Re: make -j hangs? anyone else?

[Christopher Faylor <cgf at redhat dot com>]

On Tue, Nov 14, 2000 at 11:17:16PM -0600, Chris Abbey wrote:
>At 21:06 11/14/00 -0800, Earnie Boyd wrote:
>>Chris has warned before not to use the job switches for make.
>
>hmm... not sure how I missed them. (well, ok the ~35 messages/day
>on this list alone could be part of it.... ;) Thanks for the 'minder
>Earnie.

Actually, I think DJ warns of this sometimes.  The problem is that
Windows (or maybe just Cygwin) lacks the ability to track load average,
so, in 1.1.4 and before you could end up running up to cygwin's pid table
size processes.  Since 1.1.5 uses NT's process limit that number grows
by quite a bit.  Something like 'make -j2' should work ok, though.

I've just (2 minutes ago) uncovered a problem that was due to the fact
that cygwin is now using win32 pids.  In windows, when you are starting
new processes that exit quickly, the pids can look like this

1234
5678
9012
1234
9012
5678

i.e., the pids repeat themselves.  If this happens with something that
is trying to keep track of pids, like bash it seems to lead to
confusion.  I don't know if make is the same way or not, though.  The
symptom in bash is, coincidentally, a "hang", although the bash process
is accumulating CPU.  I'll have to work around this somehow in cygwin.

Anyway, I don't like the sound of your problem.  It shouldn't hang the
way that you are seeing it hang.  If I can duplicate the problem, I'll
see if I can fix it.

But, then once I do fix it -- don't do that!  :-)

cgf

--
Want to unsubscribe from this list?
Send a message to cygwin-unsubscribe@sourceware.cygnus.com