CVE-2026-64556

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> perf/core: Detach event groups during remove_on_exec<br /> <br /> perf_event_remove_on_exec() removes events by calling<br /> perf_event_exit_event(). For top-level events, this removes the event from<br /> the context with DETACH_EXIT only.<br /> <br /> This can leave inconsistent group state when a removed event is a group<br /> leader and the group contains siblings without remove_on_exec. If the group<br /> was active, the surviving siblings can remain active and attached to the<br /> removed leader&amp;#39;s sibling list, but are no longer represented by a valid<br /> group leader on the PMU context active lists.<br /> <br /> A later close of the removed leader uses DETACH_GROUP and can promote the<br /> still-active siblings from this stale group state. The next schedule-in can<br /> then add an already-linked active_list entry again, corrupting the PMU<br /> context active list.<br /> <br /> With DEBUG_LIST enabled, this is caught as a list_add double-add in<br /> merge_sched_in().<br /> <br /> Fix this by detaching group relationships when remove_on_exec removes an<br /> event. This preserves the existing task-exit and revoke behavior, while<br /> ensuring surviving siblings are ungrouped before the removed event leaves<br /> the context.