Repackaged Patches for QNX and install note

Werner Koch wk at
Thu Mar 2 08:14:03 CET 2000

On Wed, 1 Mar 2000, Sam Roberts wrote:

> These are the same QNX patches as previous, but in a format that is
> understandable to patch, and a not on how to apply them.

Thanks for these pacthes.  I have applied them and they should show up
on CVS soon.

>   QNX doesn't have the capability of paging *any* memory to disk, so
>   it's safe to make lock_pool() a null op, *all* memory is locked in RAM.
>   Should I do this, or do I misunderstand what mlock() does (I don't
>   think I do)?

Does that mean you have other ways to control whether QNX pages or
not.  Reasonable for a Realtime OS I guess.

There is also the mlockall() but this locks the whole memory and not
only the areas for which it is required.  It is much more friendly to 
do the OS to lock only a small amount of memory.

>   cipher/rndunix.c: Interestingly, there are already QNX ifdefs in this file,
>   but it's not complete. This is.

It changed a lot in the original code by Peter Gutman and may have
introduced the need for the other header files.

>   config.guess and config.sub:
>   For some reason, after close to 20 years, QNX still hasn't made it
>   into these "standard" GNU autoconf scripts. I've also posted these diffs
>   to bug-gnu-utils (the suggested address in the autoconf docs) so maybe
>   they'll be in the next autoconf, I'm getting tired of hacking them into
>   every GNU package I compile.

I applied these pacthes.  It may happen that they get overwritten when
I apply new ones from the GNU repository - I always try to keep my
private changes in GnuPG.  If you patches did make it into the GNU
archives next time I update config.guess - I will send the diffs again
to maintainers.


More information about the Gnupg-devel mailing list