linux-kernel - [PATCH -next v2] mm/util: annotate an data race at vm_committed

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Date:   Wed, 12 Feb 2020 09:35:09 -0500
From:   Qian Cai <cai@....pw>
To:     akpm@...ux-foundation.org
Cc:     tj@...nel.org, elver@...gle.com, cl@...ux.com, dennis@...nel.org,
        linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        Qian Cai <cai@....pw>
Subject: [PATCH -next v2] mm/util: annotate an data race at vm_committed_as

"vm_committed_as.count" could be accessed concurrently as reported by
KCSAN,

 read to 0xffffffff923164f8 of 8 bytes by task 1268 on cpu 38:
  __vm_enough_memory+0x43/0x280 mm/util.c:801
  mmap_region+0x1b2/0xb90 mm/mmap.c:1726
  do_mmap+0x45c/0x700
  vm_mmap_pgoff+0xc0/0x130
  vm_mmap+0x71/0x90
  elf_map+0xa1/0x1b0
  load_elf_binary+0x9de/0x2180
  search_binary_handler+0xd8/0x2b0
  __do_execve_file+0xb61/0x1080
  __x64_sys_execve+0x5f/0x70
  do_syscall_64+0x91/0xb47
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

 write to 0xffffffff923164f8 of 8 bytes by task 1265 on cpu 41:
  percpu_counter_add_batch+0x83/0xd0 lib/percpu_counter.c:91
  exit_mmap+0x178/0x220 include/linux/mman.h:68
  mmput+0x10e/0x270
  flush_old_exec+0x572/0xfe0
  load_elf_binary+0x467/0x2180
  search_binary_handler+0xd8/0x2b0
  __do_execve_file+0xb61/0x1080
  __x64_sys_execve+0x5f/0x70
  do_syscall_64+0x91/0xb47
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

The warning is almost impossible to trigger according to the commit
82f71ae4a2b8 ("mm: catch memory commitment underflow") but leave it for
now to catch any possible unbalanced vm_unacct_memory() in the future.
Since only the read is operating as lockless, mark it as an intentional
data race using the data_race() macro to avoid modifying
percpu_counter_read() and still catch unintended races elsewhere.

Acked-by: Christoph Lameter <cl@...ux.com>
Acked-by: Dennis Zhou <dennis@...nel.org>
Signed-off-by: Qian Cai <cai@....pw>
---

v2: add some code comments.

 mm/util.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/mm/util.c b/mm/util.c
index 988d11e6c17c..cc89e2404e19 100644
--- a/mm/util.c
+++ b/mm/util.c
@@ -798,8 +798,12 @@ int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin)
 {
 	long allowed;
 
-	VM_WARN_ONCE(percpu_counter_read(&vm_committed_as) <
-			-(s64)vm_committed_as_batch * num_online_cpus(),
+	/*
+	 * A transient decrease in the value is unlikely, so no need
+	 * READ_ONCE() for vm_committed_as.count.
+	 */
+	VM_WARN_ONCE(data_race(percpu_counter_read(&vm_committed_as) <
+			-(s64)vm_committed_as_batch * num_online_cpus()),
 			"memory commitment underflow");
 
 	vm_acct_memory(pages);
-- 
1.8.3.1