Am Montag, den 29.11.2021, 22:58 -0500 schrieb Tom Rini:
On Tue, Nov 30, 2021 at 01:06:56AM +0100, Stefan Mätje wrote:
On AM4372 the SPI_GCLK input gets its clock from the PRCM module which divides the PER_CLKOUTM2 frequency (192MHz) by a fixed factor of 4. See AM437x Reference Manual in section 27 QSPI >> 27.2 Integration.
The QSPI_FCLK therefore needs to take this factor into account and becomes (192000000 / 4).
Signed-off-by: Stefan Mätje stefan.maetje@esd.eu
drivers/spi/ti_qspi.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/spi/ti_qspi.c b/drivers/spi/ti_qspi.c index 664b9cad79..bccdeeaf82 100644 --- a/drivers/spi/ti_qspi.c +++ b/drivers/spi/ti_qspi.c @@ -25,7 +25,8 @@ DECLARE_GLOBAL_DATA_PTR;
/* ti qpsi register bit masks */ #define QSPI_TIMEOUT 2000000 -#define QSPI_FCLK 192000000 +/* AM4372: QSPI gets SPI_GCLK from PRCM unit as PER_CLKOUTM2 divided by 4. */ +#define QSPI_FCLK (192000000 / 4) #define QSPI_DRA7XX_FCLK 76800000 #define QSPI_WLEN_MAX_BITS 128 #define QSPI_WLEN_MAX_BYTES (QSPI_WLEN_MAX_BITS >> 3)
How is this treated in the kernel? Thanks.
The current driver in mainline Linux @drivers/spi/spi-ti-qspi.c still has the wrong QSPI_FCLK define, but it seems to be used nowhere any more.
The driver gets its "fclk" from the device tree. But the device tree (arch/arm/boot/dts/am4372.dtsi) was broken till this patch on the mainline kernel: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/ar...
It seems to me that there the first time a valid "fclk" property was provided to the qspi module.
Best regards, Stefan Mätje