Re: brcmfmac bus level addressing issues.

[Ian Molton <ian@mnementh.co.uk>]

On 19/07/17 09:39, Hante Meuleman wrote:
> Dear Ian,

Hi,

> Stuff like " The PCIe code sets the window *regardless* of wether its
> changed, on *every single* write." Is totally incorrect. Sure if you limit
> yourself to the function brcmf_pcie_buscore_{read,write}32(). But you
> talked about performance, and msgbuf prototocol is where performance
> counts and that don't use those functions. You wrote: " The PCIe code uses
> brcmf_pcie_select_core(), which, ultimately, appears to be totally
> redundant" and that is simply not true.

This actually kinda underlines my point - at the time I was, as you say,
referring to the _{read,write}32() functions (which is where I'd been
cleaning up in the sdio code too.

> So I decided to answer that mail
> and provoke you a bit. I'm sorry for that, I shouldn't have done that.

Fair enough. I lost my cool in my reply to that too, so lets just start
again. Fair?

> You obviously spent some time on creating all these patches, but why
> provoke/agitate people? Why use such strong words? You may not consider
> them personally, but I just explained why I do. Can you at least
> understand that?

Sorry it came over as personal. Its not. And yes, it sometimes hurts
when people are critical about code you've written - and I've been on
the receiving end of it before now - and they were correct, and I was
wrong at the time. It has to be said though - that code is not pretty
(sdio side in particular - as mentioned before, the PCIe side is a fair
bit better).

> Just some word of advice and then I hope we can leave it to that.
Yes, lets.

-Ian