On 11.03.19 15:09, Tom Rini wrote:
On Mon, Mar 11, 2019 at 01:29:23PM +0100, Stefan Roese wrote:
When checking for boards that are enabling a SATA driver that isn't converted to DM yet we need to be sure to not also trip over boards that do set CONFIG_AHCI & CONFIG_BLK by itself, as that is not a bug.
This was detected when converting the Armada XP SATA driver sata_mv.c to DM with AHCI & BLK support. This warning is still printed without this patch applied here.
Signed-off-by: Stefan Roese sr@denx.de Cc: Simon Glass sjg@chromium.org Cc: Tom Rini trini@konsulko.com
Makefile | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/Makefile b/Makefile index 26db4e7b59..896e0c3350 100644 --- a/Makefile +++ b/Makefile @@ -940,6 +940,7 @@ endif endif ifeq ($(CONFIG_LIBATA)$(CONFIG_MVSATA_IDE),y) ifneq ($(CONFIG_DM_SCSI),y) +ifneq ($(CONFIG_BLK)$(CONFIG_AHCI),yy) @echo >&2 "===================== WARNING ======================" @echo >&2 "This board does not use CONFIG_DM_SCSI. Please update" @echo >&2 "the storage controller to use CONFIG_DM_SCSI before the v2019.07 release."
OK, I think we need to properly clarify the logic, and perhaps I was being too clever trying to catch all of the cases under drivers/ata that aren't converted yet. We have, after your patches, afaict:
- MVSATA_IDE
- DWC_AHSATA
- FSL_SATA
- SATA_SIL
- SATA_SIL3114
Of these, MVSATA_IDE has no further deps. Everything else depends on LIBATA. Before SATA_MV was converted, everything that was converted depends on DM_SCSI and AHCI. So I made my messages / logic like the above. But perhaps the right answer really is:
- MVSATA_IDE alone.
- LIBATA && !AHCI
So... we can just re-work the second part of the check logic to be what you added, as all converted drivers under drivers/ata will be both BLK&&AHCI enabled.
I'll work on a v2 of this patch shortly.
And also as you note, it's not quite AHCI as the controller in question here isn't AHCI-compatible, but is SATA, so a fixup / rename of some CONFIG checks in code / Kconfig is in order, somewhere, but perhaps after conversion?
I agree.
Also, probably some re-organization of drivers/block, drivers/ide, drivers/ata, drivers/nvme (and common/usb_storage.c) too, to be more consistent.
Yes, that could also be helpful.
Thanks, Stefan