From: Pavel Begunkov <asml.silence@xxxxxxxxx> commit 00927931cb630bbf8edb6d7f4dadb25139fc5e16 upstream. Only with the big sqe feature they take 128 bytes per entry, but we unconditionally advance by 128B. Fix it by using sq_shift. Fixes: 3b8fdd1dc35e3 ("io_uring/fdinfo: fix sqe dumping for IORING_SETUP_SQE128") Reported-and-tested-by: syzbot+e5198737e8a2d23d958c@xxxxxxxxxxxxxxxxxxxxxxxxx Signed-off-by: Pavel Begunkov <asml.silence@xxxxxxxxx> Link: https://lore.kernel.org/r/8b41287cb75d5efb8fcb5cccde845ddbbadd8372.1665449983.git.asml.silence@xxxxxxxxx Signed-off-by: Jens Axboe <axboe@xxxxxxxxx> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- io_uring/fdinfo.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- a/io_uring/fdinfo.c +++ b/io_uring/fdinfo.c @@ -95,7 +95,7 @@ static __cold void __io_uring_show_fdinf sq_idx = READ_ONCE(ctx->sq_array[entry & sq_mask]); if (sq_idx > sq_mask) continue; - sqe = &ctx->sq_sqes[sq_idx << 1]; + sqe = &ctx->sq_sqes[sq_idx << sq_shift]; seq_printf(m, "%5u: opcode:%s, fd:%d, flags:%x, off:%llu, " "addr:0x%llx, rw_flags:0x%x, buf_index:%d " "user_data:%llu",