blob: c5a5603c7b689fac7af8b4e918932e43f3667da9 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
|
#ifndef _HW_SPI_H_
#define _HW_SPI_H_
#include <stddef.h> /* for size_t */
enum spi_mode {
SPI_MODE_0 = 0, /* clk_polarity=0 (idle low), clk_phase=0 (sample on rise) */
SPI_MODE_1 = 1, /* clk_polarity=0 (idle low), clk_phase=1 (sample on fall) */
SPI_MODE_2 = 2, /* clk_polarity=1 (idle high), clk_phase=0 (sample on rise) */
SPI_MODE_3 = 3, /* clk_polarity=1 (idle high), clk_phase=1 (sample on fall) */
};
struct bidi_iovec {
void *iov_read_dst;
void *iov_write_src;
size_t iov_len;
};
struct spi;
/* This API assumes that an SPI frame is a multiple of 8-bits.
*
* It is my understanding that this is a common constraint of SPI
* hardware, and that the RP2040 is somewhat unusual in that it allows
* frames of any length 4-16 bits (we disconnect the CS pin from the
* PL022 SSP and manually GPIO it from the CPU in order to achieve
* longer frames).
*
* But, more relevantly: The W5500's protocol uses frames that are 4-N
* octets; so we have no need for an API that allows a
* non-multiple-of-8 number of bits.
*/
struct spi_vtable {
void (*readwritev)(struct spi *, const struct bidi_iovec *iov, int iovcnt);
};
struct spi {
struct spi_vtable *vtable;
/* This is where your implementation data goes. */
char data[0];
};
#endif /* _HW_SPI_H_ */
|
