понедельник

[Bug 2174560] Re: ACPI BIOS errors from \WBRF during s2idle on AMD systems

** Tags added: kernel-daily-bug -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2174560 Title: ACPI BIOS errors from \WBRF during s2idle on AMD systems Status in HWE Next: New Status in linux package in Ubuntu: New Status in linux-oem-7.3 package in Ubuntu: New Status in linux source package in Resolute: New Status in linux-oem-7.3 source package in Resolute: In Progress Status in linux source package in Stonking: New Status in linux-oem-7.3 source package in Stonking: Invalid Bug description: [Impact] On some AMD systems, the amd_s2idle test reports "ACPI BIOS errors found" and "Page faults found" on every suspend cycle (CID: not listed in the ticket). The WBRF driver sends the _DSM record argument as a package of integers. The spec says it must be a single buffer holding the header and the frequency ranges. Some firmware aborts the \WBRF method when it reads Arg2[0]: AE_AML_OPERAND_TYPE The error name comes from the upstream commit message. Hits 100% on affected systems (Wistron-built AMD platforms). [Fix] Pack the WBRF record into a little-endian buffer (number of ranges, action, then start/end pairs). Pass that buffer as the only element of the _DSM package. Upstream patch, not merged yet: https://lore.kernel.org/platform-driver-x86/20261102-superm1-wbrf-v1-1-b58d8d85277c@kernel.org/ platform/x86/amd: send WBRF DSM payload as buffer Author: Mario Limonciello (AMD) It fixes the original WBRF support, commit 58e82a62669d ("platform/x86/amd: Add support for AMD ACPI based Wifi band RFI mitigation feature"), which is in v6.10. [Test Plan] Run the s2idle check 3 times: $ sudo fwts s3 --s3-multiple=3 Without patch: "ACPI BIOS errors found" and "Page faults found" after each cycle. dmesg shows AE_AML_OPERAND_TYPE from \WBRF. With patch: no ACPI BIOS errors and no AE_AML_OPERAND_TYPE in dmesg. Tested on an OEM 7.3 kernel with this patch. The reporter confirmed there are no more ACPI errors. [Where problems could occur] Could break the AMD WBRF driver (drivers/platform/x86/amd/wbrf.c). That driver sends Wifi band records and retrieves them for the AMD GPU and Wifi drivers. If the buffer layout is wrong, firmware could read bad ranges or reject the call. Then Wifi band avoidance would stop working. The system could show Wifi interference with memory clocks, or the ACPI error could stay. Firmware that already accepted the old integer package could now fail with the buffer. The patch changes the argument for every system with WBRF,not just the broken ones. To manage notifications about this bug go to: https://bugs.launchpad.net/hwe-next/+bug/2174560/+subscriptions

Комментариев нет:

Отправить комментарий