Redhatter (VK4MSL)

Hi all,

The O32 stage builds for little-endian MIPS systems are now complete, and you will find the stages on my devspace for now at http://dev.gentoo.org/~redhatter/mips/cobalt/stages/.  There you will find stages for:

• MIPS-I: Suitable for all little-endian MIPS hardware
• MIPS-III: Best for Loongson-based hardware
• MIPS-IV: Best for Cobalt Qube2 and RaQ2… Possibly the earlier Qube/RaQ as well.

Note that for the latter two, there are also N32 stages in the pipeline.  N32 overcomes a lot of shortcomings of the O32 ABI and offers better performance overall.  For many years, Gentoo/MIPS has primarily used O32 as the default ABI, with N32 being considered “experimental”.  Zhang Le and others have done a lot of work in the field, and now many applications run fine on N32.  Gentoo will therefore be transitioning over to this newer ABI, leaving O32 available for legacy system support and for experimental targets.

And yes, I’ll have to get to and update the documentation.

Some erratum notes however… These stages all feature a binutils patched for Loongson 2F support.  In the case of the MIPS-I and MIPS-III stages, these were also built with the -Wa,-mfix-loongson2f-nop set in CFLAGS so that these stages will work on Lemote hardware.  (And I needed to do that, otherwise we’d be waiting until Christmas for the Qube2 to finish building.)  So Loongson users should find these stages will work without further patching.  MIPS-IV was not built with the changes to CFLAGS, however these stages will also not run due to incompatibilities with ABI.

Secondly, the issue with /dev still exists… I’m not sure where the bug is, I’m still investigating this.  The following run inside the chroot should correct this issue:

# rm /dev/null
# mknod /dev/null 1 3
# USE=build emerge --oneshot makedev

I’d be naïve to think it probably won’t happen on the big-endian stages… we shall see. Stage 1 for MIPS-I big-endian is already up, with stage 2 well underway. The O2 is just a little quicker than the Qube2 in that regard. 😉 I hope to have big-endian stages up by November.

Hi all…

After a long wait, Gentoo/MIPS is on the comeback trail. Okay, I’ll stick my hand up and admit I did a bit of carnage by pulling keywords on vim-7.0 (which might’ve worked with older releases but did not work on this latest release) however we’ll soon have that cleaned up. We’re still in discussions whether to drop the stable keywords on the packages that depended on vim (probably a wise move considering the vim keywords I pulled were stable ones), or whether to promote vim to stable early (probably also fairly safe given the circumstances).

In the meantime, my builds have plodded on. Gentoo/MIPS O32 for MIPS-I little-endian is now complete, with the stage 3 due to finish uploading in 5 hours time… after which, the stage 2 for the MIPS-III little-endian port will begin its upload. Progress can be watched here if people so desire… the upload is throttled at 8KB/sec so that I have some bandwidth left-over for other tasks (it’s a 128Kbps uplink).

MIPS-IV little-endian builds are going slowly… with a snag hit when the Qube2 couldn’t download an amended patchset for binutils (basically patchset 1.1 + patches from bug #338405) because muggins forgot to copy across the envscript specifying my local GENTOO_MIRRORS.

The O2 is also happily compiling a seed stage, for what will be a round of stage builds for big-endian systems… no issues so far and stage builds will commence there in earnest shortly.

It is hoped that we will soon be offering N32 stages to compliment this release very soon. N32 will likely be the future direction for the Gentoo/MIPS port, with O32 being provided for legacy installations and for hardware unable to run 64-bit kernels reliably (or at all).

From here, the next step will be uClibc stages and installation media for the various platforms we support. This should include the newer Lemote systems which have previously enjoyed semi-official support in the past.

Hi all…

Just to let you know, there will be new stages to replace the ones I uploaded recently. Having found a solution for the issues with building on the Lemote computers, I now have MIPS-I and MIPS-III builds going simultaneously on the two Fulong systems. MIPS-IV is being handled by my much slower Qube2. One of the hold-ups was a parallel-build with gcc-4.4.4 (and I note that distcc doesn’t seem to work under Catalyst anyway).

I hope to get big-endian ports bootstrapped soon, with my O2 up and running.

Hi all…

A little note regarding the usage of the latest MIPS-I stages on Fuloong 2E… I found I had a hard-lock compiling some applications which can be possibly traced back to an issue regarding the NOP instruction on the Loongson CPU.  I understood it to only affect 2F-series systems, and also assumed latest binutils had the patch, it seems I was wrong on both those fronts.

A workaround is described on Lemote’s site… compile the fix-nop.c then use it to patch up ld and as… then builds should go smoothly. Now that I know about this issue, I can use the faster Loongson2E machines to build the stages, which means I’ll address the other issues I’ve found and roll out a new set of stages, rather than waiting for the slow Qube2 to chug its way through the build queue.

Hi all,

Well, after a long plod the stage 3 for MIPS-I little-endian systems is uploaded. I’m working on MIPS-III and MIPS-IV stages, as well as equivalents for big-endian systems.

Two things with this stage 3 tarball. First and foremost, /dev seems to be empty. I think this is most definitely a bug, and I’ll try to rectify it, but in the meantime, just merge sys-apps/makedev. That should clear everything up.

The second is that somehow dev-vcs/git got pulled in. I guess this is a happy accident since many will probably want it for fetching kernel sources. I’m undecided whether to leave it there or not… I probably should see why it got in there and fix it as that’s probably related to makedev‘s absence. I don’t see it in the profile, so it’s a mystery why it wound up in the system. I’ll do some further checks, but for the most part, it’s harmless.

I’m just rebuilding one of the Fulong2E’s with it to check it out… and as always, feedback is appreciated.

Observant viewers may notice that I’ve begun uploading some of the stage tarballs. The first of the stages built is the stage 1 tarball for MIPS-I little-endian. Stage 2 is building as I type this, hopefully by this weekend it’ll be uploaded and stage 3 will be on its way.

I’ll let you know when all the MIPS-I little-endian builds are done, that way you won’t wind up downloading a partially uploaded file. That said, should you jump the gun, you should be able to resume the download to fetch the remainder. I’ve started with MIPS-I since that’s the lowest-common-denominator … MIPS-III will be next, followed by MIPS-IV.

I plan to get onto the big-endian port shortly. SGI owners have not been forgotten, just I’ve been busy. 🙂

Well, after a long hiatus and a few battles more recently, I now have the Qube2 building O32 stages for the little-endian port of Gentoo.

N32 will probably come soon with Matt Turner throwing some Broadcom BCM91250 heavyweights at the task. I intend to keep O32 going as there’s a lot of hardware out there that cannot use N32 that people use Gentoo on, and there are also a lot of systems which still run Gentoo O32. To provide a continued operating system for these systems, I’ll continue to maintain an O32 port.

At this point… I have produced a seed stage 3 based on modern sources, and have the Qube2 busily working through the final stage 1 for MIPS1 now (at package 55 of 60). I have the occasional hiccup such as baselayout trying to install a .keep file in a mounted /proc and dying… I’ve also had the man ebuild try to call groupadd before shadow was installed.

Other than that, things are moving forward, and I hope to have new stages to you shortly.