[gPXE] support for 8086 10D3 NIC?

Stefan Hajnoczi stefanha at gmail.com
Tue May 4 16:27:23 EDT 2010


On Tue, May 4, 2010 at 8:56 PM, Barker, Kevin <kevin.barker at intel.com> wrote:

Please Reply-All so the gPXE mailing list stays CCed and others can help.

> With the latest source I am not able to boot the image. I am able to flash the gpxe image and it downloads my NW image, bot it hangs for a minute there and reboots.

What does "NW" mean in this context?

> I tried with both latest source and the 3-17 commit. The official 1.0.0 release works fine (tried both on other network cards).
> Any Ideas? I am booting a .c32 image (elf??). I think its made using syslinux.

You tried latest source versus 1.0.0 on other cards and latest source
hung but 1.0.0 booted successfully?  In that case this .c32 issue is
probably unrelated to the NIC driver.

Which .c32 from syslinux are you booting?  With this information it
might be possible to reproduce the hang.  I checked COMBOOT changes in
gPXE since 1.0.0 and only see this one (it looks fairly innocent):

http://git.etherboot.org/?p=gpxe.git;a=commitdiff;h=12defeca99fcb522f295ba2b906f25a5a0012958

If you have time to do a few more tests then git-bisect(1) could be
used to pinpoint the commit that introduced the hang.  git-bisect
automates the process of testing commits in a binary search fashion in
order to spot which commit broke something.  It goes like this:

git clone git://git.etherboot.org/scm/gpxe.git
cd gpxe
git bisect start HEAD v1.0.0   # bad=HEAD good=v1.0.0

make # do your normal build and test to see if the hang occurs
git bisect bad # if the hang still occurs or 'good' if the hang is gone
# repeat building and telling git bisect the outcome of your test
until no more commits are left

Stefan


More information about the gPXE mailing list