On Thu, Jun 13, 2019 at 09:29:48AM -0700, Lucas De Marchi wrote:
On Thu, Jun 13, 2019 at 04:12:37PM +0300, Jani Nikula wrote:
On Wed, 12 Jun 2019, Lucas De Marchi <lucas.demarchi@xxxxxxxxx> wrote:
We are slowly converting dev_priv to i915 everywhere, spread into
smaller series. While this is good to avoid unrelated breakages to other
inflight patches, it's bad because inflight patches on nearby paths keep
breaking. Paired with other code moves and refactores this is becoming a
nightmare.
Now that I915_{READ,WRITE} are gone and implicit access to dev_priv no
longer exists we can simply sed all files and be done with the
conversion. This was generated with the following commands with no
additional fixups:
git ls-files --full-name -z -- drivers/gpu/drm/i915/ | \
xargs -0 sed -i 's/\bdev_priv\b/i915/g'
Any pending series can apply the same search and replace when rebasing.
I'm pretty strongly against renaming the implicit dev_priv local
variable before we've gotten rid of it. Renaming s/dev_priv/i915/ upon
converting a function to not use the implicit dev_priv helps us by
catching any leftover references.
I don't think this is strong enough reason to block the conversion. The
conversion is taking forever and is gonna take year+ if it continues at
this pace. It affects multiple developers rebasing their work and
introduce bugs on pending series.
Removing any file-scope reference (I didn't review yet if we still have
any) and reviewing the macros is sufficient. And if we later we find out
we missed one, we just go and fix it. I see zero advantage on slow and
forever. You have a mixed code base and new code following what's around
will just propagate more the mixed code base :(
btw, let me be clear the proposal is not "this should be done now". I do
think we should be smart and think on the best opportunity to do it.
Probably like after the big code moves gt/, gem/, display/, etc end.
Lucas De Marchi
Lucas De Marchi
BR,
Jani.
--
Jani Nikula, Intel Open Source Graphics Center
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx