Only IRIS2(_1) should enter the until-now-IS_V6() path and the
condition for skipping part of it should be IS_IRIS2_1 and not the
number of VPP pipes. Fix that.
Fixes: 4b0b6e147dc9 ("media: venus: hfi: Add 6xx AXI halt logic")
Fixes: 78d434ba8659 ("media: venus: hfi: Skip AON register programming for V6 1pipe")
Signed-off-by: Konrad Dybcio <[email protected]>
---
drivers/media/platform/qcom/venus/hfi_venus.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/media/platform/qcom/venus/hfi_venus.c b/drivers/media/platform/qcom/venus/hfi_venus.c
index 9b840440a115..ca56b1a8eb71 100644
--- a/drivers/media/platform/qcom/venus/hfi_venus.c
+++ b/drivers/media/platform/qcom/venus/hfi_venus.c
@@ -549,10 +549,10 @@ static int venus_halt_axi(struct venus_hfi_device *hdev)
u32 mask_val;
int ret;
- if (IS_V6(hdev->core)) {
+ if (IS_IRIS2(hdev->core) || IS_IRIS2_1(hdev->core)) {
writel(0x3, cpu_cs_base + CPU_CS_X2RPMH_V6);
- if (hdev->core->res->num_vpp_pipes == 1)
+ if (IS_IRIS2_1(hdev->core))
goto skip_aon_mvp_noc;
writel(0x1, aon_base + AON_WRAPPER_MVP_NOC_LPI_CONTROL);
--
2.40.1
On 5/4/2023 1:31 PM, Konrad Dybcio wrote:
> Only IRIS2(_1) should enter the until-now-IS_V6() path and the
> condition for skipping part of it should be IS_IRIS2_1 and not the
> number of VPP pipes. Fix that.
Do not see any issue with existing code. IRIS2 with single pipe is
IRIS2_1. This does not
quality as a fix to earlier implementation. Since this series introduces
VPU versions,
IRIS2 with 1 pipe is being replaced with IRIS2_1.
-Vikash
>
> Fixes: 4b0b6e147dc9 ("media: venus: hfi: Add 6xx AXI halt logic")
> Fixes: 78d434ba8659 ("media: venus: hfi: Skip AON register programming for V6 1pipe")
> Signed-off-by: Konrad Dybcio <[email protected]>
> ---
> drivers/media/platform/qcom/venus/hfi_venus.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/media/platform/qcom/venus/hfi_venus.c b/drivers/media/platform/qcom/venus/hfi_venus.c
> index 9b840440a115..ca56b1a8eb71 100644
> --- a/drivers/media/platform/qcom/venus/hfi_venus.c
> +++ b/drivers/media/platform/qcom/venus/hfi_venus.c
> @@ -549,10 +549,10 @@ static int venus_halt_axi(struct venus_hfi_device *hdev)
> u32 mask_val;
> int ret;
>
> - if (IS_V6(hdev->core)) {
> + if (IS_IRIS2(hdev->core) || IS_IRIS2_1(hdev->core)) {
> writel(0x3, cpu_cs_base + CPU_CS_X2RPMH_V6);
>
> - if (hdev->core->res->num_vpp_pipes == 1)
> + if (IS_IRIS2_1(hdev->core))
> goto skip_aon_mvp_noc;
>
> writel(0x1, aon_base + AON_WRAPPER_MVP_NOC_LPI_CONTROL);
>
On Fri, 5 May 2023 at 16:22, Vikash Garodia <[email protected]> wrote:
>
>
> On 5/4/2023 1:31 PM, Konrad Dybcio wrote:
> > Only IRIS2(_1) should enter the until-now-IS_V6() path and the
> > condition for skipping part of it should be IS_IRIS2_1 and not the
> > number of VPP pipes. Fix that.
>
> Do not see any issue with existing code. IRIS2 with single pipe is
> IRIS2_1. This does not
>
> quality as a fix to earlier implementation. Since this series introduces
> VPU versions,
>
> IRIS2 with 1 pipe is being replaced with IRIS2_1.
Could you please fix the line wrapping of your emails. It becomes hard
to read them otherwise.
--
With best wishes
Dmitry
On 5/5/2023 8:13 PM, Dmitry Baryshkov wrote:
> On Fri, 5 May 2023 at 16:22, Vikash Garodia <[email protected]> wrote:
>>
>> On 5/4/2023 1:31 PM, Konrad Dybcio wrote:
>>> Only IRIS2(_1) should enter the until-now-IS_V6() path and the
>>> condition for skipping part of it should be IS_IRIS2_1 and not the
>>> number of VPP pipes. Fix that.
>> Do not see any issue with existing code. IRIS2 with single pipe is
>> IRIS2_1. This does not
>>
>> quality as a fix to earlier implementation. Since this series introduces
>> VPU versions,
>>
>> IRIS2 with 1 pipe is being replaced with IRIS2_1.
> Could you please fix the line wrapping of your emails. It becomes hard
> to read them otherwise.
My apologies for the inconvenience. Tried to set the email wrap config, but it
does not take
an effect on my email client. Will try playing around a bit with it and fix it.
Meanwhile go through the comments and if you can align them with minimal effort, we
can continue to discuss on the same.
-Vikash
>
>
On 5.05.2023 15:21, Vikash Garodia wrote:
>
> On 5/4/2023 1:31 PM, Konrad Dybcio wrote:
>> Only IRIS2(_1) should enter the until-now-IS_V6() path and the
>> condition for skipping part of it should be IS_IRIS2_1 and not the
>> number of VPP pipes. Fix that.
>
> Do not see any issue with existing code. IRIS2 with single pipe is IRIS2_1. This does not
>
> quality as a fix to earlier implementation. Since this series introduces VPU versions,
>
> IRIS2 with 1 pipe is being replaced with IRIS2_1.
>
> -Vikash
Right, I'll drop the fixes tags.
Also, your email client seems to be inserting 2 newlines instead of 1.
If you're using thunderbird, you may want to edit:
mail.identity.(default or your mail identity idx).default.compose_html
to `false`
or you can use shift+enter as a half-measure
Konrad
>
>>
>> Fixes: 4b0b6e147dc9 ("media: venus: hfi: Add 6xx AXI halt logic")
>> Fixes: 78d434ba8659 ("media: venus: hfi: Skip AON register programming for V6 1pipe")
>> Signed-off-by: Konrad Dybcio <[email protected]>
>> ---
>> drivers/media/platform/qcom/venus/hfi_venus.c | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/media/platform/qcom/venus/hfi_venus.c b/drivers/media/platform/qcom/venus/hfi_venus.c
>> index 9b840440a115..ca56b1a8eb71 100644
>> --- a/drivers/media/platform/qcom/venus/hfi_venus.c
>> +++ b/drivers/media/platform/qcom/venus/hfi_venus.c
>> @@ -549,10 +549,10 @@ static int venus_halt_axi(struct venus_hfi_device *hdev)
>> u32 mask_val;
>> int ret;
>> - if (IS_V6(hdev->core)) {
>> + if (IS_IRIS2(hdev->core) || IS_IRIS2_1(hdev->core)) {
>> writel(0x3, cpu_cs_base + CPU_CS_X2RPMH_V6);
>> - if (hdev->core->res->num_vpp_pipes == 1)
>> + if (IS_IRIS2_1(hdev->core))
>> goto skip_aon_mvp_noc;
>> writel(0x1, aon_base + AON_WRAPPER_MVP_NOC_LPI_CONTROL);
>>