race between geom and root mount

[Poul-Henning Kamp]

In message <20040921110802.GF93460 at darkness.comp.waw.pl>, Pawel Jakub Dawidek w
rites:

>+> Yes.  We need to solve this problem correctly, and that has a fair
>+> number of issues involved.
>+>
>+> I am not going to rush in a band-aid for 5.3, the risk of breaking
>+> something is far larger than the inconvenience of not being able
>+> to use a number of new geom classes for root filesystems.
>+>
>+> When we have the correct solution, it can be MT5'ed and future 5.x
>+> releases will have it.
>
>Why can't we just use my fix temporarily, so gmirror can be fully usable
>for 5.3-RELEASE? We can always back it out and MT5 a better solution
>later.

I am not convinced that your fix will not cause other problems.

--
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.