Incompatibility between enc28j60.c and sunxi_spi.c

[vinifr]

There is a test that precludes the use enc28j60.c in its original form:

sunxi_spi.c: LINE 812:
/* check in the __spi_async if use half duplex */
if(t->tx_buf && t->rx_buf)
return -EINVAL;

This does not allow spi_read_buf() to accomplish its task:

enc28j60.c: LINE 92
struct spi_transfer t = {
.tx_buf = tx_buf,
.rx_buf = rx_buf,
.len = SPI_OPLEN + len,
};

Lorenzo proposed to change enc28j60.c this way: http://pastebin.com/ZxYkuebs

What would be the best solution?

[amery]

please submit to the linux-sunxi mailinglist.

[vinifr]

Hi amery,

Lorenzo already posted on the mailing list: https://groups.google.com/forum/?fromgroups=#!topic/linux-sunxi/unWV4VIZXoY

We can also change spi_sunxi.c.

What do you think?

[amery]

sounds fair, but please tell so on the ML and submit the git formatted commit including a description of the problem and the explaination of the fix. there are several devs which aren't subscribed to this issue tracker.

[vinifr]

Sorry for reopen.

I have a doubt... in the case of mismatch between chip driver and platform driver, which must be changed? taking into account that enc28j60.c is in the mainline.

I ask because with atmel_spi.c(for example), enc28j60.c accomplish its task normally, not required any changes.

[amery]

the sunxi driver should be fixed to coexist with mainlined code, not the other way around.