linux-dev/arch/x86/kernel/cpu/mce, branch master

x86/mce: Retrieve poison range from hardware

2022-08-29T07:33:42Z

When memory poison consumption machine checks fire, MCE notifier handlers like nfit_handle_mce() record the impacted physical address range which is reported by the hardware in the MCi_MISC MSR. The error information includes data about blast radius, i.e. how many cachelines did the hardware determine are impacted. A recent change 7917f9cdb503 ("acpi/nfit: rely on mce->misc to determine poison granularity") updated nfit_handle_mce() to stop hard coding the blast radius value of 1 cacheline, and instead rely on the blast radius reported in 'struct mce' which can be up to 4K (64 cachelines). It turns out that apei_mce_report_mem_error() had a similar problem in that it hard coded a blast radius of 4K rather than reading the blast radius from the error information. Fix apei_mce_report_mem_error() to convey the proper poison granularity. Signed-off-by: Jane Chu Signed-off-by: Borislav Petkov Reviewed-by: Dan Williams Reviewed-by: Ingo Molnar Link: https://lore.kernel.org/r/7ed50fd8-521e-cade-77b1-738b8bfb8502@oracle.com Link: https://lore.kernel.org/r/20220826233851.1319100-1-jane.chu@oracle.com

x86/mce: Check whether writes to MCA_STATUS are getting ignored

2022-06-28T10:08:10Z

The platform can sometimes - depending on its settings - cause writes to MCA_STATUS MSRs to get ignored, regardless of HWCR[McStatusWrEn]'s value. For further info see PPR for AMD Family 19h, Model 01h, Revision B1 Processors, doc ID 55898 at https://bugzilla.kernel.org/show_bug.cgi?id=206537. Therefore, probe for ignored writes to MCA_STATUS to determine if hardware error injection is at all possible. [ bp: Heavily massage commit message and patch. ] Signed-off-by: Smita Koralahalli Signed-off-by: Borislav Petkov Link: https://lore.kernel.org/r/20220214233640.70510-2-Smita.KoralahalliChannabasappa@amd.com

Merge tag 'libnvdimm-for-5.19' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm

2022-05-27T22:49:30Z

Pull libnvdimm and DAX updates from Dan Williams: "New support for clearing memory errors when a file is in DAX mode, alongside with some other fixes and cleanups. Previously it was only possible to clear these errors using a truncate or hole-punch operation to trigger the filesystem to reallocate the block, now, any page aligned write can opportunistically clear errors as well. This change spans x86/mm, nvdimm, and fs/dax, and has received the appropriate sign-offs. Thanks to Jane for her work on this. Summary: - Add support for clearing memory error via pwrite(2) on DAX - Fix 'security overwrite' support in the presence of media errors - Miscellaneous cleanups and fixes for nfit_test (nvdimm unit tests)" * tag 'libnvdimm-for-5.19' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm: pmem: implement pmem_recovery_write() pmem: refactor pmem_clear_poison() dax: add .recovery_write dax_operation dax: introduce DAX_RECOVERY_WRITE dax access mode mce: fix set_mce_nospec to always unmap the whole page x86/mce: relocate set{clear}_mce_nospec() functions acpi/nfit: rely on mce->misc to determine poison granularity testing: nvdimm: asm/mce.h is not needed in nfit.c testing: nvdimm: iomap: make __nfit_test_ioremap a macro nvdimm: Allow overwrite in the presence of disabled dimms tools/testing/nvdimm: remove unneeded flush_workqueue

Merge tag 'acpi-5.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

2022-05-24T22:46:55Z

Pull ACPI updates from Rafael Wysocki: "These update the ACPICA kernel code to upstream revision 20220331, improve handling of PCI devices that are in D3cold during system initialization, add support for a few features, fix bugs and clean up code. Specifics: - Update ACPICA code in the kernel to upstream revision 20220331 including the following changes: - Add support for the Windows 11 _OSI string (Mario Limonciello) - Add the CFMWS subtable to the CEDT table (Lawrence Hileman). - iASL: NHLT: Treat Terminator as specific_config (Piotr Maziarz). - iASL: NHLT: Fix parsing undocumented bytes at the end of Endpoint Descriptor (Piotr Maziarz). - iASL: NHLT: Rename linux specific strucures to device_info (Piotr Maziarz). - Add new ACPI 6.4 semantics to Load() and LoadTable() (Bob Moore). - Clean up double word in comment (Tom Rix). - Update copyright notices to the year 2022 (Bob Moore). - Remove some tabs and // comments - automated cleanup (Bob Moore). - Replace zero-length array with flexible-array member (Gustavo A. R. Silva). - Interpreter: Add units to time variable names (Paul Menzel). - Add support for ARM Performance Monitoring Unit Table (Besar Wicaksono). - Inform users about ACPI spec violation related to sleep length (Paul Menzel). - iASL/MADT: Add OEM-defined subtable (Bob Moore). - Interpreter: Fix some typo mistakes (Selvarasu Ganesan). - Updates for revision E.d of IORT (Shameer Kolothum). - Use ACPI_FORMAT_UINT64 for 64-bit output (Bob Moore). - Improve debug messages in the ACPI device PM code (Rafael Wysocki). - Block ASUS B1400CEAE from suspend to idle by default (Mario Limonciello). - Improve handling of PCI devices that are in D3cold during system initialization (Rafael Wysocki). - Fix BERT error region memory mapping (Lorenzo Pieralisi). - Add support for NVIDIA 16550-compatible port subtype to the SPCR parsing code (Jeff Brasen). - Use static for BGRT_SHOW kobj_attribute defines (Tom Rix). - Fix missing prototype warning for acpi_agdi_init() (Ilkka Koskinen). - Fix missing ERST record ID in the APEI code (Liu Xinpeng). - Make APEI error injection to refuse to inject into the zero page (Tony Luck). - Correct description of INT3407 / INT3532 DPTF attributes in sysfs (Sumeet Pawnikar). - Add support for high frequency impedance notification to the DPTF driver (Sumeet Pawnikar). - Make mp_config_acpi_gsi() a void function (Li kunyu). - Unify Package () representation for properties in the ACPI device properties documentation (Andy Shevchenko). - Include UUID in _DSM evaluation warning (Michael Niewöhner)" * tag 'acpi-5.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (41 commits) Revert "ACPICA: executer/exsystem: Warn about sleeps greater than 10 ms" ACPI: utils: include UUID in _DSM evaluation warning ACPI: PM: Block ASUS B1400CEAE from suspend to idle by default x86: ACPI: Make mp_config_acpi_gsi() a void function ACPI: DPTF: Add support for high frequency impedance notification ACPI: AGDI: Fix missing prototype warning for acpi_agdi_init() ACPI: bus: Avoid non-ACPI device objects in walks over children ACPI: DPTF: Correct description of INT3407 / INT3532 attributes ACPI: BGRT: use static for BGRT_SHOW kobj_attribute defines ACPI, APEI, EINJ: Refuse to inject into the zero page ACPI: PM: Always print final debug message in acpi_device_set_power() ACPI: SPCR: Add support for NVIDIA 16550-compatible port subtype ACPI: docs: enumeration: Unify Package () for properties (part 2) ACPI: APEI: Fix missing ERST record id ACPICA: Update version to 20220331 ACPICA: exsystem.c: Use ACPI_FORMAT_UINT64 for 64-bit output ACPICA: IORT: Updates for revision E.d ACPICA: executer/exsystem: Fix some typo mistakes ACPICA: iASL/MADT: Add OEM-defined subtable ACPICA: executer/exsystem: Warn about sleeps greater than 10 ms ...

mce: fix set_mce_nospec to always unmap the whole page

2022-05-16T18:46:44Z

The set_memory_uc() approach doesn't work well in all cases. As Dan pointed out when "The VMM unmapped the bad page from guest physical space and passed the machine check to the guest." "The guest gets virtual #MC on an access to that page. When the guest tries to do set_memory_uc() and instructs cpa_flush() to do clean caches that results in taking another fault / exception perhaps because the VMM unmapped the page from the guest." Since the driver has special knowledge to handle NP or UC, mark the poisoned page with NP and let driver handle it when it comes down to repair. Please refer to discussions here for more details. https://lore.kernel.org/all/CAPcyv4hrXPb1tASBZUg-GgdVs0OOFKXMXLiHmktg_kFi7YBMyQ@mail.gmail.com/ Now since poisoned page is marked as not-present, in order to avoid writing to a not-present page and trigger kernel Oops, also fix pmem_do_write(). Fixes: 284ce4011ba6 ("x86/memory_failure: Introduce {set, clear}_mce_nospec()") Reviewed-by: Christoph Hellwig Reviewed-by: Dan Williams Signed-off-by: Jane Chu Acked-by: Tony Luck Link: https://lore.kernel.org/r/165272615484.103830.2563950688772226611.stgit@dwillia2-desk3.amr.corp.intel.com Signed-off-by: Dan Williams

x86/mce: Add messages for panic errors in AMD's MCE grading

2022-04-25T10:40:48Z

When a machine error is graded as PANIC by the AMD grading logic, the MCE handler calls mce_panic(). The notification chain does not come into effect so the AMD EDAC driver does not decode the errors. In these cases, the messages displayed to the user are more cryptic and miss information that might be relevant, like the context in which the error took place. Add messages to the grading logic for machine errors so that it is clear what error it was. [ bp: Massage commit message. ] Signed-off-by: Carlos Bilbao Signed-off-by: Borislav Petkov Reviewed-by: Yazen Ghannam Link: https://lore.kernel.org/r/20220405183212.354606-3-carlos.bilbao@amd.com

x86/mce: Simplify AMD severity grading logic

2022-04-25T10:32:03Z

The MCE handler needs to understand the severity of the machine errors to act accordingly. Simplify the AMD grading logic following a logic that closely resembles the descriptions of the public PPR documents. This will help include more fine-grained grading of errors in the future. [ bp: Touchups. ] Signed-off-by: Carlos Bilbao Signed-off-by: Borislav Petkov Reviewed-by: Yazen Ghannam Link: https://lore.kernel.org/r/20220405183212.354606-2-carlos.bilbao@amd.com

ACPI: APEI: Fix missing ERST record id

2022-04-13T18:29:24Z

Read a record is cleared by others, but the deleted record cache entry is still created by erst_get_record_id_next. When next enumerate the records, get the cached deleted record, then erst_read() return -ENOENT and try to get next record, loop back to first ID will return 0 in function __erst_record_id_cache_add_one and then set record_id as APEI_ERST_INVALID_RECORD_ID, finished this time read operation. It will result in read the records just in the cache hereafter. This patch cleared the deleted record cache, fix the issue that "./erst-inject -p" shows record counts not equal to "./erst-inject -n". A reproducer of the problem(retry many times): [root@localhost erst-inject]# ./erst-inject -c 0xaaaaa00011 [root@localhost erst-inject]# ./erst-inject -p rc: 273 rcd sig: CPER rcd id: 0xaaaaa00012 rc: 273 rcd sig: CPER rcd id: 0xaaaaa00013 rc: 273 rcd sig: CPER rcd id: 0xaaaaa00014 [root@localhost erst-inject]# ./erst-inject -i 0xaaaaa000006 [root@localhost erst-inject]# ./erst-inject -i 0xaaaaa000007 [root@localhost erst-inject]# ./erst-inject -i 0xaaaaa000008 [root@localhost erst-inject]# ./erst-inject -p rc: 273 rcd sig: CPER rcd id: 0xaaaaa00012 rc: 273 rcd sig: CPER rcd id: 0xaaaaa00013 rc: 273 rcd sig: CPER rcd id: 0xaaaaa00014 [root@localhost erst-inject]# ./erst-inject -n total error record count: 6 Signed-off-by: Liu Xinpeng Reviewed-by: Tony Luck Signed-off-by: Rafael J. Wysocki

x86/MCE/AMD: Fix memory leak when threshold_create_bank() fails

2022-04-05T19:24:37Z

In mce_threshold_create_device(), if threshold_create_bank() fails, the previously allocated threshold banks array @bp will be leaked because the call to mce_threshold_remove_device() will not free it. This happens because mce_threshold_remove_device() fetches the pointer through the threshold_banks per-CPU variable but bp is written there only after the bank creation is successful, and not before, when threshold_create_bank() fails. Add a helper which unwinds all the bank creation work previously done and pass into it the previously allocated threshold banks array for freeing. [ bp: Massage. ] Fixes: 6458de97fc15 ("x86/mce/amd: Straighten CPU hotplug path") Co-developed-by: Alviro Iskandar Setiawan Signed-off-by: Alviro Iskandar Setiawan Co-developed-by: Yazen Ghannam Signed-off-by: Yazen Ghannam Signed-off-by: Ammar Faizi Signed-off-by: Borislav Petkov Cc: Link: https://lore.kernel.org/r/20220329104705.65256-3-ammarfaizi2@gnuweeb.org

x86/mce: Avoid unnecessary padding in struct mce_bank

2022-04-05T19:23:34Z

Convert struct mce_bank member "init" from bool to a bitfield to get rid of unnecessary padding. $ pahole -C mce_bank arch/x86/kernel/cpu/mce/core.o before: /* size: 16, cachelines: 1, members: 2 */ /* padding: 7 */ /* last cacheline: 16 bytes */ after: /* size: 16, cachelines: 1, members: 3 */ /* last cacheline: 16 bytes */ No functional changes. Signed-off-by: Smita Koralahalli Signed-off-by: Borislav Petkov Link: https://lore.kernel.org/r/20220225193342.215780-2-Smita.KoralahalliChannabasappa@amd.com