** Description changed: Description: During run-time, the kernel can be notified of new/latent errors in two different ways: If an application trips over a latent/unknown error, and if the system has machine check recovery, then we will be notified via the mce handler, and the app will be killed. In this case, the kernel should take the chance to go clear-errors for that page, and re-online it so it can be used again in the future. If the 'patrol scrubber' discovers an error on a yet-to-be-accessed location, it can send an ACPI notification to the nfit driver. In this case, the kernel should go clear the error if the page is not in use. If it is in use, the application that has it mapped may need to killed as in case 1 above. This additional run-time handling of errors will augment (and be complimentary to) the init-time handling in userspace, and having both will give us the best possible coverage for media errors. - Target Release: 21.04 + Target Release: 21.10 Target Kernel: TBD
** Tags removed: intel-kernel-21.04 ** Tags added: intel-kernel-21.10 -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu. https://bugs.launchpad.net/bugs/1835340 Title: [HMEM] kmem: clear poison reported via mce or ACPI patrol scrub notifications Status in intel: New Status in linux package in Ubuntu: Incomplete Bug description: Description: During run-time, the kernel can be notified of new/latent errors in two different ways: If an application trips over a latent/unknown error, and if the system has machine check recovery, then we will be notified via the mce handler, and the app will be killed. In this case, the kernel should take the chance to go clear-errors for that page, and re-online it so it can be used again in the future. If the 'patrol scrubber' discovers an error on a yet-to-be-accessed location, it can send an ACPI notification to the nfit driver. In this case, the kernel should go clear the error if the page is not in use. If it is in use, the application that has it mapped may need to killed as in case 1 above. This additional run-time handling of errors will augment (and be complimentary to) the init-time handling in userspace, and having both will give us the best possible coverage for media errors. Target Release: 21.10 Target Kernel: TBD To manage notifications about this bug go to: https://bugs.launchpad.net/intel/+bug/1835340/+subscriptions -- Mailing list: https://launchpad.net/~kernel-packages Post to : kernel-packages@lists.launchpad.net Unsubscribe : https://launchpad.net/~kernel-packages More help : https://help.launchpad.net/ListHelp
