On Fri, Aug 19, 2016 at 03:59:08PM +0200, Marcin Wojtas wrote:
Right, but I'm not obsessing over what the initial commit should look like, I'm obsessing over what the resulting code should look like.
We cannot have the consumers of said imported patches needing to redefine a bunch of datatypes in order to call them. But also, if we are not expecting to be seeing some sort of destination-independent updates to this driver from the vendor, I see little value in keeping the code non-compliant.
Frankly, if a rework could be done, I a LOT prefer to do it now, rather than it unexpectedly hit us when you merge OPP into EDK2.
*nod*
About tracking fixes and modifications in the other sources - it would be great, but I wouldn't bother that much. Basically, the bootloader's
firmware's
network is needed mostly for downloading files. Current state of Pp2Dxe is that in different phy modes (SGMII, RGMII, fixed link switch) it works stable, we can tftp 500MB file within 1 minute - so it's in a satisfying state. Performance @highest rates should not be a question here. After all, possible modifications will have to be applies manually and if they refer to particular patch somewhere else, I think caring about mentioning the source in the commit log should be sufficient.
In order to summary, I propose following view of v2 patchset:
- Import library as-is now (except for licence, which has to fit uefi
from the beginning). 2. Add style modification to pass edk2 checkpatch. 3. Add separate glue header. 4. Add driver 5. Enable driver. Pairs 1-2 and 3-4 can be squashed, but we can discuss it later.
What do you think?
I'd say 2-3 can be squashed, but otherwise this sounds like a good plan to me, so let's discuss that later.
Thanks!
/ Leif