[uClibc] Re: shm funkiness

Manuel Novoa III mjn3 at codepoet.org
Fri Aug 1 04:30:31 UTC 2003


On Thu, Jul 31, 2003 at 09:11:47PM -0700, David Wuertele wrote:
> Once again it is Manuel to the rescue!  Thanks man!

No problem.  Thanks for finding it and supplying a test app.

> Is this something I can fix myself?  If it is something basic I'll do
> it right away.

I mentioned it to Erik and he was going to look at it.  I'd expect it
to be fixed in cvs pretty quickly.  Actually for some archs, it should
be correct.

But, if you want to get a fix in now, all you need to do is supply an
an appropriate libc/sysdeps/linux/{arch}/bits/shm.h file, with the struct
definitions matching that of the kernel.  Just remember that the kernel's
idea of the size of pid_t, etc. isn't necessarily what uClibc uses.

Ideally we need at some point to write some tests to check the offsets
and sizes of various fields in both uClibc and the kernel, and flag any
differences.  That's assuming of course that we don't switch over to the
glibc approach.

Manuel



More information about the uClibc mailing list