[Lldb-commits] [PATCH]race condition calling PushProcessIOHandler

Zachary Turner zturner at google.com
Tue Jul 29 10:53:02 PDT 2014

Even better would be an std::condition_variable

On Mon, Jul 28, 2014 at 10:30 PM, Matthew Gardiner <mg11 at csr.com> wrote:

> Hi Shawn,
> I use 64-bit linux and I see this issue a lot. It usually manifests itself
> as the prompt just not being printed (or perhaps it just gets overwritten)
> - regardless - I invoke a command, and I don't see an (lldb) prompt when I
> should. So I'm well pleased that you are looking at this!
> Would it not be more robust to use a semaphore than usleep to synchronise
> the problematic threads?
> Although I've not looked too deeply into this particular issue, whenever
> I've seen similar races, I found that it's almost impossible to pick the
> right value when using a sleep command. A semaphore, though, should always
> ensure the waiting thread will wake precisely.
> I'd be happy to help to test such a fix.
> Matt
> Shawn Best wrote:
>> Hi,
>> I have attached a patch which addresses 3 related race conditions that
>> cause the command line (lldb) prompt to get displayed inappropriately and
>> make it appear it is not working correctly.  This issue can be seen on
>> linux and FreeBSD.  I can also artificailly induce the problem on OSX.
>> The issue happens when the command handler (in the main thread) issues a
>> command such as run, step or continue.  After the command finishes
>> initiating its action, it returns up the call stack and goes back into the
>> main command loop waiting for user input.  Simultaneously, as the inferior
>> process starts up, the MonitorChildProcess thread picks up the change and
>> posts to the PrivateEvent thread.  HandePrivateEvent() then calls
>> PushProcessIOHandler() which will disable the command IO handler and give
>> the inferior control of the TTY.  To observe this on OSX, put a
>> usleep(100);
>> immediately prior the PushProcessIOHandler() in HandlePrivateEvent.
>> My proposed solution is that after a 'run', 'step', or 'continue'
>> command, insert a synchronization point and wait until HandlePrivateEvent
>> knows the inferior process is running and has pushed the IO handler.  One
>> context switch (<100us) is usually all the time it takes on my machine.  As
>> an additional safety, I have a timeout (currently 1ms) so it will never
>> hang the main thread.
>> Any thoughts, or suggestions would be appreciated.
>> Regards,
>> Shawn.
>> To report this email as spam click here <https://www.mailcontrol.com/
>> sr/MZbqvYs5QwJvpeaetUwhCQ==>.
>> _______________________________________________
>> lldb-commits mailing list
>> lldb-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
> Member of the CSR plc group of companies. CSR plc registered in England
> and Wales, registered number 4187346, registered office Churchill House,
> Cambridge Business Park, Cowley Road, Cambridge, CB4 0WZ, United Kingdom
> More information can be found at www.csr.com. Keep up to date with CSR on
> our technical blog, www.csr.com/blog, CSR people blog, www.csr.com/people,
> YouTube, www.youtube.com/user/CSRplc, Facebook,
> www.facebook.com/pages/CSR/191038434253534, or follow us on Twitter at
> www.twitter.com/CSR_plc.
> New for 2014, you can now access the wide range of products powered by
> aptX at www.aptx.com.
> _______________________________________________
> lldb-commits mailing list
> lldb-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20140729/1d85cfcc/attachment.html>

More information about the lldb-commits mailing list