Message ID | 20210322173949.1156393-1-kieran.bingham+renesas@ideasonboard.com (mailing list archive) |
---|---|
State | New |
Delegated to: | Kieran Bingham |
Headers | show |
Series | [v2] media: vsp1: Add support for the V3U VSPD | expand |
Hi Kieran, Thank you for the patch. On Mon, Mar 22, 2021 at 05:39:49PM +0000, Kieran Bingham wrote: > The V3U provides two VSPD instances, with a new update to the version > register to detect the new SoC. > > Add the new version and model detection, and detail the features > available in this module. > > Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com> > --- > This patch adds in the VSPd on the V3U, and successfully probes and can > read and write to registers. > > However, as yet I have not been able to successfully validate the VSPD > using the UAPI interface (by forcing .uapi = true) Stupid question: while I have set .uapi to true to test VSPD instances in Gen3 SoCs, this isn't something we routinely do, so maybe you should first check on a Gen3 board if this hack still works ? > The observed symptoms show that the hardware halts at the first display > list queued to the device. > > Notably, however the display list has been processed, and the registers > set by the display list are updated accordingly and can be read back, > inferring that the display list was processed and the frame commenced. > > Alas the frame never completes, and no interrupts are generated. > > Investigating this, I have seen that the CPG MSSR configures the FCPVD > and VSPD on the R8A779A0_CLK_S3D1 clock, which appears to be a 266666656 > clock. This seems low, and I would expect the VSP to share the same > clocking as the VIN/VSPX, which is on R8A779A0_CLK_S1D1. > > However, changing those clocks has no effect on the operation of the > VSPD. > > drivers/media/platform/vsp1/vsp1_drv.c | 10 ++++++++++ > drivers/media/platform/vsp1/vsp1_regs.h | 3 +++ > 2 files changed, 13 insertions(+) > > diff --git a/drivers/media/platform/vsp1/vsp1_drv.c b/drivers/media/platform/vsp1/vsp1_drv.c > index aa66e4f5f3f3..0a9812206b3f 100644 > --- a/drivers/media/platform/vsp1/vsp1_drv.c > +++ b/drivers/media/platform/vsp1/vsp1_drv.c > @@ -785,6 +785,16 @@ static const struct vsp1_device_info vsp1_device_infos[] = { > .uif_count = 2, > .wpf_count = 2, > .num_bru_inputs = 5, > + }, { > + .version = VI6_IP_VERSION_MODEL_VSPD_V3U, > + .model = "VSP2-D", > + .gen = 3, > + .features = VSP1_HAS_BRU | VSP1_HAS_CLU | VSP1_HAS_EXT_DL, Unless I'm mistaken, the V3U VSPD has no CLU. It has a CLUT in RPF2, but that's a different feature. The BSP code also sets VSP1_HAS_WPF_VFLIP here, but according to the documentation, that's not correct, so I'd skip it for now. With this, Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> > + .lif_count = 1, > + .rpf_count = 5, > + .uif_count = 2, > + .wpf_count = 1, > + .num_bru_inputs = 5, > }, > }; > > diff --git a/drivers/media/platform/vsp1/vsp1_regs.h b/drivers/media/platform/vsp1/vsp1_regs.h > index fe3130db1fa2..b378ea4451ce 100644 > --- a/drivers/media/platform/vsp1/vsp1_regs.h > +++ b/drivers/media/platform/vsp1/vsp1_regs.h > @@ -766,6 +766,8 @@ > #define VI6_IP_VERSION_MODEL_VSPD_V3 (0x18 << 8) > #define VI6_IP_VERSION_MODEL_VSPDL_GEN3 (0x19 << 8) > #define VI6_IP_VERSION_MODEL_VSPBS_GEN3 (0x1a << 8) > +#define VI6_IP_VERSION_MODEL_VSPD_V3U (0x1c << 8) > + > #define VI6_IP_VERSION_SOC_MASK (0xff << 0) > #define VI6_IP_VERSION_SOC_H2 (0x01 << 0) > #define VI6_IP_VERSION_SOC_V2H (0x01 << 0) > @@ -777,6 +779,7 @@ > #define VI6_IP_VERSION_SOC_D3 (0x04 << 0) > #define VI6_IP_VERSION_SOC_M3N (0x04 << 0) > #define VI6_IP_VERSION_SOC_E3 (0x04 << 0) > +#define VI6_IP_VERSION_SOC_V3U (0x05 << 0) > > /* ----------------------------------------------------------------------------- > * RPF CLUT Registers
Hi Laurent, On 22/03/2021 19:08, Laurent Pinchart wrote: > Hi Kieran, > > Thank you for the patch. > > On Mon, Mar 22, 2021 at 05:39:49PM +0000, Kieran Bingham wrote: >> The V3U provides two VSPD instances, with a new update to the version >> register to detect the new SoC. >> >> Add the new version and model detection, and detail the features >> available in this module. >> >> Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com> >> --- >> This patch adds in the VSPd on the V3U, and successfully probes and can >> read and write to registers. >> >> However, as yet I have not been able to successfully validate the VSPD >> using the UAPI interface (by forcing .uapi = true) > > Stupid question: while I have set .uapi to true to test VSPD instances > in Gen3 SoCs, this isn't something we routinely do, so maybe you should > first check on a Gen3 board if this hack still works ? I did that already ;-) Works on the H3 Salvator-XS. >> The observed symptoms show that the hardware halts at the first display >> list queued to the device. >> >> Notably, however the display list has been processed, and the registers >> set by the display list are updated accordingly and can be read back, >> inferring that the display list was processed and the frame commenced. >> >> Alas the frame never completes, and no interrupts are generated. >> >> Investigating this, I have seen that the CPG MSSR configures the FCPVD >> and VSPD on the R8A779A0_CLK_S3D1 clock, which appears to be a 266666656 >> clock. This seems low, and I would expect the VSP to share the same >> clocking as the VIN/VSPX, which is on R8A779A0_CLK_S1D1. >> >> However, changing those clocks has no effect on the operation of the >> VSPD. >> >> drivers/media/platform/vsp1/vsp1_drv.c | 10 ++++++++++ >> drivers/media/platform/vsp1/vsp1_regs.h | 3 +++ >> 2 files changed, 13 insertions(+) >> >> diff --git a/drivers/media/platform/vsp1/vsp1_drv.c b/drivers/media/platform/vsp1/vsp1_drv.c >> index aa66e4f5f3f3..0a9812206b3f 100644 >> --- a/drivers/media/platform/vsp1/vsp1_drv.c >> +++ b/drivers/media/platform/vsp1/vsp1_drv.c >> @@ -785,6 +785,16 @@ static const struct vsp1_device_info vsp1_device_infos[] = { >> .uif_count = 2, >> .wpf_count = 2, >> .num_bru_inputs = 5, >> + }, { >> + .version = VI6_IP_VERSION_MODEL_VSPD_V3U, >> + .model = "VSP2-D", >> + .gen = 3, >> + .features = VSP1_HAS_BRU | VSP1_HAS_CLU | VSP1_HAS_EXT_DL, > > Unless I'm mistaken, the V3U VSPD has no CLU. It has a CLUT in RPF2, but > that's a different feature. > > The BSP code also sets VSP1_HAS_WPF_VFLIP here, but according to the > documentation, that's not correct, so I'd skip it for now. Yes, I saw the incorrect VFLIP in the BSP, but I had not realised CLU != CLUT ;-) I thought that was shorthand for it... Ooops. I'll update. > > With this, > > Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> > >> + .lif_count = 1, >> + .rpf_count = 5, >> + .uif_count = 2, >> + .wpf_count = 1, >> + .num_bru_inputs = 5, >> }, >> }; >> >> diff --git a/drivers/media/platform/vsp1/vsp1_regs.h b/drivers/media/platform/vsp1/vsp1_regs.h >> index fe3130db1fa2..b378ea4451ce 100644 >> --- a/drivers/media/platform/vsp1/vsp1_regs.h >> +++ b/drivers/media/platform/vsp1/vsp1_regs.h >> @@ -766,6 +766,8 @@ >> #define VI6_IP_VERSION_MODEL_VSPD_V3 (0x18 << 8) >> #define VI6_IP_VERSION_MODEL_VSPDL_GEN3 (0x19 << 8) >> #define VI6_IP_VERSION_MODEL_VSPBS_GEN3 (0x1a << 8) >> +#define VI6_IP_VERSION_MODEL_VSPD_V3U (0x1c << 8) >> + >> #define VI6_IP_VERSION_SOC_MASK (0xff << 0) >> #define VI6_IP_VERSION_SOC_H2 (0x01 << 0) >> #define VI6_IP_VERSION_SOC_V2H (0x01 << 0) >> @@ -777,6 +779,7 @@ >> #define VI6_IP_VERSION_SOC_D3 (0x04 << 0) >> #define VI6_IP_VERSION_SOC_M3N (0x04 << 0) >> #define VI6_IP_VERSION_SOC_E3 (0x04 << 0) >> +#define VI6_IP_VERSION_SOC_V3U (0x05 << 0) >> >> /* ----------------------------------------------------------------------------- >> * RPF CLUT Registers >
diff --git a/drivers/media/platform/vsp1/vsp1_drv.c b/drivers/media/platform/vsp1/vsp1_drv.c index aa66e4f5f3f3..0a9812206b3f 100644 --- a/drivers/media/platform/vsp1/vsp1_drv.c +++ b/drivers/media/platform/vsp1/vsp1_drv.c @@ -785,6 +785,16 @@ static const struct vsp1_device_info vsp1_device_infos[] = { .uif_count = 2, .wpf_count = 2, .num_bru_inputs = 5, + }, { + .version = VI6_IP_VERSION_MODEL_VSPD_V3U, + .model = "VSP2-D", + .gen = 3, + .features = VSP1_HAS_BRU | VSP1_HAS_CLU | VSP1_HAS_EXT_DL, + .lif_count = 1, + .rpf_count = 5, + .uif_count = 2, + .wpf_count = 1, + .num_bru_inputs = 5, }, }; diff --git a/drivers/media/platform/vsp1/vsp1_regs.h b/drivers/media/platform/vsp1/vsp1_regs.h index fe3130db1fa2..b378ea4451ce 100644 --- a/drivers/media/platform/vsp1/vsp1_regs.h +++ b/drivers/media/platform/vsp1/vsp1_regs.h @@ -766,6 +766,8 @@ #define VI6_IP_VERSION_MODEL_VSPD_V3 (0x18 << 8) #define VI6_IP_VERSION_MODEL_VSPDL_GEN3 (0x19 << 8) #define VI6_IP_VERSION_MODEL_VSPBS_GEN3 (0x1a << 8) +#define VI6_IP_VERSION_MODEL_VSPD_V3U (0x1c << 8) + #define VI6_IP_VERSION_SOC_MASK (0xff << 0) #define VI6_IP_VERSION_SOC_H2 (0x01 << 0) #define VI6_IP_VERSION_SOC_V2H (0x01 << 0) @@ -777,6 +779,7 @@ #define VI6_IP_VERSION_SOC_D3 (0x04 << 0) #define VI6_IP_VERSION_SOC_M3N (0x04 << 0) #define VI6_IP_VERSION_SOC_E3 (0x04 << 0) +#define VI6_IP_VERSION_SOC_V3U (0x05 << 0) /* ----------------------------------------------------------------------------- * RPF CLUT Registers
The V3U provides two VSPD instances, with a new update to the version register to detect the new SoC. Add the new version and model detection, and detail the features available in this module. Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com> --- This patch adds in the VSPd on the V3U, and successfully probes and can read and write to registers. However, as yet I have not been able to successfully validate the VSPD using the UAPI interface (by forcing .uapi = true) The observed symptoms show that the hardware halts at the first display list queued to the device. Notably, however the display list has been processed, and the registers set by the display list are updated accordingly and can be read back, inferring that the display list was processed and the frame commenced. Alas the frame never completes, and no interrupts are generated. Investigating this, I have seen that the CPG MSSR configures the FCPVD and VSPD on the R8A779A0_CLK_S3D1 clock, which appears to be a 266666656 clock. This seems low, and I would expect the VSP to share the same clocking as the VIN/VSPX, which is on R8A779A0_CLK_S1D1. However, changing those clocks has no effect on the operation of the VSPD. drivers/media/platform/vsp1/vsp1_drv.c | 10 ++++++++++ drivers/media/platform/vsp1/vsp1_regs.h | 3 +++ 2 files changed, 13 insertions(+)