On 01 May 2006 15:58, Brian Ford wrote:
> On Thu, 27 Apr 2006, Dave Korn wrote:
>
>> On 26 April 2006 20:34, Steven Hartland wrote:
>>
>>> Interestingly if I try to get a thread dump using
>>> sysinternals process explorer the rsync process goes
>>> mad using all available cpu.
>>
>> That's a known issue, and one which I'd like to try and work on, but
>> it's really complex. It's a side-effect of procexp doing an
>> RtlQueryProcessDebugInformation call on the cygwin process, which works
>> internally by injecting a remote thread into the cygwin process; this
>> new thread interacts badly with dll csrss in some way because it isn't
>> "lpc-registered".
>
> FWIW, I believe this behavior started about the time thread local storage
> at the base of the stack was implimented via thread creation interception
> by the cygwin1.dll.
>
> Just a WAG, but you might consider disabling that in some way to test.
> You might also consider disabling the code to force double stack alignment
> in same. Just in case you're open for leads that I don't have time to
> investigate ;-).
Thanks, that's just the sort of hint I was looking for. Will report back as
and when (if!) I figure it out.
cheers,
DaveK
--
Can't think of a witty .sigline today....
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
