Add libc++ __tree to ubsan blacklist.

Add libc++ __tree to ubsan blacklist.

libc++ __tree is not ubsan clean. It floods compile and run steps
with "runtime error: downcast" and "runtime error: upcast" errors,
making ubsan much less usable.
This is a KI with libc++: https://llvm.org/bugs/show_bug.cgi?id=19302

Committed: https://crrev.com/e2abe2366d707ac6ad308e2643f41423971f81fc
Cr-Commit-Position: refs/heads/master@{#402902}

[mbjorge, 2016-06-28 23:16:47]

mbjorge@chromium.org changed reviewers:
+ dpranke@chromium.org, ochang@chromium.org

[mbjorge, 2016-06-28 23:16:48]

Came across this issue with UBSAN when building ubsan chromecast builds.

Both at compile and run time, logs are filled with:

../../buildtools/third_party/libc++/trunk/include/__tree:836:16: runtime error:
downcast of address 0x7ffcfa208940 with insufficient space for an object of type
'std::__1::__tree_node<std::__1::__value_type<std::__1::basic_string<char>,
std::__1::vector<base::TestResult, std::__1::allocator<base::TestResult> > >,
void *>'
0x7ffcfa208940: note: pointer points here

I'm curious how this isn't being seen on the Chromium ubsan builders (but I
couldn't figure out where they were?)

[Dirk Pranke, 2016-06-28 23:24:00]

lgtm if ochang approves.

[Oliver Chang, 2016-06-29 16:50:17]

lgtm

[mbjorge, 2016-06-29 18:39:14]

The CQ bit was checked by mbjorge@chromium.org

[mbjorge, 2016-06-29 18:39:14]

Added map also; was able to see this error once the __tree errors were cleared
out. I can now compile + run all the unittests without any ubsan errors.

[mbjorge, 2016-06-29 18:39:14]

The patchset sent to the CQ was uploaded after l-g-t-m from ochang@chromium.org,
dpranke@chromium.org
Link to the patchset: https://codereview.chromium.org/2109663003/#ps20001
(title: "Added map also")

[commit-bot: I haz the power, 2016-06-29 18:40:33]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-06-29 19:38:58]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-06-29 19:39:21]

CQ bit was unchecked.

[commit-bot: I haz the power, 2016-06-29 19:40:47]

Description was changed from

==========
Add libc++ __tree to ubsan blacklist.

libc++ __tree is not ubsan clean. It floods compile and run steps
with "runtime error: downcast" and "runtime error: upcast" errors,
making ubsan much less usable.
This is a KI with libc++: https://llvm.org/bugs/show_bug.cgi?id=19302
==========

to

==========
Add libc++ __tree to ubsan blacklist.

libc++ __tree is not ubsan clean. It floods compile and run steps
with "runtime error: downcast" and "runtime error: upcast" errors,
making ubsan much less usable.
This is a KI with libc++: https://llvm.org/bugs/show_bug.cgi?id=19302

Committed: https://crrev.com/e2abe2366d707ac6ad308e2643f41423971f81fc
Cr-Commit-Position: refs/heads/master@{#402902}
==========

[commit-bot: I haz the power, 2016-06-29 19:40:48]

Patchset 2 (id:??) landed as
https://crrev.com/e2abe2366d707ac6ad308e2643f41423971f81fc
Cr-Commit-Position: refs/heads/master@{#402902}

[Nico, 2017-07-17 23:40:33]

thakis@chromium.org changed reviewers:
+ thakis@chromium.org

[Nico, 2017-07-17 23:40:34]

Looks like this has been fixed upstream. Should we try undoing this again?

[mbjorge, 2017-07-17 23:41:28]

On 2017/07/17 at 23:40:34, thakis wrote:
> Looks like this has been fixed upstream. Should we try undoing this again?

sgtm

[Nico, 2017-07-18 14:35:51]

On 2017/07/17 23:41:28, mbjorge wrote:
> On 2017/07/17 at 23:40:34, thakis wrote:
> > Looks like this has been fixed upstream. Should we try undoing this again?
> 
> sgtm

Cool, can you revert locally, check whatever this fixed back then remains fixed,
and then upload the revert?

[mbjorge, 2017-07-18 23:30:29]

On 2017/07/18 at 14:35:51, thakis wrote:
> On 2017/07/17 23:41:28, mbjorge wrote:
> > On 2017/07/17 at 23:40:34, thakis wrote:
> > > Looks like this has been fixed upstream. Should we try undoing this again?
> > 
> > sgtm
> 
> Cool, can you revert locally, check whatever this fixed back then remains
fixed, and then upload the revert?

Yeah, I'll test it out and see how it looks

[mbjorge, 2017-07-19 20:40:05]

A revert of this CL (patchset #2 id:20001) has been created in
https://codereview.chromium.org/2983093003/ by mbjorge@chromium.org.

The reason for reverting is: Issued fixed upstream.