Hi Adrian,

On 2/16/23 3:07 AM, John Paul Adrian Glaubitz wrote:
> Hi Stan!
> 
> On Wed, 2023-02-15 at 19:44 -0700, Stan Johnson wrote:
>> Going from 15 min to about 4 min seems worth the effort on old hardware.
>> As always, YMMV. Developers who use QEMU or other emulators likely don't
>> always realize how long it takes to boot real hardware.
> 
> I am not sure what makes you think we aren't aware of the long boot times,
> I just booted Debian unstable on my Amiga 4000/060 yesterday:
> 

Yes, I know that you are aware of the slowness. However, based on my
results (see below), I doubt that any of the systemd developers have
ever attempted to use systemd on real Mac 68030 systems. And yes, I know
that they have to focus their efforts on more modern hardware, etc.

I don't knnow anything about Amiga systems, but I should probably learn
more. If your Amiga 4000/060 has a 68060 processor and runs at 50 MHz, I
would expect it should be a good deal faster than a Mac IIfx (68030, 40
MHz, 128 MiB). As I recall from a few years ago, I was only barely able
to run a Centris 650 (68040, 25 MHz, 136 MiB) with the default Debian
SID and systemd at that time.

Here are the boot times for the systemd version of Debian SID (updated
today):

Default Debian kernel (vmlinux-6.1.0-5-m68k, initrd-6.1.0-5-m68k)
Default Debian systemd configuration
Boot times (ABC... to login prompt):
  QEMU (7.2)  --  34 sec
  Mac IIfx    --  couldn't reach multiuser mode after more than 17 min.
                  (see attached serial console log)


> root@elgar:~> systemd-analyze
> Startup finished in 1min 30.785s (kernel) + 5min 31.586s (userspace) = 7min 
> 2.371s
> graphical.target reached after 5min 26.053s in userspace
> root@elgar:~>

In QEMU, I see this:

# systemd-analyze
Startup finished in 9.544s (kernel) + 24.815s (userspace) = 34.359s
graphical.target reached after 24.643s in userspace.

The Mac IIfx couldn't reach multiuser mode, so my attempt to run
systemd-analyze there failed:

"Give root password for maintenance
(or press Control-D to continue): Reloading system manager configuration
Failed to reload daemon: Refusing to reload, not enough space available
on /run/systemd. Currently, 11.5M are free, but a safety buffer of 16.0M
is enforced."


> 
> I understand your frustration, but please keep in mind that modern software
> is more complex and therefore runs slower on older hardware.
> 

Yes, of course I expect slowness on old hardware. Currently, however,
the Debian kernel itself does not boot at all on an SE/30 (likely an
issue with keys), and the default Debian SID configuration with systemd
does not boot to multiuser mode even on a Mac IIfx with 128 MiB, which
implies that it won't boot on any 68030 Mac (all the 68030 Macs that
I've been able to test can boot with a stripped down kernel and
sysvinit, and the faster ones can use Debian's default kernel).

> We will certainly work on providing a leaner kernel in the future to help
> alleviate this issue. And rebootstrapping the whole distribution with 32-bit
> alignment (-malign-int) should improve code execution as well.
> 

ok, thanks for your help. Note that it's not just the kernel, but also
the choice of an init system. I'm fully aware that Debian developers
settled on systemd years ago, so your options may be limited unless you
are able to look at sysvinit. I suspect that the developers who settled
on systemd all those years ago never tried to boot on real Mac 68030
hardware, or maybe they had the very rare 50 MHz Dash 30fx:
"https://lowendmac.com/2016/the-68000-dash-30fx-an-accelerate-mac-iifx/";

-Stan

Attachment: maciifx_02162023.txt.xz
Description: Binary data

Reply via email to