Re: [Intel-gfx] [PATCH RESEND] drm/i915: Fix pipe/transcoder enum mismatches

[Grant Grundler]

On Fri, Jul 14, 2017 at 2:35 PM, Daniel Vetter <dan...@ffwll.ch> wrote:
> On Fri, Jul 14, 2017 at 7:32 PM, Grant Grundler <grund...@chromium.org> wrote:
>> On Fri, Jul 14, 2017 at 2:11 AM, Jani Nikula
>> <jani.nik...@linux.intel.com> wrote:
>>> On Thu, 13 Jul 2017, Stéphane Marchesin <stephane.marche...@gmail.com> 
>>> wrote:
>>>> So, if you think this is wrong, can you fix this warning in a way that
>>>> you'd like?
>>>
>>> As I replied previously [1], with more background, fixing the warnings
>>> properly, in a way that actually improves the code instead of making it
>>> worse, would mean a bunch of churn that's not just purely mechanical
>>> conversion.
>>
>> That's fair.
>>
>>> Unless you can point out a bug which is actually caused by mixing the
>>> types (which is mostly intentional, see the background) I have a hard
>>> time telling people this should be a priority.
>>
>> This feels like "can't see the forest because of the trees".
>>
>> The original patch was submitted in order to compile cleanly using
>> clang and the above suggests using clang is not important.  Using
>> clang is important to Matthias and the Chrome OS organization for many
>> good reasons - including better warnings.
>>
>> The original patch message was clear that clang was generating the
>> warning. This isn't the only patch mka has sent to kernel devs. What
>> one can infer is Chrome OS is trying to move to clang (like other
>> Google products _already_ have.)  My impression is all these products
>> are a priority to Intel - but it would be good to know otherwise.
>>
>>> Definitely something we'd
>>> like to do in the long run and pedantically correct (and I tend to
>>> prefer code that way) but we certainly have more important things to do.
>>
>> The long run is now. Everyone agrees the code should change and you
>> don't have to do it. Matthias submitted an unacceptable patch and
>> giving him some concrete guidance on what would be acceptable would
>> enable him to implement/test it (or anyone else could for that
>> matter).  Can you do that?
>>
>> Just give an example of what the "right" API looks like and see where it 
>> goes.
>
> We've replied and discussed on May 5th what that roughly should be,
> right when Matthias pinged us. The original submission unfortunately
> fell through the cracks (it happens, not much we can do with this
> flood). Matthias didn't seem to have any questions about the proposed
> solutions (we laid out both the minimal short-term fix to unconfuse
> things, and what might be done on top), I think a reasonable
> assumption was that it's all clear. Otherwise he should have asked.

Indeed!

After briefly chatting with Stephane and mka, it seems the difference
between short-term fix and "done on top" were not clear.

> Now, over 2 months later (and complete silence from your side) there's
> suddenly mass panic and multiple escalations on all available
> channels, which feels like a rather decent overreaction and not a
> terrible constructive way to collaborate on the upstream codebase.

I'm sorry - I'm not on the other channels and I didn't see any mass
panic. I agree that's not a collaborative. The previous answer in this
thread didn't seem particularly collaborative either though.

The silence was partly due to mka working on other "clang enablement" patches:

$ pwclient list -w m...@chromium.org
Patches submitted by Matthias Kaehlcke <m...@chromium.org>:
ID  StateName
--  -
9668095 Superseded   mac80211: Fix clang warning about constant
operand in logical operation
9668479 Accepted ath9k: Add cast to u8 to FREQ2FBIN macro
9668643 Accepted [v2] mac80211: Fix clang warning about constant
operand in logical operation
9679753 Accepted [v2] cfg80211: Fix array-bounds warning in fragment copy
9684547 Accepted mac80211: ibss: Fix channel type enum in
ieee80211_sta_join_ibss()
9684629 Accepted nl80211: Fix enum type of variable in
nl80211_put_sta_rate()

> Anyway, I've done the quick draft for the function declaration changes
> that would clear up the confusion, just needs a clang run to update
> all the parameters to match, and passed that on to Stéphane Marchesin.

Awesome - thanks! :)

> I expect you to follow up with the corresponding patch right away.

mka said "he would take a look at it". But knowing how he understates
things in a typical "German Engineer" way, I'm optimistic it will be
more than that. Thanks!

cheers,
grant

>
> Thanks a lot.
>
> Yours, Daniel
>
> For reference the

Re: [Intel-gfx] [PATCH RESEND] drm/i915: Fix pipe/transcoder enum mismatches

[Grant Grundler]

On Fri, Jul 14, 2017 at 2:11 AM, Jani Nikula
 wrote:
> On Thu, 13 Jul 2017, Stéphane Marchesin  wrote:
>> So, if you think this is wrong, can you fix this warning in a way that
>> you'd like?
>
> As I replied previously [1], with more background, fixing the warnings
> properly, in a way that actually improves the code instead of making it
> worse, would mean a bunch of churn that's not just purely mechanical
> conversion.

That's fair.

> Unless you can point out a bug which is actually caused by mixing the
> types (which is mostly intentional, see the background) I have a hard
> time telling people this should be a priority.

This feels like "can't see the forest because of the trees".

The original patch was submitted in order to compile cleanly using
clang and the above suggests using clang is not important.  Using
clang is important to Matthias and the Chrome OS organization for many
good reasons - including better warnings.

The original patch message was clear that clang was generating the
warning. This isn't the only patch mka has sent to kernel devs. What
one can infer is Chrome OS is trying to move to clang (like other
Google products _already_ have.)  My impression is all these products
are a priority to Intel - but it would be good to know otherwise.

> Definitely something we'd
> like to do in the long run and pedantically correct (and I tend to
> prefer code that way) but we certainly have more important things to do.

The long run is now. Everyone agrees the code should change and you
don't have to do it. Matthias submitted an unacceptable patch and
giving him some concrete guidance on what would be acceptable would
enable him to implement/test it (or anyone else could for that
matter).  Can you do that?

Just give an example of what the "right" API looks like and see where it goes.

cheers,
grant

>
> BR,
> Jani.
>
> [1] http://mid.mail-archive.com/87wp9rahjy.fsf@intel.com
>
>
> --
> Jani Nikula, Intel Open Source Technology Center
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH RESEND] drm/i915: Fix pipe/transcoder enum mismatches

[Grant Grundler]

On Fri, May 5, 2017 at 1:08 PM, Ville Syrjälä
 wrote:
...
>> > I'm not convinced the patch is making things any better really. To
>> > fix this really properly, I think we'd need to introduce a new enum
>> > pch_transcoder and thus avoid the confusion of which type of
>> > transcoder we're talking about.
>>
>> Is an enum better than coding an explicit conversion in an inline function?
>
> The point of the enum would be to make it more clear which piece of
> hardware we're talking to in each case.

Ah ok - I misunderstood - I thought this was already the case.

> But this would require going
> through the entire PCH code and changing things to use the right type
> in each case. Quite a bit of work with little measurable gain I'd say.

IMHO, one of the best things that happened to C standard was addition
of strong type checking. It's helped prevent developers from making
one class of "stupid mistakes". So while this change wouldn't improve
performance, it would allow a form of automated correctness checking
that can be enforced with every patch you get (every time the code
base is compiled).

In other words, the gain isn't currently measurable the same way
performance is but I believe it's worth doing.  Given the number of
typedefs and enums in kernel code, I suspect most kernel developers
would agree.

cheers,
grant



>
>> Then the code can do some sanity checking as well. Something like:
>>
>> enum transcoder pch_to_cpu_enum(enum pipe)
>> {
>> WARN_ON(pipe > FOO);
>> return (enum transcoder) pipe;
>> }
>
> That would have to be called pipe_to_pch_transcoder() or something like
> that.
>
>>
>> > Currently most places expect an
>> > enum pipe when dealing with PCH transcoders, and enum transcoder
>> > when dealing with CPU transcoders. But there are some exceptions
>> > of course.
>>
>> cheers,
>> grant
>> >
>> > --
>> > Ville Syrjälä
>> > Intel OTC
>
> --
> Ville Syrjälä
> Intel OTC
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [Intel-gfx] [PATCH RESEND] drm/i915: Fix pipe/transcoder enum mismatches

[Grant Grundler]

On Fri, May 5, 2017 at 10:40 AM, Ville Syrjälä
 wrote:
> On Fri, May 05, 2017 at 10:26:36AM -0700, Matthias Kaehlcke wrote:
>> El Thu, Apr 20, 2017 at 02:56:05PM -0700 Matthias Kaehlcke ha dit:
>>
>> > In several instances the driver passes an 'enum pipe' value to a
>> > function expecting an 'enum transcoder' and viceversa. Since PIPE_x and
>> > TRANSCODER_x have the same values this doesn't cause functional
>> > problems. Still it is incorrect and causes clang to generate warnings
>> > like this:
>> >
>> > drivers/gpu/drm/i915/intel_display.c:1844:34: warning: implicit
>> >   conversion from enumeration type 'enum transcoder' to different
>> >   enumeration type 'enum pipe' [-Wenum-conversion]
>> > assert_fdi_rx_enabled(dev_priv, TRANSCODER_A);
>> >
>> > Change the code to pass values of the type expected by the callee.
>> >
>> > Signed-off-by: Matthias Kaehlcke 
>> > ---
>> >  drivers/gpu/drm/i915/intel_display.c | 4 ++--
>> >  drivers/gpu/drm/i915/intel_dp.c  | 6 --
>> >  drivers/gpu/drm/i915/intel_hdmi.c| 6 --
>> >  drivers/gpu/drm/i915/intel_sdvo.c| 6 --
>> >  4 files changed, 14 insertions(+), 8 deletions(-)
>>
>> Ping, any comments on this patch?
>
> I'm not convinced the patch is making things any better really. To
> fix this really properly, I think we'd need to introduce a new enum
> pch_transcoder and thus avoid the confusion of which type of
> transcoder we're talking about.

Is an enum better than coding an explicit conversion in an inline function?
Then the code can do some sanity checking as well. Something like:

enum transcoder pch_to_cpu_enum(enum pipe)
{
WARN_ON(pipe > FOO);
return (enum transcoder) pipe;
}

> Currently most places expect an
> enum pipe when dealing with PCH transcoders, and enum transcoder
> when dealing with CPU transcoders. But there are some exceptions
> of course.

cheers,
grant
>
> --
> Ville Syrjälä
> Intel OTC
___
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx