Review of Arjan's "fast boot" patches for 2.6.28
Scott James Remnant wrote:
> Summary: STAY AWAY!
> I merged Arjan's fast boot patches from
> ssh://master.kernel.org/pub/scm/linux/kernel/git/arjan/linux-2.6-async.git into the Jaunty kernel tree and rebuilt.
> The effect was that my root filesystem failed to mount unless I boot
> without "quiet".
> At a rough stabbing guess, it finishes the kernel init when the PCI
> probe completes without waiting for the SCSI probe or something -- so if
> the kernel boots particularly fast, it won't have found your root disk.
> So, ironically, these patches make an initramfs necessary (I wasn't
> using one).
> He may have fixed this in the 2.6.29 patch set, I'll test that for
> We should probably ask him about this on Friday anyway.
Architecturally, Arjan's approach makes sense. The asynchronicity of
this startup infrastructure is exposing faults in the underlying device
drivers and subsystems. I expect it will take a couple of releases
before all of the kinks are worked out. Perhaps by 2.6.30 or .31 ...
Tim Gardner firstname.lastname@example.org
kernel-team mailing list