From 8e47a457321ca1a74ad194ab5dcbca764bc70731 Mon Sep 17 00:00:00 2001 From: Piotr Jaroszynski Date: Sun, 27 Jan 2019 08:46:45 -0800 Subject: iomap: get/put the page in iomap_page_create/release() migrate_page_move_mapping() expects pages with private data set to have a page_count elevated by 1. This is what used to happen for xfs through the buffer_heads code before the switch to iomap in commit 82cb14175e7d ("xfs: add support for sub-pagesize writeback without buffer_heads"). Not having the count elevated causes move_pages() to fail on memory mapped files coming from xfs. Make iomap compatible with the migrate_page_move_mapping() assumption by elevating the page count as part of iomap_page_create() and lowering it in iomap_page_release(). It causes the move_pages() syscall to misbehave on memory mapped files from xfs. It does not not move any pages, which I suppose is "just" a perf issue, but it also ends up returning a positive number which is out of spec for the syscall. Talking to Michal Hocko, it sounds like returning positive numbers might be a necessary update to move_pages() anyway though. Fixes: 82cb14175e7d ("xfs: add support for sub-pagesize writeback without buffer_heads") Signed-off-by: Piotr Jaroszynski [hch: actually get/put the page iomap_migrate_page() to make it work properly] Signed-off-by: Christoph Hellwig Reviewed-by: Dave Chinner Reviewed-by: Darrick J. Wong Signed-off-by: Darrick J. Wong --- fs/iomap.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/fs/iomap.c b/fs/iomap.c index a3088fae567b..cb184ff68680 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -116,6 +116,12 @@ iomap_page_create(struct inode *inode, struct page *page) atomic_set(&iop->read_count, 0); atomic_set(&iop->write_count, 0); bitmap_zero(iop->uptodate, PAGE_SIZE / SECTOR_SIZE); + + /* + * migrate_page_move_mapping() assumes that pages with private data have + * their count elevated by 1. + */ + get_page(page); set_page_private(page, (unsigned long)iop); SetPagePrivate(page); return iop; @@ -132,6 +138,7 @@ iomap_page_release(struct page *page) WARN_ON_ONCE(atomic_read(&iop->write_count)); ClearPagePrivate(page); set_page_private(page, 0); + put_page(page); kfree(iop); } @@ -569,8 +576,10 @@ iomap_migrate_page(struct address_space *mapping, struct page *newpage, if (page_has_private(page)) { ClearPagePrivate(page); + get_page(newpage); set_page_private(newpage, page_private(page)); set_page_private(page, 0); + put_page(page); SetPagePrivate(newpage); } -- cgit v1.2.3 From 4ea899ead2786a30aaa8181fefa81a3df4ad28f6 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 17 Jan 2019 08:58:58 -0800 Subject: iomap: fix a use after free in iomap_dio_rw Introduce a local wait_for_completion variable to avoid an access to the potentially freed dio struture after dropping the last reference count. Also use the chance to document the completion behavior to make the refcounting clear to the reader of the code. Fixes: ff6a9292e6 ("iomap: implement direct I/O") Reported-by: Chandan Rajendra Reported-by: Darrick J. Wong Signed-off-by: Christoph Hellwig Tested-by: Chandan Rajendra Tested-by: Darrick J. Wong Reviewed-by: Dave Chinner Reviewed-by: Darrick J. Wong Signed-off-by: Darrick J. Wong --- fs/iomap.c | 28 +++++++++++++++++++++------- 1 file changed, 21 insertions(+), 7 deletions(-) diff --git a/fs/iomap.c b/fs/iomap.c index cb184ff68680..897c60215dd1 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -1813,6 +1813,7 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, loff_t pos = iocb->ki_pos, start = pos; loff_t end = iocb->ki_pos + count - 1, ret = 0; unsigned int flags = IOMAP_DIRECT; + bool wait_for_completion = is_sync_kiocb(iocb); struct blk_plug plug; struct iomap_dio *dio; @@ -1832,7 +1833,6 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, dio->end_io = end_io; dio->error = 0; dio->flags = 0; - dio->wait_for_completion = is_sync_kiocb(iocb); dio->submit.iter = iter; dio->submit.waiter = current; @@ -1887,7 +1887,7 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, dio_warn_stale_pagecache(iocb->ki_filp); ret = 0; - if (iov_iter_rw(iter) == WRITE && !dio->wait_for_completion && + if (iov_iter_rw(iter) == WRITE && !wait_for_completion && !inode->i_sb->s_dio_done_wq) { ret = sb_init_dio_done_wq(inode->i_sb); if (ret < 0) @@ -1903,7 +1903,7 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, if (ret <= 0) { /* magic error code to fall back to buffered I/O */ if (ret == -ENOTBLK) { - dio->wait_for_completion = true; + wait_for_completion = true; ret = 0; } break; @@ -1925,8 +1925,24 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, if (dio->flags & IOMAP_DIO_WRITE_FUA) dio->flags &= ~IOMAP_DIO_NEED_SYNC; + /* + * We are about to drop our additional submission reference, which + * might be the last reference to the dio. There are three three + * different ways we can progress here: + * + * (a) If this is the last reference we will always complete and free + * the dio ourselves. + * (b) If this is not the last reference, and we serve an asynchronous + * iocb, we must never touch the dio after the decrement, the + * I/O completion handler will complete and free it. + * (c) If this is not the last reference, but we serve a synchronous + * iocb, the I/O completion handler will wake us up on the drop + * of the final reference, and we will complete and free it here + * after we got woken by the I/O completion handler. + */ + dio->wait_for_completion = wait_for_completion; if (!atomic_dec_and_test(&dio->ref)) { - if (!dio->wait_for_completion) + if (!wait_for_completion) return -EIOCBQUEUED; for (;;) { @@ -1943,9 +1959,7 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, __set_current_state(TASK_RUNNING); } - ret = iomap_dio_complete(dio); - - return ret; + return iomap_dio_complete(dio); out_free_dio: kfree(dio); -- cgit v1.2.3