All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] io_uring/kbuf: don't allow registered buffer rings on highmem pages
@ 2023-10-03  0:26 Jens Axboe
  0 siblings, 0 replies; only message in thread
From: Jens Axboe @ 2023-10-03  0:26 UTC (permalink / raw)
  To: io-uring

syzbot reports that registering a mapped buffer ring on arm32 can
trigger an OOPS. Registered buffer rings have two modes, one of them
is the application passing in the memory that the buffer ring should
reside in. Once those pages are mapped, we use page_address() to get
a virtual address. This will obviously fail on highmem pages, which
aren't mapped.

Add a check if we have any highmem pages after mapping, and fail the
attempt to register a provided buffer ring if we do. This will return
the same error as kernels do that don't support provided buffer rings
to begin with.

Link: https://lore.kernel.org/io-uring/000000000000af635c0606bcb889@google.com/
Fixes: c56e022c0a27 ("io_uring: add support for user mapped provided buffer ring")
Cc: stable@vger.kernel.org
Reported-by: syzbot+2113e61b8848fa7951d8@syzkaller.appspotmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

---

diff --git a/io_uring/kbuf.c b/io_uring/kbuf.c
index 556f4df25b0f..c5a2d4b38c8a 100644
--- a/io_uring/kbuf.c
+++ b/io_uring/kbuf.c
@@ -477,7 +477,7 @@ static int io_pin_pbuf_ring(struct io_uring_buf_reg *reg,
 {
 	struct io_uring_buf_ring *br;
 	struct page **pages;
-	int nr_pages;
+	int i, nr_pages;
 
 	pages = io_pin_pages(reg->ring_addr,
 			     flex_array_size(br, bufs, reg->ring_entries),
@@ -485,6 +485,17 @@ static int io_pin_pbuf_ring(struct io_uring_buf_reg *reg,
 	if (IS_ERR(pages))
 		return PTR_ERR(pages);
 
+	/*
+	 * Apparently some 32-bit boxes (ARM) will return highmem pages,
+	 * which then need to be mapped. We could support that, but it'd
+	 * complicate the code and slowdown the common cases quite a bit.
+	 * So just error out, returning -EINVAL just like we did on kernels
+	 * that didn't support mapped buffer rings.
+	 */
+	for (i = 0; i < nr_pages; i++)
+		if (PageHighMem(pages[i]))
+			goto error_unpin;
+
 	br = page_address(pages[0]);
 #ifdef SHM_COLOUR
 	/*
@@ -496,13 +507,8 @@ static int io_pin_pbuf_ring(struct io_uring_buf_reg *reg,
 	 * should use IOU_PBUF_RING_MMAP instead, and liburing will handle
 	 * this transparently.
 	 */
-	if ((reg->ring_addr | (unsigned long) br) & (SHM_COLOUR - 1)) {
-		int i;
-
-		for (i = 0; i < nr_pages; i++)
-			unpin_user_page(pages[i]);
-		return -EINVAL;
-	}
+	if ((reg->ring_addr | (unsigned long) br) & (SHM_COLOUR - 1))
+		goto error_unpin;
 #endif
 	bl->buf_pages = pages;
 	bl->buf_nr_pages = nr_pages;
@@ -510,6 +516,10 @@ static int io_pin_pbuf_ring(struct io_uring_buf_reg *reg,
 	bl->is_mapped = 1;
 	bl->is_mmap = 0;
 	return 0;
+error_unpin:
+	for (i = 0; i < nr_pages; i++)
+		unpin_user_page(pages[i]);
+	return -EINVAL;
 }
 
 static int io_alloc_pbuf_ring(struct io_uring_buf_reg *reg,

-- 
Jens Axboe


^ permalink raw reply related	[flat|nested] only message in thread

only message in thread, other threads:[~2023-10-03  0:26 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-10-03  0:26 [PATCH] io_uring/kbuf: don't allow registered buffer rings on highmem pages Jens Axboe

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.