CVS head bash problem

[Rolf Campbell]

Christopher Faylor wrote:
> On Wed, Apr 23, 2003 at 12:03:57AM -0400, Pierre A. Humblet wrote:
> 
>>At 11:29 PM 4/22/2003 -0400, Christopher Faylor wrote:
>>
>>>A few debugging possibilities spring to mind here.
>>>
>>>Try running it under strace with a very large buffer size in (the -b
>>>option).  Attach to the hung process in strace or gdb.
>>>
>>>Does it also have problems with cygwin snapshots?  If not, then we need
>>>to see how you've configured your version of cygwin, what version of gcc
>>>you use, etc.
>>>
>>>If this is just a cvs thing (and that isn't really clear) then the old
>>>popular binary search through cvs might be useful to find out what
>>>change caused the problem.
>>>
>>>Also, send cygcheck output (as an attachment).  I suspect that you may
>>>be using CYGWIN=tty but we don't know what your setup is.
>>
>>I had sent a cygcheck output immediately after my initial report.
>>CYGWIN is unset.
>>
>>Contrary to Rolf I have never seen a hung process in this context.
> It just occurred to me that this is probably the old bash "reuse the
> pid" problem.  Bash is confused when pid numbers are reused in quick
> succession, which can be the case on Windows.  I thought that Corinna
> had turned on something in bash to work around this problem.  I wonder
> if it is no longer on in recent versions of bash or if the workaround
> doesn't work.
> 
> You can prove if this is the error or not by defining SLOW_PID_REUSE
> when compiling fork.cc.
That seems to have fixed the problem.  I tried 4950 commands, all of 
them worked (approx 1% failed before).  I will continue to use this 
build for a couple of days of 'real' work, and see if this problem stays 
away.

-Rolf



--
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/