brd: protect page with rcu

Currently, after fetching the page by xa_load() in IO path, there is no
protection and page can be freed concurrently by discard:

cpu0
brd_submit_bio
 brd_do_bvec
  page = brd_lookup_page
                          cpu1
                          brd_submit_bio
                           brd_do_discard
                            page = __xa_erase()
                            __free_page()
  // page UAF

Fix the problem by protecting page with rcu.

Meanwhile, if page is already freed, also prevent BUG_ON() by skipping
the write, and user will get zero data later if there is no page.

Fixes: 9ead7efc6f3f ("brd: implement discard support")
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20250506061756.2970934-2-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/drivers/block/brd.c b/drivers/block/brd.c
index fa1290992a7fc0b9dd2c8949f4d739b8b857fcad..fc793d48a9c633910871bb390ebf2fd393643038 100644 (file)
--- a/drivers/block/brd.c
+++ b/drivers/block/brd.c
@@ -132,12 +132,18 @@ static bool brd_rw_bvec(struct brd_device *brd, struct bio *bio)
                }
        }
 
+       rcu_read_lock();
        page = brd_lookup_page(brd, sector);
 
        kaddr = bvec_kmap_local(&bv);
        if (op_is_write(opf)) {
-               BUG_ON(!page);
-               memcpy_to_page(page, offset, kaddr, bv.bv_len);
+               /*
+                * Page can be removed by concurrent discard, it's fine to skip
+                * the write and user will read zero data if page does not
+                * exist.
+                */
+               if (page)
+                       memcpy_to_page(page, offset, kaddr, bv.bv_len);
        } else {
                if (page)
                        memcpy_from_page(kaddr, page, offset, bv.bv_len);
@@ -145,11 +151,19 @@ static bool brd_rw_bvec(struct brd_device *brd, struct bio *bio)
                        memset(kaddr, 0, bv.bv_len);
        }
        kunmap_local(kaddr);
+       rcu_read_unlock();
 
        bio_advance_iter_single(bio, &bio->bi_iter, bv.bv_len);
        return true;
 }
 
+static void brd_free_one_page(struct rcu_head *head)
+{
+       struct page *page = container_of(head, struct page, rcu_head);
+
+       __free_page(page);
+}
+
 static void brd_do_discard(struct brd_device *brd, sector_t sector, u32 size)
 {
        sector_t aligned_sector = (sector + PAGE_SECTORS) & ~PAGE_SECTORS;
@@ -160,7 +174,7 @@ static void brd_do_discard(struct brd_device *brd, sector_t sector, u32 size)
        while (size >= PAGE_SIZE && aligned_sector < rd_size * 2) {
                page = __xa_erase(&brd->brd_pages, aligned_sector >> PAGE_SECTORS_SHIFT);
                if (page) {
-                       __free_page(page);
+                       call_rcu(&page->rcu_head, brd_free_one_page);
                        brd->brd_nr_pages--;
                }
                aligned_sector += PAGE_SECTORS;