Thread (3 messages) 3 messages, 3 authors, 2025-01-29

Re: [PATCH v2] audit: Initialize lsmctx to avoid memory allocation error

From: Paul Moore <paul@paul-moore.com>
Date: 2025-01-28 23:41:39
Also in: lkml

On Jan 25, 2025 Huacai Chen [off-list ref] wrote:
Initialize the local variable lsmctx in audit_receive_msg(), so as to
avoid memory allocation errors like:

[  258.074914] WARNING: CPU: 2 PID: 443 at mm/page_alloc.c:4727 __alloc_pages_noprof+0x4c8/0x1040
[  258.074997] pc 900000000304d588 ra 9000000003059644 tp 9000000107774000 sp 9000000107777890
[  258.075000] a0 0000000000040cc0 a1 0000000000000012 a2 0000000000000000 a3 0000000000000000
[  258.075003] a4 9000000107777bd0 a5 0000000000000280 a6 0000000000000010 a7 0000000000000000
[  258.075006] t0 9000000004b4c000 t1 0000000000000001 t2 1f3f37829c264c80 t3 000000000000002e
[  258.075009] t4 0000000000000000 t5 00000000000003f6 t6 90000001066b6310 t7 000000000000002f
[  258.075011] t8 000000000000003c u0 00000000000000b4 s9 900000010006f880 s0 9000000004a4b000
[  258.075014] s1 0000000000000000 s2 9000000004a4b000 s3 9000000106673400 s4 9000000107777af0
[  258.075017] s5 90000001066b6300 s6 0000000000000012 s7 fffffffffffff000 s8 0000000000000004
[  258.075019]    ra: 9000000003059644 ___kmalloc_large_node+0x84/0x1e0
[  258.075026]   ERA: 900000000304d588 __alloc_pages_noprof+0x4c8/0x1040
[  258.075030]  CRMD: 000000b0 (PLV0 -IE -DA +PG DACF=CC DACM=CC -WE)
[  258.075040]  PRMD: 00000004 (PPLV0 +PIE -PWE)
[  258.075045]  EUEN: 00000007 (+FPE +SXE +ASXE -BTE)
[  258.075051]  ECFG: 00071c1d (LIE=0,2-4,10-12 VS=7)
[  258.075056] ESTAT: 000c0000 [BRK] (IS= ECode=12 EsubCode=0)
[  258.075061]  PRID: 0014c010 (Loongson-64bit, Loongson-3A5000)
[  258.075064] CPU: 2 UID: 0 PID: 443 Comm: auditd Not tainted 6.13.0-rc1+ #1899
[  258.075070] Stack : ffffffffffffffff 0000000000000000 9000000002debf5c 9000000107774000
[  258.075077]         90000001077774f0 0000000000000000 90000001077774f8 900000000489e480
[  258.075083]         9000000004b380e8 9000000004b380e0 9000000107777380 0000000000000001
[  258.075089]         0000000000000001 9000000004a4b000 1f3f37829c264c80 90000001001a9b40
[  258.075094]         9000000107774000 9000000004b080e8 00000000000003d4 9000000004b080e8
[  258.075100]         9000000004a580e8 000000000000002d 0000000006ebc000 900000010006f880
[  258.075106]         00000000000000b4 0000000000000000 0000000000000004 0000000000001277
[  258.075112]         900000000489e480 90000001066b6300 0000000000000012 fffffffffffff000
[  258.075118]         0000000000000004 900000000489e480 9000000002def6a8 00007ffff2ba4065
[  258.075124]         00000000000000b0 0000000000000004 0000000000000000 0000000000071c1d
[  258.075129]         ...
[  258.075132] Call Trace:
[  258.075135] [<9000000002def6a8>] show_stack+0x30/0x148
[  258.075146] [<9000000002debf58>] dump_stack_lvl+0x68/0xa0
[  258.075152] [<9000000002e0fe18>] __warn+0x80/0x108
[  258.075158] [<900000000407486c>] report_bug+0x154/0x268
[  258.075163] [<90000000040ad468>] do_bp+0x2a8/0x320
[  258.075172] [<9000000002dedda0>] handle_bp+0x120/0x1c0
[  258.075178] [<900000000304d588>] __alloc_pages_noprof+0x4c8/0x1040
[  258.075183] [<9000000003059640>] ___kmalloc_large_node+0x80/0x1e0
[  258.075187] [<9000000003061504>] __kmalloc_noprof+0x2c4/0x380
[  258.075192] [<9000000002f0f7ac>] audit_receive_msg+0x764/0x1530
[  258.075199] [<9000000002f1065c>] audit_receive+0xe4/0x1c0
[  258.075204] [<9000000003e5abe8>] netlink_unicast+0x340/0x450
[  258.075211] [<9000000003e5ae9c>] netlink_sendmsg+0x1a4/0x4a0
[  258.075216] [<9000000003d9ffd0>] __sock_sendmsg+0x48/0x58
[  258.075222] [<9000000003da32f0>] __sys_sendto+0x100/0x170
[  258.075226] [<9000000003da3374>] sys_sendto+0x14/0x28
[  258.075229] [<90000000040ad574>] do_syscall+0x94/0x138
[  258.075233] [<9000000002ded318>] handle_syscall+0xb8/0x158
If you are going to include the backtrace, please ensure that the lines
do not wrap when viewing the commit using 'git log' in an 80 char wide
terminal.
Root cause: if we enable AUDIT but disable LSM, in the AUDIT_SIGNAL_INFO
case, lsmprop_is_set() returns false, then kmalloc() may try to allocate
a large amount of memory with an uninitialized length.
You shouldn't need to have a special "Root cause:" section, just work
that into the commit description normally.  I would suggest that the
commit description could simply be something like the following:

  When audit is enabled in a kernel build, and there are no LSMs
  active that support LSM labeling, it is possible that the lsmctx
  variable in the AUDIT_SIGNAL_INFO handler in audit_receive_msg()
  could be used before it is properly initialize.  This patch
  corrects this problem by initializing the lsmctx to a safe value
  when it is declared.
quoted hunk ↗ jump to hunk
Fixes: 6fba89813ccf333d ("lsm: ensure the correct LSM context releaser")
Signed-off-by: Huacai Chen <redacted>
---
V2: Update commit message and CC list.

 kernel/audit.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/audit.c b/kernel/audit.c
index 13d0144efaa3..5f5bf85bcc90 100644
--- a/kernel/audit.c
+++ b/kernel/audit.c
@@ -1221,7 +1221,7 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh,
 	struct audit_buffer	*ab;
 	u16			msg_type = nlh->nlmsg_type;
 	struct audit_sig_info   *sig_data;
-	struct lsm_context	lsmctx;
+	struct lsm_context	lsmctx = { NULL, 0, 0 };
 
 	err = audit_netlink_ok(skb, msg_type);
 	if (err)
-- 
2.47.1
--
paul-moore.com
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help