Current-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: gdbm-1.11 weirdness in gdbmtool.o

On Thu, Jan 02, 2014 at 08:50:16PM +0100, Rhialto wrote:
> On Thu 02 Jan 2014 at 20:22:02 +0100, Thomas Klausner wrote:
> > Can someone please try building databases/gdbm version 1.11 (just
> > update Makefile.common from 1.10 to 1.11, 'make mdi', 'make package')?
> > 
> > For me on NetBSD-6.99.28/amd64, with gcc I see a nearly unkillable
> > 100% CPU gcc process, while with clang I just get one 100% CPU process
> > (both while compiling gdbmtool.o).
> A data point:
> On my laptop with amd64-6.1 userland, and pkgsrc compiled from the
> latest branch (built inside a 6.0.1 pkg_comp chroot), I see no such issues.
> Clang produces some warnings that gcc doesn't, though.

I tried again, with clang. When I let it run longer, I have system
hangs. top reports (when the hang is over) that system uses 127% CPU
before it goes down to normal values.

The clang process sits at around 114MB, the system process around
96MB, both are stable memory-wise. Swap use doesn't change (32GB
RAM, a few Gig swap used, but doesn't change over the run).

My NetBSD is 6.99.28 from Jan 4.

What can I look at next?

Home | Main Index | Thread Index | Old Index