Re: [PATCH v3] [gdb/testsuite] test a function call by hand from pretty printer

[Bruno Larsen <blarsen@redhat.com>]

On 3/21/22 17:21, Simon Marchi wrote:
> On 2022-03-14 16:39, Bruno Larsen via Gdb-patches wrote:
>> The test case added here is testing the bug gdb/28856, where calling a
>> function by hand from a pretty printer makes GDB crash. There are 6
>> mechanisms to trigger this crash in the current test, using the commands
>> backtrace, up, down, finish, step and continue. Since the failure happens
>> because of use-after-free (more details below) the tests will always
>> have a chance of passing through sheer luck, but anecdotally they seem
>> to fail all of the time.
>>
>> The reason GDB is crashing is a use-after-free problem. The above
>> mentioned functions save a pointer to the current frame's information,
>> then calls the pretty printer, and uses the saved pointer for different
>> reasons, depending on the function. The issue happens because
>> call_function_by_hand needs to reset the obstack to get the current
>> frame, invalidating the saved pointer.
>> ---
>>
>> Changes from v2:
>>      * Make .c follow GDB's coding style 2 - electric boogaloo
>>      * .exp tests outputs, not just if GDB breaks or not
>>      * added copyright header to .py file
>>
>> Changes from v1:
>>      * make .c follow GDB's coding guidelines more closely
>>      * Documented .exp file better
>>      * renamed and refactored TCL proc to make it less confusing
> 
> Hi,
> 
> Having the test without the fix is annoying, because it crashes GDB on
> an ASan-enabled build, and causes failures that did not exist before,
> and the setup_kfails don't help for that.  This is something we
> generally don't want, as we try to strictly progress towards less
> failures.
> 
> Does the test actually test anything right now, given that everything is
> broken?  If not, I'd suggest attaching the test case to the bug, and it
> can be merged along with the fix.
> 
> 
> Simon
> 

Hi

Oh sorry, I didn't think about that. My thinking was that this looked like a pretty complex issue, so I could at least get some sort conversation going in case anyone knew how to fix the problem, and could help me find the origin of the problems. It doesn't test anything new that doesn't fail, but unfortunately, I had already pushed the change when your email arrived. If you want to remove it until I have a fix, I understand it. Sorry again.

-- 
Cheers!
Bruno Larsen