[Mesa-dev] [PATCH 1/2] gm107/ir: make use of FADD32I for all immediates

Ilia Mirkin imirkin at alum.mit.edu
Tue Jun 28 14:23:49 UTC 2016


On Tue, Jun 28, 2016 at 10:21 AM, Samuel Pitoiset
<samuel.pitoiset at gmail.com> wrote:
> On 06/28/2016 04:15 PM, Ilia Mirkin wrote:
>>
>> Again, what problem was this patch trying to solve?
>
>
> The problem is that FADD can only emits 19-bits but longIMMD() will return
> false because it only checks for the high 12-bits.
>
> I don't know if you saw my messages on IRC but I found some other issues
> with longIMMD() and emitIMMD().

Nope, it will emit 19 bits and then the 20th (high aka sign) bit as
well, just to a different location. [And the bottom 12 bits are
guaranteed to be 0.]

What's a specific example that you think it doesn't emit correctly?

 -ilia


More information about the mesa-dev mailing list