CVE-2026-74591

A flaw was found in the Linux kernel. A race condition in the __filemap_add_folio() function, responsible for managing memory pages, can occur during specific memory allocation retries. If another process resolves a memory conflict while the current operation is paused, an incorrect memory address might be used to store data. This can lead to memory corruption, potentially causing system instability (Denial of Service) or allowing an attacker to execute unauthorized code.

__filemap_add_folio can retry after xas_nomem drops the xarray lock while xas.xa_index still contains an intermediate rounded down index from xas_set_order. If another thread resolves the conflict before the retry, the folio can be inserted at the wrong file offset, causing page cache index corruption and execution of wrong file contents. Observed production symptoms include rare SIGILL/SIGSEGV and executable text found one page away from the expected location. For the CVSS the PR:L is used because a local process that can drive file backed page cache activity can potentially reach the path, while reliable triggering still requires a narrow race and memory allocation retry. The issue is not network reachable. Impact is at least local DoS and file/page cache integrity corruption. In the paranoid score, confidentiality and integrity are raised because wrong-offset page cache mapping may expose or execute unintended file data and should be reviewed as a potential local privilege escalation candidate.

The CVSS score(s) listed for this vulnerability do not reflect the associated product's status, and are included for informational purposes to better understand the severity of this vulnerability.

Mitigation for this issue is either not available or the currently available options don't meet the Red Hat Product Security criteria comprising ease of use and deployment, applicability to widespread installation base or stability.