages; i++) {
+		pbl->pg_arr[i] = dma_alloc_coherent(&pdev->dev,
+				pbl->pg_size,
+				&pbl->pg_map_arr[i],
+				GFP_KERNEL);
+		if (!pbl->pg_arr[i])
+			goto fail;
+		pbl->pg_count++;
+	}
+
+	return 0;
+fail:
+	bng_free_pbl(res, pbl);
+	return -ENOMEM;
+}
+
+void bng_re_free_hwq(struct bng_re_res *res,
+		     struct bng_re_hwq *hwq)
+{
+	int i;
+
+	if (!hwq->max_elements)
+		return;
+	if (hwq->level >= BNG_PBL_LVL_MAX)
+		return;
+
+	for (i = 0; i < hwq->level + 1; i++)
+		bng_free_pbl(res, &hwq->pbl[i]);
+
+	hwq->level = BNG_PBL_LVL_MAX;
+	hwq->max_elements = 0;
+	hwq->element_size = 0;
+	hwq->prod = 0;
+	hwq->cons = 0;
+}
+
+/* All HWQs are power of 2 in size */
+int bng_re_alloc_init_hwq(struct bng_re_hwq *hwq,
+			  struct bng_re_hwq_attr *hwq_attr)
+{
+	u32 npages, pg_size;
+	struct bng_re_sg_info sginfo = {};
+	u32 depth, stride, npbl, npde;
+	dma_addr_t *src_phys_ptr, **dst_virt_ptr;
+	struct bng_re_res *res;
+	struct pci_dev *pdev;
+	int i, rc, lvl;
+
+	res = hwq_attr->res;
+	pdev = res->pdev;
+	pg_size = hwq_attr->sginfo->pgsize;
+	hwq->level = BNG_PBL_LVL_MAX;
+
+	depth = roundup_pow_of_two(hwq_attr->depth);
+	stride = roundup_pow_of_two(hwq_attr->stride);
+
+	npages = (depth * stride) / pg_size;
+	if ((depth * stride) % pg_size)
+		npages++;
+	if (!npages)
+		return -EINVAL;
+	hwq_attr->sginfo->npages = npages;
+
+	if (npages == MAX_PBL_LVL_0_PGS && !hwq_attr->sginfo->nopte) {
+		/* This request is Level 0, map PTE */
+		rc = bng_alloc_pbl(res, &hwq->pbl[BNG_PBL_LVL_0], hwq_attr->sginfo);
+		if (rc)
+			goto fail;
+		hwq->level = BNG_PBL_LVL_0;
+		goto done;
+	}
+
+	if (npages >= MAX_PBL_LVL_0_PGS) {
+		if (npages > MAX_PBL_LVL_1_PGS) {
+			u32 flag = PTU_PTE_VALID;
+			/* 2 levels of indirection */
+			npbl = npages >> MAX_PBL_LVL_1_PGS_SHIFT;
+			if (npages % BIT(MAX_PBL_LVL_1_PGS_SHIFT))
+				npbl++;
+			npde = npbl >> MAX_PDL_LVL_SHIFT;
+			if (npbl % BIT(MAX_PDL_LVL_SHIFT))
+				npde++;
+			/* Alloc PDE pages */
+			sginfo.pgsize = npde * pg_size;
+			sginfo.npages = 1;
+			rc = bng_alloc_pbl(res, &hwq->pbl[BNG_PBL_LVL_0], &sginfo);
+			if (rc)
+				goto fail;
+
+			/* Alloc PBL pages */
+			sginfo.npages = npbl;
+			sginfo.pgsize = PAGE_SIZE;
+			rc = bng_alloc_pbl(res, &hwq->pbl[BNG_PBL_LVL_1], &sginfo);
+			if (rc)
+				goto fail;
+			/* Fill PDL with PBL page pointers */
+			dst_virt_ptr =
+				(dma_addr_t **)hwq->pbl[BNG_PBL_LVL_0].pg_arr;
+			src_phys_ptr = hwq->pbl[BNG_PBL_LVL_1].pg_map_arr;
+			for (i = 0; i < hwq->pbl[BNG_PBL_LVL_1].pg_count; i++)
+				dst_virt_ptr[0][i] = src_phys_ptr[i] | flag;
+
+			/* Alloc or init PTEs */
+			rc = bng_alloc_pbl(res, &hwq->pbl[BNG_PBL_LVL_2],
+					 hwq_attr->sginfo);
+			if (rc)
+				goto fail;
+			hwq->level = BNG_PBL_LVL_2;
+			if (hwq_attr->sginfo->nopte)
+				goto done;
+			/* Fill PBLs with PTE pointers */
+			dst_virt_ptr =
+				(dma_addr_t **)hwq->pbl[BNG_PBL_LVL_1].pg_arr;
+			src_phys_ptr = hwq->pbl[BNG_PBL_LVL_2].pg_map_arr;
+			for (i = 0; i < hwq->pbl[BNG_PBL_LVL_2].pg_count; i++) {
+				dst_virt_ptr[PTR_PG(i)][PTR_IDX(i)] =
+					src_phys_ptr[i] | PTU_PTE_VALID;
+			}
+			if (hwq_attr->type == BNG_HWQ_TYPE_QUEUE) {
+				/* Find the last pg of the size */
+				i = hwq->pbl[BNG_PBL_LVL_2].pg_count;
+				dst_virt_ptr[PTR_PG(i - 1)][PTR_IDX(i - 1)] |=
+								  PTU_PTE_LAST;
+				if (i > 1)
+					dst_virt_ptr[PTR_PG(i - 2)]
+						    [PTR_IDX(i - 2)] |=
+						    PTU_PTE_NEXT_TO_LAST;
+			}
+		} else { /* pages < 512 npbl = 1, npde = 0 */
+			u32 flag = PTU_PTE_VALID;
+
+			/* 1 level of indirection */
+			npbl = npages >> MAX_PBL_LVL_1_PGS_SHIFT;
+			if (npages % BIT(MAX_PBL_LVL_1_PGS_SHIFT))
+				npbl++;
+			sginfo.npages = npbl;
+			sginfo.pgsize = PAGE_SIZE;
+			/* Alloc PBL page */
+			rc = bng_alloc_pbl(res, &hwq->pbl[BNG_PBL_LVL_0], &sginfo);
+			if (rc)
+				goto fail;
+			/* Alloc or init  PTEs */
+			rc = bng_alloc_pbl(res, &hwq->pbl[BNG_PBL_LVL_1],
+					 hwq_attr->sginfo);
+			if (rc)
+				goto fail;
+			hwq->level = BNG_PBL_LVL_1;
+			if (hwq_attr->sginfo->nopte)
+				goto done;
+			/* Fill PBL with PTE pointers */
+			dst_virt_ptr =
+				(dma_addr_t **)hwq->pbl[BNG_PBL_LVL_0].pg_arr;
+			src_phys_ptr = hwq->pbl[BNG_PBL_LVL_1].pg_map_arr;
+			for (i = 0; i < hwq->pbl[BNG_PBL_LVL_1].pg_count; i++)
+				dst_virt_ptr[PTR_PG(i)][PTR_IDX(i)] =
+					src_phys_ptr[i] | flag;
+			if (hwq_attr->type == BNG_HWQ_TYPE_QUEUE) {
+				/* Find the last pg of the size */
+				i = hwq->pbl[BNG_PBL_LVL_1].pg_count;
+				dst_virt_ptr[PTR_PG(i - 1)][PTR_IDX(i - 1)] |=
+								  PTU_PTE_LAST;
+				if (i > 1)
+					dst_virt_ptr[PTR_PG(i - 2)]
+						    [PTR_IDX(i - 2)] |=
+						    PTU_PTE_NEXT_TO_LAST;
+			}
+		}
+	}
+done:
+	hwq->prod = 0;
+	hwq->cons = 0;
+	hwq->pdev = pdev;
+	hwq->depth = hwq_attr->depth;
+	hwq->max_elements = hwq->depth;
+	hwq->element_size = stride;
+	hwq->qe_ppg = pg_size / stride;
+	/* For direct access to the elements */
+	lvl = hwq->level;
+	if (hwq_attr->sginfo->nopte && hwq->level)
+		lvl = hwq->level - 1;
+	hwq->pbl_ptr = hwq->pbl[lvl].pg_arr;
+	hwq->pbl_dma_ptr = hwq->pbl[lvl].pg_map_arr;
+	spin_lock_init(&hwq->lock);
+
+	return 0;
+fail:
+	bng_re_free_hwq(res, hwq);
+	return -ENOMEM;
+}

drivers/infiniband/hw/bng_re/bng_res.h

@@ -0,0 +1,215 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+// Copyright (c) 2025 Broadcom.
+
+#ifndef __BNG_RES_H__
+#define __BNG_RES_H__
+
+#include "roce_hsi.h"
+
+#define BNG_ROCE_FW_MAX_TIMEOUT	60
+
+#define PTR_CNT_PER_PG		(PAGE_SIZE / sizeof(void *))
+#define PTR_MAX_IDX_PER_PG	(PTR_CNT_PER_PG - 1)
+#define PTR_PG(x)		(((x) & ~PTR_MAX_IDX_PER_PG) / PTR_CNT_PER_PG)
+#define PTR_IDX(x)		((x) & PTR_MAX_IDX_PER_PG)
+
+#define HWQ_CMP(idx, hwq)	((idx) & ((hwq)->max_elements - 1))
+#define HWQ_FREE_SLOTS(hwq)	(hwq->max_elements - \
+				((HWQ_CMP(hwq->prod, hwq)\
+				- HWQ_CMP(hwq->cons, hwq))\
+				& (hwq->max_elements - 1)))
+
+#define MAX_PBL_LVL_0_PGS		1
+#define MAX_PBL_LVL_1_PGS		512
+#define MAX_PBL_LVL_1_PGS_SHIFT		9
+#define MAX_PBL_LVL_1_PGS_FOR_LVL_2	256
+#define MAX_PBL_LVL_2_PGS		(256 * 512)
+#define MAX_PDL_LVL_SHIFT               9
+
+#define BNG_RE_DBR_VALID		(0x1UL << 26)
+#define BNG_RE_DBR_EPOCH_SHIFT	24
+#define BNG_RE_DBR_TOGGLE_SHIFT	25
+
+#define BNG_MAX_TQM_ALLOC_REQ	48
+
+struct bng_re_reg_desc {
+	u8		bar_id;
+	resource_size_t	bar_base;
+	unsigned long	offset;
+	void __iomem	*bar_reg;
+	size_t		len;
+};
+
+struct bng_re_db_info {
+	void __iomem		*db;
+	void __iomem		*priv_db;
+	struct bng_re_hwq	*hwq;
+	u32			xid;
+	u32			max_slot;
+	u32                     flags;
+	u8			toggle;
+};
+
+enum bng_re_db_info_flags_mask {
+	BNG_RE_FLAG_EPOCH_CONS_SHIFT        = 0x0UL,
+	BNG_RE_FLAG_EPOCH_PROD_SHIFT        = 0x1UL,
+	BNG_RE_FLAG_EPOCH_CONS_MASK         = 0x1UL,
+	BNG_RE_FLAG_EPOCH_PROD_MASK         = 0x2UL,
+};
+
+enum bng_re_db_epoch_flag_shift {
+	BNG_RE_DB_EPOCH_CONS_SHIFT  = BNG_RE_DBR_EPOCH_SHIFT,
+	BNG_RE_DB_EPOCH_PROD_SHIFT  = (BNG_RE_DBR_EPOCH_SHIFT - 1),
+};
+
+struct bng_re_chip_ctx {
+	u16	chip_num;
+	u16	hw_stats_size;
+	u64	hwrm_intf_ver;
+	u16	hwrm_cmd_max_timeout;
+};
+
+struct bng_re_pbl {
+	u32		pg_count;
+	u32		pg_size;
+	void		**pg_arr;
+	dma_addr_t	*pg_map_arr;
+};
+
+enum bng_re_pbl_lvl {
+	BNG_PBL_LVL_0,
+	BNG_PBL_LVL_1,
+	BNG_PBL_LVL_2,
+	BNG_PBL_LVL_MAX
+};
+
+enum bng_re_hwq_type {
+	BNG_HWQ_TYPE_CTX,
+	BNG_HWQ_TYPE_QUEUE
+};
+
+struct bng_re_sg_info {
+	u32	npages;
+	u32	pgshft;
+	u32	pgsize;
+	bool	nopte;
+};
+
+struct bng_re_hwq_attr {
+	struct bng_re_res		*res;
+	struct bng_re_sg_info		*sginfo;
+	enum bng_re_hwq_type		type;
+	u32				depth;
+	u32				stride;
+	u32				aux_stride;
+	u32				aux_depth;
+};
+
+struct bng_re_hwq {
+	struct pci_dev			*pdev;
+	/* lock to protect hwq */
+	spinlock_t			lock;
+	struct bng_re_pbl		pbl[BNG_PBL_LVL_MAX + 1];
+	/* Valid values: 0, 1, 2 */
+	enum bng_re_pbl_lvl		level;
+	/* PBL entries */
+	void				**pbl_ptr;
+	/* PBL  dma_addr */
+	dma_addr_t			*pbl_dma_ptr;
+	u32				max_elements;
+	u32				depth;
+	u16				element_size;
+	u32				prod;
+	u32				cons;
+	/* queue entry per page */
+	u16				qe_ppg;
+};
+
+struct bng_re_stats {
+	dma_addr_t			dma_map;
+	void				*dma;
+	u32				size;
+	u32				fw_id;
+};
+
+struct bng_re_res {
+	struct pci_dev			*pdev;
+	struct bng_re_chip_ctx		*cctx;
+	struct bng_re_dev_attr		*dattr;
+};
+
+static inline void *bng_re_get_qe(struct bng_re_hwq *hwq,
+				  u32 indx, u64 *pg)
+{
+	u32 pg_num, pg_idx;
+
+	pg_num = (indx / hwq->qe_ppg);
+	pg_idx = (indx % hwq->qe_ppg);
+	if (pg)
+		*pg = (u64)&hwq->pbl_ptr[pg_num];
+	return (void *)(hwq->pbl_ptr[pg_num] + hwq->element_size * pg_idx);
+}
+
+#define BNG_RE_INIT_DBHDR(xid, type, indx, toggle) \
+	(((u64)(((xid) & DBC_DBC_XID_MASK) | DBC_DBC_PATH_ROCE |  \
+		(type) | BNG_RE_DBR_VALID) << 32) | (indx) |  \
+	 (((u32)(toggle)) << (BNG_RE_DBR_TOGGLE_SHIFT)))
+
+static inline void bng_re_ring_db(struct bng_re_db_info *info,
+				  u32 type)
+{
+	u64 key = 0;
+	u32 indx;
+	u8 toggle = 0;
+
+	if (type == DBC_DBC_TYPE_CQ_ARMALL ||
+	    type == DBC_DBC_TYPE_CQ_ARMSE)
+		toggle = info->toggle;
+
+	indx = (info->hwq->cons & DBC_DBC_INDEX_MASK) |
+	       ((info->flags & BNG_RE_FLAG_EPOCH_CONS_MASK) <<
+		 BNG_RE_DB_EPOCH_CONS_SHIFT);
+
+	key =  BNG_RE_INIT_DBHDR(info->xid, type, indx, toggle);
+	writeq(key, info->db);
+}
+
+static inline void bng_re_ring_nq_db(struct bng_re_db_info *info,
+				     struct bng_re_chip_ctx *cctx,
+				     bool arm)
+{
+	u32 type;
+
+	type = arm ? DBC_DBC_TYPE_NQ_ARM : DBC_DBC_TYPE_NQ;
+	bng_re_ring_db(info, type);
+}
+
+static inline void bng_re_hwq_incr_cons(u32 max_elements, u32 *cons, u32 cnt,
+					u32 *dbinfo_flags)
+{
+	/* move cons and update toggle/epoch if wrap around */
+	*cons += cnt;
+	if (*cons >= max_elements) {
+		*cons %= max_elements;
+		*dbinfo_flags ^= 1UL << BNG_RE_FLAG_EPOCH_CONS_SHIFT;
+	}
+}
+
+static inline bool _is_max_srq_ext_supported(u16 dev_cap_ext_flags_2)
+{
+	return !!(dev_cap_ext_flags_2 & CREQ_QUERY_FUNC_RESP_SB_MAX_SRQ_EXTENDED);
+}
+
+void bng_re_free_hwq(struct bng_re_res *res,
+		     struct bng_re_hwq *hwq);
+
+int bng_re_alloc_init_hwq(struct bng_re_hwq *hwq,
+			  struct bng_re_hwq_attr *hwq_attr);
+
+void bng_re_free_stats_ctx_mem(struct pci_dev *pdev,
+			       struct bng_re_stats *stats);
+
+int bng_re_alloc_stats_ctx_mem(struct pci_dev *pdev,
+			       struct bng_re_chip_ctx *cctx,
+			       struct bng_re_stats *stats);
+#endif

drivers/infiniband/hw/bng_re/bng_sp.c

@@ -0,0 +1,131 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2025 Broadcom.
+#include <linux/interrupt.h>
+#include <linux/pci.h>
+
+#include "bng_res.h"
+#include "bng_fw.h"
+#include "bng_sp.h"
+#include "bng_tlv.h"
+
+static bool bng_re_is_atomic_cap(struct bng_re_rcfw *rcfw)
+{
+	u16 pcie_ctl2 = 0;
+
+	pcie_capability_read_word(rcfw->pdev, PCI_EXP_DEVCTL2, &pcie_ctl2);
+	return (pcie_ctl2 & PCI_EXP_DEVCTL2_ATOMIC_REQ);
+}
+
+static void bng_re_query_version(struct bng_re_rcfw *rcfw,
+				 char *fw_ver)
+{
+	struct creq_query_version_resp resp = {};
+	struct bng_re_cmdqmsg msg = {};
+	struct cmdq_query_version req = {};
+	int rc;
+
+	bng_re_rcfw_cmd_prep((struct cmdq_base *)&req,
+			     CMDQ_BASE_OPCODE_QUERY_VERSION,
+			     sizeof(req));
+
+	bng_re_fill_cmdqmsg(&msg, &req, &resp, NULL, sizeof(req), sizeof(resp), 0);
+	rc = bng_re_rcfw_send_message(rcfw, &msg);
+	if (rc)
+		return;
+	fw_ver[0] = resp.fw_maj;
+	fw_ver[1] = resp.fw_minor;
+	fw_ver[2] = resp.fw_bld;
+	fw_ver[3] = resp.fw_rsvd;
+}
+
+int bng_re_get_dev_attr(struct bng_re_rcfw *rcfw)
+{
+	struct bng_re_dev_attr *attr = rcfw->res->dattr;
+	struct creq_query_func_resp resp = {};
+	struct bng_re_cmdqmsg msg = {};
+	struct creq_query_func_resp_sb *sb;
+	struct bng_re_rcfw_sbuf sbuf;
+	struct cmdq_query_func req = {};
+	u8 *tqm_alloc;
+	int i, rc;
+	u32 temp;
+
+	bng_re_rcfw_cmd_prep((struct cmdq_base *)&req,
+			     CMDQ_BASE_OPCODE_QUERY_FUNC,
+			     sizeof(req));
+
+	sbuf.size = ALIGN(sizeof(*sb), BNG_FW_CMDQE_UNITS);
+	sbuf.sb = dma_alloc_coherent(&rcfw->pdev->dev, sbuf.size,
+				     &sbuf.dma_addr, GFP_KERNEL);
+	if (!sbuf.sb)
+		return -ENOMEM;
+	sb = sbuf.sb;
+	req.resp_size = sbuf.size / BNG_FW_CMDQE_UNITS;
+	bng_re_fill_cmdqmsg(&msg, &req, &resp, &sbuf, sizeof(req),
+			    sizeof(resp), 0);
+	rc = bng_re_rcfw_send_message(rcfw, &msg);
+	if (rc)
+		goto bail;
+	/* Extract the context from the side buffer */
+	attr->max_qp = le32_to_cpu(sb->max_qp);
+	/* max_qp value reported by FW doesn't include the QP1 */
+	attr->max_qp += 1;
+	attr->max_qp_rd_atom =
+		sb->max_qp_rd_atom > BNG_RE_MAX_OUT_RD_ATOM ?
+		BNG_RE_MAX_OUT_RD_ATOM : sb->max_qp_rd_atom;
+	attr->max_qp_init_rd_atom =
+		sb->max_qp_init_rd_atom > BNG_RE_MAX_OUT_RD_ATOM ?
+		BNG_RE_MAX_OUT_RD_ATOM : sb->max_qp_init_rd_atom;
+	attr->max_qp_wqes = le16_to_cpu(sb->max_qp_wr) - 1;
+
+	/* Adjust for max_qp_wqes for variable wqe */
+	attr->max_qp_wqes = min_t(u32, attr->max_qp_wqes, BNG_VAR_MAX_WQE - 1);
+
+	attr->max_qp_sges = min_t(u32, sb->max_sge_var_wqe, BNG_VAR_MAX_SGE);
+	attr->max_cq = le32_to_cpu(sb->max_cq);
+	attr->max_cq_wqes = le32_to_cpu(sb->max_cqe);
+	attr->max_cq_sges = attr->max_qp_sges;
+	attr->max_mr = le32_to_cpu(sb->max_mr);
+	attr->max_mw = le32_to_cpu(sb->max_mw);
+
+	attr->max_mr_size = le64_to_cpu(sb->max_mr_size);
+	attr->max_pd = 64 * 1024;
+	attr->max_raw_ethy_qp = le32_to_cpu(sb->max_raw_eth_qp);
+	attr->max_ah = le32_to_cpu(sb->max_ah);
+
+	attr->max_srq = le16_to_cpu(sb->max_srq);
+	attr->max_srq_wqes = le32_to_cpu(sb->max_srq_wr) - 1;
+	attr->max_srq_sges = sb->max_srq_sge;
+	attr->max_pkey = 1;
+	attr->max_inline_data = le32_to_cpu(sb->max_inline_data);
+	/*
+	 * Read the max gid supported by HW.
+	 * For each entry in HW  GID in HW table, we consume 2
+	 * GID entries in the kernel GID table.  So max_gid reported
+	 * to stack can be up to twice the value reported by the HW, up to 256 gids.
+	 */
+	attr->max_sgid = le32_to_cpu(sb->max_gid);
+	attr->max_sgid = min_t(u32, BNG_RE_NUM_GIDS_SUPPORTED, 2 * attr->max_sgid);
+	attr->dev_cap_flags = le16_to_cpu(sb->dev_cap_flags);
+	attr->dev_cap_flags2 = le16_to_cpu(sb->dev_cap_ext_flags_2);
+
+	if (_is_max_srq_ext_supported(attr->dev_cap_flags2))
+		attr->max_srq += le16_to_cpu(sb->max_srq_ext);
+
+	bng_re_query_version(rcfw, attr->fw_ver);
+	for (i = 0; i < BNG_MAX_TQM_ALLOC_REQ / 4; i++) {
+		temp = le32_to_cpu(sb->tqm_alloc_reqs[i]);
+		tqm_alloc = (u8 *)&temp;
+		attr->tqm_alloc_reqs[i * 4] = *tqm_alloc;
+		attr->tqm_alloc_reqs[i * 4 + 1] = *(++tqm_alloc);
+		attr->tqm_alloc_reqs[i * 4 + 2] = *(++tqm_alloc);
+		attr->tqm_alloc_reqs[i * 4 + 3] = *(++tqm_alloc);
+	}
+
+	attr->max_dpi = le32_to_cpu(sb->max_dpi);
+	attr->is_atomic = bng_re_is_atomic_cap(rcfw);
+bail:
+	dma_free_coherent(&rcfw->pdev->dev, sbuf.size,
+			  sbuf.sb, sbuf.dma_addr);
+	return rc;
+}

drivers/infiniband/hw/bng_re/bng_sp.h

@@ -0,0 +1,47 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+// Copyright (c) 2025 Broadcom.
+
+#ifndef __BNG_SP_H__
+#define __BNG_SP_H__
+
+#include "bng_fw.h"
+
+#define BNG_VAR_MAX_WQE		4352
+#define BNG_VAR_MAX_SGE		13
+
+struct bng_re_dev_attr {
+#define FW_VER_ARR_LEN			4
+	u8				fw_ver[FW_VER_ARR_LEN];
+#define BNG_RE_NUM_GIDS_SUPPORTED	256
+	u16				max_sgid;
+	u16				max_mrw;
+	u32				max_qp;
+#define BNG_RE_MAX_OUT_RD_ATOM		126
+	u32				max_qp_rd_atom;
+	u32				max_qp_init_rd_atom;
+	u32				max_qp_wqes;
+	u32				max_qp_sges;
+	u32				max_cq;
+	u32				max_cq_wqes;
+	u32				max_cq_sges;
+	u32				max_mr;
+	u64				max_mr_size;
+	u32				max_pd;
+	u32				max_mw;
+	u32				max_raw_ethy_qp;
+	u32				max_ah;
+	u32				max_srq;
+	u32				max_srq_wqes;
+	u32				max_srq_sges;
+	u32				max_pkey;
+	u32				max_inline_data;
+	u32				l2_db_size;
+	u8				tqm_alloc_reqs[BNG_MAX_TQM_ALLOC_REQ];
+	bool				is_atomic;
+	u16                             dev_cap_flags;
+	u16                             dev_cap_flags2;
+	u32                             max_dpi;
+};
+
+int bng_re_get_dev_attr(struct bng_re_rcfw *rcfw);
+#endif

drivers/infiniband/hw/bng_re/bng_tlv.h

@@ -0,0 +1,128 @@
+/* SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause */
+
+#ifndef __BNG_TLV_H__
+#define __BNG_TLV_H__
+
+#include "roce_hsi.h"
+
+struct roce_tlv {
+	struct tlv tlv;
+	u8 total_size; // in units of 16 byte chunks
+	u8 unused[7];  // for 16 byte alignment
+};
+
+/*
+ * TLV size in units of 16 byte chunks
+ */
+#define TLV_SIZE ((sizeof(struct roce_tlv) + 15) / 16)
+/*
+ * TLV length in bytes
+ */
+#define TLV_BYTES (TLV_SIZE * 16)
+
+#define HAS_TLV_HEADER(msg) (le16_to_cpu(((struct tlv *)(msg))->cmd_discr) == CMD_DISCR_TLV_ENCAP)
+#define GET_TLV_DATA(tlv)   ((void *)&((uint8_t *)(tlv))[TLV_BYTES])
+
+static inline u8 __get_cmdq_base_opcode(struct cmdq_base *req, u32 size)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		return ((struct cmdq_base *)GET_TLV_DATA(req))->opcode;
+	else
+		return req->opcode;
+}
+
+static inline void __set_cmdq_base_opcode(struct cmdq_base *req,
+					  u32 size, u8 val)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		((struct cmdq_base *)GET_TLV_DATA(req))->opcode = val;
+	else
+		req->opcode = val;
+}
+
+static inline __le16 __get_cmdq_base_cookie(struct cmdq_base *req, u32 size)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		return ((struct cmdq_base *)GET_TLV_DATA(req))->cookie;
+	else
+		return req->cookie;
+}
+
+static inline void __set_cmdq_base_cookie(struct cmdq_base *req,
+					  u32 size, __le16 val)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		((struct cmdq_base *)GET_TLV_DATA(req))->cookie = val;
+	else
+		req->cookie = val;
+}
+
+static inline __le64 __get_cmdq_base_resp_addr(struct cmdq_base *req, u32 size)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		return ((struct cmdq_base *)GET_TLV_DATA(req))->resp_addr;
+	else
+		return req->resp_addr;
+}
+
+static inline void __set_cmdq_base_resp_addr(struct cmdq_base *req,
+					     u32 size, __le64 val)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		((struct cmdq_base *)GET_TLV_DATA(req))->resp_addr = val;
+	else
+		req->resp_addr = val;
+}
+
+static inline u8 __get_cmdq_base_resp_size(struct cmdq_base *req, u32 size)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		return ((struct cmdq_base *)GET_TLV_DATA(req))->resp_size;
+	else
+		return req->resp_size;
+}
+
+static inline void __set_cmdq_base_resp_size(struct cmdq_base *req,
+					     u32 size, u8 val)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		((struct cmdq_base *)GET_TLV_DATA(req))->resp_size = val;
+	else
+		req->resp_size = val;
+}
+
+static inline u8 __get_cmdq_base_cmd_size(struct cmdq_base *req, u32 size)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		return ((struct roce_tlv *)(req))->total_size;
+	else
+		return req->cmd_size;
+}
+
+static inline void __set_cmdq_base_cmd_size(struct cmdq_base *req,
+					    u32 size, u8 val)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		((struct cmdq_base *)GET_TLV_DATA(req))->cmd_size = val;
+	else
+		req->cmd_size = val;
+}
+
+static inline __le16 __get_cmdq_base_flags(struct cmdq_base *req, u32 size)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		return ((struct cmdq_base *)GET_TLV_DATA(req))->flags;
+	else
+		return req->flags;
+}
+
+static inline void __set_cmdq_base_flags(struct cmdq_base *req,
+					 u32 size, __le16 val)
+{
+	if (HAS_TLV_HEADER(req) && size > TLV_BYTES)
+		((struct cmdq_base *)GET_TLV_DATA(req))->flags = val;
+	else
+		req->flags = val;
+}
+
+#endif /* __BNG_TLV_H__ */

drivers/infiniband/hw/bnxt_re/bnxt_re.h

@@ -224,6 +224,8 @@ struct bnxt_re_dev {
 	struct workqueue_struct		*dcb_wq;
 	struct dentry                   *cc_config;
 	struct bnxt_re_dbg_cc_config_params *cc_config_params;
+	struct dentry			*cq_coal_cfg;
+	struct bnxt_re_dbg_cq_coal_params *cq_coal_cfg_params;
 #define BNXT_VPD_FLD_LEN		32
 	char board_partno[BNXT_VPD_FLD_LEN];
 	/* RoCE mirror */

drivers/infiniband/hw/bnxt_re/debugfs.c

@@ -23,6 +23,14 @@
 
 static struct dentry *bnxt_re_debugfs_root;
 
+static const char * const bnxt_re_cq_coal_str[] = {
+	"buf_maxtime",
+	"normal_maxbuf",
+	"during_maxbuf",
+	"en_ring_idle_mode",
+	"enable",
+};
+
 static const char * const bnxt_re_cc_gen0_name[] = {
 	"enable_cc",
 	"run_avg_weight_g",
@@ -349,6 +357,123 @@ static void bnxt_re_debugfs_add_info(struct bnxt_re_dev *rdev)
 	debugfs_create_file("info", 0400, rdev->dbg_root, rdev, &info_fops);
 }
 
+static ssize_t cq_coal_cfg_write(struct file *file,
+				 const char __user *buf,
+				 size_t count, loff_t *pos)
+{
+	struct seq_file *s = file->private_data;
+	struct bnxt_re_cq_coal_param *param = s->private;
+	struct bnxt_re_dev *rdev = param->rdev;
+	int offset = param->offset;
+	char lbuf[16] = { };
+	u32 val;
+
+	if (count > sizeof(lbuf))
+		return -EINVAL;
+
+	if (copy_from_user(lbuf, buf, count))
+		return -EFAULT;
+
+	lbuf[sizeof(lbuf) - 1] = '\0';
+
+	if (kstrtou32(lbuf, 0, &val))
+		return -EINVAL;
+
+	switch (offset) {
+	case BNXT_RE_COAL_CQ_BUF_MAXTIME:
+		if (val < 1 || val > BNXT_QPLIB_CQ_COAL_MAX_BUF_MAXTIME)
+			return -EINVAL;
+		rdev->cq_coalescing.buf_maxtime = val;
+		break;
+	case BNXT_RE_COAL_CQ_NORMAL_MAXBUF:
+		if (val < 1 || val > BNXT_QPLIB_CQ_COAL_MAX_NORMAL_MAXBUF)
+			return -EINVAL;
+		rdev->cq_coalescing.normal_maxbuf = val;
+		break;
+	case BNXT_RE_COAL_CQ_DURING_MAXBUF:
+		if (val < 1 || val > BNXT_QPLIB_CQ_COAL_MAX_DURING_MAXBUF)
+			return -EINVAL;
+		rdev->cq_coalescing.during_maxbuf = val;
+		break;
+	case BNXT_RE_COAL_CQ_EN_RING_IDLE_MODE:
+		if (val > BNXT_QPLIB_CQ_COAL_MAX_EN_RING_IDLE_MODE)
+			return -EINVAL;
+		rdev->cq_coalescing.en_ring_idle_mode = val;
+		break;
+	case BNXT_RE_COAL_CQ_ENABLE:
+		if (val > 1)
+			return -EINVAL;
+		rdev->cq_coalescing.enable = val;
+		break;
+	default:
+		return -EINVAL;
+	}
+	return  count;
+}
+
+static int cq_coal_cfg_show(struct seq_file *s, void *unused)
+{
+	struct bnxt_re_cq_coal_param *param = s->private;
+	struct bnxt_re_dev *rdev = param->rdev;
+	int offset = param->offset;
+	u32 val = 0;
+
+	switch (offset) {
+	case BNXT_RE_COAL_CQ_BUF_MAXTIME:
+		val = rdev->cq_coalescing.buf_maxtime;
+		break;
+	case BNXT_RE_COAL_CQ_NORMAL_MAXBUF:
+		val = rdev->cq_coalescing.normal_maxbuf;
+		break;
+	case BNXT_RE_COAL_CQ_DURING_MAXBUF:
+		val = rdev->cq_coalescing.during_maxbuf;
+		break;
+	case BNXT_RE_COAL_CQ_EN_RING_IDLE_MODE:
+		val = rdev->cq_coalescing.en_ring_idle_mode;
+		break;
+	case BNXT_RE_COAL_CQ_ENABLE:
+		val = rdev->cq_coalescing.enable;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	seq_printf(s, "%u\n", val);
+	return 0;
+}
+DEFINE_SHOW_STORE_ATTRIBUTE(cq_coal_cfg);
+
+static void bnxt_re_cleanup_cq_coal_debugfs(struct bnxt_re_dev *rdev)
+{
+	debugfs_remove_recursive(rdev->cq_coal_cfg);
+	kfree(rdev->cq_coal_cfg_params);
+}
+
+static void bnxt_re_init_cq_coal_debugfs(struct bnxt_re_dev *rdev)
+{
+	struct bnxt_re_dbg_cq_coal_params *dbg_cq_coal_params;
+	int i;
+
+	if (!_is_cq_coalescing_supported(rdev->dev_attr->dev_cap_flags2))
+		return;
+
+	dbg_cq_coal_params = kzalloc(sizeof(*dbg_cq_coal_params), GFP_KERNEL);
+	if (!dbg_cq_coal_params)
+		return;
+
+	rdev->cq_coal_cfg = debugfs_create_dir("cq_coal_cfg", rdev->dbg_root);
+	rdev->cq_coal_cfg_params = dbg_cq_coal_params;
+
+	for (i = 0; i < BNXT_RE_COAL_CQ_MAX; i++) {
+		dbg_cq_coal_params->params[i].offset = i;
+		dbg_cq_coal_params->params[i].rdev = rdev;
+		debugfs_create_file(bnxt_re_cq_coal_str[i],
+				    0600, rdev->cq_coal_cfg,
+				    &dbg_cq_coal_params->params[i],
+				    &cq_coal_cfg_fops);
+	}
+}
+
 void bnxt_re_debugfs_add_pdev(struct bnxt_re_dev *rdev)
 {
 	struct pci_dev *pdev = rdev->en_dev->pdev;
@@ -374,10 +499,13 @@ void bnxt_re_debugfs_add_pdev(struct bnxt_re_dev *rdev)
 							 rdev->cc_config, tmp_params,
 							 &bnxt_re_cc_config_ops);
 	}
+
+	bnxt_re_init_cq_coal_debugfs(rdev);
 }
 
 void bnxt_re_debugfs_rem_pdev(struct bnxt_re_dev *rdev)
 {
+	bnxt_re_cleanup_cq_coal_debugfs(rdev);
 	debugfs_remove_recursive(rdev->qp_debugfs);
 	debugfs_remove_recursive(rdev->cc_config);
 	kfree(rdev->cc_config_params);

drivers/infiniband/hw/bnxt_re/debugfs.h

@@ -33,4 +33,23 @@ struct bnxt_re_cc_param {
 struct bnxt_re_dbg_cc_config_params {
 	struct bnxt_re_cc_param	gen0_parms[BNXT_RE_CC_PARAM_GEN0];
 };
+
+struct bnxt_re_cq_coal_param {
+	struct bnxt_re_dev	*rdev;
+	u32			offset;
+};
+
+enum bnxt_re_cq_coal_types {
+	BNXT_RE_COAL_CQ_BUF_MAXTIME,
+	BNXT_RE_COAL_CQ_NORMAL_MAXBUF,
+	BNXT_RE_COAL_CQ_DURING_MAXBUF,
+	BNXT_RE_COAL_CQ_EN_RING_IDLE_MODE,
+	BNXT_RE_COAL_CQ_ENABLE,
+	BNXT_RE_COAL_CQ_MAX
+
+};
+
+struct bnxt_re_dbg_cq_coal_params {
+	struct bnxt_re_cq_coal_param	params[BNXT_RE_COAL_CQ_MAX];
+};
 #endif

drivers/infiniband/hw/bnxt_re/ib_verbs.c

@@ -601,7 +601,8 @@ static int bnxt_re_create_fence_mr(struct bnxt_re_pd *pd)
 	mr->qplib_mr.va = (u64)(unsigned long)fence->va;
 	mr->qplib_mr.total_size = BNXT_RE_FENCE_BYTES;
 	rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, NULL,
-			       BNXT_RE_FENCE_PBL_SIZE, PAGE_SIZE);
+			       BNXT_RE_FENCE_PBL_SIZE, PAGE_SIZE,
+			       _is_alloc_mr_unified(rdev->dev_attr->dev_cap_flags));
 	if (rc) {
 		ibdev_err(&rdev->ibdev, "Failed to register fence-MR\n");
 		goto fail;
@@ -4027,7 +4028,7 @@ struct ib_mr *bnxt_re_get_dma_mr(struct ib_pd *ib_pd, int mr_access_flags)
 	mr->qplib_mr.hwq.level = PBL_LVL_MAX;
 	mr->qplib_mr.total_size = -1; /* Infinte length */
 	rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, NULL, 0,
-			       PAGE_SIZE);
+			       PAGE_SIZE, false);
 	if (rc)
 		goto fail_mr;
 
@@ -4257,7 +4258,8 @@ static struct ib_mr *__bnxt_re_user_reg_mr(struct ib_pd *ib_pd, u64 length, u64
 
 	umem_pgs = ib_umem_num_dma_blocks(umem, page_size);
 	rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, umem,
-			       umem_pgs, page_size);
+			       umem_pgs, page_size,
+			       _is_alloc_mr_unified(rdev->dev_attr->dev_cap_flags));
 	if (rc) {
 		ibdev_err(&rdev->ibdev, "Failed to register user MR - rc = %d\n", rc);
 		rc = -EIO;

drivers/infiniband/hw/bnxt_re/main.c

@@ -1453,6 +1453,7 @@ static struct bnxt_re_dev *bnxt_re_dev_add(struct auxiliary_device *adev,
 	atomic_set(&rdev->stats.res.pd_count, 0);
 	rdev->cosq[0] = 0xFFFF;
 	rdev->cosq[1] = 0xFFFF;
+	rdev->cq_coalescing.enable = 1;
 	rdev->cq_coalescing.buf_maxtime = BNXT_QPLIB_CQ_COAL_DEF_BUF_MAXTIME;
 	if (bnxt_re_chip_gen_p7(en_dev->chip_num)) {
 		rdev->cq_coalescing.normal_maxbuf = BNXT_QPLIB_CQ_COAL_DEF_NORMAL_MAXBUF_P7;

drivers/infiniband/hw/bnxt_re/qplib_fp.c

@@ -2226,7 +2226,8 @@ int bnxt_qplib_create_cq(struct bnxt_qplib_res *res, struct bnxt_qplib_cq *cq)
 	req.cq_handle = cpu_to_le64(cq->cq_handle);
 	req.cq_size = cpu_to_le32(cq->max_wqe);
 
-	if (_is_cq_coalescing_supported(res->dattr->dev_cap_flags2)) {
+	if (_is_cq_coalescing_supported(res->dattr->dev_cap_flags2) &&
+	    cq->coalescing->enable) {
 		req.flags |= cpu_to_le16(CMDQ_CREATE_CQ_FLAGS_COALESCING_VALID);
 		coalescing |= ((cq->coalescing->buf_maxtime <<
 				CMDQ_CREATE_CQ_BUF_MAXTIME_SFT) &