Hi Ezequiel,
On 06/09/14 22:06, Ezequiel Garcia wrote:
Hi Greg,
(Ccing mvebu guys, full discussion here: http://www.spinics.net/lists/devicetree/msg47806.html)
On 05 Sep 11:11 PM, Greg Ungerer wrote:
On 05/09/14 17:36, Geert Uytterhoeven wrote:
No, the other way around (from most-specific to least-specific):
compatible = "marvell,armada-spi", "marvell,orion-spi"
So the Armada version is used if available, and the Orion version
is used as a fallback.
Since you are doing another spin, please split this in two patches.
The driver and devicetree binding documentation should be in one patch,
and the devicetree .dtsi change in another patch.
By following Geert's request about the compatible, you make sure both
patches can be merged through different maintainers, and that an old devicetree
blob can be used safely.
Sure, yes, I would prefer that too. I didn't do it for this because
checkpatch
complained for the spi-orion.c change in its own:
WARNING: DT compatible string "marvell,armada-spi" appears un-documented
-- check ./Documentation/devicetree/bindings/
#170: FILE: drivers/spi/spi-orion.c:412:
+ { .compatible = "marvell,armada-spi", .data = &armada_spi_dev_data, },
But I am happy to ignore that :-)
Also what about armada370-xp, armada375, armada380
and any other variants out there?
I don't know. I don't have access to functional specs for anything
other than the 370. I expect they are probably the same though.
(Can anyone confirm?)
Confirmed. Judging from the specs, seems the same SPI IP is on all the
Armada SoCs; you can do the 'armada-spi' compatible addition on 375 and 38x.
Did a few tests on the AXP-GP board (which has an SPI flash) using
'orion-spi' and 'armada-spi' compatibles.
Tested-by: Ezequiel Garcia <ezequiel.garcia@xxxxxxxxxxxxxxxxxx>
Reviewed-by: Ezequiel Garcia <ezequiel.garcia@xxxxxxxxxxxxxxxxxx>
Thanks, appreciate that.
I haven't been able to spot any speed difference, though (perhaps my
test is too limited?), so I wonder if there's any improvement with this
patch?
Try setting a speed like 2MHz for an SPI device. It will fail because that
is below the minimum divisor it can calculate with the old orion style
behavior. I have an armada 370 based board with an ethernet switch
that is controlled by SPI. Its typical SPI bus speed is 2MHz, that is how I
came up against this.
Regards
Greg
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
