summaryrefslogtreecommitdiff
path: root/drivers
AgeCommit message (Collapse)Author
2026-10-15HID: roccat: koneplus: reject short button reportsJiale Yao
The raw event callback runs before HID core validates and zero-pads the report. It passes the buffer to helpers which inspect data[0] and, for a button report, read the complete struct koneplus_mouse_report_button. A truncated report can therefore cause an out-of-bounds read. Require enough data for the button report before calling either helper. Commit 47669bec44fe ("HID: asus: refactor the two workqueues and init sequence") added the same kind of raw-event length validation to hid-asus. Fixes: 47dbdbffe15b ("HID: roccat: Add support for Roccat Kone[+] v2") Cc: stable@vger.kernel.org Signed-off-by: Jiale Yao <yaojiale02@163.com> Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>
2026-10-15HID: roccat: isku: reject short button reportsJiale Yao
The raw event callback runs before HID core validates and zero-pads the report. It passes the buffer to helpers which inspect data[0] and, for a button report, read the complete struct isku_report_button. A truncated report can therefore cause an out-of-bounds read. Require enough data for the button report before calling either helper. Commit 47669bec44fe ("HID: asus: refactor the two workqueues and init sequence") added the same kind of raw-event length validation to hid-asus. Fixes: d41c2a7011df ("HID: roccat: Add support for Isku keyboard") Cc: stable@vger.kernel.org Signed-off-by: Jiale Yao <yaojiale02@163.com> Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>
2026-10-15leds: pca955x: Drop platform data and allocate LEDs with the deviceRosen Penev
struct pca955x_platform_data has been private to this driver since commit ed1f4b9676a8 ("leds: pca955x: add device tree support"), so nothing can pass it in through dev_get_platdata() and it only serves as an intermediate copy of the per-LED type, default state and fwnode. That copy is a full array of struct pca955x_led, the same runtime structure that struct pca955x already holds, so the data is stored twice. Remove the platform data and parse the child firmware nodes directly into pca955x->leds[]. Make leds[] a flexible array member so it is allocated together with struct pca955x. This removes three allocations and the num_leds sanity check, which could no longer fail. struct pca955x_led moves above struct pca955x, as a flexible array member needs the complete element type. Assisted-by: LLM Signed-off-by: Rosen Penev <rosenp@gmail.com> Link: https://patch.msgid.link/20261005200906.188636-1-rosenp@gmail.com Signed-off-by: Lee Jones <lee@kernel.org>
2026-10-15iommu/hyperv: Make hv_iommu_device staticYu Zhang
Make hv_iommu_device static to give it internal linkage. This pointer is only used within hv-iommu-guest.c, but previously has external linkage without a prior declaration. This causes sparse to warn that the symbol was not declared. Fixes: 2690d4156b51 ("iommu/hyperv: Add para-virtualized IOMMU support for Hyper-V guest") Reported-by: kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202610092126.TuSXnTYR-lkp@intel.com/ Signed-off-by: Yu Zhang <zhangyu1@linux.microsoft.com> Reviewed-by: Jacob Pan <jacob.pan@linux.microsoft.com> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
2026-10-15iommu/hyperv: Add missing PCI dependency for the guest IOMMU driverYu Zhang
Make HYPERV_PVIOMMU depend on PCI, to fix the possible link failure of pci_request_acs(), if HYPERV_PVIOMMU is enabled yet with PCI disabled. Such configuration shall be prevented. Fixes: 2690d4156b51 ("iommu/hyperv: Add para-virtualized IOMMU support for Hyper-V guest") Reported-by: kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202610110511.Nk1sZ8SG-lkp@intel.com/ Signed-off-by: Yu Zhang <zhangyu1@linux.microsoft.com> Reviewed-by: Jacob Pan <jacob.pan@linux.microsoft.com> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
2026-10-15RDMA/irdma: Check MR contiguity before PBLE allocJacob Moroni
The original logic would attempt to allocate PBLEs, then would check whether the region was contiguous _after_, and if so, would free the PBLEs and use a direct mapping instead. These unnecessary temporary PBLE allocations can technically fail (PBLEs are a finite resource), which can cause an MR registration failure even if the region is contiguous and should have succeeded. Fix this by checking whether the region is contiguous before allocating any PBLEs. This also provides an opportunity to consolidate some of the rereg_mr logic and use the existing ib_umem_is_contiguous() helper. Tested on GEN3 HW with a test program that does an extensive sweep of MR size/alignment/hugepage combinations, but most importantly, first/last byte reachable and -1/+1 unreachable when using contig regions with both aligned and unaligned start addresses, plus a kprobe on irdma_get_pble() to ensure we're actually hitting the new path. Ring MRs still have their own logic. That will be addressed next. Signed-off-by: Jacob Moroni <jmoroni@google.com> Link: https://patch.msgid.link/20261014162144.4053279-1-jmoroni@google.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15mfd: intel-lpss: Add Dell Latitude 7275 to the resource conflict quirkD. Manresa
The Dell Latitude 7275 2-in-1 has the same BIOS bug as the Latitude 5285: the ACPI GEXP device declares an OperationRegion (BAR0) on the GNVS variable SB04 and the I2C4 controller (INT3446) builds its _CRS from that same variable, so intel_lpss_acpi fails to bind the controller with -EBUSY and the front camera sensor (OV5670, INT3479) is never registered unless acpi_enforce_resources=lax is passed. Add the machine to the DMI table so QUIRK_IGNORE_RESOURCE_CONFLICTS is applied to INT3446 on it as well. Assisted-by: LLM Reviewed-by: Hans de Goede <johannes.goede@oss.qualcomm.com> Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: D. Manresa <dmanresa@gmail.com> Signed-off-by: Thierry Chatard <tchatard@gmail.com> Link: https://patch.msgid.link/20261002163350.1248431-3-tchatard@gmail.com Signed-off-by: Lee Jones <lee@kernel.org>
2026-10-15mfd: intel-lpss: Add resource conflict quirk for Dell Latitude 5285Thierry Chatard
The Dell Latitude 5285 2-in-1 has a BIOS bug where the ACPI GEXP device and the I2C4 controller (INT3446) both claim the same MMIO region via the shared SB04 variable. This causes intel_lpss_acpi to fail binding to I2C4 with -EBUSY, preventing the front camera (OV5670) sensor from being registered. Match the affected machine by DMI and the affected device by ACPI ID, carrying QUIRK_IGNORE_RESOURCE_CONFLICTS in the ACPI match data and applying it to the per-device info in probe. This follows the quirk organisation used by the PCI counterpart. Assisted-by: LLM Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by: Thierry Chatard <tchatard@gmail.com> Link: https://patch.msgid.link/20261002163350.1248431-2-tchatard@gmail.com Signed-off-by: Lee Jones <lee@kernel.org>
2026-10-15IB/mlx4: Avoid requeueing executing CM timeout workPraveen Kumar Kannoju
id->scheduled_delete is set by schedule_delayed() itself. It is used in two cases after the RTU timeout support was added: * A new id_map_entry is created, and a 60-second RTU timeout is scheduled. * DREQ or REJ is received for an existing entry, and the normal 30-second teardown timeout is scheduled or adjusted. The following race is possible when the RTU timeout expires while a DREQ or REJ is handled: CPU 0: timeout work CPU 1: DREQ/REJ handler -------------------- ----------------------- RTU timer expires timeout callback starts waits for id_map_lock takes id_map_lock calls schedule_delayed() calls mod_delayed_work() drops id_map_lock takes id_map_lock removes mapping frees id_map_entry requeued timeout runs accesses freed id_map_entry mod_delayed_work() may requeue the work after the timeout callback has started. The original timeout callback then frees the id_map_entry, leaving the requeued work with a dangling delayed_work pointer. When a timeout is already scheduled, first cancel its pending work. A successful cancellation means the callback has not started and the timeout can safely be scheduled again. If cancellation fails, the callback is executing and will free the entry after id_map_lock is released; do not requeue it. cancel_delayed_work() is deliberately used instead of cancel_delayed_work_sync(), because the timeout callback may be waiting for id_map_lock, which is held by the caller. Fixes: fa09adf4231f ("IB/mlx4: Fix stale CM id_map entries when RTU is never received") Signed-off-by: Praveen Kumar Kannoju <praveen.kannoju@oracle.com> Link: https://patch.msgid.link/20261014121344.100048-1-praveen.kannoju@oracle.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15mfd: tps65217: Use devres for IRQ domain and wake teardownRadiy Zhambakiev
Removing the parent device used to tear the IRQ domain down in tps65217_remove() before the MFD children are unbound: the i2c core only releases the parent's devres after the remove callback returns, whereas devm_mfd_add_devices() registered its own devres action which unbinds the children. Children that requested interrupts from the domain (e.g. tps65217-charger) therefore call free_irq() on virtual IRQs whose descriptors have already been disposed. With CONFIG_SPARSE_IRQ free_irq() then returns early without stopping the threaded handler, leaking the irqaction, the IRQ kthread and a module reference. Register the domain teardown as a devres action right after the domain is created instead. Devres actions are released in reverse order of registration, so the MFD children release their IRQs first, then the parent interrupt is freed and its thread stopped, and only then is the domain torn down. The same ordering covers probe failure, which no longer needs manual cleanup, and the remove callback can be dropped entirely. Balance the enable_irq_wake() in tps65217_irq_init() by registering disable_irq_wake() as a devres action as well. Only do so when enabling actually succeeded, since the parent chip may not support setting wake-up. Fixes: 6556bdacf646fcaa ("mfd: tps65217: Add support for IRQs") Cc: stable@vger.kernel.org Signed-off-by: Radiy Zhambakiev <r.zhambakiev@prosoftsystems.ru> Link: https://patch.msgid.link/20261002113624.178588-4-r.zhambakiev@prosoftsystems.ru Signed-off-by: Lee Jones <lee@kernel.org>
2026-10-15mfd: tps65217: Check return value when masking interrupt sourcesRadiy Zhambakiev
tps65217_irq_init() ignores the error returned by tps65217_set_bits() when masking all interrupt sources. A failed register write leaves the driver's software mask out of sync with the hardware and may result in spurious interrupts. Check the return value and propagate the error to the caller. Found by Linux Verification Center (linuxtesting.org) with SVACE. Fixes: 6556bdacf646fcaa ("mfd: tps65217: Add support for IRQs") Cc: stable@vger.kernel.org Signed-off-by: Radiy Zhambakiev <r.zhambakiev@prosoftsystems.ru> Link: https://patch.msgid.link/20261002113624.178588-3-r.zhambakiev@prosoftsystems.ru Signed-off-by: Lee Jones <lee@kernel.org>
2026-10-15mfd: tps65217: Fix NULL pointer dereference on IRQ init failureRadiy Zhambakiev
tps65217_probe() ignores the return value of tps65217_irq_init(), so when the irq domain creation fails the probe still completes and the driver ends up bound with a NULL tps->irq_domain. Unloading the module then makes tps65217_remove() call irq_domain_remove() on the NULL pointer and oops the kernel. On top of that, irq_find_mapping() may fall back to the default irq domain and dispose of mappings that belong to other interrupt controllers. Check the return value and abort the probe on failure so the error is reported and no inconsistent state is left for removal. Fixes: 6556bdacf646fcaa ("mfd: tps65217: Add support for IRQs") Cc: stable@vger.kernel.org Signed-off-by: Radiy Zhambakiev <r.zhambakiev@prosoftsystems.ru> Reviewed-by: Andreas Kemnade <andreas@kemnade.info> Link: https://patch.msgid.link/20261002113624.178588-2-r.zhambakiev@prosoftsystems.ru Signed-off-by: Lee Jones <lee@kernel.org>
2026-10-15PCI: endpoint: pci-epf-vntb: Track link state from both sidesKoichiro Den
The control-region link status is currently updated only by COMMAND_LINK_UP and COMMAND_LINK_DOWN from the HOST. The virtual NTB link callbacks are empty. Consequently, the HOST can see the link as up before the VHOST has enabled it, and ntb_link_disable() on the VHOST leaves LINK_STATUS_UP set without notifying either NTB client. ntb_netdev can hide this device-level state bug because ntb_transport uses a separate per-QP LINK_DOWN_FLAG message. Use ntb_tool to observe the device link state directly: After bringing both sides up, start one waiter on each side: HOST# echo N > /sys/kernel/debug/ntb_tool/<H-device>/peer0/link_event VHOST# echo N > /sys/kernel/debug/ntb_tool/<V-device>/peer0/link_event While both waiters are blocked, disable the link on the VHOST: VHOST# echo N > /sys/kernel/debug/ntb_tool/<V-device>/link Without this patch, the link-disable command succeeds, but peer0/link remains Y on both sides and neither waiter returns. Track HOST and VHOST enablement separately and report the effective link as up only when both sides have enabled it, as pci-epf-ntb does for its two physical hosts. Both NTB clients observe this effective state, so notify both sides only when it changes, including the side whose request completes the transition. Serialize the two paths because HOST commands run from delayed work while the VHOST callbacks may run concurrently. With this patch, peer0/link changes to N on both sides and both waiters return. Fixes: e35f56bb0330 ("PCI: endpoint: Support NTB transfer between RC and EP") Signed-off-by: Koichiro Den <den@valinux.co.jp> Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@oss.qualcomm.com> Reviewed-by: Frank Li <Frank.Li@nxp.com> Link: https://lore.kernel.org/r/20261014071448.2473208-1-den@valinux.co.jp/ Link: https://patch.msgid.link/20261015142716.3452900-1-den@valinux.co.jp
2026-10-15Merge tag 'net-7.3-rc8' of ↵Linus Torvalds
git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net Pull networking fixes from Paolo Abeni: "This is considerably smaller than in recent history, mostly due to reduced capacity and increased backlog. Also we actively redirected more stuff towards next. I'm not aware of any pending regression. Current release - regressions: - eth: move netops initialization to fix crash Previous releases - regressions: - sched: fix use-after-free in __tcf_action_put() - rds: fix uninitialized trans dereference in CM event handler Previous releases - always broken: - vsock: treat TCP_CLOSING as once-established - netlink: avoid hashing the network namespace pointer - eth: mlx5: HWS, release L2 to L3 tunnel reformats to their own pool" * tag 'net-7.3-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (23 commits) net: stmmac: dwmac1000: clear the RGMII status interrupt net/sched: fix use-after-free in __tcf_action_put() calipso: keep BIG TCP payload_len in calipso_skbuff_setattr() MAINTAINERS: add Gustavo Luiz Duarte as netconsole reviewer net: bcmgenet: unshare the skb before writing the control block microchip: lan865x: move setting of netops to fix crash net: ti: icssm-prueth: Move setting of netops to fix crash net: ethernet: aeroflex: move setting of netops to fix crash net: ethoc: move setting of netops to fix crash net: au1000: move setting of netops to fix crash net: mv643xx: move setting of netops to fix crash net: ethernet: adi: Move setting of netops to fix crash net: ethernet: actions: Move setting of netops to fix crash net: hisilicon: hip04_eth: Move setting of netops to fix crash net: hisilicon: hisi_femac: Move setting of netops to fix crash netlink: avoid hashing the network namespace pointer net/mlx5: HWS, fix flow meter offset within its bulk net/mlx5: HWS, release L2 to L3 tunnel reformats to their own pool netlink: reset skb headers before unicast and multicast filtering sctp: revalidate output stream after association connect wait ...
2026-10-15Merge branch 'for-7.4/cxl-fixes' into cxl-for-nextDave Jiang
cxl/events: Test CME counter expiry with equality, not bitwise AND
2026-10-15cxl/events: Test CME counter expiry with equality, not bitwise ANDGuixin Liu
The Memory Event Type field in the General Media and DRAM event records is an enumeration, not a bitmask. The validity checks for the advanced programmable CME counter expiration event (05h) test the field with a bitwise AND, so types 01h, 03h, 04h and 06h match it as well. A device reporting one of those types trips WARN_ON_ONCE once the general media or DRAM tracepoint is enabled, which is the case anywhere rasdaemon collects CXL events. Under panic_on_warn=1 that kills the machine. Compare the type field for equality instead. Type 05h records that do violate the spec conditions still warn. Fixes: cd3b36cfc659 ("cxl/events: Add extra validity checks for corrected memory error count in General Media Event Record") Fixes: d8145bb8af5c ("cxl/events: Add extra validity checks for CVME count in DRAM Event Record") Cc: stable@vger.kernel.org Assisted-by: LLM Signed-off-by: Guixin Liu <kanie@linux.alibaba.com> Reviewed-by: Dave Jiang <dave.jiang@intel.com> Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com> Link: https://patch.msgid.link/20260922082214.3526355-1-kanie@linux.alibaba.com Signed-off-by: Dave Jiang <dave.jiang@intel.com>
2026-10-15Merge tag 'hyperv-fixes-signed-20261015' of ↵Linus Torvalds
git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux Pull Hyper-V fixes from Wei Liu: - Typo and comment fixes (Hemanth Selam, Jiangshan Yi) - Update maintainers list (Wei Liu, Long Li) - Two Hyper-V PCI driver fixes (Naman Jain, Sahil Chandna) - Small fixes for core Hyper-V code (Hardik Garg, Michael Kelley, Waiman Long) * tag 'hyperv-fixes-signed-20261015' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux: MAINTAINERS: drop KY from Hyper-V maintainers list PCI: hv: Warn when wait_for_response() waits indefinitely PCI: hv: Probe vPCI buses asynchronously MAINTAINERS: remove myself from the Hyper-V entries mshv_vtl: Check per-CPU register page before mmap tools/hv: fix typo "retrive" in comment Drivers: hv: vmbus: fix typo "substract" in comment Drivers: hv: Don't reset VMBus connection state on error/exit paths x86/hyperv: Avoid using per-cpu output page in hv_apicid_to_vp_index() x86/hyperv: Fix stale comments after stimer message mode removal Drivers: hv: Avoid infinite retry loop in init_vp_index()
2026-10-15mfd: bq257xx: Mark self-modifying BQ25792 registers volatileAlexey Charkov
Some BQ25792 configuration registers change without being written: input source detection sets VINDPM and IINDPM at plug-in, plug-in also clears EN_HIZ, the FORCE_* bits clear once their operation starts or completes, EN_ACDRV1/2 follow input path changes, and ADC_EN clears after a one-shot conversion. Reading them from the cache returns stale values. Mark these registers volatile, keeping the cache for the rest. Cc: stable@vger.kernel.org Fixes: 9adbe53e32cd ("mfd: bq257xx: Add BQ25792 support") Signed-off-by: Alexey Charkov <alchark@flipper.net> Link: https://patch.msgid.link/20261001-bq25792-volatile-v1-1-cb3d4ec8a414@flipper.net Signed-off-by: Lee Jones <lee@kernel.org>
2026-10-15Revert "platform/x86: thinkpad_acpi: Add X9-14 Gen 1 to non-standard ↵Huang Wei
thermal/fan quirk tables" This reverts commit bb0f3a5d6b9844b70ad696b9470762028f6d86ec. Lenovo's firmware team has verified that the X9-14 Gen 1 uses EC register offsets that are unique to this platform, rather than the shared non-standard ECFW layout of the models already covered by the quirk tables. Applying the generic quirk would make the driver read the wrong registers and report bogus temperatures. Lenovo is working on the correct solution based on an updated EC register specification. Link: https://lore.kernel.org/all/000acb24-7aef-4718-89ea-0dffd1ab34d9@app.fastmail.com/ Signed-off-by: Huang Wei <huangwei@kylinos.cn> Link: https://patch.msgid.link/20261013070740.789934-1-huangwei@kylinos.cn Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86: msi-ec: Add MSI Creator Z16 A11UE EC firmwareAlexey Subbota
Add EC firmware version 1571EMS1.106 (MSI Creator Z16 A11UE) to the list of firmwares handled by CONF14. The EC register layout matches CONF14 field for field (verified against the out-of-tree msi-ec configuration for this firmware, where it is part of CONF_G2_10). Tested on a Creator Z16 A11UE with this driver built against 7.2.9: the module loads, charge_control_start_threshold and charge_control_end_threshold read the values set by the EC, and writing charge_control_end_threshold is applied and reads back. Assisted-by: LLM Signed-off-by: Alexey Subbota <subbota-job@yandex.ru> Link: https://patch.msgid.link/20261013212228.17349-1-subbota-job@yandex.ru Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15drm/panthor: Remove redundant drm_gem_create_mmap_offset() call from ↵Akash Goel
BO_MMAP_OFFSET ioctl This commit removes the call to drm_gem_create_mmap_offset() from the BO_MMAP_OFFSET ioctl handler. The call was redundant and didn't have any effect as file offset range is allocated unconditionally at the BO creation time. v2: - New patch Suggested-by: Boris Brezillon <boris.brezillon@collabora.com> Signed-off-by: Akash Goel <akash.goel@arm.com> Reviewed-by: Steven Price <steven.price@arm.com> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com> Signed-off-by: Liviu Dudau <liviu.dudau@arm.com> Link: https://patch.msgid.link/20261013125433.26432-2-akash.goel@arm.com
2026-10-15drm/panthor: Fix handling of NO_MMAP BOsAkash Goel
The BOs created with DRM_PANTHOR_BO_NO_MMAP flag are not supposed to be mmappable by the Userspace. But after the rework to support shrinker, it became possible to mmap NO_MMAP BOs. Userspace could second guess the file offset range, allocated unconditionally at the BO creation time, and create a CPU mapping. There may not be any adverse effect of such a CPU mapping though. This commit adds a check in the mmap callback of Panthor driver to fail the mmap() attempts on BOs created with NO_MMAP flag. The check covers private BOs as well as the prime exported BOs that are mmapped using prime fd. v2: - Fix breakage of mmap() on exported BOs that didn't have BO_MMAP_OFFSET ioctl invoked for them. Fixes: 68cbf96b1e9b ("drm/panthor: Part ways with drm_gem_shmem_object") Suggested-by: Boris Brezillon <boris.brezillon@collabora.com> Signed-off-by: Akash Goel <akash.goel@arm.com> Reviewed-by: Steven Price <steven.price@arm.com> Reviewed-by: Liviu Dudau <liviu.dudau@arm.com> Signed-off-by: Liviu Dudau <liviu.dudau@arm.com> Link: https://patch.msgid.link/20261013125433.26432-1-akash.goel@arm.com
2026-10-15Revert "nvme-pci: clamp max_hw_sectors based on DMA optimized limitation"Christoph Hellwig
This turns down to cause up to 15% performance regressions on standard large transfer workloads. Revert it and hope the IOVA allocator can eventually be fixed to remove the need for the original workaround. This reverts commit 3710e2b056cb92ad816e4d79fa54a6a5b6ad8cbd. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Keith Busch <kbusch@kernel.org>
2026-10-15bnxt_en: Pad short SW USO segments to BNXT_MIN_PKT_SIZEJoe Damato
Each SW USO segment is hdr_len + seg_payload bytes. With IPv4, hdr_len is 42 bytes, so a segment carrying less than 10 bytes of UDP payload is sent shorter than BNXT_MIN_PKT_SIZE. This happens with a short last segment (e.g. gso_size 1400 and 1405 bytes of payload) or with a gso_size below 10 bytes. Fix the short last segment in bnxt_sw_udp_gso_xmit(): zero the pad bytes after the header in the segment's inline header slot and add a pad BD pointing at them after the payload BDs. Only the last segment can need padding, and its pad BD fits within the existing bound in bds_needed and BNXT_SW_USO_MAX_DESCS, since payload BDs are at most num_segs + nr_frags. If gso_size itself is too small, every segment would need a pad BD, exceeding the descriptor bound. This case is unlikely, so bnxt_features_check() clears GSO features for these skbs and the stack segments them instead; bnxt_start_xmit() then pads each segment. Reviewed-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Joe Damato <joe@dama.to> Link: https://patch.msgid.link/20261009215426.2377946-3-joe@dama.to Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-10-15bnxt_en: Add helper to fill SW USO payload BDsJoe Damato
Move the code that fills a payload BD and its software ring entry in bnxt_sw_udp_gso_xmit() into a helper, bnxt_sw_gso_data_bd(). No functional change. A following patch uses the helper to add a pad BD to short SW USO segments. Reviewed-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Joe Damato <joe@dama.to> Link: https://patch.msgid.link/20261009215426.2377946-2-joe@dama.to Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-10-15spi: npcm-fiu: limit UMA data_size to the 16-byte staging bufferWeibin Liu
npcm_fiu_uma_read() and npcm_fiu_uma_write() use a 16-byte on-stack staging buffer (u32 data_reg[4]) and move data_size bytes to or from it. data_size comes straight from op->data.nbytes, which is supplied by the SPI MEM operation and is not bounded by the driver: npcm_fiu_exec_op() only validates op->addr.nbytes, and no adjust_op_size callback truncates the transfer to the controller FIFO size. A SPI MEM operation with addr.nbytes == 0 and data.nbytes > 16 therefore overflows data_reg[] on the kernel stack (write path) or reads past it (read path), corrupting the stack. Reject data_size larger than the staging buffer in both UMA helpers by returning -EOPNOTSUPP, so spi_mem falls back to the generic SPI message path. The chunked path in npcm_fiu_manualwrite() always passes at most CHUNK_SIZE bytes, so it is unaffected. Fixes: ace55c411b11 ("spi: npcm-fiu: add NPCM FIU controller driver") Cc: stable@vger.kernel.org # 5.4+ Reviewed-by: Guanglei Zhu <zhugl3@xiaopeng.com> Signed-off-by: Weibin Liu <liuwb@xiaopeng.com> Link: https://patch.msgid.link/20261015-npcm-fiu-uma-datasize-v1-1-7dee91cadee6@xiaopeng.com Signed-off-by: Mark Brown <broonie@kernel.org>
2026-10-15RDMA/rxe: bound the WQE opcode before indexing rxe_wr_opcode_info[]Youngsung Ahn
wr_opcode_mask() indexes the global rxe_wr_opcode_info[] array with a caller-supplied opcode and no bounds check. For a user QP the WQE comes from an mmap'd SQ ring, so wqe->wr.opcode is an attacker-controlled __u32 read back by the requester (req_next_wqe() -> rxe_requester()); rxe_post_send() takes the qp->is_user branch and never runs validate_send_wr(), whose only opcode check is itself !wr_opcode_mask() (it indexes before it checks). rxe_wr_opcode_info[] has entries only up to IB_WR_REG_MR, so an out-of-range opcode reads out of bounds and the result is used as a mask and dereferenced; observed as a KASAN global-out-of-bounds "Read of size 4" and a wild-pointer oops. Return a zero mask for an out-of-range opcode, matching how validate_send_wr() already treats a zero mask (an invalid WR), so no path indexes the array out of bounds. Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Youngsung Ahn <ays511.kr@gmail.com> Link: https://patch.msgid.link/20261002164552.294749-1-ays511.kr@gmail.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15RDMA/rxe: Fix out-of-range unsigned-to-signed conversion for RDMA message in ↵Honggang LI
2GiB size When RDMA READ request for 2GiB in single WR, res->read.resid is u32 0x80000000, which is INT_MIN (-2147483648). payload = min_t(int, res->read.resid, mtu); The `min_t` function will return -2147483648 for payload. The wrong size is propagated through call chain, `read_reply` -> `prepare_ack_packet` -> `rxe_init_packet` -> `alloc_skb` . `alloc_skb` failed because of invalid size. The passive side failed to emit response packet for RDMA READ request. After fixed it, the active side of RDMA READ failed with error code "IB_WC_LOC_PROT_ERR". When RDMA_READ_RESPONSE_FIRST packet recived by the active side, `do_read` call `copy_data`. dma->resid is u32 0x80000000. int resid = dma->resid; This conversion set resid to -2147483648. `copy_data` abort as length greater than resid. RDMA SEND and WRITE 2GiB message works too, after fixed these two bugs. Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Honggang LI <honggangli@163.com> Link: https://patch.msgid.link/20261002142250.657151-1-honggangli@163.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15RDMA/rxe: validate inline sge_offset in finish_packet() before indexing ↵Youngsung Ahn
inline_data[] For a user QP, rxe_post_send() takes the qp->is_user branch and never runs init_send_wqe(), so dma.length/resid/num_sge/cur_sge/sge_offset all come straight from the mmap'd SQ ring. In the inline send path finish_packet() uses sge_offset directly as an index: u8 *tmp = &wqe->dma.inline_data[wqe->dma.sge_offset]; memcpy(payload_addr(pkt), tmp, payload); inline_data is a flex array whose capacity is qp->sq.max_inline, but sge_offset is a u32 from user space, so the source pointer is displaced arbitrarily past the WQE and the memcpy reads out of bounds. commit 126c757e4cd4 ("RDMA/rxe: Validate num_sge/cur_sge before indexing wqe->dma.sge[]") added num_sge/cur_sge checks on the send path but not sge_offset, the third attacker-controlled index in the same struct. Read sge_offset once with READ_ONCE() and bound it (and the payload that follows it) by qp->sq.max_inline before indexing, so the value that is validated is the value that is used. Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Youngsung Ahn <ays511.kr@gmail.com> Link: https://patch.msgid.link/20261001152433.3259573-1-ays511.kr@gmail.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15RDMA/rxe: bound the ODP page index against the umem in rxe_check_pagefault()Youngsung Ahn
An ODP MR has a page array, map.pfn_list[]. Its size is computed from the umem start address. But rxe_check_pagefault() builds the index into that array from an iova that is relative to the same start, while the access itself was range-checked against the MR's iova (ibmr.iova). Those two bases do not have to match. reg_mr only requires that start and hca_va have the same page offset, so a user can register an MR with hca_va = start + N*PAGE. The difference then becomes an index skew of (hca_va - start) >> PAGE_SHIFT. Nothing compares the result with the size of pfn_list[], so the read goes past the end of the array. An unprivileged local user can reach this with a single RDMA operation on a self-connected rxe QP. KASAN reports a slab-out-of-bounds read. If the out-of-bounds qword happens to look like a writable HMM pfn, rxe then copies the payload into a page the user never registered. Reject an index that falls outside the umem and take the fault path instead. That path rejects the out-of-range access in ib_umem_odp_map_dma_and_lock(), which does check the umem range. The other users of pfn_list[], __rxe_odp_mr_copy() and the atomic helpers, run only after a successful map, so this single check covers them too. Fixes: 2fae67ab63db ("RDMA/rxe: Add support for Send/Recv/Write/Read with ODP") Signed-off-by: Youngsung Ahn <ays511.kr@gmail.com> Link: https://patch.msgid.link/20261004153246.2951263-1-ays511.kr@gmail.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15RDMA/rxe: reject IB_WR_REG_MR from user QPs in rxe_reg_fast_mr()Youngsung Ahn
IB_WR_REG_MR is documented as kernel-only ("can not be issued by userspace"), but nothing enforces it. rxe_wr_opcode_info[IB_WR_REG_MR] sets WR_LOCAL_OP_MASK for IB_QPT_RC, a QP type userspace can create, so a user QP that writes opcode = IB_WR_REG_MR into its mmap'd SQ ring reaches rxe_do_local_ops() -> rxe_reg_fast_mr(). That function then trusts the ring-supplied fields: struct rxe_mr *mr = to_rmr(wqe->wr.wr.reg.mr); ... if (unlikely(mr->state != RXE_MR_STATE_FREE)) /* deref */ ... mr->access = access; /* write */ For a user QP wqe->wr.wr.reg.mr is a fully attacker-chosen pointer, so this is an arbitrary-kernel-address dereference (arbitrary read confirmed, oops). No existing check distinguishes a user QP on this path. Reject IB_WR_REG_MR for user QPs before dereferencing the supplied pointer. Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Youngsung Ahn <ays511.kr@gmail.com> Link: https://patch.msgid.link/20261001152426.3259522-1-ays511.kr@gmail.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15RDMA/rxe: refuse to destroy a QP with type 2 MWs still boundYoungsung Ahn
Binding a type 2 memory window takes a reference on the QP: rxe_do_bind_mw() does rxe_get(qp) and stores it in mw->qp. That reference is released only when the window is invalidated (rxe_do_invalidate_mw() on IB_WR_LOCAL_INV) or the window itself is destroyed (rxe_mw_cleanup()). rxe_qp_chk_destroy() does not account for those references -- it only refuses when qp->mcg_num is non-zero. So an unprivileged user can bind a type 2 MW to a QP and then destroy the QP while the binding is still live. __rxe_cleanup() cannot drop the QP refcount to zero (the MW still holds a reference), hits its -ETIMEDOUT path, and frees the QP anyway; mw->qp is left dangling. When the MW is later invalidated or deallocated, rxe_mw_cleanup()/rxe_do_invalidate_mw() do rxe_put(mw->qp) on the freed struct rxe_qp -- a use-after-free write (refcount_dec, and complete()/list work if it reaches zero) reachable from userspace. IB_WR_BIND_MW is a local operation, so no peer is required. Track the number of type 2 MWs bound to a QP and refuse to destroy the QP while that count is non-zero, returning -EBUSY exactly as the existing multicast-attachment check does. The window must be invalidated or deallocated, which drops the reference and the count, before the QP can be destroyed. Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Youngsung Ahn <ays511.kr@gmail.com> Link: https://patch.msgid.link/20260930182223.2511145-1-ays511.kr@gmail.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-10-15platform/x86/intel/tpmi: Describe the device creation policy on the functionKuppuswamy Sathyanarayanan
The comment explaining the error policy of tpmi_create_devices() sits inside the loop, between the call and the check it describes. Move it above the function, where it describes the policy as a whole. Also split the check into its two cases, skipping unsupported features and failing on any other error, so that each reads on its own. No functional change intended. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Suggested-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-21-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Declare the loop counters in their loopsKuppuswamy Sathyanarayanan
Declare the loop counters in the for statements that use them, so that they are scoped to their loops. In tpmi_create_device() this also replaces the tmp cursor, which moved in step with the counter, with res[i]. While at it, use post-increment in all of them, as the other loops in the file already do. No functional change intended. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-20-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Use DEFINE_RES_MEM() for the feature resourcesKuppuswamy Sathyanarayanan
tpmi_create_device() fills in each feature resource field by field, deriving the end from the start and setting IORESOURCE_MEM by hand. That is what DEFINE_RES_MEM() exists for, so use it. The resources come from kzalloc_flex(), so the fields the macro sets that the open coded version left alone, name and desc, were already zero. No functional change intended. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-19-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Use scnprintf() to build the debugfs namesKuppuswamy Sathyanarayanan
The kernel has both snprintf() and scnprintf() for formatting into a fixed size buffer. The preference is to settle on scnprintf() and stop adding snprintf() users, so that only one of the two remains in use. Convert the two callers in tpmi_dbgfs_register(). Neither of them looks at the return value, so the difference in what the two functions return does not matter here. No functional change intended. Suggested-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Link: https://patch.msgid.link/20261013192935.2243291-18-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Use DEFINE_SHOW_STORE_ATTRIBUTE() for mem_writeKuppuswamy Sathyanarayanan
The mem_write debugfs file open handler and file_operations are written out by hand, which is what DEFINE_SHOW_STORE_ATTRIBUTE() exists for. Use it, and rename the write callback to mem_write_write() to match the name the macro derives. The other two files in this driver already come from DEFINE_SHOW_ATTRIBUTE(), so this also gives mem_write the .owner = THIS_MODULE that the open coded file_operations was missing. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-17-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Build the TPMI control words in one expressionKuppuswamy Sathyanarayanan
Both mailbox words in tpmi_read_feature_status() are assembled as an assignment followed by an or-assign of a second field, even though neither value is used in between. Build each of them in a single expression, so that a word and the fields it is made of read as one thing. The two comments on the command word described one value between them, so fold them together as well. No functional change intended. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-16-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Use scope based cleanup in the memory dumpKuppuswamy Sathyanarayanan
tpmi_mem_dump_show() frees its dump buffer and drops tpmi_dev_lock by hand at the end of the function, so the ioremap() failure in the middle of the loop cannot just return. It breaks out with an error stashed in ret, which is the only reason ret exists. Attach the buffer to __free(kfree) at its allocation and take the mutex with guard(), then return the error directly from the loop. mem_write() in this file already does both, so this makes the two debugfs paths consistent. No functional change intended. Suggested-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-15-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Use guard() for tpmi_dev_lock in the status readKuppuswamy Sathyanarayanan
tpmi_read_feature_status() takes tpmi_dev_lock explicitly and unwinds through an err_unlock label, which is what forced the two error paths before the mailbox command is issued to be gotos rather than returns. Take the mutex with guard() and return directly from those two paths. The done_proc label stays, because the paths that reach it have already handed the mailbox to the hardware and still owe it the CPL write before dropping the lock. That is unchanged: the two early failures never wrote CPL before either. No functional change intended. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Suggested-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-14-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Turn TPMI_GET_SINGLE_ENTRY_SIZE() into a functionKuppuswamy Sathyanarayanan
The macro computing a feature instance size has a few problems. Its comment says the hardware reports the size in u32 units, but the conversion to bytes is spelled as a shift by 2, so the comment and the code have to be read together to see that they agree. Its body is a single conditional expression wrapped in a statement expression that it never needed. And it evaluates its argument twice. Make it a plain static function, tpmi_get_single_entry_size(), which fixes all three: the conversion becomes a multiply by sizeof(u32), there is nothing left to wrap, and the argument is evaluated once and type checked. No functional change intended. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Suggested-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-13-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Include the headers tpmi_common.c relies onKuppuswamy Sathyanarayanan
tpmi_common.c uses a number of interfaces whose headers it never includes directly and only gets by chance through other headers: BIT_ULL(), GENMASK_ULL() bits.h __packed compiler_attributes.h EXPORT_SYMBOL_NS_GPL() export.h DEFINE_IDA() idr.h IORESOURCE_MEM ioport.h DEFINE_MUTEX(), mutex_lock() mutex.h seq_printf(), seq_hex_dump() seq_file.h kmalloc(), kzalloc_flex() slab.h USEC_PER_MSEC, USEC_PER_SEC time64.h uXX types.h Include them, so that the file does not break when an unrelated header stops pulling them in. Drop io.h while at it, as iopoll.h already includes it, and sort the block alphabetically: debugfs.h and cleanup.h were the wrong way round. No functional change intended. Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Suggested-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-12-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Split off the PCI VSEC enumerationKuppuswamy Sathyanarayanan
Everything the TPMI driver does once the PFS table has been located is independent of how the TPMI MMIO region was discovered, from walking the PFS to creating the per feature devices. Only the entry point is VSEC specific and it is now reduced to filling in the arguments of intel_tpmi_init(). Move the enumeration independent code to a new intel-tpmi_common module and leave the auxiliary driver binding to the "intel_vsec.tpmi" device created by the Intel VSEC driver in vsec_tpmi.c. The code is moved as is, apart from a stray blank line and a typo in a comment that are fixed in passing. tpmi_common.h exposes what an enumeration method needs, which is intel_tpmi_init(), intel_tpmi_deinit() and struct intel_tpmi_info. The structure is a complete type because the enumeration method allocates it and is expected to extend it with its own data. The PFS, TPMI_INFO and feature state layouts stay private to tpmi_common.c, only the hardware parsing touches them. CONFIG_INTEL_TPMI_COMMON is a hidden symbol selected by the enumeration method, in the same way as CONFIG_INTEL_TPMI_POWER_DOMAINS. The TPMI client API is now exported by intel-tpmi_common, so the client drivers depend on that module instead of intel-vsec_tpmi. That is what allows a second, non PCI enumeration method to be added later without the client drivers knowing which one is loaded. No functional change. Co-developed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-11-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Split out enumeration independent init and exitKuppuswamy Sathyanarayanan
Once the two devices and the MMIO resources are known, the rest of what the init path does is independent of how the TPMI instance was found. Turn that part into intel_tpmi_init(), taking them as arguments, and add intel_tpmi_deinit() for what removal has to undo. The auxiliary bus and intel_vsec_device handling stays in tpmi_probe() and tpmi_remove(). intel_tpmi_info is allocated by the caller instead of by intel_tpmi_init(), so that an enumeration method can embed it in its own state and get back to that with container_of(). The TPMI_CORE_INIT and TPMI_CORE_EXIT notifications now pass the struct device of the TPMI instance instead of its auxiliary device, so that the generic code does not have to know about the auxiliary bus. No in tree notifier callback looks at that argument and one that needs the auxiliary device can use to_auxiliary_dev(). The driver data is set with dev_set_drvdata() on the same device that auxiliary_set_drvdata() used, so tpmi_remove() still finds it. No functional change. Co-developed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-10-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Do not assume TPMI is enumerated from PCIKuppuswamy Sathyanarayanan
The init path converts the enumerating device into a pci_dev without checking and always registers the OOBMSM package to PCI device mapping. That works today because TPMI is only enumerated from a PCI VSEC capability, but ACPI based enumeration is being worked on and the device it enumerates from is not a PCI device. Do both only for a PCI device. The bus number read from the pci_dev is a fallback that tpmi_process_info() overwrites with the value reported by the TPMI_INFO feature, so there is nothing to do without a pci_dev. The mapping stored by intel_vsec_set_mapping() is looked up by pci_dev and cannot be consumed by another enumeration method, and calling it anyway would fail the probe with -ENODEV. While here, stop describing intel_tpmi_info as holding the TPMI devices of one PCI device. No functional change for PCI enumerated TPMI. Co-developed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-9-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Drop unused arg from tpmi_set_control_base()Kuppuswamy Sathyanarayanan
tpmi_set_control_base() takes the device for its devm_ioremap() from tpmi_info->tpmi_dev, so the auxiliary device argument is unused. Remove it. No functional change. Co-developed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-8-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Describe a TPMI instance by its two devicesKuppuswamy Sathyanarayanan
The intel_vsec_device pointer in intel_tpmi_info is by now only used to reach two devices, the device the TPMI driver is bound to and the device TPMI was enumerated from. Store those two directly and call them tpmi_dev and parent. The devm allocations, the driver data and the TPMI feature devices all hang off tpmi_dev. The parent device names the debugfs directory, is recorded in every TPMI feature device so that the enumerating bus can find those devices again during PCI error recovery, and is the device intel_vsec_set_mapping() registers the mapping for. Both are assigned once during init and used everywhere below it, so intel_tpmi_info no longer refers to an intel_vsec_device. tpmi_to_dev() went away with its only user. There is no functional change, tpmi_dev is &auxdev->dev and parent is vsec_dev->dev, which is what the affected call sites used before. Co-developed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-7-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Keep the feature resources in intel_tpmi_infoKuppuswamy Sathyanarayanan
The PFS walk and tpmi_create_devices() go back to the intel_vsec_device for the MMIO resource array and its count, although the count is already cached in intel_tpmi_info. Cache the resource array as well and use it everywhere. The array belongs to the enumerating device and outlives the TPMI instance, so holding a pointer to it is safe. Fewer places need an intel_vsec_device, which enumeration methods other than PCI VSEC do not have. No functional change. Co-developed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-6-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Get tpmi_info directly from the parent deviceKuppuswamy Sathyanarayanan
tpmi_get_feature_status() and tpmi_get_debugfs_dir() are called by the TPMI feature drivers with their own auxiliary device. Both convert the parent device into an intel_vsec_device and then read the driver data of the auxiliary device it was derived from, which is the parent device they started with. Read the driver data from the parent device directly. Apart from being shorter, this drops an assumption. dev_to_ivdev() is a container_of(), so it only holds while the device the TPMI driver binds to is an auxiliary device created by the Intel VSEC driver. Another enumeration method binds to a different kind of device and the cast would hand out a bogus intel_vsec_device. The driver data is set on that device in either case. The TPMI feature devices are not affected. They remain intel_vsec_device instances allocated by the TPMI driver, since the auxiliary bus is how the feature drivers are reached and not a property of the enumeration. No functional change. Co-developed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-5-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
2026-10-15platform/x86/intel/tpmi: Remove unused vsec_dev from intel_tpmi_pm_featureKuppuswamy Sathyanarayanan
The intel_vsec_device pointer stored in every PFS entry is never read back. Remove the field and its assignment. No functional change. Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Reviewed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com> Link: https://patch.msgid.link/20261013192935.2243291-4-sathyanarayanan.kuppuswamy@linux.intel.com Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>