On Thu, Mar 10, 2022 at 08:29:30PM +0300, Serge Semin wrote:
> On Wed, Mar 09, 2022 at 03:12:02PM -0600, Frank Li wrote:
> > From: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxx>
> >
> > The "direction" member of the "dma_slave_config" structure is deprecated.
> > The clients no longer use this field to specify the direction of the slave
> > channel. But in the eDMA core, this field is used to differentiate between the
> > Root complex (remote) and Endpoint (local) DMA accesses.
> >
> > Nevertheless, we can't differentiate between local and remote accesses without
> > a dedicated flag. So let's get rid of the old check and add a new check for
> > verifying the DMA operation between local and remote memory instead.
> >
> > Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxx>
> > Signed-off-by: Frank Li <Frank.Li@xxxxxxx>
> > ---
> > no chang between v1 to v4
> > drivers/dma/dw-edma/dw-edma-core.c | 17 ++---------------
> > 1 file changed, 2 insertions(+), 15 deletions(-)
> >
> > diff --git a/drivers/dma/dw-edma/dw-edma-core.c b/drivers/dma/dw-edma/dw-edma-core.c
> > index 507f08db1aad3..47c6a52929fcd 100644
> > --- a/drivers/dma/dw-edma/dw-edma-core.c
> > +++ b/drivers/dma/dw-edma/dw-edma-core.c
> > @@ -341,22 +341,9 @@ dw_edma_device_transfer(struct dw_edma_transfer *xfer)
> > if (!chan->configured)
> > return NULL;
> >
> > - switch (chan->config.direction) {
> > - case DMA_DEV_TO_MEM: /* local DMA */
> > - if (dir == DMA_DEV_TO_MEM && chan->dir == EDMA_DIR_READ)
> > - break;
> > - return NULL;
> > - case DMA_MEM_TO_DEV: /* local DMA */
> > - if (dir == DMA_MEM_TO_DEV && chan->dir == EDMA_DIR_WRITE)
> > - break;
>
> > + /* eDMA supports only read and write between local and remote memory */
>
> The comment is a bit confusing because both cases are named as
> "memory" while the permitted directions contains DEV-part, which
> means "device". What I would suggest to write here is something like:
> "DW eDMA supports transferring data from/to the CPU/Application memory
> to/from the PCIe link partner device by injecting the PCIe MWr/MRd TLPs."
>
End of the day, you'd be transferring data between remote and local memory
only and the terms (local and remote) are also used in the databook. So I think
the comment is fine.
Thanks,
Mani
> -Sergey
>
> > + if (dir != DMA_DEV_TO_MEM && dir != DMA_MEM_TO_DEV)
> > return NULL;
> > - default: /* remote DMA */
> > - if (dir == DMA_MEM_TO_DEV && chan->dir == EDMA_DIR_READ)
> > - break;
> > - if (dir == DMA_DEV_TO_MEM && chan->dir == EDMA_DIR_WRITE)
> > - break;
> > - return NULL;
> > - }
> >
> > if (xfer->type == EDMA_XFER_CYCLIC) {
> > if (!xfer->xfer.cyclic.len || !xfer->xfer.cyclic.cnt)
> > --
> > 2.24.0.rc1
> >
