Booting a new kernel after "make installkernel"

Roger Leigh rleigh at codelibre.net
Mon Apr 18 22:27:20 UTC 2016


I have built 10.3 world and then the kernel with and without the change 
below.  In both cases, I get a working system, so at least with my 
hardware, I don't think I'm tripping up on the change causing the 
regression.  I can only assume I was hitting some other issue in the 
10.0/10.1/early 10.2 releases, or getting something wrong on my side.


Kind regards,
Roger

On 14/04/2016 01:42, Bill Sorenson wrote:
> If this fixes your problem I'll file a bug report. I've been working on
> coming up with a real patch to properly fix this in my spare time.
>
> Thanks,
> Bill S.
>
> On Wed, Apr 13, 2016 at 7:37 PM, Bill Sorenson <instructionset at gmail.com
> <mailto:instructionset at gmail.com>> wrote:
>
>     Hopefully this makes sense. What i want you to do is replace
>     "sys/dev/ofw/ofw_iicbus.c" in 10.3 with the
>     "sys/dev/ofw/ofw_iicbus.c" from 10.2. There was a commit about 2
>     months ago to 10-stable and now 10.3 that I think broke the way Open
>     Firmware machines read temp sensors, so your kernel wont boot
>     because it thinks your system is overheating and resets.
>
>     If you check out 10.3 and copy the 10.2 ofw_iicbus.c over into the
>     tree replacing the newer one, I'll bet you'll build and boot fine.
>
>     If it solves your problem I'll know I'm on the right track. (I think
>     the breakage is actually in a driver somewhere, and the fix to
>     ofw_iicbus.c caused a regression).
>
>     Anyway, you should be able to just try building the kernel rather
>     than the whole system if you want to save time. Especially if your
>     10.3 userland already booted back up with your kernel.old before.
>
>     Thanks,
>     Bill S.
>
>
>     On Wed, Apr 13, 2016 at 7:23 PM, Roger Leigh <rleigh at codelibre.net
>     <mailto:rleigh at codelibre.net>> wrote:
>
>         I've retried with 10.2-releng (r297915) built from a
>         10.2-RELEASE-p0 install, and it built and rebooted just fine,
>         now on p14.  It took 14 hours, but great success!
>
>         I'd be happy to retry with 10.3 or 11 either vanilla or with any
>         patches you'd like testing.
>
>
>         Thanks,
>         Roger
>
>         On 12/04/2016 20:41, Bill Sorenson wrote:
>
>             I have had an issue with 10.3 on PowerPC that causes an
>             unbootable
>             situation. Try 10.2-releng and if that works I have
>             something for you to
>             try.
>
>             On Apr 12, 2016 2:36 PM, "Roger Leigh" <rleigh at codelibre.net
>             <mailto:rleigh at codelibre.net>
>             <mailto:rleigh at codelibre.net <mailto:rleigh at codelibre.net>>>
>             wrote:
>
>                  Hi folks,
>
>                  Sorry for the awfully naive question, but I've tried on
>             several
>                  occasions to build world using the instructions here:
>             https://www.freebsd.org/doc/en/books/handbook/makeworld.html
>
>                  They work fine on amd64, but on powerpc I'm always left
>             with an
>                  unbootable system.  It looks like it can't load or boot
>             the kernel.
>                  I've not customised anything; I've been trying to build
>             10.x-stable.
>
>                  My previous experience was on Debian which required
>             running yaboot
>                  to update the Open Firmware bootloader configuration.
>             Are there any
>                  additional steps required on FreeBSD/powerpc which
>             aren't in the
>                  generic instructions?
>
>
>
>



More information about the freebsd-ppc mailing list