Gentoo Updates

[immolo]

Been a while however I have been doing some work in the background with MIPS on Gentoo to make the process a lot smoother.

My main project has been getting the Catalystt build system to build mipsel3 musl stage3 files which has been successful and awaiting a PR into the main project.

If you are interested in the meantime then I have a write at https://wiki.gentoo.org/wiki/Cataylst_Musl_Stages_Creation which takes you through all the steps, obviously the tuple needs to be changed here for the PS2. There is also https://wiki.gentoo.org/wiki/User:Immolo/GentooPS2 which needs a little more work before it usable but it will provide helpful hints with things I have discovered.

I may release an update to the LiveUSB using the new codebase however I'll most likely just wait until hard drive support is added to the kernel so Gentoo is actually worth using on the PS2.

[AKuHAK]

HDD support isnt that easy, so I advice to not wait for it. BTW where you post your LiveUsb images?

[immolo]

http://www.immolo.co.uk/gentoo-ps2-liveusb-alpha1.tar.xz

Not really much point rebasing it right now but I'll assess it when the need arises.

[frno7]

Ooh, that’s very nice! :-) Maybe I’m aiming a bit high with multi-buffer hard disk read/write DMA for a first implementation. Performance would be quite good though, as mentioned in frno7/linux#18 (comment).

...
modprobe hid-generic
sleep 5

#Mount USB
mount -t ext4 -o ro -L PS2Root /mnt || echo "Failed to mount root device"
sleep 30
...

Are the sleep in initramfs/init for slowing down kernel prints, to make them easier to read? Things that genuinely need delays?

[immolo]

Does that mean, that your work will never be merged upstream as considered obsolete for Linux team?

That seems to be the case, most likely, for several reasons:

* MIPS Linux kernel maintainer review capacity has been declining sharply for several years, and 10,000+ lines of code for PlayStation 2 support is apparently much too much code to consider for inclusion;

* MIPS hardware, software, and interest (in general) have been declining for several years, meaning especially software related to MIPS tend to break more often;

* vintage hardware (in general) and their related subsystems, are being purged from the Linux kernel, for the benefit of expanding and simplifying maintenance of modern hardware.

Staying on a stable kernel version (5.4, for instance) indefinitely has some advantages. So, why not? :-)

That's huge shame they won't as it makes getting the PS2 an official Gentoo port a harder sell for me :P That said it's pretty easy to workaround old versions when using Musl as a libc.

[frno7]

That's huge shame they won't as it makes getting the PS2 an official Gentoo port a harder sell for me :P

The -mfix-r5900 option was a very good step forward, as it means we’ve got official Gentoo/MIPS compatibility. Gentoo supplies downloads for hardware in even more dire situations than the PlayStation 2, I should think. For instance HP-PA and IA-64?

That said it's pretty easy to workaround old versions when using Musl as a libc.

Hmm, work around old kernel versions? Old Gentoo versions? Please elaborate. :-)

[immolo]

It's not impossible but from experience they aren't as keen.

Old kernel versions on latest Gentoo, I've done a lot of work on this topic in the past and even have a guide for going back as far realistically usable (2.6.39 btw).

https://wiki.gentoo.org/wiki/User:Immolo/OldKernelInstall if you want to see how it is done.

[frno7]

It's not impossible but from experience they aren't as keen.

That’s understandable. Would people want to upgrade regularly to recent kernels, if the PlayStation 2 port actually was officially merged? Security fixes exempted, upgrading doesn’t seem very compelling, since newer kernels often mean loss of features, lower performance, and upgrade breakages, compared to older ones? Has the kernel had any new features in, say, the last few years, that would make upgrading meaningful?

These questions may be difficult to fully answer, but to me, playing with retro/hobby hardware is much more worthwhile than struggling with kernel deprecation policies. :-)

Old kernel versions on latest Gentoo, I've done a lot of work on this topic in the past and even have a guide for going back as far realistically usable (2.6.39 btw). https://wiki.gentoo.org/wiki/User:Immolo/OldKernelInstall if you want to see how it is done.

Heh, that kernel is 10+ years old. Sure, systemd, pam, etc. may need to be replaced.

[immolo]

It's not impossible but from experience they aren't as keen.

That’s understandable. Would people want to upgrade regularly to recent kernels, if the PlayStation 2 port actually was officially merged? Security fixes exempted, upgrading doesn’t seem very compelling, since newer kernels often mean loss of features, lower performance, and upgrade breakages, compared to older ones? Has the kernel had any new features in, say, the last few years, that would make upgrading meaningful?

These questions may be difficult to fully answer, but to me, playing with retro/hobby hardware is much more worthwhile than struggling with kernel deprecation policies. :-)

Judging by the Wii being stuck on 3.10.5 I would assume not, I think the last change was around 4.14 where some core applications started relying on a feature although GLIBC can be a pain rasing the minimum kernel version something. I would have to look deeper though as these are just off my head.

Old kernel versions on latest Gentoo, I've done a lot of work on this topic in the past and even have a guide for going back as far realistically usable (2.6.39 btw). https://wiki.gentoo.org/wiki/User:Immolo/OldKernelInstall if you want to see how it is done.

Heh, that kernel is 10+ years old. Sure, systemd, pam, etc. may need to be replaced.

Had to make it fun somehow :)