New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.
Starred by 4 users

Issue metadata

Status: Fixed
Closed: Apr 2011
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 2
Type: Bug-Security

  • Only users with EditIssue permission may comment.

Sign in to add a comment

race in base/third_party/xdg_mime (crasher)

Reported by, Feb 1 2011

Issue description

Opening suitable pages sometimes causes Chrome to crash at various places on startup. These kind of crashes have been happening for a while now, but have been difficult to isolate and reproduce. The last browser test run found several files which seem to trigger this fairly often. One such file is attached. I'll try to narrow down the cause later if others can also reproduce and this is not a duplicate issue.

The crashes look like there could be a memory corruption. Many of the traces have had hosed pointers, often with freeing or allocating going on nearby, so reporting this conservatively as a security issue.

Chrome Version: 9.0.597.83 beta (also current dev)
Operating System: Linux, Ubuntu 10.10 x64_64 

The crash should occur at least once a minute when repeatedly opening the attached file. The following loop should trigger a browser crash about once a minute:
 $ while true; do google-chrome a.html a.html a.html a.html & sleep 1.5; pkill -9 chrome; done

Type of crash: browser
Crash State: 
Crash state varies a lot. One from 9.0.597.83 beta:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffe7b62700 (LWP 24653)]
__strcmp_sse2 () at ../sysdeps/x86_64/multiarch/../strcmp.S:107
107	../sysdeps/x86_64/multiarch/../strcmp.S: No such file or directory.
	in ../sysdeps/x86_64/multiarch/../strcmp.S
(gdb) bt
#0  __strcmp_sse2 () at ../sysdeps/x86_64/multiarch/../strcmp.S:107
#1  0x00007ffff1ac43b4 in msort_with_tmp (p=<value optimized out>, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:105
#2  0x00007ffff1ac4188 in msort_with_tmp (p=<value optimized out>, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:54
#3  0x00007ffff1ac4188 in msort_with_tmp (p=<value optimized out>, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:54
#4  0x00007ffff1ac4188 in msort_with_tmp (p=<value optimized out>, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:54
#5  0x00007ffff1ac4178 in msort_with_tmp (p=0x4b9da40, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:53
#6  0x00007ffff1ac4178 in msort_with_tmp (p=0x4b9da40, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:53
#7  0x00007ffff1ac4178 in msort_with_tmp (p=0x4b9da40, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:53
#8  0x00007ffff1ac4178 in msort_with_tmp (p=0x4b9da40, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:53
#9  0x00007ffff1ac4178 in msort_with_tmp (p=0x4b9da40, 
    b=<value optimized out>, n=<value optimized out>) at msort.c:53
#10 0x00007ffff1ac47ec in qsort_r (b=<value optimized out>, 
    n=<value optimized out>, s=0, cmp=0xbd5860, arg=0x0) at msort.c:294
#11 0x0000000000bd5812 in ?? ()
#12 0x0000000000bd5533 in ?? ()
(gdb) info registers
rax            0x0	0
rbx            0x4bad080	79351936
rcx            0x2f	47
rdx            0x0	0
rsi            0x6d00750063006f	30681274979123311
rdi            0x4b9da40	79288896
rbp            0x4bab800	0x4bab800
rsp            0x7fffe7b60da8	0x7fffe7b60da8
r8             0x4ba0880	79300736
r9             0x4ba0940	79300928
r10            0x4ba0920	79300896
r11            0x206	518
r12            0x1	1
r13            0x4bab810	79345680
r14            0x4bab810	79345680
r15            0x10	16
rip            0x7ffff1b0d42a	0x7ffff1b0d42a <__strcmp_sse2+26>
eflags         0x10283	[ CF SF IF RF ]
cs             0x33	51
ss             0x2b	43
ds             0x0	0
es             0x0	0
fs             0x0	0
gs             0x0	0
(gdb) disas $rip-32, $rip+32
Dump of assembler code from 0x7ffff1b0d40a to 0x7ffff1b0d44a:
   0x00007ffff1b0d40a <strcmp+58>:	add    %al,%bl
   0x00007ffff1b0d40c:	jmp    0x7ffff1b0d410 <__strcmp_sse2>
   0x00007ffff1b0d40e:	nop
   0x00007ffff1b0d40f:	nop
   0x00007ffff1b0d410 <__strcmp_sse2+0>:	mov    %esi,%ecx
   0x00007ffff1b0d412 <__strcmp_sse2+2>:	mov    %edi,%eax
   0x00007ffff1b0d414 <__strcmp_sse2+4>:	and    $0x3f,%rcx
   0x00007ffff1b0d418 <__strcmp_sse2+8>:	and    $0x3f,%rax
   0x00007ffff1b0d41c <__strcmp_sse2+12>:	cmp    $0x30,%ecx
   0x00007ffff1b0d41f <__strcmp_sse2+15>:	ja     0x7ffff1b0d460 <__strcmp_sse2+80>
   0x00007ffff1b0d421 <__strcmp_sse2+17>:	cmp    $0x30,%eax
   0x00007ffff1b0d424 <__strcmp_sse2+20>:	ja     0x7ffff1b0d460 <__strcmp_sse2+80>
   0x00007ffff1b0d426 <__strcmp_sse2+22>:	movlpd (%rdi),%xmm1
=> 0x00007ffff1b0d42a <__strcmp_sse2+26>:	movlpd (%rsi),%xmm2
   0x00007ffff1b0d42e <__strcmp_sse2+30>:	movhpd 0x8(%rdi),%xmm1
   0x00007ffff1b0d433 <__strcmp_sse2+35>:	movhpd 0x8(%rsi),%xmm2
   0x00007ffff1b0d438 <__strcmp_sse2+40>:	pxor   %xmm0,%xmm0
   0x00007ffff1b0d43c <__strcmp_sse2+44>:	pcmpeqb %xmm1,%xmm0
   0x00007ffff1b0d440 <__strcmp_sse2+48>:	pcmpeqb %xmm2,%xmm1
   0x00007ffff1b0d444 <__strcmp_sse2+52>:	psubb  %xmm0,%xmm1
   0x00007ffff1b0d448 <__strcmp_sse2+56>:	pmovmskb %xmm1,%edx
End of assembler dump.
(gdb) p $rsi
$1 = 30681274979123311

The next one was:
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffe7b62700 (LWP 26063)]
0x0000000000b76464 in ?? ()
(gdb) bt
#0  0x0000000000b76464 in ?? ()
#1  0x0000000000b7673d in ?? ()
#2  0x00007fffdc494e9e in sqlite3_free () from /usr/lib/
#3  0x00007fffdc4c52fd in ?? () from /usr/lib/
#4  0x00007fffdc4d200c in ?? () from /usr/lib/
#5  0x00007fffdc4f1caa in sqlite3_exec () from /usr/lib/
#6  0x00007fffdc7428a3 in tableExists (sqlDB=0x4bf0408, tableName=<value optimized out>) at sdb.c:1657
10.1 KB Download

Comment 1 by, Feb 2 2011

A more symbolic trace using 11.0.657.0 (Developer Build 73427). This seems to be the most common place where the crash occurs.
7.9 KB View Download
@aohelin: that's an interesting trace in #c1, because both of those strings being compared look valid to me. It leads one to suspect a data race condition. Could you provide the output for "info threads" at the time of the crash? If any thread looks like it is also doing something MIME related, please provide its backtrace.
Also, it'd be useful to see the actual faulting instruction (plus registers) inside strcmp() itself.
Oh, I see the crashes are all over the shop... this may need valgrind :-/
It would still be useful to sample what the other threads are doing at the exact time of the SIGSEGV. Whoever may have stomped memory won't have moved on very far, especially on a multi-core machine.

Comment 5 by, Feb 2 2011

Took a while to get it again to gdb. Loading symbols is a bit slow. I'll add the thread backtraces soon.
6.0 KB View Download

Comment 6 by, Feb 2 2011

Thread backtraces. This is in one of the other common crash sites. Thread 10 crashed, and 5 looks interesting.

(gdb) thread 5
[Switching to thread 5 (Thread 0x7fffe9478700 (LWP 26507))]#0  __strcmp_sse2 ()
    at ../sysdeps/x86_64/multiarch/../strcmp.S:107
107	in ../sysdeps/x86_64/multiarch/../strcmp.S
(gdb) info registers
rax            0x0	0
rbx            0x7fffe470eb00	140737025993472
rcx            0x0	0
rdx            0x100000000	4294967296
rsi            0x100000000	4294967296
rdi            0x7fffdff89b40	140736950999872
rbp            0x7fffe9476250	0x7fffe9476250
rsp            0x7fffe9476238	0x7fffe9476238
r8             0x7fffdff8cd80	140736951012736
r9             0x7fffdff8cd60	140736951012704
r10            0x7fffdff8cd40	140736951012672
r11            0x206	518
r12            0x1	1
r13            0x7fffdff99810	140736951064592
r14            0x7fffdff99810	140736951064592
r15            0x10	16
rip            0x7ffff0da142a	0x7ffff0da142a <__strcmp_sse2+26>
eflags         0x10283	[ CF SF IF RF ]
cs             0x33	51
ss             0x2b	43
ds             0x0	0
es             0x0	0
fs             0x0	0
gs             0x0	0
(gdb) thread 10
[Switching to thread 10 (Thread 0x7fffe6c73700 (LWP 26512))]#0  __strcmp_sse2
    () at ../sysdeps/x86_64/multiarch/../strcmp.S:106
106	in ../sysdeps/x86_64/multiarch/../strcmp.S
(gdb) info registers
rax            0x0	0
rbx            0x7fffdff94010	140736951042064
rcx            0x0	0
rdx            0x7fffdff8cf40	140736951013184
rsi            0x7fffdff8cf40	140736951013184
rdi            0x0	0
rbp            0x7fffe6c71170	0x7fffe6c71170
rsp            0x7fffe6c71158	0x7fffe6c71158
r8             0x30	48
r9             0x101010101010101	72340172838076673
r10            0x7fffdff8f2c0	140736951022272
r11            0x7ffff0da5756	140737234229078
r12            0x1	1
r13            0x7fffdff98010	140736951058448
r14            0x7fffdff98010	140736951058448
r15            0x10	16
rip            0x7ffff0da1426	0x7ffff0da1426 <__strcmp_sse2+22>
eflags         0x10283	[ CF SF IF RF ]
cs             0x33	51
ss             0x2b	43
ds             0x0	0
es             0x0	0
fs             0x0	0
gs             0x0	0

23.3 KB View Download

Comment 7 by, Feb 2 2011

Reduced testcase. Same effect and reproduction instructions. Some a's can probably be removed, but the dot seems to be required. Always a bit disappointing to see fun test data reduce to something like this :)
67 bytes View Download
Hmm. Might be a race with the plugin process at startup?

Comment 9 by, Feb 2 2011

Mime strings and sorting would fit multiple threads having raced to populate a shared mime database at the same time, and breaking the browser process in the... process.
Tom, you're a bit of a race wizard. Any chance you're interesting in looking at this bug?

Comment 11 by, Feb 3 2011

I just had a quick look here. Might be worth starting at base/third_party/xdg_mime/xdgmime.c. Many functions call xdg_mime_init to make sure the mime data is loaded, and the flag whether it needs to be loaded seems to be a non-protected global variable. It is set to FALSE after the data has been loaded, so there is plenty of time for two threads to race and start reading the data.
Nice. I wonder which two threads? We only have one file IO thread, and IO tasks (such as loading the MIME db) are supposed to be strictly serialized on that thread. This design is what makes the Chrome UI so responsive.

Maybe a strategic sprinkling of DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO)); in a debug build would help?

Comment 13 by, Feb 3 2011

At least asserts checking if someone has changed need_reread after starting to read the data fire pretty often, as in "base/third_party/xdg_mime/xdgmime.c:467: xdg_mime_init: Assertion `need_reread == (!(0))' failed." I'll have another look at the thread issue with DCHECKs after work unless this has been resolved already. Seems to take a while for me to figure out who is doing what and sharing what with whom :)

Comment 14 by, Feb 3 2011

One of the threads seems to be handling a MimeRegistryMessageFilter::OnGetMimeTypeFromExtension, and does an xdg_mime_init() since it has not yet been loaded, and the other thread is doing a net::URLRequestFileJob and does xdg_mime_init() via URLRequestFileJob::GetMimeType. The latter usually seems to have finished loading mime data and has moved on, while the former usually crashes or aborts if the need_reread check is in place.

Different runs show a slightly different position, but this seems to be the common pattern. Adding a mutex or some other exclusion mechanism to mime reading would probably be enough to fix this.

Comment 15 by, Feb 4 2011

Crash does not seem to happen if a mutex is added around need_reread. This is just an ugly workaround though, as the whole xdg library is not thread-safe. Chrome itself should probably be checking that no two threads are in the air at the same time calling xdg code. I didn't check yet whether the other thread is bypassing the mime registry and causing the race that way, or if both go through a same point which could enforce single threaded access.

Not sure if this helps. I seem to have ended up spending a moment with morning coffee and evening tea searching for or looking at bugs :)
1.5 KB View Download
Labels: -Pri-0 -Area-Undefined Pri-2 Area-Internals Feature-Plugins OS-Linux SecSeverity-Low
Given that this appears specific to startup and probably not web triggerable, I'm marking as low severity now.

@tsepez - Could you investigate and work out a fix when you get a chance?
Adding the asserts as suggested in comment 12 rapidly results in:

[] Check failed: BrowserThread::CurrentlyOn(BrowserThread::IO). 
	base::debug::StackTrace::StackTrace() [0x126dd76]
	logging::LogMessage::~LogMessage() [0x1282a22]
	mime_util::GetFileMimeType() [0x129c6e0]
	net::PlatformMimeUtil::GetPlatformMimeTypeFromExtension() [0x181252b]
	net::MimeUtil::GetMimeTypeFromExtension() [0x17fa1d0]
	net::GetMimeTypeFromExtension() [0x17fb518]
	MimeRegistryMessageFilter::OnGetMimeTypeFromExtension() [0xfa3593]
	DispatchToMethod<>() [0xfa3d2d]
	IPC::MessageWithReply<>::Dispatch<>() [0xfa39d2]
	MimeRegistryMessageFilter::OnMessageReceived() [0xfa34c9]
	BrowserMessageFilter::DispatchMessage() [0xa48d3d]
	DispatchToMethod<>() [0xa494e1]
	RunnableMethod<>::Run() [0xa49300]
	MessageLoop::RunTask() [0x1285acc]
	MessageLoop::DeferOrRunPendingTask() [0x1285bba]
	MessageLoop::DoWork() [0x12864ec]
	base::MessagePumpLibevent::Run() [0x125ce5b]
	MessageLoop::RunInternal() [0x12858f0]
	MessageLoop::RunHandler() [0x128578c]
	MessageLoop::Run() [0x128514b]
	base::Thread::Run() [0x12cdb1e]
	base::Thread::ThreadMain() [0x12cdcd9]
	base::(anonymous namespace)::ThreadFunc() [0x12c8cc7]
	start_thread [0x7fc99c3989ca]

as expected.  Probably a layering violation to allow files in base/ to call into browser/ ...
I take it back.  Looks like a small typo in comment #12: IO vs. FILE.  Would seem to want to happen in the file thread via chrome/browser/'s 

void MimeRegistryMessageFilter::OverrideThreadForMessage(
    const IPC::Message& message,
    BrowserThread::ID* thread) {
  if (IPC_MESSAGE_CLASS(message) == MimeRegistryMsgStart)
    *thread = BrowserThread::FILE;

Changing the asserts to FILE and things are no longer asserting.  Time for more digging.

Comment 19 by, Feb 10 2011

To summarize aohelin's analysis:

- MimeRegistryMessageFilter forwards all messages on the FILE thread (comment #18)
- URLRequestFileJob::GetMimeType lives on the IO thread (that's where URLRequests live)
- both call into xdg_mime_init via net::GetMimeTypeFromExtension.

It seems clear to me that this is bad.  A mutex solves it, but we usually don't like to block the IO thread on the FILE thread.  It'd be better to make the relevant calls async.

I think it's totally wrong that URLRequestFileJob is calling a blocking disk function from the IO thread anyway.  And looking at the code...

bool URLRequestFileJob::GetMimeType(std::string* mime_type) const {
  // URL requests should not block on the disk!  On Windows this goes to the
  // registry.
  base::ThreadRestrictions::ScopedAllowIO allow_io;
  return GetMimeTypeFromFile(file_path_, mime_type);


Comment 20 by, Feb 10 2011

Will and I discussed this in person.

We think:
1) Fix the race.  We should put a lock into base/ because the code it's calling (the xdg stuff) has tons of global state.  Having URLRequestFileJob::GetMimeType() block on a lock is the same amount of jank as the code currently has, so it's not such a bad thing to introduce it first.

2) Fix the jank.  This is a separate issue, and can be done as part of 59849.  I'll comment there with what needs to be done.

Comment 21 Deleted

Comment 22 by, Feb 10 2011

Issue 72042 has been merged into this issue.

Comment 23 by, Feb 10 2011

Status: Untriaged
Summary: race in base/third_party/xdg_mime (crasher)
The dup'd issue is a ThreadSanitizer report showing that we have this race.  We shoulda caught this earlier.  :~(
Adding Kostya who found the issue 72042

Comment 25 by, Feb 11 2011

As we now have an easy reproducer (do we?), I'd suggest running it under ThreadSanitizer, before and after the fix. 
Labels: -SecSeverity-Low SecSeverity-Medium
Bumping up the severity a bit because the other crash stacks make it look like a malicious extension could potentially use this for privilege escalation (although that may be a longshot).

Comment 27 by, Mar 20 2011

Labels: Stability-Memory Stability-ThreadSanitizer Stability-Valgrind
I start seeing this as a Memcheck report (in addition to ThreadSanitizer report in  issue 74042 ): 
Invalid write of size 8                                                                                                                                                                                                                                                                     
   at 0xD1365C: _xdg_mime_alias_read_from_file (base/third_party/xdg_mime/xdgmimealias.c:153)                                                                                                                                                                                               
   by 0xD12F9A: xdg_mime_init_from_directory (base/third_party/xdg_mime/xdgmime.c:201)                                                                                                                                                                                                      
   by 0xD12AED: xdg_run_command_on_dirs (base/third_party/xdg_mime/xdgmime.c:288)                                                                                                                                                                                                           
   by 0xD13405: xdg_mime_init (base/third_party/xdg_mime/xdgmime.c:456)                                                                                                                                                                                                                     
   by 0xD13490: xdg_mime_get_mime_type_from_file_name (base/third_party/xdg_mime/xdgmime.c:567)                                                                                                                                                                                             
   by 0xCDF4FD: mime_util::GetFileMimeType(FilePath const&) (base/                                                                                                                                                                                                     
   by 0x1098BCB: net::PlatformMimeUtil::GetPlatformMimeTypeFromExtension(std::string const&, std::string*) const (net/base/                                                                                                                                  
   by 0x1086975: net::MimeUtil::GetMimeTypeFromExtension(std::string const&, std::string*) const (net/base/                                                                                                                                                                
   by 0x1086A55: net::MimeUtil::GetMimeTypeFromFile(FilePath const&, std::string*) const (net/base/                                                                                                                                                                        
   by 0xFA6800: net::URLRequestFileJob::GetMimeType(std::string*) const (net/url_request/                                                                                                                                                                       
   by 0xFB04C2: net::URLRequestJob::NotifyHeadersComplete() (net/url_request/                                                                                                                                                                                        
   by 0xFA64E6: net::URLRequestFileJob::DidResolve(bool, base::PlatformFileInfo const&) (net/url_request/                                                                                                                                                       
   by 0xCCE1B8: MessageLoop::RunTask(Task*) (base/                                                                                                                                                                                                                      
   by 0xCCEE17: MessageLoop::DeferOrRunPendingTask(MessageLoop::PendingTask const&) (base/                                                                                                                                                                              
   by 0xCCFB52: MessageLoop::DoWork() (base/                                                                                                                                                                                                                            
   by 0xCB03A8: base::MessagePumpLibevent::Run(base::MessagePump::Delegate*) (base/                                                                                                                                                                            
   by 0xCD05CD: MessageLoop::RunInternal() (base/                                                                                                                                                                                                                       
   by 0xCD0769: MessageLoop::Run() (base/                                                                                                                                                                                                                               
   by 0xCFB5AB: base::Thread::ThreadMain() (base/threading/                                                                                                                                                                                                                   
   by 0xCFAAE9: base::(anonymous namespace)::ThreadFunc(void*) (base/threading/                                                                                                                                                                                 
 Address 0x540 is not stack'd, malloc'd or (recently) free'd                        
Labels: Type-Security
Status: WillMerge
Project Member

Comment 30 by, Mar 30 2011

The following revision refers to this bug:

r79851 | | Wed Mar 30 09:59:46 PDT 2011

Changed paths:

XDG isn't thread safe.  Add a lock around calls to it.

BUG= 71586 
Review URL:
Labels: -Restrict-View-SecurityTeam Restrict-View-SecurityNotify reward-topanel
Status: FixUnreleased
Merged to M11: 80097
Project Member

Comment 32 by, Mar 31 2011

The following revision refers to this bug:

r80097 | | Thu Mar 31 16:20:07 PDT 2011

Changed paths:

Merge 79851 - XDG isn't thread safe.  Add a lock around calls to it.

BUG= 71586 
Review URL:
Review URL:

Comment 33 by, Apr 14 2011

Status: Available
I am testing yet another dynamic tool for detecting memory crashers. 
The tool reports similar stacks on Chromium: r81383 / WebKit: r83605
And I do see a similar stack in crash on Chrome 12. 

So, is this really fixed? 

  #0 0x0110b2e3  << _xdg_mime_cache_new_from_file  base/third_party/xdg_mime/xdgmimecache.c:156                                                                                                     
  #1 0x01104bd8  << xdg_mime_init_from_directory  base/third_party/xdg_mime/xdgmime.c:148                                                                                                           
  #2 0x011045c5  << xdg_run_command_on_dirs  base/third_party/xdg_mime/xdgmime.c:250                                                                                                                
  #3 0x011062bd  << xdg_mime_init  base/third_party/xdg_mime/xdgmime.c:456                                                                                                                          
  #4 0x01106900  << xdg_mime_get_mime_type_from_file_name  base/third_party/xdg_mime/xdgmime.c:567                                                                                                  
  #5 0x0109525a  << mime_util::GetFileMimeType(FilePath const0x109525a )  /usr/lib/llvm/gcc-4.2/lib/gcc/x86_64-linux-gnu/4.2.1/../../../../include/c++/4.2.1/bits/stl_tree.h:184                    
  #6 0x01973815  << net::PlatformMimeUtil::GetPlatformMimeTypeFromExtension(std::string const0x1973815 , std::string*) const  net/base/                               
  #7 0x0194a1ba  << net::MimeUtil::GetMimeTypeFromExtension(std::string const0x194a1ba , std::string*) const  /usr/lib/llvm/gcc-4.2/lib/gcc/x86_64-linux-gnu/4.2.1/../../../../include/c++/4.2.1/bi»
  #8 0x0194c318  << net::MimeUtil::GetMimeTypeFromFile(FilePath const0x194c318 , std::string*) const  /usr/lib/llvm/gcc-4.2/lib/gcc/x86_64-linux-gnu/4.2.1/../../../../include/c++/4.2.1/bits/stl_v»
  #9 0x0194c367  << net::GetMimeTypeFromFile(FilePath const0x194c367 , std::string*)  /usr/lib/llvm/gcc-4.2/lib/gcc/x86_64-linux-gnu/4.2.1/../../../../include/c++/4.2.1/bits/stl_vector.h:536      
  #10 0x017a25c8  << net::URLRequestFileJob::GetMimeType(std::string*) const  ./base/memory/weak_ptr.h:66                                                                                           
  #11 0x0179adc8  << net::URLRequest::GetMimeType(std::string*)  ./base/memory/ref_counted.h:95                                                                                                     
  #12 0x03f24e8c  << ResourceDispatcherHost::CompleteResponseStarted(net::URLRequest*)  ./base/memory/ref_counted.h:143                                                                             
  #13 0x03f25033  << ResourceDispatcherHost::CompleteResponseStarted(net::URLRequest*)  ./base/memory/ref_counted.h:143                                                                             
  #14 0x03f276b9  << ResourceDispatcherHost::CancelRequestsForRoute(int, int)  ./base/memory/ref_counted.h:143                                                                                      
  #15 0x0179c3a3  << net::URLRequest::ResponseStarted()  ./base/memory/ref_counted.h:95                                                                                                             
  #16 0x017acc14  << net::URLRequestJob::NotifyHeadersComplete()  ./base/memory/ref_counted.h:241                                                                                                   
  #17 0x017a2061  << net::URLRequestFileJob::DidResolve(bool, base::PlatformFileInfo const0x17a2061 )  ./base/memory/weak_ptr.h:66                                                                  
  #18 0x01071226  << MessageLoop::RunTask(Task*)  ./base/synchronization/lock.h:104                                                                                                                 
  #19 0x01071610  << MessageLoop::DeferOrRunPendingTask(MessageLoop::PendingTask const0x1071610 )  ./base/synchronization/lock.h:104                                                                
  #20 0x0107242a  << MessageLoop::DoWork()  ./base/synchronization/lock.h:104                                                                                                                       
  #21 0x0103bcb7  << base::MessagePumpLibevent::Run(base::MessagePump::Delegate*)  /usr/lib/llvm/gcc-4.2/lib/gcc/x86_64-linux-gnu/4.2.1/../../../../include/c++/4.2.1/bits/stl_algobase.h:189       
  #22 0x01072aef  << MessageLoop::RunInternal()  ./base/synchronization/lock.h:104                                                                                                                  
  #23 0x01072ca4  << MessageLoop::Run()  ./base/synchronization/lock.h:104                                                                                                                          
  #24 0x010cf797  << base::Thread::ThreadMain()  ./base/threading/platform_thread.h:57                                                                                                              
  #25 0x010cf368  << base::(anonymous namespace)::ThreadFunc(void*)  base/threading/    
Status: FixUnreleased
The version of Chrome of referenced in http://crash/reportdetail?reportid=e70f9fc500b5a7e1 is 12.0.712.0, which is r79102 and therefore does not contain Tom's fix.

Could you go in to more detail on what your new dynamic tool is reporting, exactly? It's hard to tell from just a stack trace.

Comment 36 by, Apr 14 2011

The new tool tells me that the memory accessed in _xdg_mime_cache_new_from_file has been freed long ago. 
So long ago, that the tool lost track of where it has been free-ed. 
Could be a false positive of course, the tool is just a few days old. :) 
I am rerunning with different settings (larger free-delay buffer and red zones)... 
Let me update this bug once I have a better report.

Labels: -reward-topanel reward-500 Mstone-11 reward-unpaid
@aohelin: we're happy to be without this race, it was also causing stability issues. Accordingly, please have a $500 Chromium Security Reward :)

Boilerplate text:
Please do NOT publicly disclose details until a fix has been released to all our
users. Early public disclosure may cancel the provisional reward.
Also, please be considerate about disclosure when the bug affects a core library
that may be used by other products.
Please do NOT share this information with third parties who are not directly
involved in fixing the bug. Doing so may cancel the provisional reward.
Please be honest if you have already disclosed anything publicly or to third parties.

Comment 38 by, Apr 15 2011

Excellent, thanks :)
Labels: CVE-2011-1434
Labels: -reward-unpaid
Invoice finalized; payment is in e-payment system; it can take a couple of weeks.
Labels: SecImpacts-Stable
Batch update.
Labels: -Restrict-View-SecurityNotify
Lifting view restrictions.
Status: Fixed
Project Member

Comment 44 by, Oct 13 2012

Labels: Restrict-AddIssueComment-Commit
This issue has been closed for some time. No one will pay attention to new comments.
If you are seeing this bug or have new data, please click New Issue to start a new bug.
Project Member

Comment 45 by, Mar 10 2013

Labels: -Area-Internals -Feature-Plugins -SecSeverity-Medium -Stability-ThreadSanitizer -Stability-Valgrind -Type-Security -Mstone-11 -SecImpacts-Stable M-11 Cr-Internals Security-Severity-Medium Cr-Content-Plugins Security-Impact-Stable Performance-Valgrind Performance-ThreadSanitizer Type-Bug-Security
Project Member

Comment 46 by, Mar 11 2013

Labels: -Stability-Memory Performance-Memory
Project Member

Comment 47 by, Mar 13 2013

Labels: -Restrict-AddIssueComment-Commit Restrict-AddIssueComment-EditIssue
Project Member

Comment 48 by, Mar 21 2013

Labels: -Security-Impact-Stable Security_Impact-Stable
Project Member

Comment 49 by, Mar 21 2013

Labels: -Security-Severity-Medium Security_Severity-Medium
Project Member

Comment 50 by, Apr 1 2013

Labels: -Performance-ThreadSanitizer Stability-ThreadSanitizer
Project Member

Comment 51 by, Apr 1 2013

Labels: -Performance-Memory Stability-Memory
Project Member

Comment 52 by, Apr 1 2013

Labels: -Performance-Valgrind Stability-Valgrind
Project Member

Comment 53 by, Apr 6 2013

Labels: Cr-Blink
Project Member

Comment 54 by, Apr 6 2013

Labels: -Cr-Content-Plugins Cr-Internals-Plugins
Issue 72042 has been merged into this issue.
Issue 72042 has been merged into this issue.
Labels: hasTestcase
Project Member

Comment 58 by, Oct 1 2016

This bug has been closed for more than 14 weeks. Removing security view restrictions.

For more details visit - Your friendly Sheriffbot
Project Member

Comment 59 by, Oct 2 2016

This bug has been closed for more than 14 weeks. Removing security view restrictions.

For more details visit - Your friendly Sheriffbot
Labels: allpublic
Labels: CVE_description-submitted

Sign in to add a comment