%% [EMAIL PROTECTED] writes:
m> Been using a "suite" of makefiles to compile a java
m> project for sometime now. When using gnumake version
m> 3.79.1 built for sparc-sun-solaris2.6, make dies with a:
m> "ld.so.1: echo: fatal: libc.so.1: open failed: Too many
m> open files"
m> error, which granted looks a lot like a shell error and
m> not necessarily a make error. When using 3.74 make,
m> thouh, the project compiles on as it has been since the
m> last time any major changes were made to the makefiles.
m> Anyone know what is different between the version 3.74
m> and 3.79 of make that could be causing this problem?
I haven't seen anything like this, so you'll have to provide more
information. It looks like a file descriptor leak. The only possible
thing I can think of that might be involved with that in make itself is
the new jobserver support in GNU make.
Are you running these builds in parallel (using make -j N)?
If so, please give a little information about how you are structuring
your recursive invocations of make. How you are invoking sub-makes, for
example (show a sample rule).
Does this happen at a top-level make, or only deep into a recursive make
sequence? How deep?
What are the commands that appear right before this one?
Are you using the "+" prefix to make command scripts a lot?
Maybe you can cobble up a little program that gives info about all its
open file descriptors, and run that at each submake to see how many FDs
you have as you recurse deeper: it should be constant after the first
level.
Other than that, I can't think of anything.
--
-------------------------------------------------------------------------------
Paul D. Smith <[EMAIL PROTECTED]> Find some GNU make tips at:
http://www.gnu.org http://www.paulandlesley.org/gmake/
"Please remain calm...I may be mad, but I am a professional." --Mad Scientist
_______________________________________________
Help-make mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/help-make
