DMA safe buffers for I2C.

DMA safe buffers for I2C.

[Jonathan Cameron]

Hi Wolfram,

Just looking to get up to date info on whether we need to be careful
with cachelines on buffers passed to i2c function calls or not.
i.e. do they need to DMA safe buffers.

I have it in the back of my head somewhere that i2c insisted that
if necessary this was sorted by the i2c master drivers, but I
suspect I am completely wrong.

There are 'lots' of drivers out there that aren't taking care of
this.  We have traditionally been much better with SPI where there
were explicit warnings about this.

Anyhow, could you clarify the situation?  Even if we audited existing
master drivers to check we were fine, if it's not coming from you
the chances are another one would sneak in sometime in the future
which does need dma safe buffers.

Certainly a superficial glance suggests that the at91 i2c master
is mapping these directly as DMA buffers.

Anyhow, what I'm really looking for is a quick email that I can
point people at in future ;)  or a reference to one somewhere else.

Thanks,

Jonathan

Re: DMA safe buffers for I2C.

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 420 bytes --]

Hi Jonathan,

> Just looking to get up to date info on whether we need to be careful
> with cachelines on buffers passed to i2c function calls or not.
> i.e. do they need to DMA safe buffers.

I did document this recently and forgot to add you to CC. Please check
https://lkml.org/lkml/2017/6/6/189 (or I can bounce you the mails if you
are interested) and kindly let me know what you think.

Kind regards,

   Wolfram


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]