Re: [PATCH v6 0/6] Fix support of dw-edma HDMA NATIVE IP in remote setup

[Serge Semin]

On Tue, Nov 21, 2023 at 05:38:28PM +0530, Manivannan Sadhasivam wrote:
> On Tue, Nov 21, 2023 at 01:55:22PM +0300, Serge Semin wrote:
> > Hi Mani
> > 
> > On Tue, Nov 21, 2023 at 11:56:29AM +0530, Manivannan Sadhasivam wrote:
> > > On Fri, Nov 17, 2023 at 11:03:48AM +0100, Kory Maincent wrote:
> > > > This patch series fix the support of dw-edma HDMA NATIVE IP.
> > > > I can only test it in remote HDMA IP setup with single dma transfer, but
> > > > with these fixes it works properly.
> > > > 
> > > > Few fixes has also been added for eDMA version. Similarly to HDMA I have
> > > > tested only eDMA in remote setup.
> > > > 
> > > 
> > > Just out of curiosity, can you share how you are setting EDMA_MF_HDMA_NATIVE?
> > 
> > This topic has already been concerned on v1 (in another context
> > though):
> > https://lore.kernel.org/dmaengine/20230621151948.36125997@kmaincent-XPS-13-7390/
> > 
> > Here is the repo with the out-of-tree driver Kory said he was using
> > together with the kernel's version of the DW eDMA/hDMA driver:
> > https://github.com/Brainchip-Inc/akida_dw_edma
> > 
> 

> Thanks Sergey, I missed it! But looks like we are not focusing on the HDMA
> integration in designware-ep.c. Have you/anyone thought about it? Was it
> discussed previously that I missed?

No. We haven't discussed that in the framework of this patchset.

> 
> HDMA is used in one of the recent Qcom SoCs (SA8775) that Qcom folks are
> bringing up and I'd like to have a common solution like we have for eDMA.

AFAICS it won't be that easy to do for HDMA. Unlike eDMA, HDMA doesn't
have a handy global config registers to determine the number of R/W
channels.  Kory also said that auto-detecting them by dummy-writing to
all the CH_EN registers didn't work either because all, even
unavailable, channels CSRs were writable. This part was discussed
earlier:
https://lore.kernel.org/lkml/20230607144014.6356a197@kmaincent-XPS-13-7390/
So if you don't come up with some more clever solution, then alas the
number of R/W channels will need to be specified by the platform
code/driver.

Regarding how to auto-detect HDMA. I can't be absolutely sure whether
it will work but if we assume that:
1. HDMA reg-space is always unrolled (mapped over a separate reg-space),
2. Lowest 16 bits of base+0x8 are RO in EDMA (DMA_CTRL_OFF) and RW in HDMA
(prefetch CSR),
then we can implement a procedure like this:

1. If iATU/xDMA reg-space is specified and it's writable at the
xDMA-base+0x8 then it's HDMA controller and amount of channels is
supposed to be pre-initialized by the low-level platform driver,
otherwise it's eDMA and the read value can be used to determine the
number of channels.
2. If iATU/xDMA reg-space isn't specified then the viewport-based eDMA
auto-detection procedure will be executed.

For all of that you'll need to fix the
dw_pcie_edma_find_chip()/dw_pcie_edma_detect() method somehow.

Alternatively, to keep things simple you can convert the
dw_pcie_edma_find_chip()/dw_pcie_edma_detect() methods to just relying
on the HDMA settings being fully specified by the low-level drivers.

-Serge(y)

> 
> - Mani
> 
> > -Serge(y)
> > 
> > > 
> > > - Mani
> > > 
> > > > Changes in v2:
> > > > - Update comments and fix typos.
> > > > - Removed patches that tackle hypothetical bug and then were not pertinent.
> > > > - Add the similar HDMA race condition in remote setup fix to eDMA IP driver.
> > > > 
> > > > Changes in v3:
> > > > - Fix comment style.
> > > > - Split a patch in two to differ bug fix and simple harmless typo.
> > > > 
> > > > Changes in v4:
> > > > - Update patch git commit message.
> > > > - Link to v3: https://lore.kernel.org/r/20231011-b4-feature_hdma_mainline-v3-0-24ee0c979c6f@xxxxxxxxxxx
> > > > 
> > > > Changes in v5:
> > > > - No change
> > > > - Rebase to mainline 6.7-rc1
> > > > - Link to v4: https://lore.kernel.org/r/20231011-b4-feature_hdma_mainline-v4-0-43d417b93138@xxxxxxxxxxx
> > > > 
> > > > Changes in v6:
> > > > - Fix several commit messages and comments.
> > > > - Link to v5: https://lore.kernel.org/r/20231114-b4-feature_hdma_mainline-v5-0-7bc86d83c6f7@xxxxxxxxxxx
> > > > 
> > > > Signed-off-by: Kory Maincent <kory.maincent@xxxxxxxxxxx>
> > > > ---
> > > > Kory Maincent (6):
> > > >       dmaengine: dw-edma: Fix the ch_count hdma callback
> > > >       dmaengine: dw-edma: Fix wrong interrupt bit set for HDMA
> > > >       dmaengine: dw-edma: HDMA_V0_REMOTEL_STOP_INT_EN typo fix
> > > >       dmaengine: dw-edma: Add HDMA remote interrupt configuration
> > > >       dmaengine: dw-edma: HDMA: Add sync read before starting the DMA transfer in remote setup
> > > >       dmaengine: dw-edma: eDMA: Add sync read before starting the DMA transfer in remote setup
> > > > 
> > > >  drivers/dma/dw-edma/dw-edma-v0-core.c | 17 +++++++++++++++
> > > >  drivers/dma/dw-edma/dw-hdma-v0-core.c | 39 +++++++++++++++++++++++------------
> > > >  drivers/dma/dw-edma/dw-hdma-v0-regs.h |  2 +-
> > > >  3 files changed, 44 insertions(+), 14 deletions(-)
> > > > ---
> > > > base-commit: b85ea95d086471afb4ad062012a4d73cd328fa86
> > > > change-id: 20231011-b4-feature_hdma_mainline-b6c57f8e3b5d
> > > > 
> > > > Best regards,
> > > > -- 
> > > > Köry Maincent, Bootlin
> > > > Embedded Linux and kernel engineering
> > > > https://bootlin.com
> > > > 
> > > 
> > > -- 
> > > மணிவண்ணன் சதாசிவம்
> 
> -- 
> மணிவண்ணன் சதாசிவம்