We have been exercizing the RZ/V2H(P) IVC block quite intensly these
last two months.

Here it is a collection of fixes and improvements to the driver.

The first 4 patches in the series address a few registers writes that
do not respect the documentation.

The 5th and 6th patches fixes concurrent access to the list of queued
buffers and fix a WARN() visible under heavy system load conditions
caused by concurrent buffer transfers.

The last patch is actually up for discussion. It is my opinion that the
trouble of setting up a workqueue item is not justified by the
relatively small amount of work that has to be carried out in interrupt
context. In any case, there shouldn't be any functional change
introduced by this patch.

Patch #7 makes patch #6 reduntant: if we use direct function
calls, then the issue of concurrently running workqueue items cannot
happen. However, I actually think patch #6 has value regardless as it
makes the code more robust.

Signed-off-by: Jacopo Mondi <jacopo.mondi+renesas@ideasonboard.com>
---
Barnabás Pőcze (4):
      media: rzv2h-ivc: Fix AXIRX_VBLANK register write
      media: rzv2h-ivc: Write AXIRX_PIXFMT once
      media: rzv2h-ivc: Fix FM_STOP register write
      media: rzv2h-ivc: Fix concurrent buffer list access

Daniel Scally (1):
      media: rzv2h-ivc: Revise default VBLANK formula

Jacopo Mondi (2):
      media: rzv2h-ivc: Avoid double job scheduling
      media: rzv2h-ivc: Replace workqueue with direct function call

 .../platform/renesas/rzv2h-ivc/rzv2h-ivc-dev.c     |  2 +-
 .../platform/renesas/rzv2h-ivc/rzv2h-ivc-video.c   | 63 +++++++++++-----------
 .../media/platform/renesas/rzv2h-ivc/rzv2h-ivc.h   | 13 ++---
 3 files changed, 39 insertions(+), 39 deletions(-)
---
base-commit: f6390408a846aacc2171c17d88b062e202d84e86
change-id: 20260311-mali-ivc-fixes-v7-0-43fc33b87793

Best regards,
-- 
Jacopo Mondi <jacopo.mondi@ideasonboard.com>

Hi Jacopo,

Thank you for the patches.

On Fri, Mar 13, 2026 at 11:14 AM Jacopo Mondi
<jacopo.mondi@ideasonboard.com> wrote:
>
> We have been exercizing the RZ/V2H(P) IVC block quite intensly these
> last two months.
>
> Here it is a collection of fixes and improvements to the driver.
>
> The first 4 patches in the series address a few registers writes that
> do not respect the documentation.
>
> The 5th and 6th patches fixes concurrent access to the list of queued
> buffers and fix a WARN() visible under heavy system load conditions
> caused by concurrent buffer transfers.
>
> The last patch is actually up for discussion. It is my opinion that the
> trouble of setting up a workqueue item is not justified by the
> relatively small amount of work that has to be carried out in interrupt
> context. In any case, there shouldn't be any functional change
> introduced by this patch.
>
> Patch #7 makes patch #6 reduntant: if we use direct function
> calls, then the issue of concurrently running workqueue items cannot
> happen. However, I actually think patch #6 has value regardless as it
> makes the code more robust.
>
> Signed-off-by: Jacopo Mondi <jacopo.mondi+renesas@ideasonboard.com>
> ---
> Barnabás Pőcze (4):
>       media: rzv2h-ivc: Fix AXIRX_VBLANK register write
>       media: rzv2h-ivc: Write AXIRX_PIXFMT once
>       media: rzv2h-ivc: Fix FM_STOP register write
>       media: rzv2h-ivc: Fix concurrent buffer list access
>
> Daniel Scally (1):
>       media: rzv2h-ivc: Revise default VBLANK formula
>
> Jacopo Mondi (2):
>       media: rzv2h-ivc: Avoid double job scheduling
>       media: rzv2h-ivc: Replace workqueue with direct function call
>
Tested the patches on RZ/V2H EVK with IMX708 sensor on next-20260319.

Tested-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com> #
On RZ/V2H EVK

Cheers,
Prabhakar