Re: [PATCH v4 net-next 15/16] net: dsa: felix: setup MMIO filtering rules for PTP when using tag_8021q

On Wed, 20 Jan 2021 19:32:41 +0200 Vladimir Oltean wrote:

On Wed, Jan 20, 2021 at 08:40:42AM -0800, Jakub Kicinski wrote:

quoted

drivers/net/dsa/ocelot/felix.c:464:12: warning: variable ‘err’ set but not used [-Wunused-but-set-variable]
  464 |  int port, err;
      |            ^~~
drivers/net/dsa/ocelot/felix.c:265:53: warning: incorrect type in assignment (different base types)
drivers/net/dsa/ocelot/felix.c:265:53:    expected unsigned short [usertype]
drivers/net/dsa/ocelot/felix.c:265:53:    got restricted __be16 [usertype]


Please build test the patches locally, the patchwork testing thing is
not keeping up with the volume, and it's running on the largest VM
available thru the provider already :/  

I updated my compiler now, so that W=1 C=1 builds would not fail.
That should hopefully prevent this from happening in the future.

Thanks.

quoted

I need to add this "don't post your patches to get them build tested
or you'll make Kuba very angry" to the netdev FAQ.  

Since I definitely don't want to upset Kuba,

:)

how bad is it to exceed the 15 patches per series limit? Do I need to
do something about it?

It's not a hard rule IIUC, if you have 16, 17 patches as an atomic
series which is hard to split, I'd think that's acceptable from time 
to time. Especially if the patches themselves are not huge.
If you're already splitting a larger effort, keeping it < 15 is best.
In general if you can split a smaller logically contained series out
that's always preferred. The point is if the series is too large
reviewers are likely to postpone reviewing it until they can allocate
sufficiently large continuous block of time, which may be never.
It's all about efficient code review.

At least that's my recollection / understanding. There may be more
reasons, we'd have to ask Dave.