[PATCH 28/30] spi: Enable missing CONFIG_SPL_DM_SPI support

[Tom Rini]

On Wed, Jun 10, 2020 at 05:01:21PM -0400, Tom Rini wrote:
> On Wed, Jun 10, 2020 at 10:38:45PM +0200, Marek Vasut wrote:
> > On 6/10/20 10:16 PM, Tom Rini wrote:
> > [...]
> > >  configs/socfpga_dbm_soc1_defconfig                             | 1 +
> > >  configs/socfpga_de0_nano_soc_defconfig                         | 1 +
> > >  configs/socfpga_de10_nano_defconfig                            | 1 +
> > >  configs/socfpga_de1_soc_defconfig                              | 1 +
> > 
> > I don't think those de*_soc boards have a SPI NOR at all.
> > And I'm also afraid that enabling this will make those boards overflow
> > SPL size limits.
> 
> There is zero size change from this patch on any platform included in
> this patch.
> 
> Today each of the boards you mention enables CONFIG_SPL_SPI_SUPPORT,
> CONFIG_SPL_SPI_FLASH_SUPPORT and then CONFIG_SPL_SPI_FLASH_TINY.
> 
> That said, that means you aren't using SPL_DM_SPI for real, so I'll go
> take a look at what to change instead in the previous patch, thanks!

Nope, check your platforms again.  Based on looking over
socfpga_de1_soc/spl/u-boot-spl.map it's not discarding a bunch of the DM
SPI stuff as we're in fact building it today, and using the few
functions that spi-mem-nodm.c provides from spi-mem.c instead.  So this
is correct for what's happening today.

With Jagan's removal patches that I believe allow us to nuke
drivers/spi/spi.c (and so the unconditional building of it) as it nukes
the only platform that uses the only video driver that calls
spi_set_wordlen() (spi_do_alloc_slave() is unused globally) we could
then go ahead and do non-DM_SPI here, fully.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200610/5c54088d/attachment-0001.sig>