> Date: Sat, 13 Nov 2021 15:45:19 -0700 > From: Ted Bullock <[email protected]> > > On 2021-11-04 4:04 p.m., Ted Bullock wrote: > > On 2021-11-02 5:36 p.m., Ted Bullock wrote: > >> Reporting an issue that is causing my SunBlade 100 (ultrasparc64) > >> workstation to be unable too boot. > > > > I've identified the malfunctioning component that is causing installs > > after 6.7 to fail for this system, although I'm not certain yet which > > patch broke things for this system. > I've done further testing on ofwboot. The version in 7.0 and snapshots > is very broken, and appears to entirely mangle the root filesystem of > sparc64 machines that use wd(4) boot devices to the point that it's > necessary to reinstall entirely between testing attempts. (I've had a > friend test on an ultra 10 and encountered the same fault) > > The current workaround I am using is to install and before the first > boot, manually copy a working version of ofwboot from the 6.7 release. > Note that ofwboot from 6.8 and 6.9, while broken, don't seem to mangle > the filesystem entirely like the version in snapshots does. > > I gather from the lack of response to the bug reports that the platform > and impacted systems are low priority, so I don't want to impose too > much on the devs here. I would like to bisect the various patches to > ofwboot to find the fault but I need some guidance on how to build the > program (in sys/arch/sparc64/stand) since my first several attempts > don't seem to have produced a working binary. > > Thanks guys
I suspect that ofdev.c rev 1.30 is what breaks things. So maybe try a version from before 2020/05/26 and one after and see if things break. If you check out the source tree with -D2020/05/25 and do: # cd /usr/src/sys/arch/sparc64/stand # make obj # make you should end up with an ofwboot binary in /usr/src/sys/arch/sparc64/stand/ofwboot/obj that you can test.
