linux-kernel - uprobes/perf: KASAN: use-after-free in uprobe_perf

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>] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 22 Feb 2018 14:08:30 +0900
From:   Prashant Bhole <bhole_prashant_q7@....ntt.co.jp>
To:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Jiri Olsa <jolsa@...hat.com>,
        Namhyung Kim <namhyung@...nel.org>
Cc:     linux-kernel@...r.kernel.org
Subject: uprobes/perf: KASAN: use-after-free in uprobe_perf_close

Hi,
I encountered following BUG caught by KASAN with recent kernels when 
trying out [BCC project] bcc/testing/python/test_usdt2.py
Tried with v4.12, it was reproducible.

--- KASAN log ---
BUG: KASAN: use-after-free in uprobe_perf_close+0x118/0x1a0
Read of size 4 at addr ffff8800bb2db4cc by task test_usdt2.py/1265

CPU: 2 PID: 1265 Comm: test_usdt2.py Not tainted 
4.16.0-rc2-next-20180220+ #38
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 
1.10.2-1.fc26 04/01/2014
Call Trace:
  dump_stack+0x5c/0x80
  print_address_description+0x73/0x290
  kasan_report+0x257/0x380
  ? uprobe_perf_close+0x118/0x1a0
  uprobe_perf_close+0x118/0x1a0
  perf_uprobe_destroy+0x54/0x90
  _free_event+0x1a5/0x5c0
  perf_event_release_kernel+0x35e/0x620
  ? put_event+0x20/0x20
  perf_release+0x1c/0x20
  __fput+0x182/0x360
  task_work_run+0x9c/0xc0
  exit_to_usermode_loop+0xc2/0xd0
  do_syscall_64+0x244/0x250
  entry_SYSCALL_64_after_hwframe+0x3d/0xa2
[...]
Allocated by task 1265:
  kasan_kmalloc+0xa0/0xd0
  kmem_cache_alloc_node+0x123/0x210
  copy_process.part.32+0xb9d/0x3050
  _do_fork+0x178/0x630
  do_syscall_64+0xe7/0x250
  entry_SYSCALL_64_after_hwframe+0x3d/0xa2

Freed by task 1265:
  __kasan_slab_free+0x135/0x180
  kmem_cache_free+0xaf/0x230
  rcu_process_callbacks+0x559/0xd90
  __do_softirq+0x125/0x3a2

The buggy address belongs to the object at ffff8800bb2db480
  which belongs to the cache task_struct of size 12928
-----------------


After debugging, found that uprobe_perf_close() is called after task has 
been terminated and uprobe_perf_close() tries to access task_struct of 
the terminated process.

As fix I came up with following changes. Basically it gets a refcount on 
task_struct in uprobe_perf_open() and releases in uprobe_perf_close(). 
If this is a correct fix, I will submit a proper patch.


Signed-off-by: Prashant Bhole <bhole_prashant_q7@....ntt.co.jp>
---
  kernel/trace/trace_uprobe.c | 9 +++++++--
  1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/trace_uprobe.c b/kernel/trace/trace_uprobe.c
index 2014f4351ae0..b81e0a88136a 100644
--- a/kernel/trace/trace_uprobe.c
+++ b/kernel/trace/trace_uprobe.c
@@ -1039,6 +1039,7 @@ uprobe_filter_event(struct trace_uprobe *tu, 
struct perf_event *event)

  static int uprobe_perf_close(struct trace_uprobe *tu, struct 
perf_event *event)
  {
+    int err = 0;
      bool done;

      write_lock(&tu->filter.rwlock);
@@ -1054,9 +1055,12 @@ static int uprobe_perf_close(struct trace_uprobe 
*tu, struct perf_event *event)
      write_unlock(&tu->filter.rwlock);

      if (!done)
-        return uprobe_apply(tu->inode, tu->offset, &tu->consumer, false);
+        err =  uprobe_apply(tu->inode, tu->offset, &tu->consumer, false);

-    return 0;
+    if (event->hw.target)
+        put_task_struct(event->hw.target);
+
+    return err;
  }

  static int uprobe_perf_open(struct trace_uprobe *tu, struct perf_event 
*event)
@@ -1077,6 +1081,7 @@ static int uprobe_perf_open(struct trace_uprobe 
*tu, struct perf_event *event)
          done = tu->filter.nr_systemwide ||
              event->parent || event->attr.enable_on_exec ||
              uprobe_filter_event(tu, event);
+        get_task_struct(event->hw.target);
          list_add(&event->hw.tp_list, &tu->filter.perf_events);
      } else {
          done = tu->filter.nr_systemwide;
-- 
2.14.3


-Prashant