linux-toradex.git/drivers/cxl, branch master Linux kernel for Apalis and Colibri modules Merge tag 'cxl-for-7.4' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl 2026-10-26T16:50:38+00:00 Linus Torvalds torvalds@linux-foundation.org 2026-10-26T16:50:38+00:00 d6bbdb67a4b9f2366debd8b5b05b1822490982ef Pull CXL updates from Dave Jiang: "The headline item is the first part of support for Dynamic Capacity Devices (DCD). It covers flagging DCD support in the mailbox, reading the dynamic capacity configuration from the device, gathering DSMAS data for the DCD partitions, and configuring the dynamic capacity interrupts. It also adds a documentation update and a cleanup that enforces partition ordering. DCD is not enabled yet with these code changes. Patches are applied to reduce the whole set in order to relief the burden of very large patch set review. Zero sized HDM decoders are now allowed, along with cxl_test support for them. There is also a set of decoder hardening patches for switch decoder enumeration. Latency calculations now account for link width and use consistent units. The GPF timeout programming has been reworked. The pull also has a number of misc patches. A large number of bug fixes against pre-existing issues make up the rest of the pull. Many of these were flagged by review of patches posted to the mailing list by sashiko. Also adding Richard Cheng as a reviewer (R) to the CXL sub-system" * tag 'cxl-for-7.4' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl: (49 commits) cxl/events: Test CME counter expiry with equality, not bitwise AND Documentation/cxl: Document DPA partition layout and ordering rules cxl/core: Enforce partition order/simplify partition calls cxl/mem: Configure dynamic capacity interrupts cxl/pci: Factor out interrupt policy check cxl/events: Split event msgnum configuration from irq setup cxl/cdat: Gather DSMAS data for DCD partitions cxl/mem: Read dynamic capacity configuration from the device cxl/mbox: Flag support for Dynamic Capacity Devices (DCD) cxl/acpi: Convert QTG _DSM latencies to picoseconds cxl: Account for link width in latency calculation cxl/port: Convert PCIe link latency to nanoseconds cxl/pci: Update only the Port GPF timeout fields cxl/pci: Program the Port GPF timeouts before caching the DVSEC cxl/pci: Use the Device GPF DVSEC for restricted endpoints cxl/port: Attach endpoints to a dport under the lock that pins it cxl/core: Hold the dport host lock across dport lookup and use cxl/port: Clear cached dport pointers when a dport is removed cxl/region: Warn on user region position mismatch cxl/acpi: Check ACPI companion before use ... 
Pull CXL updates from Dave Jiang:
 "The headline item is the first part of support for Dynamic Capacity
  Devices (DCD). It covers flagging DCD support in the mailbox, reading
  the dynamic capacity configuration from the device, gathering DSMAS
  data for the DCD partitions, and configuring the dynamic capacity
  interrupts. It also adds a documentation update and a cleanup that
  enforces partition ordering. DCD is not enabled yet with these code
  changes. Patches are applied to reduce the whole set in order to
  relief the burden of very large patch set review.

  Zero sized HDM decoders are now allowed, along with cxl_test support
  for them. There is also a set of decoder hardening patches for switch
  decoder enumeration. Latency calculations now account for link width
  and use consistent units. The GPF timeout programming has been
  reworked. The pull also has a number of misc patches.

  A large number of bug fixes against pre-existing issues make up the
  rest of the pull. Many of these were flagged by review of patches
  posted to the mailing list by sashiko.

  Also adding Richard Cheng as a reviewer (R) to the CXL sub-system"

* tag 'cxl-for-7.4' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl: (49 commits)
  cxl/events: Test CME counter expiry with equality, not bitwise AND
  Documentation/cxl: Document DPA partition layout and ordering rules
  cxl/core: Enforce partition order/simplify partition calls
  cxl/mem: Configure dynamic capacity interrupts
  cxl/pci: Factor out interrupt policy check
  cxl/events: Split event msgnum configuration from irq setup
  cxl/cdat: Gather DSMAS data for DCD partitions
  cxl/mem: Read dynamic capacity configuration from the device
  cxl/mbox: Flag support for Dynamic Capacity Devices (DCD)
  cxl/acpi: Convert QTG _DSM latencies to picoseconds
  cxl: Account for link width in latency calculation
  cxl/port: Convert PCIe link latency to nanoseconds
  cxl/pci: Update only the Port GPF timeout fields
  cxl/pci: Program the Port GPF timeouts before caching the DVSEC
  cxl/pci: Use the Device GPF DVSEC for restricted endpoints
  cxl/port: Attach endpoints to a dport under the lock that pins it
  cxl/core: Hold the dport host lock across dport lookup and use
  cxl/port: Clear cached dport pointers when a dport is removed
  cxl/region: Warn on user region position mismatch
  cxl/acpi: Check ACPI companion before use
  ...
Merge tag 'acpi-7.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm 2026-10-21T22:08:18+00:00 Linus Torvalds torvalds@linux-foundation.org 2026-10-21T22:08:18+00:00 4df4d2a577db2d3d57b888bcfb1206fc74661d16 Pull ACPI support updates from Rafael Wysocki: "The most significant change here is the rework of the core ACPI device enumeration code to avoid initializing ACPI power management prematurely for devices with unmet dependencies. This essentially is a problem fix, but it requires taking PCI device enumeration ordering into account among other things, so it is far from straightforward. There was an attempt to do this in the 7.3 time frame that had to be reverted, so this time it is done much more carefully. There is also an ACPI thermal control update rearranging the code to register thermal class cooling devices under specific parents, which is visible in sysfs. This helps to figure out what is represented by those devices much more easily, but it also allows the binding of cooling devices to trip points in thermal zones to be adjusted so it need not rely on specific cooling device data usage in thermal drivers. From the number of commits perspective, the most visible part is the one replacing acpi_get_first_physical_node() with acpi_bus_get_primary_device() everywhere and dropping the former that was inherently racy and generally unsafe to use. Fortunately, the maintainers involved responded timely and all of that could be done within one development cycle. Another quite visible part is Dave Jiang's series fixing multiple issues reported by Sashiko in the ACPI APEI support code with a PCI AER improvement on top. It took several iterations of this to make Sashiko happy and the code in question should be much more robust going forward. One other thing worth mentioning separately is that the ACPI bus type is now going to be registered as a "companion" one and it will not be possible to register a driver binding directly to a struct acpi_device. All of the existing users of those drivers in the tree were converted to platform drivers in the previous cycles, so this is just a final step in a longer journey. Apart from the above, there are many fixes, mostly in drivers, with a noticeable exception of a fix in the code binding devices to their ACPI companions, new IOVT IOMMU support in ACPICA, new _Ixx object support for suspend-to-idle on x86, multiple new quirks, mostly related to unused ACPI power resources and backlight handling, and assorted cleanups. Specifics: - Make the core ACPI device enumeration code stop attempting to initialize ACPI power management for devices before they are known to be present and functional (Rafael Wysocki) - Simplify and clean up the core ACPI device enumeration code on top of the previous adjustment and reduce debug noise coming from it (Rafael Wysocki) - Fix locking in the code binding devices to their ACPI companions, prevent it from replacing a valid ACPI companion pointer with NULL incidentally, improve its messaging (including debug noise reduction), and reduce its overhead by making it skip the cases in which no ACPI companion may exist early (Rafael Wysocki) - Reduce runtime memory footprint of struct acpi_device by rearranging fields in it (Andy Shevchenko) - Update the driver core to allow bus types without drivers to be registered as "companion bus types" without driver-related sysfs attributes and register the ACPI bus type as a companion one (Rafael Wysocki) - Use strscpy() instead of strcpy() in the ACPI sysfs interface (Ruslan Vagner) - Remove __meminitdata from hmat_adist_nb declaration preserving it even when CONFIG_MEMORY_HOTPLUG is not set to prevent notifier chain corruption in some cases (Alejandro Lucero) - Prevent a NULL pointer dereference when attempting to read the table length after a mapping bookkeeping failure during an initrd override invocation with permanent table mappings enabled (Slavin Liu) - Use for_each_populated_zone() in get_node_num() in the MRRM table pareser (Sang-Heon Jeon) - Use vsnprintf() in acpi_os_vprintf() to prevent memory corruption resulting from overly long diagnostic messages that may not fit into a static buffer used by ACPICA and may be produced by a malformed ACPI table (Vlatko Kosturjak) - Ignore leading root scope prefix in string _UID match to allow AMD IOMMU matching to work as expected in some cases (Mario Limonciello) - Adjust message printing macros for ACPI objects to use dev_fmt() for more consistency with the analogous dev_*() macros (Rafael Wysocki) - Add _Ixx GPE wake indicator support to the x86 suspend-to-idle handling code (Saranya Gopal) - Clear driver_data before free in the ACPI PCI Link driver to prevent a dangling pointer from being used later (Jiale Yao) - Accumulate the queued flag across error sections of a CPER record in ghes_do_proc() to prevent it from being cleared by a later section or entry that does not queue memory-failure work (Breno Leitao) - Map a raw PCI AER Capability image field by field to prevent information loss from occurring while processing it (Dave Jiang) - Update a stale driver source reference in the ACPI APEI EINJ driver documentation (Prabhakar Pujeri) - Reject CPER records with an out-of-range error_data_length or an error status block length wrapping a u32 in the EFI CPER driver (Dave Jiang) - Address multiple assorted issues in the ACPI extlog driver (Dave Jiang) - Address two issues related to CXL in the ACPI APEI GHES driver (Dave Jiang) - Read only validated fields in cper_mem_err_pack() (Dave Jiang) - Validate memory error section length before payload access, prevent writes past array bounds from occurring and sanitize software metadata in the ACPI APEI GHES driver (Dave Jiang) - Make cxl_cper_handle_prot_err() static (Dave Jiang) - Add IOVT IOMMU support to ACPICA (Bibo Mao) - Fix PCC OperationRegion command offsets in ACPICA (Sudeep Holla) - Fix memory leak in acpi_ds_create_field() related to the handling of internal_pcc_buffer allocations (Breno Leitao) - Fix memory leak due to leftover devm_kcalloc() argument in the ACPI fan driver (Rafael Wysocki) - Fix backlight unregistration ordering in the ACPI video bus device driver (Rafael Wysocki) - Update the thermal core to allow the registration of thermal class cooling devices under specific parents and convert the ACPI fan, ACPI processor, and ACPI video bus device drivers to do so (Rafael Wysocki) - Adjust the ACPI thermal zone driver to use thermal class cooling device parent information for binding cooling devices to trip points in thermal zones (Rafael Wysocki) - Use cooling device data that is a better match for a given driver instead of a struct acpi_device pointer that had to be used for this purpose previously in the ACPI fan and processor drivers (Rafael Wysocki) - Fix acpi_thermal_guess_offset() in the ACPI thermal zone driver for thermal zone devices with a 273.0°K offset (Hans de Goede) - Define a free for ACPI objects allocated by ACPICA helpers and use it to simplify AML error handling in the ACPI fan driver (Rafael Wysocki) - Switch over the users of acpi_get_first_physical_node(), that was inherently racy and therefore not really safe to use, to use its replacement acpi_bus_get_primary_device(), and eliminate the former (Rafael Wysocki) - Add !CONFIG_ACPI stubs for acpi_find_child_device() and acpi_device_uid() (Mario Limonciello and Robin Murphy) - Fix bogus cycle count values on systems without _BIX in the ACPI battery driver (Armin Wolf) - Fix NULL pointer dereference on allocation failure and report relative state of charge as CAPACITY in the ACPI SBS driver (Itay Shem-tov and Guangshuo Li) - Add lid_init_state quirk for Razer Blade Stealth 13 (Early 2020) RZ09-0310 to the ACPI button driver (Ian Gallagher) - Add new quirks to force the native backlight on Dell OptiPlex AIO 7410 35W and Acer Nitro AN515-58 in the ACPI video bus device driver (Chitransh Mandhaniya and Dirga Yuza) - Fix the shared memory signature handling in the ACPI PCC address space handler, free the PCC mailbox channel on operation region deactivation, and avoid unnecessary repeated computations of the PCC operation region command timeout (Sudeep Holla) - Fix a log message typo in the ACPI NUMA support code (Hemanth Selam) - Zero-initialize capability query result in the ACPI Platform Firmware Runtime Update and Telemetry driver to avoid disclosure of kernel stack data to user space (Karl Mehltretter) - Replace the remaining built-in/module check for CONFIG_ACPI_WMI in a header file with IS_ENABLED() macro (Lalit Shankar Chowdhury) - Retry probing battery docks after hotplug in the ACPI docking station driver to avoid missing ACPI add notifications (Vibhor Goel) - Detach device after resource transition failure in pnp_device_probe() to avoid blocking later manual PNP configuration attempts (Myeonghun Pak)" * tag 'acpi-7.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (90 commits) ACPI: PM: Add Acer Aspire A515-53G to dmi_leave_unused_power_resources_on[] ACPI: video: force native backlight on Dell OptiPlex AIO 7410 35W ACPI: PM: sleep: Drop unused sleep_no_lps0 define ACPI: PM: Add unused power resource quirk for Casper Excalibur G900 ACPI: dock: Retry probing battery docks after hotplug ACPI: PCI: pci_link: clear driver_data before free ACPI: APEI: accumulate the queued flag across error sections ACPICA: actbl2.h: Add bit constants of the Flags field in the IOVT IOMMU structure ACPI: bus: Drop acpi_get_first_physical_node() efi/dev-path-parser: Switch to use acpi_bus_get_primary_device() ALSA: hda: tas2781: Switch to use acpi_bus_get_primary_device() ALSA: hda: cs35l41: Switch to use acpi_bus_get_primary_device() iommu/arm-smmu-v3: Switch to use acpi_bus_get_primary_device() PCI/AER: Map a raw AER Capability image field by field ACPI: NUMA: fix typo in log message ACPI: Add stub for acpi_find_child_device() driver core/ACPI: Introduce companion_bus_register() ACPI: PCC: Cache OpRegion command timeout ACPI: PCC: Free channel on OpRegion deactivation ACPI: PCC: Preserve shared memory signature in OpRegion handler ... 
Pull ACPI support updates from Rafael Wysocki:
 "The most significant change here is the rework of the core ACPI device
  enumeration code to avoid initializing ACPI power management
  prematurely for devices with unmet dependencies. This essentially is a
  problem fix, but it requires taking PCI device enumeration ordering
  into account among other things, so it is far from straightforward.
  There was an attempt to do this in the 7.3 time frame that had to be
  reverted, so this time it is done much more carefully.

  There is also an ACPI thermal control update rearranging the code to
  register thermal class cooling devices under specific parents, which
  is visible in sysfs. This helps to figure out what is represented by
  those devices much more easily, but it also allows the binding of
  cooling devices to trip points in thermal zones to be adjusted so it
  need not rely on specific cooling device data usage in thermal
  drivers.

  From the number of commits perspective, the most visible part is the
  one replacing acpi_get_first_physical_node() with
  acpi_bus_get_primary_device() everywhere and dropping the former that
  was inherently racy and generally unsafe to use. Fortunately, the
  maintainers involved responded timely and all of that could be done
  within one development cycle.

  Another quite visible part is Dave Jiang's series fixing multiple
  issues reported by Sashiko in the ACPI APEI support code with a PCI
  AER improvement on top. It took several iterations of this to make
  Sashiko happy and the code in question should be much more robust
  going forward.

  One other thing worth mentioning separately is that the ACPI bus type
  is now going to be registered as a "companion" one and it will not be
  possible to register a driver binding directly to a struct
  acpi_device. All of the existing users of those drivers in the tree
  were converted to platform drivers in the previous cycles, so this is
  just a final step in a longer journey.

  Apart from the above, there are many fixes, mostly in drivers, with a
  noticeable exception of a fix in the code binding devices to their
  ACPI companions, new IOVT IOMMU support in ACPICA, new _Ixx object
  support for suspend-to-idle on x86, multiple new quirks, mostly
  related to unused ACPI power resources and backlight handling, and
  assorted cleanups.

  Specifics:

   - Make the core ACPI device enumeration code stop attempting to
     initialize ACPI power management for devices before they are known
     to be present and functional (Rafael Wysocki)

   - Simplify and clean up the core ACPI device enumeration code on top
     of the previous adjustment and reduce debug noise coming from it
     (Rafael Wysocki)

   - Fix locking in the code binding devices to their ACPI companions,
     prevent it from replacing a valid ACPI companion pointer with NULL
     incidentally, improve its messaging (including debug noise
     reduction), and reduce its overhead by making it skip the cases in
     which no ACPI companion may exist early (Rafael Wysocki)

   - Reduce runtime memory footprint of struct acpi_device by
     rearranging fields in it (Andy Shevchenko)

   - Update the driver core to allow bus types without drivers to be
     registered as "companion bus types" without driver-related sysfs
     attributes and register the ACPI bus type as a companion one
     (Rafael Wysocki)

   - Use strscpy() instead of strcpy() in the ACPI sysfs interface
     (Ruslan Vagner)

   - Remove __meminitdata from hmat_adist_nb declaration preserving it
     even when CONFIG_MEMORY_HOTPLUG is not set to prevent notifier
     chain corruption in some cases (Alejandro Lucero)

   - Prevent a NULL pointer dereference when attempting to read the
     table length after a mapping bookkeeping failure during an initrd
     override invocation with permanent table mappings enabled (Slavin
     Liu)

   - Use for_each_populated_zone() in get_node_num() in the MRRM table
     pareser (Sang-Heon Jeon)

   - Use vsnprintf() in acpi_os_vprintf() to prevent memory corruption
     resulting from overly long diagnostic messages that may not fit
     into a static buffer used by ACPICA and may be produced by a
     malformed ACPI table (Vlatko Kosturjak)

   - Ignore leading root scope prefix in string _UID match to allow AMD
     IOMMU matching to work as expected in some cases (Mario
     Limonciello)

   - Adjust message printing macros for ACPI objects to use dev_fmt()
     for more consistency with the analogous dev_*() macros (Rafael
     Wysocki)

   - Add _Ixx GPE wake indicator support to the x86 suspend-to-idle
     handling code (Saranya Gopal)

   - Clear driver_data before free in the ACPI PCI Link driver to
     prevent a dangling pointer from being used later (Jiale Yao)

   - Accumulate the queued flag across error sections of a CPER record
     in ghes_do_proc() to prevent it from being cleared by a later
     section or entry that does not queue memory-failure work (Breno
     Leitao)

   - Map a raw PCI AER Capability image field by field to prevent
     information loss from occurring while processing it (Dave Jiang)

   - Update a stale driver source reference in the ACPI APEI EINJ driver
     documentation (Prabhakar Pujeri)

   - Reject CPER records with an out-of-range error_data_length or an
     error status block length wrapping a u32 in the EFI CPER driver
     (Dave Jiang)

   - Address multiple assorted issues in the ACPI extlog driver (Dave
     Jiang)

   - Address two issues related to CXL in the ACPI APEI GHES driver
     (Dave Jiang)

   - Read only validated fields in cper_mem_err_pack() (Dave Jiang)

   - Validate memory error section length before payload access, prevent
     writes past array bounds from occurring and sanitize software
     metadata in the ACPI APEI GHES driver (Dave Jiang)

   - Make cxl_cper_handle_prot_err() static (Dave Jiang)

   - Add IOVT IOMMU support to ACPICA (Bibo Mao)

   - Fix PCC OperationRegion command offsets in ACPICA (Sudeep Holla)

   - Fix memory leak in acpi_ds_create_field() related to the handling
     of internal_pcc_buffer allocations (Breno Leitao)

   - Fix memory leak due to leftover devm_kcalloc() argument in the ACPI
     fan driver (Rafael Wysocki)

   - Fix backlight unregistration ordering in the ACPI video bus device
     driver (Rafael Wysocki)

   - Update the thermal core to allow the registration of thermal class
     cooling devices under specific parents and convert the ACPI fan,
     ACPI processor, and ACPI video bus device drivers to do so (Rafael
     Wysocki)

   - Adjust the ACPI thermal zone driver to use thermal class cooling
     device parent information for binding cooling devices to trip
     points in thermal zones (Rafael Wysocki)

   - Use cooling device data that is a better match for a given driver
     instead of a struct acpi_device pointer that had to be used for
     this purpose previously in the ACPI fan and processor drivers
     (Rafael Wysocki)

   - Fix acpi_thermal_guess_offset() in the ACPI thermal zone driver for
     thermal zone devices with a 273.0°K offset (Hans de Goede)

   - Define a free for ACPI objects allocated by ACPICA helpers and use
     it to simplify AML error handling in the ACPI fan driver (Rafael
     Wysocki)

   - Switch over the users of acpi_get_first_physical_node(), that was
     inherently racy and therefore not really safe to use, to use its
     replacement acpi_bus_get_primary_device(), and eliminate the former
     (Rafael Wysocki)

   - Add !CONFIG_ACPI stubs for acpi_find_child_device() and
     acpi_device_uid() (Mario Limonciello and Robin Murphy)

   - Fix bogus cycle count values on systems without _BIX in the ACPI
     battery driver (Armin Wolf)

   - Fix NULL pointer dereference on allocation failure and report
     relative state of charge as CAPACITY in the ACPI SBS driver (Itay
     Shem-tov and Guangshuo Li)

   - Add lid_init_state quirk for Razer Blade Stealth 13 (Early 2020)
     RZ09-0310 to the ACPI button driver (Ian Gallagher)

   - Add new quirks to force the native backlight on Dell OptiPlex AIO
     7410 35W and Acer Nitro AN515-58 in the ACPI video bus device
     driver (Chitransh Mandhaniya and Dirga Yuza)

   - Fix the shared memory signature handling in the ACPI PCC address
     space handler, free the PCC mailbox channel on operation region
     deactivation, and avoid unnecessary repeated computations of the
     PCC operation region command timeout (Sudeep Holla)

   - Fix a log message typo in the ACPI NUMA support code (Hemanth
     Selam)

   - Zero-initialize capability query result in the ACPI Platform
     Firmware Runtime Update and Telemetry driver to avoid disclosure of
     kernel stack data to user space (Karl Mehltretter)

   - Replace the remaining built-in/module check for CONFIG_ACPI_WMI in
     a header file with IS_ENABLED() macro (Lalit Shankar Chowdhury)

   - Retry probing battery docks after hotplug in the ACPI docking
     station driver to avoid missing ACPI add notifications (Vibhor
     Goel)

   - Detach device after resource transition failure in
     pnp_device_probe() to avoid blocking later manual PNP configuration
     attempts (Myeonghun Pak)"

* tag 'acpi-7.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (90 commits)
  ACPI: PM: Add Acer Aspire A515-53G to dmi_leave_unused_power_resources_on[]
  ACPI: video: force native backlight on Dell OptiPlex AIO 7410 35W
  ACPI: PM: sleep: Drop unused sleep_no_lps0 define
  ACPI: PM: Add unused power resource quirk for Casper Excalibur G900
  ACPI: dock: Retry probing battery docks after hotplug
  ACPI: PCI: pci_link: clear driver_data before free
  ACPI: APEI: accumulate the queued flag across error sections
  ACPICA: actbl2.h: Add bit constants of the Flags field in the IOVT IOMMU structure
  ACPI: bus: Drop acpi_get_first_physical_node()
  efi/dev-path-parser: Switch to use acpi_bus_get_primary_device()
  ALSA: hda: tas2781: Switch to use acpi_bus_get_primary_device()
  ALSA: hda: cs35l41: Switch to use acpi_bus_get_primary_device()
  iommu/arm-smmu-v3: Switch to use acpi_bus_get_primary_device()
  PCI/AER: Map a raw AER Capability image field by field
  ACPI: NUMA: fix typo in log message
  ACPI: Add stub for acpi_find_child_device()
  driver core/ACPI: Introduce companion_bus_register()
  ACPI: PCC: Cache OpRegion command timeout
  ACPI: PCC: Free channel on OpRegion deactivation
  ACPI: PCC: Preserve shared memory signature in OpRegion handler
  ...
Merge branch 'for-7.4/cxl-fixes' into cxl-for-next 2026-10-15T15:13:53+00:00 Dave Jiang dave.jiang@intel.com 2026-10-15T15:13:53+00:00 cf15f9f2361dd4002203716b7b878393c0eee4bf cxl/events: Test CME counter expiry with equality, not bitwise AND 
cxl/events: Test CME counter expiry with equality, not bitwise AND
cxl/events: Test CME counter expiry with equality, not bitwise AND 2026-10-15T15:13:17+00:00 Guixin Liu kanie@linux.alibaba.com 2026-09-22T08:22:14+00:00 b94ba9a1f2d5c57f2eec67d502bb8fdd1e53c87a 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> 
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>
Merge branch 'for-7.4/cxl-dcd' into cxl-for-next 2026-10-13T16:01:39+00:00 Dave Jiang dave.jiang@intel.com 2026-10-13T16:01:39+00:00 d304df5f329302fff4658cd6ed8f93aec4b94d27 Part 1 of DCD support. Functionality disabled. Documentation/cxl: Document DPA partition layout and ordering rules cxl/core: Enforce partition order/simplify partition calls cxl/mem: Configure dynamic capacity interrupts cxl/pci: Factor out interrupt policy check cxl/events: Split event msgnum configuration from irq setup cxl/cdat: Gather DSMAS data for DCD partitions cxl/mem: Read dynamic capacity configuration from the device cxl/mbox: Flag support for Dynamic Capacity Devices (DCD) 
Part 1 of DCD support. Functionality disabled.

Documentation/cxl: Document DPA partition layout and ordering rules
cxl/core: Enforce partition order/simplify partition calls
cxl/mem: Configure dynamic capacity interrupts
cxl/pci: Factor out interrupt policy check
cxl/events: Split event msgnum configuration from irq setup
cxl/cdat: Gather DSMAS data for DCD partitions
cxl/mem: Read dynamic capacity configuration from the device
cxl/mbox: Flag support for Dynamic Capacity Devices (DCD)
cxl/core: Enforce partition order/simplify partition calls 2026-10-12T16:44:52+00:00 Ira Weiny iweiny@kernel.org 2026-10-10T22:54:58+00:00 c801107605f39b5a8632fbc835e0da65a6fb6b51 Device partitions share one flat DPA space in a fixed order: static volatile capacity at DPA 0, static persistent capacity immediately above it, and dynamic capacity above all static capacity. The static pair is mandated by CXL r4.0 8.2.10.9.2.1 "Get Partition Info", Table 8-310: the device "shall provide this volatile capacity starting at DPA 0" and "shall provide this persistent capacity starting at the DPA immediately following the volatile capacity". Dynamic capacity has no such rule. Table 8-347 only requires the region base to be 256M aligned, so placing it directly above static capacity is a Linux requirement, enforced by cxl_configure_dcd(). DPA 0 end +---------------+---------------+---------------------------+ | ram | pmem | dynamic_ram_1 | +---------------+---------------+---------------------------+ part[0] part[1] part[2] Signed-off-by: Ira Weiny <iweiny@kernel.org> Co-developed-by: Anisa Su <anisa.su@samsung.com> Signed-off-by: Anisa Su <anisa.su@samsung.com> Tested-by: Wonjae Lee <wj28.lee@samsung.com> Tested-by: Junhee Park <jh9934.park@samsung.com> Tested-by: Heesoo Kim <habil.kim@samsung.com> Reviewed-by: Dave Jiang <dave.jiang@intel.com> Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com> Reviewed-by: Alison Schofield <alison.schofield@intel.com> Link: https://patch.msgid.link/20261010225852.10727-8-anisa.su@samsung.com Signed-off-by: Dave Jiang <dave.jiang@intel.com> 
Device partitions share one flat DPA space in a fixed order: static
volatile capacity at DPA 0, static persistent capacity immediately above
it, and dynamic capacity above all static capacity.  The static pair is
mandated by CXL r4.0 8.2.10.9.2.1 "Get Partition Info", Table 8-310: the
device "shall provide this volatile capacity starting at DPA 0" and
"shall provide this persistent capacity starting at the DPA immediately
following the volatile capacity".  Dynamic capacity has no such rule.
Table 8-347 only requires the region base to be 256M aligned, so placing
it directly above static capacity is a Linux requirement,
enforced by cxl_configure_dcd().

    DPA 0                                                     end
    +---------------+---------------+---------------------------+
    |      ram      |     pmem      |       dynamic_ram_1       |
    +---------------+---------------+---------------------------+
    part[0]         part[1]         part[2]

Signed-off-by: Ira Weiny <iweiny@kernel.org>
Co-developed-by: Anisa Su <anisa.su@samsung.com>
Signed-off-by: Anisa Su <anisa.su@samsung.com>
Tested-by: Wonjae Lee <wj28.lee@samsung.com>
Tested-by: Junhee Park <jh9934.park@samsung.com>
Tested-by: Heesoo Kim <habil.kim@samsung.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Link: https://patch.msgid.link/20261010225852.10727-8-anisa.su@samsung.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
cxl/mem: Configure dynamic capacity interrupts 2026-10-12T16:44:52+00:00 Ira Weiny iweiny@kernel.org 2026-10-10T22:54:57+00:00 9c45dd3684a51ef93afe326b33a16551ee9a74e5 Dynamic Capacity Devices (DCD) support extent change notifications through the event log mechanism. The interrupt mailbox commands were extended in CXL 3.1 to support these notifications. Firmware can't configure DCD events to be FW controlled but can retain control of memory events. Configure DCD event log interrupts on devices supporting dynamic capacity. Disable DCD if interrupts are not supported. The DCD event log is not drained here. cxl_event_drain_mask() reports only the logs the driver can service, and the DCD bit is added by the patch introducing DCD event handling. Until then a DCD interrupt wakes the event thread, which finds no log it owns and returns. Care is taken to preserve the interrupt policy set by the FW if FW first has been selected by the BIOS. Based on an original patch by Navneet Singh. Signed-off-by: Ira Weiny <iweiny@kernel.org> Co-developed-by: Anisa Su <anisa.su@samsung.com> Signed-off-by: Anisa Su <anisa.su@samsung.com> Tested-by: Wonjae Lee <wj28.lee@samsung.com> Tested-by: Junhee Park <jh9934.park@samsung.com> Tested-by: Heesoo Kim <habil.kim@samsung.com> Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com> Link: https://patch.msgid.link/20261010225852.10727-7-anisa.su@samsung.com Signed-off-by: Dave Jiang <dave.jiang@intel.com> 
Dynamic Capacity Devices (DCD) support extent change notifications
through the event log mechanism. The interrupt mailbox commands were
extended in CXL 3.1 to support these notifications. Firmware can't
configure DCD events to be FW controlled but can retain control of
memory events.

Configure DCD event log interrupts on devices supporting dynamic
capacity. Disable DCD if interrupts are not supported.

The DCD event log is not drained here. cxl_event_drain_mask() reports
only the logs the driver can service, and the DCD bit is added by
the patch introducing DCD event handling. Until then a DCD interrupt
wakes the event thread, which finds no log it owns and returns.

Care is taken to preserve the interrupt policy set by the FW if FW first
has been selected by the BIOS.

Based on an original patch by Navneet Singh.

Signed-off-by: Ira Weiny <iweiny@kernel.org>
Co-developed-by: Anisa Su <anisa.su@samsung.com>
Signed-off-by: Anisa Su <anisa.su@samsung.com>
Tested-by: Wonjae Lee <wj28.lee@samsung.com>
Tested-by: Junhee Park <jh9934.park@samsung.com>
Tested-by: Heesoo Kim <habil.kim@samsung.com>
Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com>
Link: https://patch.msgid.link/20261010225852.10727-7-anisa.su@samsung.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
cxl/pci: Factor out interrupt policy check 2026-10-12T16:44:52+00:00 Ira Weiny iweiny@kernel.org 2026-10-10T22:54:56+00:00 726ac97330a47642dca3b6322d59f038634cdc19 Dynamic Capacity Devices (DCD) require event interrupts to process memory addition or removal. BIOS may have control over non-DCD event processing. DCD interrupt configuration needs to be separate from memory event interrupt configuration. Factor out check for firmware ownership of memory event logs. Link: https://lore.kernel.org/all/663922b475e50_d54d72945b@dwillia2-xfh.jf.intel.com.notmuch/ [1] Suggested-by: Dan Williams <djbw@kernel.org> Signed-off-by: Ira Weiny <iweiny@kernel.org> Signed-off-by: Anisa Su <anisa.su@samsung.com> Tested-by: Wonjae Lee <wj28.lee@samsung.com> Tested-by: Junhee Park <jh9934.park@samsung.com> Tested-by: Heesoo Kim <habil.kim@samsung.com> Reviewed-by: Dave Jiang <dave.jiang@intel.com> Reviewed-by: Jonathan Cameron <jic23@kernel.org> Reviewed-by: Fan Ni <nifan.cxl@gmail.com> Reviewed-by: Li Ming <ming.li@zohomail.com> Reviewed-by: Davidlohr Bueso <dave@stgolabs.net> Reviewed-by: Alison Schofield <alison.schofield@intel.com> Reviewed-by: Richard Cheng <icheng@nvidia.com> Tested-by: Alison Schofield <alison.schofield@intel.com> Link: https://patch.msgid.link/20261010225852.10727-6-anisa.su@samsung.com Signed-off-by: Dave Jiang <dave.jiang@intel.com> 
Dynamic Capacity Devices (DCD) require event interrupts to process
memory addition or removal.  BIOS may have control over non-DCD event
processing.  DCD interrupt configuration needs to be separate from
memory event interrupt configuration.

Factor out check for firmware ownership of memory event logs.

Link: https://lore.kernel.org/all/663922b475e50_d54d72945b@dwillia2-xfh.jf.intel.com.notmuch/ [1]
Suggested-by: Dan Williams <djbw@kernel.org>
Signed-off-by: Ira Weiny <iweiny@kernel.org>
Signed-off-by: Anisa Su <anisa.su@samsung.com>
Tested-by: Wonjae Lee <wj28.lee@samsung.com>
Tested-by: Junhee Park <jh9934.park@samsung.com>
Tested-by: Heesoo Kim <habil.kim@samsung.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <jic23@kernel.org>
Reviewed-by: Fan Ni <nifan.cxl@gmail.com>
Reviewed-by: Li Ming <ming.li@zohomail.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Richard Cheng <icheng@nvidia.com>
Tested-by: Alison Schofield <alison.schofield@intel.com>
Link: https://patch.msgid.link/20261010225852.10727-6-anisa.su@samsung.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
cxl/events: Split event msgnum configuration from irq setup 2026-10-12T16:44:52+00:00 Ira Weiny iweiny@kernel.org 2026-10-10T22:54:55+00:00 f9c672229ca8a8d272c5e5c8ea3105238d558cb0 Dynamic Capacity Devices (DCD) require event interrupts to process memory addition or removal. BIOS may have control over non-DCD event processing. DCD interrupt configuration needs to be separate from memory event interrupt configuration. Split cxl_event_config_msgnums() from irq setup in preparation for separate DCD interrupts configuration. Signed-off-by: Ira Weiny <iweiny@kernel.org> Signed-off-by: Anisa Su <anisa.su@samsung.com> Tested-by: Wonjae Lee <wj28.lee@samsung.com> Tested-by: Junhee Park <jh9934.park@samsung.com> Tested-by: Heesoo Kim <habil.kim@samsung.com> Reviewed-by: Jonathan Cameron <jic23@kernel.org> Reviewed-by: Fan Ni <nifan.cxl@gmail.com> Reviewed-by: Dave Jiang <dave.jiang@intel.com> Reviewed-by: Li Ming <ming.li@zohomail.com> Reviewed-by: Alison Schofield <alison.schofield@intel.com> Reviewed-by: Davidlohr Bueso <dave@stgolabs.net> Reviewed-by: Richard Cheng <icheng@nvidia.com> Tested-by: Alison Schofield <alison.schofield@intel.com> Link: https://patch.msgid.link/20261010225852.10727-5-anisa.su@samsung.com Signed-off-by: Dave Jiang <dave.jiang@intel.com> 
Dynamic Capacity Devices (DCD) require event interrupts to process
memory addition or removal.  BIOS may have control over non-DCD event
processing.  DCD interrupt configuration needs to be separate from
memory event interrupt configuration.

Split cxl_event_config_msgnums() from irq setup in preparation for
separate DCD interrupts configuration.

Signed-off-by: Ira Weiny <iweiny@kernel.org>
Signed-off-by: Anisa Su <anisa.su@samsung.com>
Tested-by: Wonjae Lee <wj28.lee@samsung.com>
Tested-by: Junhee Park <jh9934.park@samsung.com>
Tested-by: Heesoo Kim <habil.kim@samsung.com>
Reviewed-by: Jonathan Cameron <jic23@kernel.org>
Reviewed-by: Fan Ni <nifan.cxl@gmail.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Li Ming <ming.li@zohomail.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Richard Cheng <icheng@nvidia.com>
Tested-by: Alison Schofield <alison.schofield@intel.com>
Link: https://patch.msgid.link/20261010225852.10727-5-anisa.su@samsung.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
cxl/cdat: Gather DSMAS data for DCD partitions 2026-10-12T16:44:52+00:00 Ira Weiny iweiny@kernel.org 2026-10-10T22:54:54+00:00 a7d2b5e9c4f51a17912c1a4470a24e038c944067 Additional DCD partition (AKA region) information is contained in the DSMAS CDAT tables, including performance, shareable and read-only attributes. Match DCD partitions with DSMAS tables and store the meta data. A DCD shall describe its DC partition in a DSMAS entry (CXL r4.0 9.13.3), so if the CDAT is absent, fails to parse, or has no entry matching the DC partition, disable DCD rather than run with unknown attributes. DSLBIS is optional, but cxl_cdat_endpoint_process() treated zero DSLBIS entries as a parse failure, which would now disable DCD on a device with a valid DSMAS. Accept an empty DSLBIS, and skip DSMAS entries with no DSLBIS coordinates. Signed-off-by: Ira Weiny <iweiny@kernel.org> Co-developed-by: Anisa Su <anisa.su@samsung.com> Signed-off-by: Anisa Su <anisa.su@samsung.com> Tested-by: Wonjae Lee <wj28.lee@samsung.com> Tested-by: Junhee Park <jh9934.park@samsung.com> Tested-by: Heesoo Kim <habil.kim@samsung.com> Reviewed-by: Alison Schofield <alison.schofield@intel.com> Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com> Link: https://patch.msgid.link/20261010225852.10727-4-anisa.su@samsung.com Signed-off-by: Dave Jiang <dave.jiang@intel.com> 
Additional DCD partition (AKA region) information is contained in the
DSMAS CDAT tables, including performance, shareable and read-only
attributes.

Match DCD partitions with DSMAS tables and store the meta data. A DCD
shall describe its DC partition in a DSMAS entry (CXL r4.0 9.13.3), so
if the CDAT is absent, fails to parse, or has no entry matching the DC
partition, disable DCD rather than run with unknown attributes.

DSLBIS is optional, but cxl_cdat_endpoint_process() treated zero DSLBIS
entries as a parse failure, which would now disable DCD on a device with
a valid DSMAS. Accept an empty DSLBIS, and skip DSMAS entries with no
DSLBIS coordinates.

Signed-off-by: Ira Weiny <iweiny@kernel.org>
Co-developed-by: Anisa Su <anisa.su@samsung.com>
Signed-off-by: Anisa Su <anisa.su@samsung.com>
Tested-by: Wonjae Lee <wj28.lee@samsung.com>
Tested-by: Junhee Park <jh9934.park@samsung.com>
Tested-by: Heesoo Kim <habil.kim@samsung.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com>
Link: https://patch.msgid.link/20261010225852.10727-4-anisa.su@samsung.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>