Re: cygwin-1.7.10-1 fork - address space needed by ... already in use

[Denis Excoffier <cygwin at Denis-Excoffier dot org>]

On 2012-03-16 18:14, Christopher Faylor wrote:

> On Fri, Mar 09, 2012 at 10:48:30AM -0500, Christopher Faylor wrote:
>> On Thu, Mar 08, 2012 at 10:55:35AM +0100, Corinna Vinschen wrote:
>>> On Mar  8 09:50, Denis Excoffier wrote:
>>>> On Wed, Mar 07, 2012 at 06:47:49PM +0100, Corinna Vinschen wrote:
>>>>>> Hi Denis,
>>>>>> 
>>>>>> can you please test this again using the latest developer snapshot or
>>>>>> the current from CVS if you build Cygwin by yourself?  It provides a bit
>>>>>> more information to find the reason for the permission denied error in
>>>>>> _pinfo::dup_proc_pipe.
>>>> Thank you cgf (the committer and snapshot maker at least).
>>>>>> 
>>>>>> In theory, the user should have permissions to duplicate handles into
>>>>>> every own process, if the handle has been opened with these permissions,
>>>>>> so it's quite interesting to find the reason.
>>>>>> 
>>>> 
>>>> After about 3 hours of exercising the new snapshot (and shaking it a
>>>> little), i met the "something failed" instance only twice:
>>>> 
>>>>      1 [main] tcsh 7648! _pinfo::dup_proc_pipe: (child_info_spawn::worker) process synchronization failed for pid 7648/0x754, wr_proc_pipe 0x0 vs. 0x764: DuplicateHandle winerr 5, WFSO returned 258, Win32 error 5
>>>>    503 [main] tcsh 6148! _pinfo::dup_proc_pipe: (child_info_spawn::worker) process synchronization failed for pid 6148/0x758, wr_proc_pipe 0x0 vs. 0x768: DuplicateHandle winerr 5, WFSO returned 258, Win32 error 5
>>>> 
>>>> I continue, of course.
>>> 
>>> Thanks, I don't think it's necessary to try further.  What this shows is
>>> that the process handle returned by the call to CreateProcess sometimes,
>>> for some reason, does not allow handle duplication.  That's weird.
>> 
>> I have a vague idea about why this is happening.  I'll look into it within
>> the next 48 hours.
> 
> My vague idea about this proved to be incorrect.  I did manage to make a
> royal mess of the exec synchronization code before I figured that out
> though so at least that's something.
> 
> Denis, did you see this while running the STC in
> 
> http://cygwin.com/ml/cygwin/2012-02/msg00187.html
> 
> ?
No i didn't. This STC was another (probably independent, but much more important)
problem, solved by Corinna in February. For the moment i could not find
any reproducible scheme that leads to "process synchronization failed".
However, a full night of various compilations (using make -j 3) produces about
a dozen of such messages (using make with no -j produces no message, that's
all what i know for the moment).
> 
> Also, I checked back through the archives but didn't see any cygcheck
> output from you.  Maybe I just missed it but would you mind sending it
> here?
Included cygcheck.out and cygcheck.err (from last Tuesday).

Other problems i still have not fully reported:
- ldd produces "??? => ???" on some DLLs (/usr/bin/cygoctave-1.dll produces 6 such lines)
- strace produces "too many environment variables" always
- sometimes a process get frozen (not reproducible), if i kill it, the
  killing process also gets frozen

Denis Excoffier.

Attachment: cygcheck.out
Description: Binary data

Attachment: cygcheck.err
Description: Binary data

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple