[lldb-dev] [RFC] [PATCH] Self Debugging on Linux

[Todd Fiala]

Hey Russell,

-1 for Todd.  No, sorry I haven't.  We've had a larger than normal surge of
activity over here.

But - thanks for prodding.  I will make a point of walking through them as
soon as I can carve out some time.  (And don't be bashful about poking
again if I seem to have stalled).

I'm OOO tomorrow (working but off site) so it might be Friday before I get
a chance.

On Wed, Oct 8, 2014 at 4:15 PM, Russell Harmon <eatnumber1 at google.com>
wrote:

> Hey Todd,
>
> Did you get a chance to look at the patches?
>
> Thanks,
> Russ
>
> On Mon Sep 29 2014 at 9:12:53 AM Todd Fiala <tfiala at google.com> wrote:
>
>> Hey Russell,
>>
>> I'll have a look at those patches sometime over the next couple days and
>> get back here.
>>
>> Thanks for sharing them!
>>
>> Sincerely,
>> Todd Fiala
>>
>> On Fri, Sep 26, 2014 at 9:56 AM, <eatnumber1 at google.com> wrote:
>>
>>> Hey everyone,
>>>
>>> I've got some preliminary but functioning work in place in order to
>>> support
>>> self-debugging. Immediately forthcoming is two patches, summarized
>>> below, which
>>> implement basic support for self-debugging on Linux. I've leveraged the
>>> ProcessMonitor architecture on Linux in order to do this, so the first
>>> patch is
>>> a refactor of ProcessMonitor to allow me to create a different
>>> implementation of
>>> it. The second patch actually fills in the self-debugging implementation.
>>>
>>> [PATCH 1/2] Refactor ProcessMonitor into an abstract base class.
>>> [PATCH 2/2] Support debugging yourself on Linux.
>>>
>>> There's still some issues:
>>>  * I needed to disable JITLoaderGDB [1] in order to prevent lldb from
>>> attempting
>>>    to set a breakpoint at attach. I'm also not clear on what
>>> JITLoaderGDB does
>>>    even after perusing the code, so I'm not sure if disabling it is the
>>> right
>>>    thing to do.
>>>  * I needed to disable LLDB copying it's stdin to it's inferior's [1] as
>>> it
>>>    would result in multiple threads reading from stdin - making it
>>> impossible to
>>>    use the prompt. This behavior makes sense to me if the debugee is
>>> launched
>>>    from LLDB, but I'm not clear on why LLDB attempts to do it when you've
>>>    attached to your debugee.
>>>  * "thread list" returns nothing. I think there's a larger problem here.
>>> Needs
>>>    investigation.
>>>
>>> Thanks,
>>> Russ Harmon
>>>
>>> [1]: https://gist.github.com/anonymous/188533a776226257206f
>>>
>> _______________________________________________
>>> lldb-dev mailing list
>>> lldb-dev at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-dev
>>>
>>
>>
>>
>> --
>> Todd Fiala | Software Engineer | tfiala at google.com | 650-943-3180
>>
>


-- 
Todd Fiala | Software Engineer | tfiala at google.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-dev/attachments/20141008/030929f2/attachment.html>