RE: [PATCH v3 2/2] spi: cadence: add support for Cadence XSPI controller

From: Parshuram Raju Thombare
Date: Fri Sep 03 2021 - 04:10:54 EST


Hi Mark,

Thanks for your reply.

>Please make the entire comment a C++ so things look more intentional.

Sure

>The core already validates this, are you seeing it happen? If so we
>should fix the core and either way just remove setup() entirely.

Oh, right. spi_add_device() seems to be already doing that check.
I will remove it from the driver.

>We will just silently ignore any unknown interrupts here. It would be
>better to either only ack known interrupts (so genirq can notice if
>there's a problem with other interrupts) or at least log that we're
>seeing unexpected interrupts. The current code will cause trouble if
>this is deployed in a system with the interrupt line shared (which the
>driver claims to support), or if something goes wrong and the IP starts
>asserting some interrupt that isn't expected.

Ok, I will modify this to return IRQ_HANDLED only for enabled interrupts.

>> + master->mode_bits = SPI_3WIRE | SPI_TX_DUAL | SPI_TX_QUAD |
>> + SPI_RX_DUAL | SPI_RX_QUAD | SPI_TX_OCTAL | SPI_RX_OCTAL
>|
>> + SPI_MODE_0 | SPI_MODE_3;
>
>I don't see any handling of these in the code?

This is just to declare controller's capability, so that spi_setup() can modify
spi->mode according to the capability of attached device.

Regards,
Parshuram Thombare