linux-kernel - [PATCH 5.9 05/74] io_uring: stash ctx task reference for SQPOLL

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20201031113500.297738073@linuxfoundation.org>
Date:   Sat, 31 Oct 2020 12:35:47 +0100
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     linux-kernel@...r.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        stable@...r.kernel.org, Pavel Begunkov <asml.silence@...il.com>,
        Jens Axboe <axboe@...nel.dk>
Subject: [PATCH 5.9 05/74] io_uring: stash ctx task reference for SQPOLL

From: Jens Axboe <axboe@...nel.dk>

commit 2aede0e417db846793c276c7a1bbf7262c8349b0 upstream.

We can grab a reference to the task instead of stashing away the task
files_struct. This is doable without creating a circular reference
between the ring fd and the task itself.

Reviewed-by: Pavel Begunkov <asml.silence@...il.com>
Signed-off-by: Jens Axboe <axboe@...nel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
---
 fs/io_uring.c |   47 ++++++++++++++++++++++++++++++++++-------------
 1 file changed, 34 insertions(+), 13 deletions(-)

--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -265,7 +265,16 @@ struct io_ring_ctx {
 	/* IO offload */
 	struct io_wq		*io_wq;
 	struct task_struct	*sqo_thread;	/* if using sq thread polling */
-	struct mm_struct	*sqo_mm;
+
+	/*
+	 * For SQPOLL usage - we hold a reference to the parent task, so we
+	 * have access to the ->files
+	 */
+	struct task_struct	*sqo_task;
+
+	/* Only used for accounting purposes */
+	struct mm_struct	*mm_account;
+
 	wait_queue_head_t	sqo_wait;
 
 	/*
@@ -969,9 +978,10 @@ static int __io_sq_thread_acquire_mm(str
 {
 	if (!current->mm) {
 		if (unlikely(!(ctx->flags & IORING_SETUP_SQPOLL) ||
-			     !mmget_not_zero(ctx->sqo_mm)))
+			     !ctx->sqo_task->mm ||
+			     !mmget_not_zero(ctx->sqo_task->mm)))
 			return -EFAULT;
-		kthread_use_mm(ctx->sqo_mm);
+		kthread_use_mm(ctx->sqo_task->mm);
 	}
 
 	return 0;
@@ -7591,11 +7601,11 @@ static void io_unaccount_mem(struct io_r
 	if (ctx->limit_mem)
 		__io_unaccount_mem(ctx->user, nr_pages);
 
-	if (ctx->sqo_mm) {
+	if (ctx->mm_account) {
 		if (acct == ACCT_LOCKED)
-			ctx->sqo_mm->locked_vm -= nr_pages;
+			ctx->mm_account->locked_vm -= nr_pages;
 		else if (acct == ACCT_PINNED)
-			atomic64_sub(nr_pages, &ctx->sqo_mm->pinned_vm);
+			atomic64_sub(nr_pages, &ctx->mm_account->pinned_vm);
 	}
 }
 
@@ -7610,11 +7620,11 @@ static int io_account_mem(struct io_ring
 			return ret;
 	}
 
-	if (ctx->sqo_mm) {
+	if (ctx->mm_account) {
 		if (acct == ACCT_LOCKED)
-			ctx->sqo_mm->locked_vm += nr_pages;
+			ctx->mm_account->locked_vm += nr_pages;
 		else if (acct == ACCT_PINNED)
-			atomic64_add(nr_pages, &ctx->sqo_mm->pinned_vm);
+			atomic64_add(nr_pages, &ctx->mm_account->pinned_vm);
 	}
 
 	return 0;
@@ -7918,9 +7928,12 @@ static void io_ring_ctx_free(struct io_r
 {
 	io_finish_async(ctx);
 	io_sqe_buffer_unregister(ctx);
-	if (ctx->sqo_mm) {
-		mmdrop(ctx->sqo_mm);
-		ctx->sqo_mm = NULL;
+
+	if (ctx->sqo_task) {
+		put_task_struct(ctx->sqo_task);
+		ctx->sqo_task = NULL;
+		mmdrop(ctx->mm_account);
+		ctx->mm_account = NULL;
 	}
 
 	io_sqe_files_unregister(ctx);
@@ -8665,8 +8678,16 @@ static int io_uring_create(unsigned entr
 	ctx->user = user;
 	ctx->creds = get_current_cred();
 
+	ctx->sqo_task = get_task_struct(current);
+
+	/*
+	 * This is just grabbed for accounting purposes. When a process exits,
+	 * the mm is exited and dropped before the files, hence we need to hang
+	 * on to this mm purely for the purposes of being able to unaccount
+	 * memory (locked/pinned vm). It's not used for anything else.
+	 */
 	mmgrab(current->mm);
-	ctx->sqo_mm = current->mm;
+	ctx->mm_account = current->mm;
 
 	/*
 	 * Account memory _before_ installing the file descriptor. Once