Fix the GCC warning -Werror=shift-negative-value

[dm0-]

I am cross-compiling from Fedora 24 x86_64 to GNU Hurd i686 with GCC 6.1.0. This compiler warning causes the build to fail.

[anttikantee]

Thanks! Please add yourself to AUTHORS.

[dm0-]

@anttikantee Thanks for applying this fix. A similar change had to be made to the NetBSD source. The specific revisions are mentioned here:

https://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=51371

Would it be possible to update the src-netbsd repo with those changes as well?

[anttikantee]

Thanks, I'll have a look soon. Sometimes it's not entirely trivial to pull up individual fixes, since the update mechanism uses a timestamp-based method, and the files may have changed in other ways too, so one would have to do a blanket timestamp update. While a blanket timestamp update might not be a bad idea, testing that there are no regressions takes time.

Meanwhile, you might be able to get away with a "broken" src-netbsd with your compiler by using -V NOGCCERROR=1.

[anttikantee]

hmm, apparently pull reqs can't be re-opened once they're merged.

I'll try to have a look at this tonight. If I don't get around to it, please open another issue.

[anttikantee]

Ok, seems like everything is in order for updating src-netbsd. I'm planning to change the timestamp to 20160728 1100UTC, which should contain your fixes.

@dm0- could you do a favour and test that rumprun and rumpctrl build with your compiler? You need to fetch the current NetBSD sources manually and use the -s flag to the build-rr.sh and buildnb.sh scripts for the aforementioned repos.

[dm0-]

Sorry, but I have not yet been able to successfully build rumpctrl or rumprun with any version of the NetBSD sources using the Hurd cross-compiler. So far, I've only looked into building the pci-userspace libraries, which worked with a 20160728 1100UTC CVS checkout (after replacing mig with ${MIG:Umig} in Makefile.inc so I could specify a host-prefixed version).

I don't know when I'll have enough time to track down all the build failures I was seeing in rumpctrl and rumprun, but in case you or anyone else wants to look into it first, the cross-compiler I'm using is from my build scripts at https://github.com/dm0-/gnuxc. GNU Guix also has GCC 6.1.0 and a Hurd version, so perhaps the issue could be reproduced there as well, though I haven't tried it.

[anttikantee]

Ah, I didn't know you were building for Hurd. While rumpctrl might some day be relevant for Hurd, it's hard to imagine that Rumprun would ever be. Anyway, thanks, I'll update the sources when I have a few hours of contiguous time; experience has shown that even with everything working in local tests, all hell breaks loose when I push a src-netbsd update to github :-/

Feel free to submit a PR for your ${MIG} change to pci-userspace. I wonder if it should be called something like HURD_MIG, though.

[anttikantee]

Oh, heh, I was just about to close this window, and noticed what you wrote at the top ;-)
(reading is very difficult)

[anttikantee]

Ok your patches are now part of src-netbsd. Thanks!