Vulnerabilidades

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdgpu/vcn3: Prevent OOB reads when parsing dec msg<br /> <br /> Check bounds against the end of the BO whenever we access the msg.

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdkfd: Clear VRAM on allocation to prevent stale data exposure<br /> <br /> KFD VRAM allocations set AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE<br /> but not AMDGPU_GEM_CREATE_VRAM_CLEARED, leaving freshly allocated<br /> VRAM with stale data from prior use observable by compute kernels.<br /> <br /> The GEM ioctl path already sets VRAM_CLEARED for all userspace<br /> allocations via amdgpu_gem_create_ioctl() and<br /> amdgpu_mode_dumb_create(). The KFD path was missing this flag,<br /> allowing stale page table remnants to leak into user buffers.<br /> <br /> This causes crashes in RCCL P2P transport where non-zero data in<br /> ptrExchange/head/tail fields corrupts the protocol handshake.

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> spi: ch341: fix devres lifetime<br /> <br /> USB drivers bind to USB interfaces and any device managed resources<br /> should have their lifetime tied to the interface rather than parent USB<br /> device. This avoids issues like memory leaks when drivers are unbound<br /> without their devices being physically disconnected (e.g. on probe<br /> deferral or configuration changes).<br /> <br /> Fix the controller and driver data lifetime so that they are released<br /> on driver unbind.<br /> <br /> Note that this also makes sure that the SPI controller is placed<br /> correctly under the USB interface in the device tree.

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> sctp: revalidate list cursor after sctp_sendmsg_to_asoc() in SCTP_SENDALL<br /> <br /> The SCTP_SENDALL path in sctp_sendmsg() iterates ep-&gt;asocs with<br /> list_for_each_entry_safe(), which caches the next entry in @tmp before<br /> the loop body runs. The body calls sctp_sendmsg_to_asoc(), which may<br /> drop the socket lock inside sctp_wait_for_sndbuf().<br /> <br /> While the lock is dropped, another thread can SCTP_SOCKOPT_PEELOFF the<br /> association cached in @tmp, migrating it to a new endpoint via<br /> sctp_sock_migrate() (list_del_init() + list_add_tail() to<br /> newep-&gt;asocs), and optionally close the new socket which frees the<br /> association via kfree_rcu(). The cached @tmp can also be freed by a<br /> network ABORT for that association, processed in softirq while the<br /> lock is dropped.<br /> <br /> sctp_wait_for_sndbuf() revalidates @asoc (the current entry) on re-lock<br /> via the "sk != asoc-&gt;base.sk" and "asoc-&gt;base.dead" checks, but nothing<br /> revalidates @tmp. After a successful return, the iterator advances to<br /> the stale @tmp, yielding either a use-after-free (if the peeled socket<br /> was closed) or a list-walk onto the new endpoint&amp;#39;s list head (type<br /> confusion of &amp;newep-&gt;asocs as a struct sctp_association *).<br /> <br /> Both are reachable from CapEff=0; the type-confusion path gives<br /> controlled indirect call via the outqueue.sched-&gt;init_sid pointer.<br /> <br /> Fix by re-deriving @tmp from @asoc after sctp_sendmsg_to_asoc()<br /> returns. @asoc is known to still be on ep-&gt;asocs at that point: the<br /> only callers that list_del an association from ep-&gt;asocs are<br /> sctp_association_free() (which sets asoc-&gt;base.dead) and<br /> sctp_assoc_migrate() (which changes asoc-&gt;base.sk), and<br /> sctp_wait_for_sndbuf() checks both under the lock before any<br /> successful return; a tripped check propagates as err

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> spi: mpc52xx: fix use-after-free on unbind<br /> <br /> The state machine work is scheduled by the interrupt handler and<br /> therefore needs to be cancelled after disabling interrupts to avoid a<br /> potential use-after-free.

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdgpu/sdma4: replace BUG_ON with WARN_ON in fence emission<br /> <br /> sdma_v4_0_ring_emit_fence() contains two BUG_ON(addr &amp; 0x3) assertions<br /> that verify fence writeback addresses are dword-aligned. These<br /> assertions can be reached from unprivileged userspace via crafted<br /> DRM_IOCTL_AMDGPU_CS submissions, causing a fatal kernel panic in a<br /> scheduler worker thread.<br /> <br /> Replace both BUG_ON() calls with WARN_ON() to log the condition without<br /> crashing the kernel. A misaligned fence address at this point indicates<br /> a driver bug, but crashing the kernel is never the correct response when<br /> the assertion is reachable from userspace.<br /> <br /> The CS IOCTL path is the correct place to filter invalid submissions;<br /> the ring emission callback is too late to do anything about it.<br /> <br /> (cherry picked from commit b90250bd933afd1ba94d86d6b13821997b22b18e)

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> EDAC/versalnet: Fix device name memory leak<br /> <br /> The device name allocated via kzalloc() in init_one_mc() is assigned to<br /> dev-&gt;init_name but never freed on the normal removal path. device_register()<br /> copies init_name and then sets dev-&gt;init_name to NULL, so the name pointer<br /> becomes unreachable from the device. Thus leaking memory.<br /> <br /> Use a stack-local char array instead of using kzalloc() for name.

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> media: rockchip: rkcif: Add missing MUST_CONNECT flag to pads<br /> <br /> The pads missed checks for connected devices which may a null dereference<br /> when the stream is enabled.<br /> <br /> Unable to handle kernel NULL pointer dereference at virtual address<br /> 0000000000000020<br /> pc : rkcif_interface_enable_streams+0x48/0xf0<br /> lr : rkcif_interface_enable_streams+0x44/0xf0<br /> Call trace:<br /> rkcif_interface_enable_streams+0x48/0xf0<br /> v4l2_subdev_enable_streams+0x26c/0x3f0<br /> rkcif_stream_start_streaming+0x140/0x278<br /> vb2_start_streaming+0x74/0x188<br /> vb2_core_streamon+0xe0/0x1d8<br /> vb2_ioctl_streamon+0x60/0xa8<br /> v4l_streamon+0x2c/0x40<br /> __video_do_ioctl+0x34c/0x400<br /> video_usercopy+0x2d0/0x800<br /> video_ioctl2+0x20/0x60<br /> v4l2_ioctl+0x48/0x78

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdgpu: Add bounds checking to ib_{get,set}_value<br /> <br /> The uvd/vce/vcn code accesses the IB at predefined offsets without<br /> checking that the IB is large enough. Check the bounds here. The caller<br /> is responsible for making sure it can handle arbitrary return values.<br /> <br /> Also make the idx a uint32_t to prevent overflows causing the condition<br /> to fail.

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/xe/hdcp: Add NULL check for media_gt in intel_hdcp_gsc_check_status()<br /> <br /> When media GT is disabled via configfs, there is no allocation for<br /> media_gt, which is kept as NULL. In such scenario,<br /> intel_hdcp_gsc_check_status() results in a kernel pagefault error due to<br /> &amp;gt-&gt;uc.gsc being evaluated as an invalid memory address.<br /> <br /> Fix that by introducing a NULL check on media_gt and bailing out early<br /> if so.<br /> <br /> While at it, also drop the NULL check for gsc, since it can&amp;#39;t be NULL if<br /> media_gt is not NULL.<br /> <br /> v2:<br /> - Get address for gsc only after checking that gt is not NULL.<br /> (Shuicheng)<br /> - Drop the NULL check for gsc. (Shuicheng)<br /> v3:<br /> - Add "Fixes" and "Cc: " tags. (Matt)<br /> <br /> (cherry picked from commit bfaf87e84ca3ca3f6e275f9ae56da47a8b55ffd1)

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> vsock/virtio: fix accept queue count leak on transport mismatch<br /> <br /> virtio_transport_recv_listen() calls sk_acceptq_added() before<br /> vsock_assign_transport(). If vsock_assign_transport() fails or<br /> selects a different transport, the error path returns without<br /> calling sk_acceptq_removed(), permanently incrementing<br /> sk_ack_backlog.<br /> <br /> After approximately backlog+1 such failures, sk_acceptq_is_full()<br /> returns true, causing the listener to reject all new connections.<br /> <br /> Fix by moving sk_acceptq_added() to after the transport validation,<br /> matching the pattern used by vmci_transport and hyperv_transport.

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> cgroup: Defer css percpu_ref kill on rmdir until cgroup is depopulated<br /> <br /> A chain of commits going back to v7.0 reworked rmdir to satisfy the<br /> controller invariant that a subsystem&amp;#39;s -&gt;css_offline() must not run while<br /> tasks are still doing kernel-side work in the cgroup.<br /> <br /> [1] d245698d727a ("cgroup: Defer task cgroup unlink until after the task is done switching out")<br /> [2] a72f73c4dd9b ("cgroup: Don&amp;#39;t expose dead tasks in cgroup")<br /> [3] 1b164b876c36 ("cgroup: Wait for dying tasks to leave on rmdir")<br /> [4] 4c56a8ac6869 ("cgroup: Fix cgroup_drain_dying() testing the wrong condition")<br /> [5] 13e786b64bd3 ("cgroup: Increment nr_dying_subsys_* from rmdir context")<br /> <br /> [1] moved task cset unlink from do_exit() to finish_task_switch() so a<br /> task&amp;#39;s cset link drops only after the task has fully stopped scheduling.<br /> That made tasks past exit_signals() linger on cset-&gt;tasks until their final<br /> context switch, which led to a series of problems as what userspace expected<br /> to see after rmdir diverged from what the kernel needs to wait for. [2]-[5]<br /> tried to bridge that divergence: [2] filtered the exiting tasks from<br /> cgroup.procs; [3] had rmdir(2) sleep in TASK_UNINTERRUPTIBLE for them; [4]<br /> fixed the wait&amp;#39;s condition; [5] made nr_dying_subsys_* visible<br /> synchronously.<br /> <br /> The cgroup_drain_dying() wait in [3] turned out to be a dead end. When the<br /> rmdir caller is also the reaper of a zombie that pins a pidns teardown (e.g.<br /> host PID 1 systemd reaping orphan pids that were re-parented to it during<br /> the same teardown), rmdir blocks in TASK_UNINTERRUPTIBLE waiting for those<br /> pids to free, the pids can&amp;#39;t free because PID 1 is the reaper and it&amp;#39;s stuck<br /> in rmdir, and the system A-A deadlocks. No internal lock ordering breaks<br /> this; the wait itself is the bug.<br /> <br /> The css killing side that drove the original reorder, however, can be made<br /> cleanly asynchronous: -&gt;css_offline() is already async, run from<br /> css_killed_work_fn() driven by percpu_ref_kill_and_confirm(). The fix is to<br /> make that chain start only after all tasks have left the cgroup. rmdir&amp;#39;s<br /> user-visible side then returns as soon as cgroup.procs and friends are<br /> empty, while -&gt;css_offline() still runs only after the cgroup is fully<br /> drained.<br /> <br /> Verified by the original reproducer (pidns teardown + zombie reaper, runs<br /> under vng) which hangs vanilla and succeeds here, and by per-commit<br /> deterministic repros for [2], [3], [4], [5] with a boot parameter that<br /> widens the post-exit_signals() window so each state is reliably reachable.<br /> Some stress tests on top of that.<br /> <br /> cgroup_apply_control_disable() has the same shape of pre-existing race:<br /> when a controller is disabled via subtree_control, kill_css() ran<br /> synchronously while tasks past exit_signals() could still be linked to<br /> the cgroup&amp;#39;s csets, and -&gt;css_offline() could fire before they drained.<br /> This patch preserves the existing synchronous behavior at that call site<br /> (kill_css_sync() + kill_css_finish() back-to-back) and a follow-up patch<br /> will defer kill_css_finish() there using a per-css trigger.<br /> <br /> This seems like the right approach and I don&amp;#39;t see problems with it. The<br /> changes are somewhat invasive but not excessively so, so backporting to<br /> -stable should be okay. If something does turn out to be wrong, the fallback<br /> is to revert the entire chain ([1]-[5]) and rework in the development branch<br /> instead.<br /> <br /> v2: Pin cgrp across the deferred destroy work with explicit<br /> cgroup_get()/cgroup_put() around queue_work() and the work_fn. v1<br /> wasn&amp;#39;t actually broken (ordered cgroup_offline_wq + queue_work order<br /> in cgroup_task_dead() saved it) but the explicit ref removes the<br /> dependency on those non-obvious invariants. Also note the<br /> pre-existing cgroup_apply_control_disable() race in the description;<br /> a follow-up will defer kill_css_finish() there.