Re: [PATCH 6/6] block: Remove "select BLK" from non-block drivers

[Tom Rini <trini@konsulko.com>]

[-- Attachment #1: Type: text/plain, Size: 1660 bytes --]

On Tue, Jan 14, 2025 at 02:53:48PM +0100, Quentin Schulz wrote:
> Hi Tom,
> 
> On 12/20/24 11:22 PM, Tom Rini wrote:
> > Now that block drivers are all selecting the BLK symbol, there's no need
> > for other options to be select'ing BLK so that other required
> > functionality can be enabled. Remove these places.
> > 
> 
> We have multiple commands depending on the BLK symbol.

Yes.

> BOOTSTD also depends on it, but I assume we should be able to network boot
> without HW block drivers?

Correct. That's part of the motivation for this series (which I wasn't
clear enough about on its own). Without something like this series if we
remove the BLK dependency from BOOTSTD then some other platforms fail to
build or grow a bunch in size (as BOOTSTD is default y and now it's
enabled on those platforms).

> CMD_UFETCH wouldn't be usable without those drivers as well.
> 
> Should we do something about that by making them not depend on BLK e.g. use
> CONFIG_IS_ENABLED in the right places? Not sure if all devicess based on
> those archs have at least one HW block driver enabled. I guess checking if
> all .config before and after that change are identical would help us figure
> out if this could introduce a regression?

There's a few options, depending on what the command is. For CMD_UFETCH
it's likely that a small restructure would be needed to not try and
print block devices while preserving the rest of the formatting. For
CMD_LSBLK / CMD_CLONE it's an unfortunate short-hand for "some block
device exists" which is functionally what those commands require.

Thanks for reviewing the whole series.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]