Re: [PATCH] btrfs: fix allocation of bitmap pages.

From: Christophe Leroy
Date: Mon Aug 19 2019 - 16:32:41 EST

Le 19/08/2019 Ã 19:46, David Sterba a ÃcritÂ:
On Sat, Aug 17, 2019 at 07:44:39AM +0000, Christophe Leroy wrote:
Various notifications of type "BUG kmalloc-4096 () : Redzone
overwritten" have been observed recently in various parts of
the kernel. After some time, it has been made a relation with
the use of BTRFS filesystem.

[ 22.809700] BUG kmalloc-4096 (Tainted: G W ): Redzone overwritten
[ 22.809971] -----------------------------------------------------------------------------

[ 22.810286] INFO: 0xbe1a5921-0xfbfc06cd. First byte 0x0 instead of 0xcc
[ 22.810866] INFO: Allocated in __load_free_space_cache+0x588/0x780 [btrfs] age=22 cpu=0 pid=224
[ 22.811193] __slab_alloc.constprop.26+0x44/0x70
[ 22.811345] kmem_cache_alloc_trace+0xf0/0x2ec
[ 22.811588] __load_free_space_cache+0x588/0x780 [btrfs]
[ 22.811848] load_free_space_cache+0xf4/0x1b0 [btrfs]
[ 22.812090] cache_block_group+0x1d0/0x3d0 [btrfs]
[ 22.812321] find_free_extent+0x680/0x12a4 [btrfs]
[ 22.812549] btrfs_reserve_extent+0xec/0x220 [btrfs]
[ 22.812785] btrfs_alloc_tree_block+0x178/0x5f4 [btrfs]
[ 22.813032] __btrfs_cow_block+0x150/0x5d4 [btrfs]
[ 22.813262] btrfs_cow_block+0x194/0x298 [btrfs]
[ 22.813484] commit_cowonly_roots+0x44/0x294 [btrfs]
[ 22.813718] btrfs_commit_transaction+0x63c/0xc0c [btrfs]
[ 22.813973] close_ctree+0xf8/0x2a4 [btrfs]
[ 22.814107] generic_shutdown_super+0x80/0x110
[ 22.814250] kill_anon_super+0x18/0x30
[ 22.814437] btrfs_kill_super+0x18/0x90 [btrfs]
[ 22.814590] INFO: Freed in proc_cgroup_show+0xc0/0x248 age=41 cpu=0 pid=83
[ 22.814841] proc_cgroup_show+0xc0/0x248
[ 22.814967] proc_single_show+0x54/0x98
[ 22.815086] seq_read+0x278/0x45c
[ 22.815190] __vfs_read+0x28/0x17c
[ 22.815289] vfs_read+0xa8/0x14c
[ 22.815381] ksys_read+0x50/0x94
[ 22.815475] ret_from_syscall+0x0/0x38

Commit 69d2480456d1 ("btrfs: use copy_page for copying pages instead
of memcpy") changed the way bitmap blocks are copied. But allthough
bitmaps have the size of a page, they were allocated with kzalloc().

Most of the time, kzalloc() allocates aligned blocks of memory, so
copy_page() can be used. But when some debug options like SLAB_DEBUG
are activated, kzalloc() may return unaligned pointer.

On powerpc, memcpy(), copy_page() and other copying functions use
'dcbz' instruction which provides an entire zeroed cacheline to avoid
memory read when the intention is to overwrite a full line. Functions
like memcpy() are writen to care about partial cachelines at the start
and end of the destination, but copy_page() assumes it gets pages.

This assumption is not documented nor any pitfalls mentioned in
include/asm-generic/page.h that provides the generic implementation. I
as an API user cannot check each arch implementation for additional
constraints or I would expect that it deals with the boundary cases the
same way as arch-specific memcpy implementations.

For me, copy_page() is there to ... copy pages. Not to copy any piece of RAM having the size of a page.

But it happened to others. See commit

Another thing that is lost is the slub debugging support for all
architectures, because get_zeroed_pages lacking the red zones and sanity

I find working with raw pages in this code a bit inconsistent with the
rest of btrfs code, but that's rather minor compared to the above.

What about using kmem_cache instead ? I see kmem_cache is already widely used in BTRFS, so using it also for block of memory of size PAGE_SIZE should be ok ?

AFAICS, kmem_cache has the red zones and sanity checks.

Summing it up, I think that the proper fix should go to copy_page
implementation on architectures that require it or make it clear what
are the copy_page constraints.

I guess anybody using copy_page() to copy something else than a page is on his/her own.

But following that (bad) experience, I propose a patch to at least detect it early, see