Re: [PATCH 0/2] firmware: cs_dsp: Remove usage of GFP_DMA

[Richard Fitzgerald <rf@opensource.cirrus.com>]

On 11/02/2025 5:21 pm, Richard Fitzgerald wrote:
> On 11/02/2025 5:03 pm, Thomas Weißschuh wrote:
>> Also drop the bounce buffer in cs_dsp_coeff_write_ctrl_raw().
>>
>> The bounce buffer in cs_dsp_coeff_write_ctrl_raw() could theoretically
>> also be removed. That would be a functional change as the output may be
>> modified in error cases.
>> As I don't know the driver very well I left that part out.
>>
>> Not tested on real hardware.
>> This came up while porting kunit to mips64.
>> Apparently GFP_DMA does not work there, but IMO the usage of GFP_DMA by

I would say that is a bug in mips64 that should be fixed in mips64.
It is not reasonable to expect generic drivers to have special cases for
platforms that don't handle GFP_DMA.


>> cs_dsp is unnecessary in the first place.
>>
> 
> You're sure that all I2C and SPI bus controllers now handle non-DMA-safe
> buffers correctly?
> 
I just tested this.

The spi kernel doc says this:

  * struct spi_transfer - a read/write buffer pair
  * @tx_buf: data to be written (DMA-safe memory), or NULL
  * @rx_buf: data to be read (DMA-safe memory), or NULL

On x86_64() a spi_write() fails with -EINVAL if I pass it a non-dma-safe
buffer of data. But it works if I pass it a buffer allocated with
GFP_DMA.

So I think it is a bad idea to remove GFP_DMA to workaround mips64.

>> Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
>> ---
>> Thomas Weißschuh (2):
>>        firmware: cs_dsp: Remove usage of GFP_DMA
>>        firmware: cs_dsp: Remove bounce buffer in 
>> cs_dsp_coeff_write_ctrl_raw()
>>
>>   drivers/firmware/cirrus/cs_dsp.c | 15 +++------------
>>   1 file changed, 3 insertions(+), 12 deletions(-)
>> ---
>> base-commit: 2014c95afecee3e76ca4a56956a936e23283f05b
>> change-id: 20250211-cs_dsp-gfp_dma-0581bdd09dd5
>>
>> Best regards,
>