CVE-2026-99096

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> io_uring/register: don&amp;#39;t exempt disabled rings from task restrictions<br /> <br /> A ring created with IORING_SETUP_R_DISABLED is exempt from its register<br /> restrictions until it is enabled. This lets the creator register the<br /> restrictions and then enable the ring with IORING_REGISTER_ENABLE_RINGS,<br /> whether or not that opcode is in the allowed set.<br /> <br /> Since commit ed82f35b926b ("io_uring: allow registration of per-task<br /> restrictions"), a ring also inherits restrictions from the task that<br /> creates it, and the same exemption applies to those. A task whose<br /> register opcodes are restricted can therefore create a disabled ring,<br /> issue any register opcode on it, and then enable it.<br /> <br /> Mark rings whose restrictions were inherited from the task, and keep<br /> checking register opcodes for them while the ring is disabled. Enabling<br /> such a ring now requires IORING_REGISTER_ENABLE_RINGS to be in the<br /> task&amp;#39;s allowed set. Per-ring restrictions registered on a disabled ring<br /> work as before.<br /> <br /> [axboe: amend with inheritance flag fixup]