drm2 deprecation in base: problems with gpu-firmware-kmod port

[Niclas Zeising]

On 08/29/18 15:03, Alexey Dokuchaev wrote:
> On Wed, Aug 29, 2018 at 02:32:01PM +0200, Niclas Zeising wrote:
>> On 08/29/18 14:40, Alexey Dokuchaev wrote:
>>> On Wed, Aug 29, 2018 at 02:06:34PM +0200, Niclas Zeising wrote:
>>>> On 08/29/18 14:25, Alexey Dokuchaev wrote:
>>>>>    -radeonkmsfw_RV710_pfp: could not load firmware image, error 2
>>>>>    +radeonkmsfw_RV710_pfp: firmware image loaded, but did not register
>>>>
>>>> Just for clarification, is that a change in the message (with the + and
>>>> - as in diff) or the entire message?
>>>
>>> It's a diff :) sorry for not mentioning; I've omitted some other nearby
>>> changes which are due to [drm:pid????] being different.
>>>
>>>> I have to look into this a bit more.
>>
>> This is quite strange.  Did you install the firmware port from ports, or
>> did you build it yourself?
> 
>  From the ports of course, via good old "svn up && make install".

This is what i meant by "build it yourself" :).  s/ports/pkg/ in my 
sentence above and it might make more sense.
> 
>> Looking at the actual firmware files (the uuencoded binary blobs) they
>> are the same (same sha256 sum) in both the source for the
>> gpu-firmware-kmod package and from what's in [-CURRENT].
> 
> I can't update to the tip of the HEAD due to my world being June 9ish and
> I'd rather not rebuild everything right now, but I'd assume that firmware
> for old RV710 did not change and should be the same between kernels from
> June 9 vs. when it was recently reverted.
> 
>> It might be that they are built slightly differnt though, I guess.
> 
> I'll have a closer look myself as well once I have some spare time cycles;
> things are a bit hectic here at $work ATM.

Something is strange wrt. the differences in firmware sizes.  I can see 
it on my system as well, even though the uuencoded bits are the same. 
Might be that it's just some sort of "branding" since one is built at 
the same time as the kernel and one is not (different paths, debug info 
maybe?).  There is also an issue with firmware naming which I have to 
look into.  I still find it strange that it wasn't enough to rename the 
firmware blob, but maybe something else is going on.

What exact ATI/AMD graphics card are you using?  I might be able to dig 
something up from the pile behind me, otherwise I need help testing as well.

Regards
-- 
Niclas