CVE-2025-38708

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drbd: add missing kref_get in handle_write_conflicts<br /> <br /> With `two-primaries` enabled, DRBD tries to detect "concurrent" writes<br /> and handle write conflicts, so that even if you write to the same sector<br /> simultaneously on both nodes, they end up with the identical data once<br /> the writes are completed.<br /> <br /> In handling "superseeded" writes, we forgot a kref_get,<br /> resulting in a premature drbd_destroy_device and use after free,<br /> and further to kernel crashes with symptoms.<br /> <br /> Relevance: No one should use DRBD as a random data generator, and apparently<br /> all users of "two-primaries" handle concurrent writes correctly on layer up.<br /> That is cluster file systems use some distributed lock manager,<br /> and live migration in virtualization environments stops writes on one node<br /> before starting writes on the other node.<br /> <br /> Which means that other than for "test cases",<br /> this code path is never taken in real life.<br /> <br /> FYI, in DRBD 9, things are handled differently nowadays. We still detect<br /> "write conflicts", but no longer try to be smart about them.<br /> We decided to disconnect hard instead: upper layers must not submit concurrent<br /> writes. If they do, that&amp;#39;s their fault.