From 542b8f05a1ac7d0569453540108fc0fdab1fad18 Mon Sep 17 00:00:00 2001 From: Hagar Hemdan Date: Wed, 31 Jul 2024 14:15:46 +0800 Subject: [PATCH] io_uring: fix possible deadlock in io_register_iowq_max_workers() mainline inclusion from mainline-v6.10-rc3 commit 73254a297c2dd094abec7c9efee32455ae875bdf category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/IAGEK8 CVE: CVE-2024-41080 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=73254a297c2dd094abec7c9efee32455ae875bdf -------------------------------- The io_register_iowq_max_workers() function calls io_put_sq_data(), which acquires the sqd->lock without releasing the uring_lock. Similar to the commit 009ad9f0c6ee ("io_uring: drop ctx->uring_lock before acquiring sqd->lock"), this can lead to a potential deadlock situation. To resolve this issue, the uring_lock is released before calling io_put_sq_data(), and then it is re-acquired after the function call. This change ensures that the locks are acquired in the correct order, preventing the possibility of a deadlock. Suggested-by: Maximilian Heyne Signed-off-by: Hagar Hemdan Link: https://lore.kernel.org/r/20240604130527.3597-1-hagarhem@amazon.com Signed-off-by: Jens Axboe Conflicts: io_uring/io_uring.c io_uring/register.c [Commit c43203154d8ac moves related functions from io_uring/io_uring.c to io_uring/register.c.] Signed-off-by: Wang Zhaolong --- io_uring/io_uring.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c index c284e9865826..2f5dbbc36b3b 100644 --- a/io_uring/io_uring.c +++ b/io_uring/io_uring.c @@ -10637,8 +10637,10 @@ static int io_register_iowq_max_workers(struct io_ring_ctx *ctx, } if (sqd) { + mutex_unlock(&ctx->uring_lock); mutex_unlock(&sqd->lock); io_put_sq_data(sqd); + mutex_lock(&ctx->uring_lock); } if (copy_to_user(arg, new_count, sizeof(new_count))) @@ -10663,8 +10665,10 @@ static int io_register_iowq_max_workers(struct io_ring_ctx *ctx, return 0; err: if (sqd) { + mutex_unlock(&ctx->uring_lock); mutex_unlock(&sqd->lock); io_put_sq_data(sqd); + mutex_lock(&ctx->uring_lock); } return ret; } -- Gitee