Mike Frysinger wrote on 2010-05-26:
> On Wed, May 26, 2010 at 04:46, Hennerich, Michael wrote:
>> Mike Frysinger wrote on 2010-05-26:
>>> 2010/5/26 Hennerich, Michael :
>>>> Song, Barry wrote on 2010-05-26:
>>>>> I don't think that fits with spi spec. For spi, changing CS means
>>>>> two different spi_message in one spi_transfer. You should use two
>>>>> spi_message with cs_change=1 for those devices. Thanks barry
>>>>
>>>> This means we can drop the bfin_spi_adc driver and remove the
>>> autobuffer DMA hack from the SPI bus driver as well.
>>>
>>> but does this address the touchscreen parts ?
>>
>> Well that checkin is missing a part that addresses the boards
>> resources. Take a look at struct spi_board_info. Every device that
>> doesn't set .mode uses SPI_MODE_0 or explicitly
> sets MODE2 or MODE0, and doesn't declare an GPIO SSEL is broken.
>
> i still think it sucks we block what the hardware can support, but i
> commented in the bug rather than e-mail:
> http://blackfin.uclinux.org/gf/tracker/5903
>
> to fix our resources, i think we just need to change it from using the
> pin as a peripheral CS to a GPIO ?
> -mike
Two different things:
The DMA mode where the SPI controls the SSEL is gone.
For other more typical devices the board resources can be fixed with something
like this:
Index: arch/blackfin/mach-bf548/boards/ezkit.c
===================================================================
--- arch/blackfin/mach-bf548/boards/ezkit.c (revision 8840)
+++ arch/blackfin/mach-bf548/boards/ezkit.c (working copy)
@@ -1041,7 +1041,7 @@
.irq = IRQ_PB4, /* old boards (<=Rev
1.3) use IRQ_PJ11 */
.max_speed_hz = 12500000, /* max spi clock (SCK)
speed in HZ */
.bus_num = 0,
- .chip_select = 2,
+ .chip_select = P_IDENT(P_SPI0_SSEL2) + MAX_CTRL_CS,
.controller_data = &spi_ad7877_chip_info,
},
#endif
@@ -1108,7 +1108,7 @@
/* SPI controller data */
static struct bfin5xx_spi_master bf54x_spi_master_info0 = {
- .num_chipselect = 3,
+ .num_chipselect = MAX_CTRL_CS + MAX_BLACKFIN_GPIOS,
.enable_dma = 1, /* master has the ability to do dma transfer */
.pin_req = {P_SPI0_SCK, P_SPI0_MISO, P_SPI0_MOSI, 0},
};
@@ -1124,7 +1124,7 @@
};
static struct bfin5xx_spi_master bf54x_spi_master_info1 = {
- .num_chipselect = 3,
+ .num_chipselect = MAX_CTRL_CS + MAX_BLACKFIN_GPIOS,
.enable_dma = 1, /* master has the ability to do dma transfer */
.pin_req = {P_SPI1_SCK, P_SPI1_MISO, P_SPI1_MOSI, 0},
};
However if we introduce this limitation - I see no reason why we would use the
SPI SSEL at all.
We can only use GPIO controlled SSEL, this would then allow us also to turn the
SPI off between transfers.
Greetings,
Michael
Analog Devices GmbH Wilhelm-Wagenfeld-Str. 6 80807 Muenchen
Sitz der Gesellschaft Muenchen, Registergericht Muenchen HRB 4036
Geschaeftsfuehrer Thomas Wessel, William A. Martin, Margaret Seif
_______________________________________________
Linux-kernel-commits mailing list
[email protected]
https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits
