Re: [PATCH v2 08/10] cxl: Allow the kernel to trust that an image won't change on PERST.

[Daniel Axtens <dja@axtens.net>]

[-- Attachment #1: Type: text/plain, Size: 1713 bytes --]

So after further offline conversations,

Yes, Cyril, you are right, I don't need the ifdef in my code. I just
need the symbol. I will amend my patches appropriately.

On Tue, 2015-08-11 at 21:22 +1000, Daniel Axtens wrote:
> > So I'm not super all over the putting all sorts of code inside CONFIG_CXL_EEH,
> > I understand that there is another driver being merged and they'll use
> > CONFIG_CXL_EEH so that both this driver and the other driver can go in the same
> > merge window but does this mean you need to put it around everything here?
> > 
> > I may have misunderstood what you've told me but if the other driver depends on
> > work done in this one (and not the other way around), if they depend on
> > CONFIG_CXL_EEH which you create in the last patch, then they cannot be built
> > until this series exists, so they can't have issues.
> > 
> > The one catch is that this series as is waits untill the last patch to actually
> > create the symbol, and therefore compile everything so lets be sure you don't
> > break bisecting. You might need to rethink the order of things in 8/10 and 9/10,
> > I can't see anything obvious if it helps...
> > 
> 
> Yeah, so you're right. I've taken the guards off everything except the
> new API function. I still want to leave the patch that adds the symbol
> at the end: that way you don't get the function unless it is actually
> going to make a difference in the EEH process. 
> 
> The other driver (cxlflash) just guards the API function, inserting a
> stub if it's not defined. So this setup will make our code cleaner and
> will still let their code merge cleanly.
> 
> Thanks again for the review.
> 

-- 
Regards,
Daniel

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 860 bytes --]