Re: Regression in -current

[Chavdar Ivanov <ci4ic4%gmail.com@localhost>]

Well, it turned out to be PEBKAC, I was at some stage thinking to
change the displayed font and the corresponding option was set in
GENERIC.local. For some reason this worked on the desktop system and
on the virtual machines I tested it on, but failed when ran on this
laptop. The issue with the USB stick not showing also turned out to be
connected to the stick in use - it works fine with a 8GB Sony stick,
but fails on a no-name one (which, again. works OK on the desktop
system).

It might be mildly interesting that when I use the "bad" stick to boot
and it is not found to continue the installation, some perhaps timing
issue gets me to the first trace, after about a minute or so, when I
am trying to figure out the disk spec to continue the boot:
....
breakpoint() at netbsd:breakpoint+0x5
vpanic() at netbsd:vpanic+0x173
panic() at netbsd:panic+0x3c
heartbeat() at netbsd:heartbeat+0x1ec
hardclock() at netbsd:hardclock+0x8b
Xresume_lapic_ltimer() at netbsd:Xresume_lapic_ltimer+0x1e
--- interrupt ---
x86_mwait() at netbsd:x86_mwait+0xd
acpicpu_cstate_idle() at netbsd:acpicpu_cstate_idle+0x19a
idle_loop() at netbsd:idle_loop+0x14c
cpu_hatch() at netbsd:cpu_hatch+0x180
....
and in some occasions to the second one, I don't know
....
panic: kernel diagnostic assertion "addr >= 1 && addr <= 127" failed:
file "...../src/sys/dev/usb/xhci.c", line 2948 addr 0
cpu3: begin traceback...
vpanic() at netbsd:vpanic+0x173
kern_assert() at netbsd:kern_assert+0x4b
xhci_new_device() at netbsd:xhci_new_device+0x705
uhub_explore() at netbsd:uhub_explore+0x4bd
usb_discover() at netbsd:usb_discover+0x4f
usb_event_thread() at netbsd:usb_event_thread+0x46
cpu3: End traceback...
fatal breakpoint trap in supervisor mode
trap type 1 code 0 rp 0xffffffff80235425 cs 0x8 rflags 0x202 cr2 0
ilevel 0 rsp 0xffffd48446026de0
curlwp 0xffff812f10282500 pid 0.218 lowest kstack 0xffffd484460222c0
breakpoint() at netbsd:breakpoint+0x5
.....

But otherwise there is no problem with any regression, as far as this
hardware is concerned.

Chavdar

On Wed, 19 Jul 2023 at 22:32, Robert Nestor <rnestor%mac.com@localhost> wrote:
>
> I’ve been doing some testing running NetBSD 9.x and 10.x in a VM environment and have noticed some differences in booting that might be related.   The testing I’ve done was using a fairly recent version of 10.0_BETA so similar results _might_ be present in -current as well.
>
> #
> # Install differences between 9.3 and 10.0 in a virtual machine setup
> #  installing NetBSD for an amd64. The virtual environment used is KVM
> #  in LinuxMint 21.1 runing on an S600 with an Intel i9-12900H
> #
> 9.3: (using distribution CDROM image attached as SATA CDROM)
>         BIOS boot - installs and runs fine
>         UEFI boot - installs and runs fine
> 10.0: (using distribution CDROM image - NetBSD-10.0_BETA-amd64.iso)
>         CD attached as SATA CDROM:
>             BIOS boot - installs and runs fine
>             UEFI boot - fails to boot - can't find root device
>         CD attached as USB CDROM:
>             BIOS boot - fails to boot - no BIOS boot on CD
>             UEFI boot - fails to boot - can't find netbsd image
>         CD attached as USB R/O disk:
>             BIOS boot - fails to boot - no BIOS boot on CD
>             UEFI boot - fails to boot - can't find netbsd image on CD
>         CD attached as USB R/W disk:
>             BIOS boot - fails to boot - no BIOS boot on CD
>             UEFI boot - fails to boot - can't find netbsd image on CD
> 10.0: (using installation CDROM image - NetBSD-10.0_BETA-amd64-install.img.gz)
>         CD attached as SATA CDROM:
>             BIOS boot - fails to boot - no BIOS boot on CD
>             UEFI boot - fails to boot - no UEFI boot on CD
>         CD attached as USB CDROM:
>             BIOS boot - fails to boot - no BIOS boot on CD
>             UEFI boot - complains about CD not being writable for /etc/gettytab
>         CD attached as USB RO disk:
>             BIOS boot - fails to boot - no BIOS boot on CD
>             UEFI boot - complains about not being writable for /etc/gettytab
>         CD attached as USB RW disk:
>             BIOS boot - fails to boot - no BIOS boot on CD
>             UEFI boot - boots, installs and runs
>
> So, if wanting to install 10.0 with a BIOS boot setup it's necessary to
> use the amd64 CD.  For installing with a UEFI boot setup it's necessary
> to use the amd64 install image.  I didn't test with the install image
> built for BIOS, so the above observations are not entirely complete.
>
> Notes:
> There doesn't appear to be any way to get into userconf when booting the
> installation images.  Would be nice to have that option during install.
>
> In a KVM virtual environment the keyboard and mouse aren't available until
> the kernel gets loaded, so even with an option in boot.cfg to boot up
> with the "-c" option, there's no way to use it interactively. This may be
> an issue with KVM and not NetBSD as this is the same in both 9.3 and 10.0
> and this behaviour isn't present when booting on real hardware.
>
> Because of the above it's almost impossible to get a list of the "gop"
> resolutions available.  Adding a "gop" to a boot line in /boot.cfg is an
> option, but there doesn't appear to be a way to delay booting the kernel
> after this command is executed, so the "gop" display flashes on the screen
> before being replaced with the NetBSD boot log display.
>   FYI, "gop 23" yields a 1920x1200 screen in my virtual environment.
>
>


-- 
----