Re: [PATCH v2 2/2] ASoC: fsl_sai: Refine enable and disable sequence for synchronous mode
On Wed, Aug 05, 2020 at 01:03:37PM +0800, Shengjiu Wang wrote: > > Btw, the new fsl_sai_dir_is_synced() can be probably applied to > > other places with a followup patch. > Do you mean move it to the beginning of this file? There are other existing places testing "sync[tx] && !sync[!tx]" so you may submit another change to replace them. But, yea, will be a good idea to move that helper function to the top.
Re: [PATCH v2 2/2] ASoC: fsl_sai: Refine enable and disable sequence for synchronous mode
On Wed, Aug 5, 2020 at 12:13 PM Nicolin Chen wrote: > > On Wed, Aug 05, 2020 at 10:23:53AM +0800, Shengjiu Wang wrote: > > Tx synchronous with Rx: > > The TCSR.TE is no need to enabled when only Rx is going to be enabled. > > Check if need to disable RSCR.RE before disabling TCSR.TE. > > > > Rx synchronous with Tx: > > The RCSR.RE is no need to enabled when only Tx is going to be enabled. > > Check if need to disable TSCR.RE before disabling RCSR.TE. > > Please add to the commit log more context such as what we have > discussed: what's the problem of the current driver, and why we > _have_to_ apply this change though it's sightly against what RM > recommends. > > (If thing is straightforward, it's okay to make the text short. > Yet I believe that this change deserves more than these lines.) > > One info that you should mention -- also the main reason why I'm > convinced to add this change: trigger() is still in the shape of > the early version where we only supported one operation mode -- > Tx synchronous with Rx. So we need an update for other modes. > > > Signed-off-by: Shengjiu Wang > > The git-diff part looks good, please add this in next ver.: > > Reviewed-by: Nicolin Chen > > Btw, the new fsl_sai_dir_is_synced() can be probably applied to > other places with a followup patch. Do you mean move it to the beginning of this file? best regards wang shengjiu
Re: [PATCH v2 2/2] ASoC: fsl_sai: Refine enable and disable sequence for synchronous mode
On Wed, Aug 05, 2020 at 10:23:53AM +0800, Shengjiu Wang wrote: > Tx synchronous with Rx: > The TCSR.TE is no need to enabled when only Rx is going to be enabled. > Check if need to disable RSCR.RE before disabling TCSR.TE. > > Rx synchronous with Tx: > The RCSR.RE is no need to enabled when only Tx is going to be enabled. > Check if need to disable TSCR.RE before disabling RCSR.TE. Please add to the commit log more context such as what we have discussed: what's the problem of the current driver, and why we _have_to_ apply this change though it's sightly against what RM recommends. (If thing is straightforward, it's okay to make the text short. Yet I believe that this change deserves more than these lines.) One info that you should mention -- also the main reason why I'm convinced to add this change: trigger() is still in the shape of the early version where we only supported one operation mode -- Tx synchronous with Rx. So we need an update for other modes. > Signed-off-by: Shengjiu Wang The git-diff part looks good, please add this in next ver.: Reviewed-by: Nicolin Chen Btw, the new fsl_sai_dir_is_synced() can be probably applied to other places with a followup patch.