linux-kernel - Re: 2.6.18-mm1 make oldconfig missed SATA

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Sun, 24 Sep 2006 20:38:14 -0400
From:	Valdis.Kletnieks@...edu
To:	Jeff Garzik <jeff@...zik.org>
Cc:	Grant Coady <gcoady.lk@...il.com>, Sam Ravnborg <sam@...nborg.org>,
	linux-kernel@...r.kernel.org
Subject: Re: 2.6.18-mm1 make oldconfig missed SATA

On Sun, 24 Sep 2006 20:12:10 EDT, Jeff Garzik said:
> Grant Coady wrote:
> > 2.6.18-mm1 make oldconfig didn't pull SATA config from 2.6.18 old screen to

> > the new libata screen, caught me out -- this may be an issue for 2.6.19 
> > upgraders that a quick make oldconfig rebuild will fail to boot?
> 
> The symbols changed.  No facility for upgrading .config symbols... 
> people who config their own kernels are expected to handle such things...

I remember getting hit with this several times in the last few -mm's as I
did bisecting and kept crossing over the patch that did that. Fortunately,
'make oldconfig' was nice enough to keep me honest and prompt me for the new
symbol names.

What the Kbuild system *could* use is, for the end of 'make oldconfig', a
report like this for 'y' or 'm' symbols that have evaporated:

The following enabled symbols found in the .config were not defined in any Kconfig file:
CONFIG_FROOBY
CONFIG_DEBUG_FROOBY

So people aren't totally mystified   Admittedly, I've only gotten bit by
silently dissapearing symbols 4-5 times since 2.5.55 or so, but the times
it happened it would have been nice to know....

(OK, I admit it - at least once of those 4-5 times it was because I ran
'make oldconfig' and I hadn't applied a local patch that added a Kconfig option.
Even in that case, a "Yo, dood, something went poof" msg would have saved me
a lot of bang-head-against-wall debugging. :)

Content of type "application/pgp-signature" skipped