lkml.org 
[lkml]   [2018]   [Nov]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: UBSAN: Undefined behaviour in mm/page_alloc.c
On Thu 08-11-18 23:09:23, Kyungtae Kim wrote:
> We report a bug in v4.19-rc2 (4.20-rc1 as well, I guess):
>
> kernel config: https://kt0755.github.io/etc/config_v2-4.19
> repro: https://kt0755.github.io/etc/repro.c4074.c
>
> In the middle of page request, this arose because order is too large to handle
> (mm/page_alloc.c:3119). It actually comes from that order is
> controllable by user input
> via raw_cmd_ioctl without its sanity check, thereby causing memory problem.
> To stop it, we can use like MAX_ORDER for bounds check before using it.

Yes, we do only check the max order in the slow path. We have already
discussed something similar with Konstantin [1][2]. Basically kvmalloc
for a large size might get to the page allocator with an out of bound
order and warn during direct reclaim.

I am wondering whether really want to check for the order in the fast
path instead. I have hard time to imagine this could cause a measurable
impact.

The full patch is below

[1] http://lkml.kernel.org/r/154109387197.925352.10499549042420271600.stgit@buzz
[2] http://lkml.kernel.org/r/154106356066.887821.4649178319705436373.stgit@buzz

>
> =========================================
> UBSAN: Undefined behaviour in mm/page_alloc.c:3117:19
> shift exponent 51 is too large for 32-bit type 'int'
> CPU: 0 PID: 6520 Comm: syz-executor1 Not tainted 4.19.0-rc2 #1
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
> Call Trace:
> __dump_stack lib/dump_stack.c:77 [inline]
> dump_stack+0xd2/0x148 lib/dump_stack.c:113
> ubsan_epilogue+0x12/0x94 lib/ubsan.c:159
> __ubsan_handle_shift_out_of_bounds+0x2b6/0x30b lib/ubsan.c:425
> __zone_watermark_ok+0x2c7/0x400 mm/page_alloc.c:3117
> zone_watermark_fast mm/page_alloc.c:3216 [inline]
> get_page_from_freelist+0xc49/0x44c0 mm/page_alloc.c:3300
> __alloc_pages_nodemask+0x21e/0x640 mm/page_alloc.c:4370
> alloc_pages_current+0xcc/0x210 mm/mempolicy.c:2093
> alloc_pages include/linux/gfp.h:509 [inline]
> __get_free_pages+0x12/0x60 mm/page_alloc.c:4414
> dma_mem_alloc+0x36/0x50 arch/x86/include/asm/floppy.h:156
> raw_cmd_copyin drivers/block/floppy.c:3159 [inline]
> raw_cmd_ioctl drivers/block/floppy.c:3206 [inline]
> fd_locked_ioctl+0xa00/0x2c10 drivers/block/floppy.c:3544
> fd_ioctl+0x40/0x60 drivers/block/floppy.c:3571
> __blkdev_driver_ioctl block/ioctl.c:303 [inline]
> blkdev_ioctl+0xb3c/0x1a30 block/ioctl.c:601
> block_ioctl+0x105/0x150 fs/block_dev.c:1883
> vfs_ioctl fs/ioctl.c:46 [inline]
> do_vfs_ioctl+0x1c0/0x1150 fs/ioctl.c:687
> ksys_ioctl+0x9e/0xb0 fs/ioctl.c:702
> __do_sys_ioctl fs/ioctl.c:709 [inline]
> __se_sys_ioctl fs/ioctl.c:707 [inline]
> __x64_sys_ioctl+0x7e/0xc0 fs/ioctl.c:707
> do_syscall_64+0xc4/0x510 arch/x86/entry/common.c:290
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
> RIP: 0033:0x4497b9
> Code: e8 8c 9f 02 00 48 83 c4 18 c3 0f 1f 80 00 00 00 00 48 89 f8 48
> 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d
> 01 f0 ff ff 0f 83 9b 6b fc ff c3 66 2e 0f 1f 84 00 00 00 00
> RSP: 002b:00007fb5ef0e2c68 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
> RAX: ffffffffffffffda RBX: 00007fb5ef0e36cc RCX: 00000000004497b9
> RDX: 0000000020000040 RSI: 0000000000000258 RDI: 0000000000000014
> RBP: 000000000071bea0 R08: 0000000000000000 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000246 R12: 00000000ffffffff
> R13: 0000000000005490 R14: 00000000006ed530 R15: 00007fb5ef0e3700
> =========================================================
>
>
> Thanks,
> Kyungtae Kim


From 7110220512be16054f2c8ee16bdd076c77c2456c Mon Sep 17 00:00:00 2001
From: Michal Hocko <mhocko@suse.com>
Date: Fri, 9 Nov 2018 09:35:29 +0100
Subject: [PATCH] mm, page_alloc: check for max order in hot path

Konstantin has noticed that kvmalloc might trigger the following warning
[Thu Nov 1 08:43:56 2018] WARNING: CPU: 0 PID: 6676 at mm/vmstat.c:986 __fragmentation_index+0x54/0x60
[...]
[Thu Nov 1 08:43:56 2018] Call Trace:
[Thu Nov 1 08:43:56 2018] fragmentation_index+0x76/0x90
[Thu Nov 1 08:43:56 2018] compaction_suitable+0x4f/0xf0
[Thu Nov 1 08:43:56 2018] shrink_node+0x295/0x310
[Thu Nov 1 08:43:56 2018] node_reclaim+0x205/0x250
[Thu Nov 1 08:43:56 2018] get_page_from_freelist+0x649/0xad0
[Thu Nov 1 08:43:56 2018] ? get_page_from_freelist+0x2d4/0xad0
[Thu Nov 1 08:43:56 2018] ? release_sock+0x19/0x90
[Thu Nov 1 08:43:56 2018] ? do_ipv6_setsockopt.isra.5+0x10da/0x1290
[Thu Nov 1 08:43:56 2018] __alloc_pages_nodemask+0x12a/0x2a0
[Thu Nov 1 08:43:56 2018] kmalloc_large_node+0x47/0x90
[Thu Nov 1 08:43:56 2018] __kmalloc_node+0x22b/0x2e0
[Thu Nov 1 08:43:56 2018] kvmalloc_node+0x3e/0x70
[Thu Nov 1 08:43:56 2018] xt_alloc_table_info+0x3a/0x80 [x_tables]
[Thu Nov 1 08:43:56 2018] do_ip6t_set_ctl+0xcd/0x1c0 [ip6_tables]
[Thu Nov 1 08:43:56 2018] nf_setsockopt+0x44/0x60
[Thu Nov 1 08:43:56 2018] SyS_setsockopt+0x6f/0xc0
[Thu Nov 1 08:43:56 2018] do_syscall_64+0x67/0x120
[Thu Nov 1 08:43:56 2018] entry_SYSCALL_64_after_hwframe+0x3d/0xa2

the problem is that we only check for an out of bound order in the slow
path and the node reclaim might happen from the fast path already. This
is fixable by making sure that kvmalloc doesn't ever use kmalloc for
requests that are larger than KMALLOC_MAX_SIZE but this also shows that
the code is rather fragile. A recent UBSAN report just underlines that
by the following report

UBSAN: Undefined behaviour in mm/page_alloc.c:3117:19
shift exponent 51 is too large for 32-bit type 'int'
CPU: 0 PID: 6520 Comm: syz-executor1 Not tainted 4.19.0-rc2 #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0xd2/0x148 lib/dump_stack.c:113
ubsan_epilogue+0x12/0x94 lib/ubsan.c:159
__ubsan_handle_shift_out_of_bounds+0x2b6/0x30b lib/ubsan.c:425
__zone_watermark_ok+0x2c7/0x400 mm/page_alloc.c:3117
zone_watermark_fast mm/page_alloc.c:3216 [inline]
get_page_from_freelist+0xc49/0x44c0 mm/page_alloc.c:3300
__alloc_pages_nodemask+0x21e/0x640 mm/page_alloc.c:4370
alloc_pages_current+0xcc/0x210 mm/mempolicy.c:2093
alloc_pages include/linux/gfp.h:509 [inline]
__get_free_pages+0x12/0x60 mm/page_alloc.c:4414
dma_mem_alloc+0x36/0x50 arch/x86/include/asm/floppy.h:156
raw_cmd_copyin drivers/block/floppy.c:3159 [inline]
raw_cmd_ioctl drivers/block/floppy.c:3206 [inline]
fd_locked_ioctl+0xa00/0x2c10 drivers/block/floppy.c:3544
fd_ioctl+0x40/0x60 drivers/block/floppy.c:3571
__blkdev_driver_ioctl block/ioctl.c:303 [inline]
blkdev_ioctl+0xb3c/0x1a30 block/ioctl.c:601
block_ioctl+0x105/0x150 fs/block_dev.c:1883
vfs_ioctl fs/ioctl.c:46 [inline]
do_vfs_ioctl+0x1c0/0x1150 fs/ioctl.c:687
ksys_ioctl+0x9e/0xb0 fs/ioctl.c:702
__do_sys_ioctl fs/ioctl.c:709 [inline]
__se_sys_ioctl fs/ioctl.c:707 [inline]
__x64_sys_ioctl+0x7e/0xc0 fs/ioctl.c:707
do_syscall_64+0xc4/0x510 arch/x86/entry/common.c:290
entry_SYSCALL_64_after_hwframe+0x49/0xbe

Note that this is not a kvmalloc path. It is just that the fast path
really depends on having sanitzed order as well. Therefore move the
order check to the fast path.

Reported-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Reported-by: Kyungtae Kim <kt0755@gmail.com>
Signed-off-by: Michal Hocko <mhocko@suse.com>
---
mm/page_alloc.c | 22 +++++++++++-----------
1 file changed, 11 insertions(+), 11 deletions(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index a919ba5cb3c8..9fc10a1029cf 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -4060,17 +4060,6 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
unsigned int cpuset_mems_cookie;
int reserve_flags;

- /*
- * In the slowpath, we sanity check order to avoid ever trying to
- * reclaim >= MAX_ORDER areas which will never succeed. Callers may
- * be using allocators in order of preference for an area that is
- * too large.
- */
- if (order >= MAX_ORDER) {
- WARN_ON_ONCE(!(gfp_mask & __GFP_NOWARN));
- return NULL;
- }
-
/*
* We also sanity check to catch abuse of atomic reserves being used by
* callers that are not in atomic context.
@@ -4364,6 +4353,17 @@ __alloc_pages_nodemask(gfp_t gfp_mask, unsigned int order, int preferred_nid,
gfp_t alloc_mask; /* The gfp_t that was actually used for allocation */
struct alloc_context ac = { };

+ /*
+ * In the slowpath, we sanity check order to avoid ever trying to
+ * reclaim >= MAX_ORDER areas which will never succeed. Callers may
+ * be using allocators in order of preference for an area that is
+ * too large.
+ */
+ if (order >= MAX_ORDER) {
+ WARN_ON_ONCE(!(gfp_mask & __GFP_NOWARN));
+ return NULL;
+ }
+
gfp_mask &= gfp_allowed_mask;
alloc_mask = gfp_mask;
if (!prepare_alloc_pages(gfp_mask, order, preferred_nid, nodemask, &ac, &alloc_mask, &alloc_flags))
--
2.19.1
--
Michal Hocko
SUSE Labs

\
 
 \ /
  Last update: 2018-11-09 09:45    [W:0.123 / U:8.492 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site