{"cve":"CVE-2026-74555","nvd":{"published":"2026-08-15T13:18:00.927","modified":"2026-08-19T17:21:10.613","status":"Received","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nscsi: libsas: Fix HA resume deadlock and hisi_sas disk-wake race\n\nCommit fbefe22811c3 (\"scsi: libsas: Don't always drain event workqueue\nfor HA resume\") introduced sas_resume_ha_no_sync() to avoid a deadlock:\nthe PHYE_RESUME_TIMEOUT handler, running on the HA event workqueue,\ncalls sas_deform_port() -> sas_destruct_devices(), which removes SCSI\ndevices and waits for the host to become runtime-active. But the host\ncannot resume until sas_resume_ha() -> sas_drain_work() returns, and the\ndrain is blocked on that very handler.\n\nHowever skipping the drain reintroduces a race: hisi_sas returns from\nresume before all PHY UP work and libsas discovery work finish. The\ncontroller may then autosuspend while disks are still waking up. The\ndisks issue IO to a suspended controller, the IO fails, and the disks\nget disabled.\n\nFix the deadlock at its source by moving the PHYE_RESUME_TIMEOUT\nnotification to after sas_drain_work(). By then the host resume is about\nto complete, so device removal through device_link no longer blocks on\nthe resume and the cycle is broken.\n\nWith the deadlock gone, restore sas_resume_ha() (the draining variant)\nin hisi_sas and remove sas_resume_ha_no_sync().\n\nThe reorder is safe for the other libsas consumers (isci, pm8001,\naic94xx, mvsas). During suspend, sas_suspend_devices() calls\nsas_notify_lldd_dev_gone() for each device, which sets dev->lldd_dev to\nNULL. When scsi_unblock_requests re-enables I/O in resume, any I/O to a\ntimed-out phy's disk is immediately rejected by the LLDD before reaching\nhardware: isci returns SAS_DEVICE_UNKNOWN (mapped to DID_BAD_TARGET),\nand pm8001 returns SAS_PHY_DOWN (mapped to DID_NO_CONNECT). Both\ncomplete directly via scsi_done() without entering SCSI EH. This is\nidentical in both the old and new ordering since lldd_dev_gone runs\nduring suspend, before resume. The reorder only affects when the\nPHYE_RESUME_TIMEOUT handler runs (synchronized by sas_drain_work()\nvs. asynchronous after resume returns), not whether I/O can reach the\ndevice. aic94xx and mvsas do not register any PM ops and never reach\nthis code path.","cvss_v4":null,"cvss_v3":null,"cwes":[],"references":["https://git.kernel.org/stable/c/09357f067122e2e28ec52e27013a1660a1571618","https://git.kernel.org/stable/c/3dbbbf656b850c9c8de05df6ad4a1dfc6ff02845","https://git.kernel.org/stable/c/9e24b47ef81d43b3fb1b14294f09991640c79fcc","https://git.kernel.org/stable/c/b9c44a14062093e9fc2d6bddc696cfceadb482d7","https://git.kernel.org/stable/c/c391b5899dd46485a5893696c12ae3e95a3a7325"],"products":[],"source":"nvd","synced_at":"2026-10-19T22:49:10.689564601+00:00"},"epss":{"cve":"CVE-2026-74555","score":0.00161,"percentile":0.05252,"date":"2026-10-31"},"kev":null,"osv":[],"attribution":["This product uses the NVD API but is not endorsed or certified by the NVD.","EPSS scores: FIRST Exploit Prediction Scoring System, https://www.first.org/epss","Known exploited vulnerabilities: CISA Known Exploited Vulnerabilities Catalog, https://www.cisa.gov/known-exploited-vulnerabilities-catalog","Package advisories: OSV (https://osv.dev), from the GitHub Advisory Database, PyPA, Go, RustSec and other ecosystem databases, under CC-BY 4.0 or CC0","CVE records: the CVE Program's CVE List (https://www.cve.org), under its terms of use (https://www.cve.org/Legal/TermsOfUse). CVE is a registered trademark of The MITRE Corporation; this product is not affiliated with the CVE Program."]}