From e873c74132f0c5f1452816cd9bb26208f0bba1e1 Mon Sep 17 00:00:00 2001 From: Shivank Garg Date: Sat, 22 Aug 2026 19:22:05 +0000 Subject: dmaengine: fix use-after-free in dma_chan_put() and dma_release_channel() When dma_device_put() drops the last reference on chan->device->ref, dma_device_release() runs and may free the dma_device along with its channels. dma_chan_put() then still reads chan->device->owner via dma_chan_to_owner() for the trailing module_put(). KASAN catches it: slab-use-after-free in dma_chan_put+0x3e6/0x4c0 Read of size 8 by task insmod/6319 Freed by task 6319: kfree+0x225/0x470 dma_chan_put+0x395/0x4c0 dmaengine_put+0xf8/0x160 Cache the module owner in dma_chan_put() before the put so the trailing module_put() does not need chan->device. Fixes: 8ad342a86359 ("dmaengine: Add reference counting to dma_device struct") Suggested-by: Sashiko Link: https://sashiko.dev/#/patchset/20260518-dmaengine-kref-fix-v1-1-4d6125048fb7@amd.com Reviewed-by: Frank Li Reviewed-by: Logan Gunthorpe Signed-off-by: Shivank Garg Link: https://patch.msgid.link/20260822-dmaengine-kref-fix-v5-3-d4a4ee47d927@amd.com Signed-off-by: Vinod Koul --- io_uring/eventfd.c | 173 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 173 insertions(+) create mode 100644 io_uring/eventfd.c (limited to 'io_uring/eventfd.c') diff --git a/io_uring/eventfd.c b/io_uring/eventfd.c new file mode 100644 index 000000000..63fe6e5d7 --- /dev/null +++ b/io_uring/eventfd.c @@ -0,0 +1,173 @@ +// SPDX-License-Identifier: GPL-2.0 +#include +#include +#include +#include +#include +#include +#include +#include + +#include "io-wq.h" +#include "eventfd.h" + +struct io_ev_fd { + struct eventfd_ctx *cq_ev_fd; + unsigned int eventfd_async; + /* protected by ->completion_lock */ + unsigned last_cq_tail; + refcount_t refs; + atomic_t ops; + struct rcu_head rcu; +}; + +enum { + IO_EVENTFD_OP_SIGNAL_BIT, +}; + +static void io_eventfd_free(struct rcu_head *rcu) +{ + struct io_ev_fd *ev_fd = container_of(rcu, struct io_ev_fd, rcu); + + eventfd_ctx_put(ev_fd->cq_ev_fd); + kfree(ev_fd); +} + +static void io_eventfd_put(struct io_ev_fd *ev_fd) +{ + if (refcount_dec_and_test(&ev_fd->refs)) + call_rcu(&ev_fd->rcu, io_eventfd_free); +} + +static void io_eventfd_do_signal(struct rcu_head *rcu) +{ + struct io_ev_fd *ev_fd = container_of(rcu, struct io_ev_fd, rcu); + + atomic_andnot(BIT(IO_EVENTFD_OP_SIGNAL_BIT), &ev_fd->ops); + eventfd_signal_mask(ev_fd->cq_ev_fd, EPOLL_URING_WAKE); + io_eventfd_put(ev_fd); +} + +/* + * Returns true if the caller should put the ev_fd reference, false if not. + */ +static bool __io_eventfd_signal(struct io_ev_fd *ev_fd, bool defer) +{ + if (!defer && eventfd_signal_allowed()) { + eventfd_signal_mask(ev_fd->cq_ev_fd, EPOLL_URING_WAKE); + return true; + } + if (!atomic_fetch_or(BIT(IO_EVENTFD_OP_SIGNAL_BIT), &ev_fd->ops)) { + call_rcu_hurry(&ev_fd->rcu, io_eventfd_do_signal); + return false; + } + return true; +} + +/* + * Trigger if eventfd_async isn't set, or if it's set and the caller is + * an async worker. + */ +static bool io_eventfd_trigger(struct io_ev_fd *ev_fd) +{ + return !ev_fd->eventfd_async || io_wq_current_is_worker(); +} + +void io_eventfd_signal(struct io_ring_ctx *ctx, bool cqe_event, bool defer) +{ + bool skip = false; + struct io_ev_fd *ev_fd; + struct io_rings *rings; + + guard(rcu)(); + + rings = rcu_dereference(ctx->rings_rcu); + if (!rings) + return; + if (READ_ONCE(rings->cq_flags) & IORING_CQ_EVENTFD_DISABLED) + return; + ev_fd = rcu_dereference(ctx->io_ev_fd); + /* + * Check again if ev_fd exists in case an io_eventfd_unregister call + * completed between the NULL check of ctx->io_ev_fd at the start of + * the function and rcu_read_lock. + */ + if (!ev_fd) + return; + if (!io_eventfd_trigger(ev_fd) || !refcount_inc_not_zero(&ev_fd->refs)) + return; + + if (cqe_event) { + /* + * Eventfd should only get triggered when at least one event + * has been posted. Some applications rely on the eventfd + * notification count only changing IFF a new CQE has been + * added to the CQ ring. There's no dependency on 1:1 + * relationship between how many times this function is called + * (and hence the eventfd count) and number of CQEs posted to + * the CQ ring. + */ + spin_lock(&ctx->completion_lock); + skip = ctx->cached_cq_tail == ev_fd->last_cq_tail; + ev_fd->last_cq_tail = ctx->cached_cq_tail; + spin_unlock(&ctx->completion_lock); + } + + if (skip || __io_eventfd_signal(ev_fd, defer)) + io_eventfd_put(ev_fd); +} + +int io_eventfd_register(struct io_ring_ctx *ctx, void __user *arg, + unsigned int eventfd_async) +{ + struct io_ev_fd *ev_fd; + __s32 __user *fds = arg; + int fd; + + ev_fd = rcu_dereference_protected(ctx->io_ev_fd, + lockdep_is_held(&ctx->uring_lock)); + if (ev_fd) + return -EBUSY; + + if (copy_from_user(&fd, fds, sizeof(*fds))) + return -EFAULT; + + ev_fd = kmalloc_obj(*ev_fd); + if (!ev_fd) + return -ENOMEM; + + ev_fd->cq_ev_fd = eventfd_ctx_fdget(fd); + if (IS_ERR(ev_fd->cq_ev_fd)) { + int ret = PTR_ERR(ev_fd->cq_ev_fd); + + kfree(ev_fd); + return ret; + } + + spin_lock(&ctx->completion_lock); + ev_fd->last_cq_tail = ctx->cached_cq_tail; + spin_unlock(&ctx->completion_lock); + + ev_fd->eventfd_async = eventfd_async; + ctx->int_flags |= IO_RING_F_HAS_EVFD; + refcount_set(&ev_fd->refs, 1); + atomic_set(&ev_fd->ops, 0); + rcu_assign_pointer(ctx->io_ev_fd, ev_fd); + return 0; +} + +int io_eventfd_unregister(struct io_ring_ctx *ctx) +{ + struct io_ev_fd *ev_fd; + + ev_fd = rcu_dereference_protected(ctx->io_ev_fd, + lockdep_is_held(&ctx->uring_lock)); + if (ev_fd) { + ctx->int_flags &= ~IO_RING_F_HAS_EVFD; + rcu_assign_pointer(ctx->io_ev_fd, NULL); + io_eventfd_put(ev_fd); + return 0; + } + + return -ENXIO; +} -- cgit v1.3.1