[Sound-open-firmware] [RFC PATCH 6/6] trace: remove trace dmac id definition

[Ranjani Sridharan]

On Tue, 2018-06-05 at 19:15 +0800, Keyon Jie wrote:
> 
> On 2018年06月05日 12:23, Ranjani Sridharan wrote:
> > The new dma_get() API procures a DMAC based on user, copy direction
> > and other flags. So no need to define the DMAC ID for
> > trace DMA anymore.
> > 
> > Signed-off-by: Ranjani Sridharan <ranjani.sridharan at linux.intel.com
> > >
> > ---
> >   src/platform/apollolake/include/platform/platform.h | 3 ---
> >   src/platform/baytrail/include/platform/platform.h   | 3 ---
> >   src/platform/cannonlake/include/platform/platform.h | 3 ---
> >   src/platform/haswell/include/platform/platform.h    | 3 ---
> >   4 files changed, 12 deletions(-)
> > 
> > diff --git a/src/platform/apollolake/include/platform/platform.h
> > b/src/platform/apollolake/include/platform/platform.h
> > index aefa9b2..e77bf6f 100644
> > --- a/src/platform/apollolake/include/platform/platform.h
> > +++ b/src/platform/apollolake/include/platform/platform.h
> > @@ -103,9 +103,6 @@ struct sof;
> >    */
> >   #define DMA_TRACE_RESCHEDULE_TIME	5000
> >   
> > -/* DMAC used for trace DMA */
> > -#define PLATFORM_TRACE_DMAC	DMA_HOST_IN_DMAC
> > -
> >   /* DSP should be idle in this time frame */
> >   #define PLATFORM_IDLE_TIME	750000
> >   
> > diff --git a/src/platform/baytrail/include/platform/platform.h
> > b/src/platform/baytrail/include/platform/platform.h
> > index 6a7ebc9..8a7d0a3 100644
> > --- a/src/platform/baytrail/include/platform/platform.h
> > +++ b/src/platform/baytrail/include/platform/platform.h
> > @@ -95,9 +95,6 @@ struct sof;
> >    */
> >   #define DMA_TRACE_RESCHEDULE_TIME	5000
> >   
> > -/* DMAC used for trace DMA */
> > -#define PLATFORM_TRACE_DMAC	DMA_ID_DMAC0
> 
> as commented in 4/6, you need to make sure the correct dmac is used
> for 
> DMA trace on byt.

Keyon, are you saying we cannot use DMAC1 for trace dma on BYT?
Because the way the new API works on BYT is: it ignores the first 2
arguments and picks the DMAC with the least number of busy channels.
I've tested it on BYT and tracw works fine. 

But, if I  cannot DMAC1 for trace on BYT like you;re saying. then I'd
have to change the logic accordingly. So could you please confirm?

> 
> thanks,
> ~Keyon
> 
> > -
> >   /* DSP should be idle in this time frame */
> >   #define PLATFORM_IDLE_TIME	750000
> >   
> > diff --git a/src/platform/cannonlake/include/platform/platform.h
> > b/src/platform/cannonlake/include/platform/platform.h
> > index 35219d7..91c1f71 100644
> > --- a/src/platform/cannonlake/include/platform/platform.h
> > +++ b/src/platform/cannonlake/include/platform/platform.h
> > @@ -109,9 +109,6 @@ struct sof;
> >    */
> >   #define DMA_TRACE_RESCHEDULE_TIME	5000
> >   
> > -/* DMAC used for trace DMA */
> > -#define PLATFORM_TRACE_DMAC	DMA_HOST_IN_DMAC
> > -
> >   /* DSP should be idle in this time frame */
> >   #define PLATFORM_IDLE_TIME	750000
> >   
> > diff --git a/src/platform/haswell/include/platform/platform.h
> > b/src/platform/haswell/include/platform/platform.h
> > index d30f329..768a63c 100644
> > --- a/src/platform/haswell/include/platform/platform.h
> > +++ b/src/platform/haswell/include/platform/platform.h
> > @@ -94,9 +94,6 @@ struct sof;
> >    */
> >   #define DMA_TRACE_RESCHEDULE_TIME	5000
> >   
> > -/* DMAC used for trace DMA */
> > -#define PLATFORM_TRACE_DMAC	DMA_ID_DMAC1
> > -
> >   /* DSP should be idle in this time frame */
> >   #define PLATFORM_IDLE_TIME	750000
> >   
> > 
> 
> _______________________________________________
> Sound-open-firmware mailing list
> Sound-open-firmware at alsa-project.org
> http://mailman.alsa-project.org/mailman/listinfo/sound-open-firmware