(another wrong list resend)
Martin Lucina writes ("Re: rumprun-xen: Cleaning up the Mini-OS namespace"):
> [email protected] said:
> > Martin Lucina writes ("rumprun-xen: Cleaning up the Mini-OS namespace"):
> > > These changes also remove the old (non-app-tools based) demo from the
> > > rumprun-xen build. I've added in a simple "Hello World" demo using
> > > Ian's app-tools for testing. We can improve the demos later.
> >
> > That demo is used as a smoke test by our CI system.
>
> Which demo? What CI system? I only disabled the old rumpkern_demo.c from
> the build, not the app-tools tests/configure which is used by the Travis CI
> configuration.
By the Xen Project automated testing system, osstest.
http://sourceforge.net/p/rumpkernel/mailman/rumpkernel-builds/thread/osstest-31437-mainreport%40xen.org/#msg33016759
> > We definitely want these changes upstream.
>
> Okay, that needs a bit more work (see the other relevant threads on the
> list). Also, the rumprun-xen build makes extraneous changes to the Mini-OS
> Makefile, which we shouldn't be doing if we don't want to diverge.
Yes. It's all a bit unfortunately.
> ie. We should import the xen.git mini-os into rumprun-xen verbatim...
That would be brilliant if it can be achieved.
> My patches handle this differently - Mini-OS is linked using ld -r into
> a discrete .o file and objcopy is run on the resulting minios.o file at
> build time to make all symbols except some select prefixes local to that
> file:
Ah. That's fine then, I think.
Ian.
------------------------------------------------------------------------------
_______________________________________________
rumpkernel-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rumpkernel-users
