Fred: Comments below.
On 2/25/2016 7:11 AM, Frederic Barrat wrote:
Le 24/02/2016 21:03, Manoj Kumar a écrit :
From: Christophe Lombard <clomb...@linux.vnet.ibm.com>
+#define CXL_DEV_MINORS 13 /* 1 control + 4 AFUs * 3
(dedicated/master/shared) */
Where does this limit of 4 AFUs come from?
Is this related to CXL_MAX_SLICES?
Should this be a computed value, in case the number of AFUs/slices
is increased at a future date?
The architecture document (CAIA) limits the number of AFUs to 4, though
I don't think anybody as tried with more than 1 so far.
So yes, we could have reused CXL_MAX_SLICES. Since we were just moving
the definition from another file and this is not likely to vary until a
major revision of the architecture, I don't intend to address it in this
patchset, but I've added it to my list of ideas for future cleanup (you
had already mentioned something about hard-coded constants in the
previous series).
Since it was a carryover from existing code, it is fine to address
later.
memset(0) after kzalloc() is redundant.
yep! Will fix.
If this is resolved in v6, you may add
Reviewed-by: Manoj Kumar <ma...@linux.vnet.ibm.com>
---
Manoj Kumar
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/linuxppc-dev