Re: [PATCH 2/2] i2c: omap: fix "Too much work in one IRQ" irq handling

[Felipe Balbi <balbi-l0cyMroinI0@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 1612 bytes --]

On Tue, Nov 18, 2014 at 07:31:44PM +0300, Alexander Kochetkov wrote:
> 
> > Still, as of now we can't consider what Alexander mentions a bug. Good
> > to get it sorted out, but not -rc material.
> > 
> 
> Actually, I focused on fixing issues then i2c-omap acts as master in a
> multi master environment.

but that's not supported either. If you wanna do that, then go through
the TRM and implement what's missing. Then give a good commit log
explaining the issues and how you solved them :-)

Still not -rc material :-)

> And must say, that current linux/mainline driver work perfectly with
> minor fixes.

that's good to know. I know some folks still have a few hick ups. The IP
isn't all that great, I'm afraid.

> I could take a look how to switch the driver into mixed master/slave
> mode, but later.
> 
> Sad to say, I encountered undocumented i2c- hardware issue.
> In short: BB-bit doesn't reflect I2C-bus state after soft reset if SDA
> or SCL line was low during reset until i2c-controller again detect
> START or STOP condition on the wire.

this is another of those which really deserves a mention in code comment
or commit log :-)

> More over, if SDA was low during reset, than next submitted transfer
> will not start ("controller timeout"), if SCL was low during reset,
> than next submitted transfer will start and corrupt signals on wire :(

another thing to add as comment to code :-)

> I'm checking it now. And going to post results to ti forum.
> 
> BB-bit reflect I2C-bus state, between power-down/power-up modes.

alright, thanks.

-- 
balbi

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]