[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [PATCH] Fix condition for adding IDE or SCSI disk driver when MODULES=dep



On Sun, 12 Dec 2010, Ben Hutchings wrote:

> The transition from old-style IDE to libata-based drivers means the
> new kernel may need different drivers from the running system.  We
> already handle the change of controller driver correctly since we find
> driver modules via module aliases.  However for the disk driver
> (sd_mod or ide-disk) we look at which types of devices exist in the
> running system, so we may not select the one we need.
> 
> Since all the controller driver modules depend on the core modules
> under drivers/ide or drivers/scsi, we can work out which disk
> driver(s) may be needed by checking whether those module
> subdirectories have been created under the target directory.
> 
> Reported-by: Gordon Farquharson <gordonfarquharson@gmail.com>
> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

Thank you a lot for the excellent bug hunt.

> ---
> I've given this a very little testing and it seems to do the right
> thing.  This obviously needs careful review!
> 
> Ben.
> 
>  hook-functions |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/hook-functions b/hook-functions
> index 0684296..da16478 100644
> --- a/hook-functions
> +++ b/hook-functions
> @@ -349,7 +349,7 @@ dep_add_modules()
>  	sys_walk_mod_add ${root_dev_path}
>  
>  	# catch old-style IDE
> -	if [ -e /sys/bus/ide/devices/ ]; then
> +	if [ -d "${DESTDIR}/lib/modules/${version}/kernel/drivers/ide" ]; then

Not sure I like this change, as this is a noop in any of our linux-2.6

>  		sys_walk_modalias ${root_dev_path}
>  		manual_add_modules ide-gd_mod
>  		# FIXME: remove post Squeeze
> @@ -357,7 +357,7 @@ dep_add_modules()
>  		manual_add_modules ide-cd
>  	fi
>  
> -	if [ -e /sys/bus/scsi/devices/ ]; then
> +	if [ -d "${DESTDIR}/lib/modules/${version}/kernel/drivers/scsi" ]; then
same here.
>  		manual_add_modules sd_mod
>  	fi
>  
> -- 
> 1.7.2.3
> 
> 
> 

so instead of overoptimising here, I'd just remove both conditions.
Of course they can be left or readded post-squeeze.

what do you think?

-- 
maks


Reply to: