PowerMac G4 booting from memstick image

[Brandon Bergren]

On Mon, Apr 19, 2021, at 8:05 PM, Mehul Sanghvi wrote:
> 
> Thank you for the explanation.  I had seen something regarding SMP on 
> the mailing list back in March
> and so thought that it might be part of the 13.0 release.

SMP timebase fixes for G4s are a WIP still currently. I will MFC them to stable/13 once I have finished them though.

https://reviews.freebsd.org/D29136

(IIRC I need to do another rebase because of some recent changes. Will try to get back to it soon.)

I am looking for someone who has a "newer" (i.e. Quicksilver) SMP G4 to assist with submitting the platform function so I can analyze it, as I only have an ancient Graphite model that uses hardcoded timebase controls.

In the meantime, on SMP G4 I recommend continuing to boot with SMP disabled until I can get all cases sorted out.

> 
> I have made USB flash drive using the 13.0 memstick image, but am 
> unable to boot from there consistently.

G4s (especially the powerbooks) can be *really* picky about which stick you're using. Of the stack of USB sticks that I own, exactly *1* consistently works. I started looking into the OpenFirmware ROM to see if I could figure out what is going on there, but so far I haven't had any ideas.

Assuming you have a stick that works, the boot command I generally use (with the iso version!) is:

boot usb1/disk at 1:2,\ppc\loader usb1/disk at 1:0

> 
> The first time, I got to the dialog box asking to choose Live CD, Shell 
> or Install.  I had waited
> A while to get to that stage, having gone through several minutes of:
> 
>    Root mount waiting for: CAM
> 
> and also
> 
>    aprobe1:ata1:0:1:0 CAM status: Command timeout 
> 
> 
> type of messages scrolling by on the screen.  

This may be caused by a descynced timebase. At the loader prompt, do "boot kern.smp.disabled=1" (or set it in the loader environment) and see if that fixes it.

> 
> Having gotten to the point of selecting Install,  the keyboard wouldn’t 
> work.  An Apple keyboard, hooked up via KVM switch, which was working 
> just fine upto that moment.  

I've seen this for some reason on ADB. I can usually work around it by using a USB keyboard. But it could be due to screwed up timers (due to timebase desync)

> I attached the keyboard directly and tried again and now I’ve got the
> 
>    mountroot> 
> 
> prompt as it couldn’t mount /dev/da0s3 the second time around.

My guess is that the timebase problems caused the controller interrupt threads to go comatose or something similar.
> 
> 
> Making a third attempt now, so lets see what happens.  I might look at 
> what Alfredo suggested and write the ISO to the USB stick and see if 
> that works rather than the memstick.img

Only the iso gets tested. memstick.img is likely going away soon, it doesn't work very well and the iso works everywhere.

> 
> 
> Otherwise will just keep running NetBSD till 13.1 comes out.
> 
> 
> cheers,
> 
>     mehul
> 
> 
> 
> > On Apr 19, 2021, at 20:46 , Mark Millard <marklmi at yahoo.com> wrote:
> > 
> > 
> > 
> > On 2021-Apr-19, at 16:34, Mehul Sanghvi <mehul.sanghvi at gmail.com> wrote:
> > 
> >> . . .
> >> 
> >> I’ll go ahead and create an image with 13.0 and see if that works.
> > 
> > 13.0-RELEASE was a major change, in that 32-bit PowrePC is finally using
> > modern llvm/clang-based toochain instead of a gcc 4.2.1 based toolchain.
> > Effectively, pre-13.0 is not as well unsupported at this point, if I
> > understand right.

Right.

> > 
> > In other words, it is unlikely that you would want to use a 12.x
> > version at all unless you have reasons to be using the old toolchain.
> > 
> >>  As I understand it 13.0 handles SMP on the G4 better, and since I have a G$ with 2 CPUs, that would be better anyway.
> > 
> > 13.0-RELEASE does not contain code changes to deal with FreeBSD's
> > PowerPC problems in that area as far as I know. There is work
> > going on in/for main in the area but some of it has not been
> > checked into main yet, if I understand right. The range of 32-bit
> > PowerMacs covered and if your hardware is (would be) I do not know.
> > 
> > There are other issues as well. For example:
> > 
> > Unfortunately, the FreeBSD 32-bit kernel that covers PowerMacs (G3/G4)
> > slowly zeros out pages of user processes that it should not touch.
> > This ends up causing some processes to crash and probably other, less
> > obvious, issues. It is not a reliable environment. To my knowledge,
> > this area is not under active development currently but the basic
> > issue of what type of thing is missing was identified some time back.
> > (The 64-bit kernel that covers G5s does not have this problem.)

I still haven't been able to reproduce this on G3/G4, but definitely concur that there is a page zeroing issue when attempting to use the 32 bit kernel on a G5 (which has a different MMU -- the kernel interacts with it via bridge mode compatibility. I've tried several times to figure out what the heck is going on but failed every time.)

By the way, I *FINALLY* figured out what is going on with moused freaking out. It ended up being an oversight in forking that lead to floating point registers getting out of sync.
My current in-testing patch is http://drop.rtk0.net/fork_rework_3.patch -- will be submitting it soon once I finish my local testing.

This issue actually affects ppc64 as well and has probably been the source of some mystery crashes when forking.

> > There is more, but the above two are fairly major items out of the
> > overall list.
> > 
> > These issues are not unique to 13.0+ as far as I know. Using 12.x
> > likely would not avoid them.

IMO 13 is in better shape than 12.x. I am currently dogfooding it on a G4 PowerBook on a daily basis and will be continuing to fix issues here and there that have been cropping up.

> >>> On Apr 19, 2021, at 19:10 , Alfredo Dal'Ava Júnior <alfredo at freebsd.org> wrote:
[snip]

-- 
  Brandon Bergren
  bdragon at FreeBSD.org