{"id":"AZL-96102","summary":"CVE-2026-74481 affecting package kernel 6.6.150.1-1","details":"In the Linux kernel, the following vulnerability has been resolved:\n\nmm/page_reporting: use system_freezable_wq to fix UAF during suspend\n\nDuring PM freeze (e.g.  S3 suspend or S4 hibernation), device drivers like\nvirtio_balloon reset their underlying virtio devices and delete their\nvirtqueues via vdev-\u003econfig-\u003edel_vqs().\n\nHowever, page reporting work (page_reporting_process) was scheduled on the\nglobal system_wq.  Because system_wq lacks the WQ_FREEZABLE flag, the PM\nfreezer skips it, leaving page_reporting_process active during suspend.\n\nIf pages are freed into the buddy allocator while suspending (for example,\nwhen core MM invokes the balloon shrinker during S4 hibernation image\nsaving), page reporting triggers virtballoon_free_page_report() on deleted\nvirtqueues, resulting in a Use-After-Free / General Protection Fault:\n\n    [  196.795226] general protection fault, probably for non-canonical address 0xaa1436fe70dae6df: 0000 [#1] SMP NOPTI\n    [  196.825967] Workqueue: events page_reporting_process\n    [  196.831038] RIP: 0010:virtqueue_add_split+0x233/0x4c0 [virtio_ring]\n    [  196.927073] virtballoon_free_page_report+0x3a/0xe0 [virtio_balloon]\n    [  196.946943] page_reporting_process+0x370/0x4f0\n\nFix this by switching page reporting work to system_freezable_wq.  This\nensures that the PM freezer pauses page_reporting_process before device\ndrivers destroy their reporting virtqueues.  Because the reporting worker\nis frozen, memory reclamation/freeing (e.g.  via shrinker execution) can\nsafely return pages to MM during freeze without triggering unfrozen\nreporting work on deleted virtqueues.\n\nThis aligns with the driver's existing design. The comment in\nvirtballoon_freeze() states:\n    /*\n     * The workqueue is already frozen by the PM core before this\n     * function is called.\n     */\n\nTesting:\nI have verified these fixes using Google’s virtualization infrastructure\nby running continuous suspend/resume iterations (40+ cycles) while\nchurning memory using stress-ng (`stress-ng --vm 4 --vm-bytes 60%\n--timeout 1`) to constantly create free pages for the buddy allocator.  We\nalso set the `page_reporting_order` parameter to 0 to make the page\nreporting worker highly sensitive, forcing it to pick up any 4K free\npages.  This confirmed that the UAF crashes are no longer reproducible.","modified":"2026-08-31T05:26:27Z","published":"2026-08-15T13:17:52Z","upstream":["CVE-2026-74481"],"references":[{"type":"WEB","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-74481"}],"affected":[{"package":{"name":"kernel","ecosystem":"Azure Linux:3","purl":"pkg:rpm/azure-linux/kernel"},"ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"last_affected":"6.6.150.1-1"}]}],"database_specific":{"source":"https://github.com/microsoft/AzureLinuxVulnerabilityData/blob/main/osv/AZL-96102.json"}}],"schema_version":"1.9.0"}