diff mbox series

[V1] Revert "PCI: tegra194: Enable support for 256 Byte payload"

Message ID 20230608093652.1409485-1-vidyas@nvidia.com
State Handled Elsewhere
Headers show
Series [V1] Revert "PCI: tegra194: Enable support for 256 Byte payload" | expand

Commit Message

Vidya Sagar June 8, 2023, 9:36 a.m. UTC
This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
support for 256 Byte payload")

Consider a PCIe hierarchy with a PCIe switch and a device connected
downstream of the switch that has support for MPS which is the minimum
in the hierarchy, and root port programmed with an MPS in its DevCtl
register that is greater than the minimum. In this scenario, the default
bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
configure the MPS settings in the hierarchy correctly resulting in the
device with support for minimum MPS in the hierarchy receiving the TLPs
of size more than that. Although this can be addresed by appending
"pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
good idea to always have this commandline argument even for the basic
functionality to work.
Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
Byte payload") avoids this requirement and ensures that the basic
functionality of the devices irrespective of the hierarchy and the MPS of
the devices in the hierarchy.
To reap the benefits of having support for higher MPS, optionally, one can
always append the kernel command line with "pci=pcie_bus_perf".

Fixes: 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte payload")
Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
---
 drivers/pci/controller/dwc/pcie-tegra194.c | 13 -------------
 1 file changed, 13 deletions(-)

Comments

Bjorn Helgaas June 8, 2023, 4:33 p.m. UTC | #1
On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
> This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
> support for 256 Byte payload")
> 
> Consider a PCIe hierarchy with a PCIe switch and a device connected
> downstream of the switch that has support for MPS which is the minimum
> in the hierarchy, and root port programmed with an MPS in its DevCtl
> register that is greater than the minimum. In this scenario, the default
> bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
> configure the MPS settings in the hierarchy correctly resulting in the
> device with support for minimum MPS in the hierarchy receiving the TLPs
> of size more than that. Although this can be addresed by appending
> "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
> good idea to always have this commandline argument even for the basic
> functionality to work.
> Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
> Byte payload") avoids this requirement and ensures that the basic
> functionality of the devices irrespective of the hierarchy and the MPS of
> the devices in the hierarchy.
> To reap the benefits of having support for higher MPS, optionally, one can
> always append the kernel command line with "pci=pcie_bus_perf".

Please add blank lines between paragraphs and wrap to fill 75 columns.
Also add a period at the end of the very first sentence.

s/addresed/addressed/

I guess that without 4fb8e46c1bc4, Linux configured everything with
128 byte MPS, and 4fb8e46c1bc4 was intended as an optimization to
allow 256 byte MPS.

If the Root Port advertises Max_Payload_Size Supported as 256 bytes in
DevCap, and the PCI core doesn't configure MPS=256 when possible, I'd
argue that should be fixed in the PCI core without a driver change
like this.

Bjorn
Vidya Sagar June 9, 2023, 2:23 a.m. UTC | #2
On 6/8/2023 10:03 PM, Bjorn Helgaas wrote:
> External email: Use caution opening links or attachments
> 
> 
> On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
>> This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
>> support for 256 Byte payload")
>>
>> Consider a PCIe hierarchy with a PCIe switch and a device connected
>> downstream of the switch that has support for MPS which is the minimum
>> in the hierarchy, and root port programmed with an MPS in its DevCtl
>> register that is greater than the minimum. In this scenario, the default
>> bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
>> configure the MPS settings in the hierarchy correctly resulting in the
>> device with support for minimum MPS in the hierarchy receiving the TLPs
>> of size more than that. Although this can be addresed by appending
>> "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
>> good idea to always have this commandline argument even for the basic
>> functionality to work.
>> Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
>> Byte payload") avoids this requirement and ensures that the basic
>> functionality of the devices irrespective of the hierarchy and the MPS of
>> the devices in the hierarchy.
>> To reap the benefits of having support for higher MPS, optionally, one can
>> always append the kernel command line with "pci=pcie_bus_perf".
> 
> Please add blank lines between paragraphs and wrap to fill 75 columns.
> Also add a period at the end of the very first sentence.
> 
> s/addresed/addressed/
> 
I'll address your comments in the next patch.

> I guess that without 4fb8e46c1bc4, Linux configured everything with
> 128 byte MPS, and 4fb8e46c1bc4 was intended as an optimization to
> allow 256 byte MPS.
Correct.

> 
> If the Root Port advertises Max_Payload_Size Supported as 256 bytes in
> DevCap, and the PCI core doesn't configure MPS=256 when possible, I'd
> argue that should be fixed in the PCI core without a driver change
> like this.
Well, kernel does configure MPS=256 but only if the 'perf' configuration
option is selected. 'perf' configuration option also changes the MRRS,
to extract the maximum performance. I'm not sure about the reasons for 
not making 'perf' configuration as the default configuration though.
(IIUC, this is what you are coming to, right?)

The current patch which is a revert of an earlier patch is to keep 
things working for different PCIe hierarchies given the default
configuration that kernel is using at the moment.

-Vidya Sagar

> 
> Bjorn
Jon Hunter June 14, 2023, 10:39 a.m. UTC | #3
Hi Bjorn,

On 09/06/2023 03:23, Vidya Sagar wrote:
> 
> 
> On 6/8/2023 10:03 PM, Bjorn Helgaas wrote:
>> External email: Use caution opening links or attachments
>>
>>
>> On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
>>> This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
>>> support for 256 Byte payload")
>>>
>>> Consider a PCIe hierarchy with a PCIe switch and a device connected
>>> downstream of the switch that has support for MPS which is the minimum
>>> in the hierarchy, and root port programmed with an MPS in its DevCtl
>>> register that is greater than the minimum. In this scenario, the default
>>> bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
>>> configure the MPS settings in the hierarchy correctly resulting in the
>>> device with support for minimum MPS in the hierarchy receiving the TLPs
>>> of size more than that. Although this can be addresed by appending
>>> "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
>>> good idea to always have this commandline argument even for the basic
>>> functionality to work.
>>> Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
>>> Byte payload") avoids this requirement and ensures that the basic
>>> functionality of the devices irrespective of the hierarchy and the 
>>> MPS of
>>> the devices in the hierarchy.
>>> To reap the benefits of having support for higher MPS, optionally, 
>>> one can
>>> always append the kernel command line with "pci=pcie_bus_perf".
>>
>> Please add blank lines between paragraphs and wrap to fill 75 columns.
>> Also add a period at the end of the very first sentence.
>>
>> s/addresed/addressed/
>>
> I'll address your comments in the next patch.
> 
>> I guess that without 4fb8e46c1bc4, Linux configured everything with
>> 128 byte MPS, and 4fb8e46c1bc4 was intended as an optimization to
>> allow 256 byte MPS.
> Correct.
> 
>>
>> If the Root Port advertises Max_Payload_Size Supported as 256 bytes in
>> DevCap, and the PCI core doesn't configure MPS=256 when possible, I'd
>> argue that should be fixed in the PCI core without a driver change
>> like this.
> Well, kernel does configure MPS=256 but only if the 'perf' configuration
> option is selected. 'perf' configuration option also changes the MRRS,
> to extract the maximum performance. I'm not sure about the reasons for 
> not making 'perf' configuration as the default configuration though.
> (IIUC, this is what you are coming to, right?)
> 
> The current patch which is a revert of an earlier patch is to keep 
> things working for different PCIe hierarchies given the default
> configuration that kernel is using at the moment.


Any more feedback on this? If not, would be great to queue for v6.5. 
Feel free to add my ...

Acked-by: Jon Hunter <jonathanh@nvidia.com>

Cheers
Jon
Manivannan Sadhasivam July 25, 2023, 7:51 a.m. UTC | #4
On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
> This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
> support for 256 Byte payload")
> 
> Consider a PCIe hierarchy with a PCIe switch and a device connected
> downstream of the switch that has support for MPS which is the minimum
> in the hierarchy, and root port programmed with an MPS in its DevCtl
> register that is greater than the minimum. In this scenario, the default
> bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
> configure the MPS settings in the hierarchy correctly resulting in the
> device with support for minimum MPS in the hierarchy receiving the TLPs
> of size more than that. Although this can be addresed by appending
> "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
> good idea to always have this commandline argument even for the basic
> functionality to work.
> Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
> Byte payload") avoids this requirement and ensures that the basic
> functionality of the devices irrespective of the hierarchy and the MPS of
> the devices in the hierarchy.
> To reap the benefits of having support for higher MPS, optionally, one can
> always append the kernel command line with "pci=pcie_bus_perf".
> 
> Fixes: 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte payload")
> Signed-off-by: Vidya Sagar <vidyas@nvidia.com>

I know that this patch is merged. But I happen to test a similar change on Qcom
platform during a patch review and found that the PCI core changes MPS to 128
when a 128byte supported device is found:

[    3.174290] pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 128 (was 256, max 128)
[    3.186538] pci 0000:01:00.0: Max Payload Size set to 128 (was 128, max 128)

This was just randomly tested on a platform whose Root port DEVCAP was 128, but
it shouldn't matter. And I didn't change the default bus configuration.

Wondering how you ended up facing issues with it.

- Mani

> ---
>  drivers/pci/controller/dwc/pcie-tegra194.c | 13 -------------
>  1 file changed, 13 deletions(-)
> 
> diff --git a/drivers/pci/controller/dwc/pcie-tegra194.c b/drivers/pci/controller/dwc/pcie-tegra194.c
> index 4fdadc7b045f..877d81b13334 100644
> --- a/drivers/pci/controller/dwc/pcie-tegra194.c
> +++ b/drivers/pci/controller/dwc/pcie-tegra194.c
> @@ -892,7 +892,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
>  	struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
>  	struct tegra_pcie_dw *pcie = to_tegra_pcie(pci);
>  	u32 val;
> -	u16 val_16;
>  
>  	pp->bridge->ops = &tegra_pci_ops;
>  
> @@ -900,11 +899,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
>  		pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
>  							      PCI_CAP_ID_EXP);
>  
> -	val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
> -	val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
> -	val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
> -	dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
> -
>  	val = dw_pcie_readl_dbi(pci, PCI_IO_BASE);
>  	val &= ~(IO_BASE_IO_DECODE | IO_BASE_IO_DECODE_BIT8);
>  	dw_pcie_writel_dbi(pci, PCI_IO_BASE, val);
> @@ -1756,7 +1750,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>  	struct device *dev = pcie->dev;
>  	u32 val;
>  	int ret;
> -	u16 val_16;
>  
>  	if (pcie->ep_state == EP_STATE_ENABLED)
>  		return;
> @@ -1887,11 +1880,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>  	pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
>  						      PCI_CAP_ID_EXP);
>  
> -	val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
> -	val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
> -	val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
> -	dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
> -
>  	/* Clear Slot Clock Configuration bit if SRNS configuration */
>  	if (pcie->enable_srns) {
>  		val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base +
> @@ -1900,7 +1888,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>  		dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_LNKSTA,
>  				   val_16);
>  	}
> -
>  	clk_set_rate(pcie->core_clk, GEN4_CORE_CLK_FREQ);
>  
>  	val = (ep->msi_mem_phys & MSIX_ADDR_MATCH_LOW_OFF_MASK);
> -- 
> 2.25.1
>
Vidya Sagar July 25, 2023, 8:19 a.m. UTC | #5
On 7/25/2023 1:21 PM, Manivannan Sadhasivam wrote:
> External email: Use caution opening links or attachments
> 
> 
> On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
>> This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
>> support for 256 Byte payload")
>>
>> Consider a PCIe hierarchy with a PCIe switch and a device connected
>> downstream of the switch that has support for MPS which is the minimum
>> in the hierarchy, and root port programmed with an MPS in its DevCtl
>> register that is greater than the minimum. In this scenario, the default
>> bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
>> configure the MPS settings in the hierarchy correctly resulting in the
>> device with support for minimum MPS in the hierarchy receiving the TLPs
>> of size more than that. Although this can be addresed by appending
>> "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
>> good idea to always have this commandline argument even for the basic
>> functionality to work.
>> Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
>> Byte payload") avoids this requirement and ensures that the basic
>> functionality of the devices irrespective of the hierarchy and the MPS of
>> the devices in the hierarchy.
>> To reap the benefits of having support for higher MPS, optionally, one can
>> always append the kernel command line with "pci=pcie_bus_perf".
>>
>> Fixes: 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte payload")
>> Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
> 
> I know that this patch is merged. But I happen to test a similar change on Qcom
> platform during a patch review and found that the PCI core changes MPS to 128
> when a 128byte supported device is found:
> 
> [    3.174290] pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 128 (was 256, max 128)
> [    3.186538] pci 0000:01:00.0: Max Payload Size set to 128 (was 128, max 128)
> 
> This was just randomly tested on a platform whose Root port DEVCAP was 128, but
> it shouldn't matter. And I didn't change the default bus configuration.
> 
> Wondering how you ended up facing issues with it.

If the endpiont device that has support only for 128byte MPS is 
connected directly to the root port, then, I agree that the PCIe 
sub-system takes care of changing the MPS to the common lowest MPS, but 
if the endpoint device is connected behind a PCIe switch, then the PCIe 
subsystem doesn't configure the MPS properly.

-Vidya Sagar

> 
> - Mani
> 
>> ---
>>   drivers/pci/controller/dwc/pcie-tegra194.c | 13 -------------
>>   1 file changed, 13 deletions(-)
>>
>> diff --git a/drivers/pci/controller/dwc/pcie-tegra194.c b/drivers/pci/controller/dwc/pcie-tegra194.c
>> index 4fdadc7b045f..877d81b13334 100644
>> --- a/drivers/pci/controller/dwc/pcie-tegra194.c
>> +++ b/drivers/pci/controller/dwc/pcie-tegra194.c
>> @@ -892,7 +892,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
>>        struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
>>        struct tegra_pcie_dw *pcie = to_tegra_pcie(pci);
>>        u32 val;
>> -     u16 val_16;
>>
>>        pp->bridge->ops = &tegra_pci_ops;
>>
>> @@ -900,11 +899,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
>>                pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
>>                                                              PCI_CAP_ID_EXP);
>>
>> -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
>> -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
>> -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
>> -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
>> -
>>        val = dw_pcie_readl_dbi(pci, PCI_IO_BASE);
>>        val &= ~(IO_BASE_IO_DECODE | IO_BASE_IO_DECODE_BIT8);
>>        dw_pcie_writel_dbi(pci, PCI_IO_BASE, val);
>> @@ -1756,7 +1750,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>>        struct device *dev = pcie->dev;
>>        u32 val;
>>        int ret;
>> -     u16 val_16;
>>
>>        if (pcie->ep_state == EP_STATE_ENABLED)
>>                return;
>> @@ -1887,11 +1880,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>>        pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
>>                                                      PCI_CAP_ID_EXP);
>>
>> -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
>> -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
>> -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
>> -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
>> -
>>        /* Clear Slot Clock Configuration bit if SRNS configuration */
>>        if (pcie->enable_srns) {
>>                val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base +
>> @@ -1900,7 +1888,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>>                dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_LNKSTA,
>>                                   val_16);
>>        }
>> -
>>        clk_set_rate(pcie->core_clk, GEN4_CORE_CLK_FREQ);
>>
>>        val = (ep->msi_mem_phys & MSIX_ADDR_MATCH_LOW_OFF_MASK);
>> --
>> 2.25.1
>>
> 
> --
> மணிவண்ணன் சதாசிவம்
Manivannan Sadhasivam July 25, 2023, 8:30 a.m. UTC | #6
On Tue, Jul 25, 2023 at 01:49:35PM +0530, Vidya Sagar wrote:
> 
> 
> On 7/25/2023 1:21 PM, Manivannan Sadhasivam wrote:
> > External email: Use caution opening links or attachments
> > 
> > 
> > On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
> > > This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
> > > support for 256 Byte payload")
> > > 
> > > Consider a PCIe hierarchy with a PCIe switch and a device connected
> > > downstream of the switch that has support for MPS which is the minimum
> > > in the hierarchy, and root port programmed with an MPS in its DevCtl
> > > register that is greater than the minimum. In this scenario, the default
> > > bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
> > > configure the MPS settings in the hierarchy correctly resulting in the
> > > device with support for minimum MPS in the hierarchy receiving the TLPs
> > > of size more than that. Although this can be addresed by appending
> > > "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
> > > good idea to always have this commandline argument even for the basic
> > > functionality to work.
> > > Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
> > > Byte payload") avoids this requirement and ensures that the basic
> > > functionality of the devices irrespective of the hierarchy and the MPS of
> > > the devices in the hierarchy.
> > > To reap the benefits of having support for higher MPS, optionally, one can
> > > always append the kernel command line with "pci=pcie_bus_perf".
> > > 
> > > Fixes: 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte payload")
> > > Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
> > 
> > I know that this patch is merged. But I happen to test a similar change on Qcom
> > platform during a patch review and found that the PCI core changes MPS to 128
> > when a 128byte supported device is found:
> > 
> > [    3.174290] pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 128 (was 256, max 128)
> > [    3.186538] pci 0000:01:00.0: Max Payload Size set to 128 (was 128, max 128)
> > 
> > This was just randomly tested on a platform whose Root port DEVCAP was 128, but
> > it shouldn't matter. And I didn't change the default bus configuration.
> > 
> > Wondering how you ended up facing issues with it.
> 
> If the endpiont device that has support only for 128byte MPS is connected
> directly to the root port, then, I agree that the PCIe sub-system takes care
> of changing the MPS to the common lowest MPS, but if the endpoint device is
> connected behind a PCIe switch, then the PCIe subsystem doesn't configure
> the MPS properly.
> 

Ah, I missed the fact that your issue only happens with a PCIe switch. But
shouldn't this be fixed in the PCI core instead?

I mean, PCI core should use 128byte in your case for Root port unless it is not
allowed in the spec (which I doubt).

- Mani

> -Vidya Sagar
> 
> > 
> > - Mani
> > 
> > > ---
> > >   drivers/pci/controller/dwc/pcie-tegra194.c | 13 -------------
> > >   1 file changed, 13 deletions(-)
> > > 
> > > diff --git a/drivers/pci/controller/dwc/pcie-tegra194.c b/drivers/pci/controller/dwc/pcie-tegra194.c
> > > index 4fdadc7b045f..877d81b13334 100644
> > > --- a/drivers/pci/controller/dwc/pcie-tegra194.c
> > > +++ b/drivers/pci/controller/dwc/pcie-tegra194.c
> > > @@ -892,7 +892,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
> > >        struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
> > >        struct tegra_pcie_dw *pcie = to_tegra_pcie(pci);
> > >        u32 val;
> > > -     u16 val_16;
> > > 
> > >        pp->bridge->ops = &tegra_pci_ops;
> > > 
> > > @@ -900,11 +899,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
> > >                pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
> > >                                                              PCI_CAP_ID_EXP);
> > > 
> > > -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
> > > -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
> > > -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
> > > -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
> > > -
> > >        val = dw_pcie_readl_dbi(pci, PCI_IO_BASE);
> > >        val &= ~(IO_BASE_IO_DECODE | IO_BASE_IO_DECODE_BIT8);
> > >        dw_pcie_writel_dbi(pci, PCI_IO_BASE, val);
> > > @@ -1756,7 +1750,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
> > >        struct device *dev = pcie->dev;
> > >        u32 val;
> > >        int ret;
> > > -     u16 val_16;
> > > 
> > >        if (pcie->ep_state == EP_STATE_ENABLED)
> > >                return;
> > > @@ -1887,11 +1880,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
> > >        pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
> > >                                                      PCI_CAP_ID_EXP);
> > > 
> > > -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
> > > -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
> > > -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
> > > -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
> > > -
> > >        /* Clear Slot Clock Configuration bit if SRNS configuration */
> > >        if (pcie->enable_srns) {
> > >                val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base +
> > > @@ -1900,7 +1888,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
> > >                dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_LNKSTA,
> > >                                   val_16);
> > >        }
> > > -
> > >        clk_set_rate(pcie->core_clk, GEN4_CORE_CLK_FREQ);
> > > 
> > >        val = (ep->msi_mem_phys & MSIX_ADDR_MATCH_LOW_OFF_MASK);
> > > --
> > > 2.25.1
> > > 
> > 
> > --
> > மணிவண்ணன் சதாசிவம்
Vidya Sagar July 25, 2023, 9:21 a.m. UTC | #7
On 7/25/2023 2:00 PM, Manivannan Sadhasivam wrote:
> External email: Use caution opening links or attachments
> 
> 
> On Tue, Jul 25, 2023 at 01:49:35PM +0530, Vidya Sagar wrote:
>>
>>
>> On 7/25/2023 1:21 PM, Manivannan Sadhasivam wrote:
>>> External email: Use caution opening links or attachments
>>>
>>>
>>> On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
>>>> This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
>>>> support for 256 Byte payload")
>>>>
>>>> Consider a PCIe hierarchy with a PCIe switch and a device connected
>>>> downstream of the switch that has support for MPS which is the minimum
>>>> in the hierarchy, and root port programmed with an MPS in its DevCtl
>>>> register that is greater than the minimum. In this scenario, the default
>>>> bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
>>>> configure the MPS settings in the hierarchy correctly resulting in the
>>>> device with support for minimum MPS in the hierarchy receiving the TLPs
>>>> of size more than that. Although this can be addresed by appending
>>>> "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
>>>> good idea to always have this commandline argument even for the basic
>>>> functionality to work.
>>>> Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
>>>> Byte payload") avoids this requirement and ensures that the basic
>>>> functionality of the devices irrespective of the hierarchy and the MPS of
>>>> the devices in the hierarchy.
>>>> To reap the benefits of having support for higher MPS, optionally, one can
>>>> always append the kernel command line with "pci=pcie_bus_perf".
>>>>
>>>> Fixes: 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte payload")
>>>> Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
>>>
>>> I know that this patch is merged. But I happen to test a similar change on Qcom
>>> platform during a patch review and found that the PCI core changes MPS to 128
>>> when a 128byte supported device is found:
>>>
>>> [    3.174290] pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 128 (was 256, max 128)
>>> [    3.186538] pci 0000:01:00.0: Max Payload Size set to 128 (was 128, max 128)
>>>
>>> This was just randomly tested on a platform whose Root port DEVCAP was 128, but
>>> it shouldn't matter. And I didn't change the default bus configuration.
>>>
>>> Wondering how you ended up facing issues with it.
>>
>> If the endpiont device that has support only for 128byte MPS is connected
>> directly to the root port, then, I agree that the PCIe sub-system takes care
>> of changing the MPS to the common lowest MPS, but if the endpoint device is
>> connected behind a PCIe switch, then the PCIe subsystem doesn't configure
>> the MPS properly.
>>
> 
> Ah, I missed the fact that your issue only happens with a PCIe switch. But
> shouldn't this be fixed in the PCI core instead?
> 
> I mean, PCI core should use 128byte in your case for Root port unless it is not
> allowed in the spec (which I doubt).
well, if the RP's DevCtl is set to 256MPS by default, then, the core 
wouldn't do it automatically unless either 'pcie_bus_safe' or 
'pcie_bus_perf' is passed.


> 
> - Mani
> 
>> -Vidya Sagar
>>
>>>
>>> - Mani
>>>
>>>> ---
>>>>    drivers/pci/controller/dwc/pcie-tegra194.c | 13 -------------
>>>>    1 file changed, 13 deletions(-)
>>>>
>>>> diff --git a/drivers/pci/controller/dwc/pcie-tegra194.c b/drivers/pci/controller/dwc/pcie-tegra194.c
>>>> index 4fdadc7b045f..877d81b13334 100644
>>>> --- a/drivers/pci/controller/dwc/pcie-tegra194.c
>>>> +++ b/drivers/pci/controller/dwc/pcie-tegra194.c
>>>> @@ -892,7 +892,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
>>>>         struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
>>>>         struct tegra_pcie_dw *pcie = to_tegra_pcie(pci);
>>>>         u32 val;
>>>> -     u16 val_16;
>>>>
>>>>         pp->bridge->ops = &tegra_pci_ops;
>>>>
>>>> @@ -900,11 +899,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
>>>>                 pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
>>>>                                                               PCI_CAP_ID_EXP);
>>>>
>>>> -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
>>>> -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
>>>> -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
>>>> -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
>>>> -
>>>>         val = dw_pcie_readl_dbi(pci, PCI_IO_BASE);
>>>>         val &= ~(IO_BASE_IO_DECODE | IO_BASE_IO_DECODE_BIT8);
>>>>         dw_pcie_writel_dbi(pci, PCI_IO_BASE, val);
>>>> @@ -1756,7 +1750,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>>>>         struct device *dev = pcie->dev;
>>>>         u32 val;
>>>>         int ret;
>>>> -     u16 val_16;
>>>>
>>>>         if (pcie->ep_state == EP_STATE_ENABLED)
>>>>                 return;
>>>> @@ -1887,11 +1880,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>>>>         pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
>>>>                                                       PCI_CAP_ID_EXP);
>>>>
>>>> -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
>>>> -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
>>>> -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
>>>> -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
>>>> -
>>>>         /* Clear Slot Clock Configuration bit if SRNS configuration */
>>>>         if (pcie->enable_srns) {
>>>>                 val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base +
>>>> @@ -1900,7 +1888,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
>>>>                 dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_LNKSTA,
>>>>                                    val_16);
>>>>         }
>>>> -
>>>>         clk_set_rate(pcie->core_clk, GEN4_CORE_CLK_FREQ);
>>>>
>>>>         val = (ep->msi_mem_phys & MSIX_ADDR_MATCH_LOW_OFF_MASK);
>>>> --
>>>> 2.25.1
>>>>
>>>
>>> --
>>> மணிவண்ணன் சதாசிவம்
> 
> --
> மணிவண்ணன் சதாசிவம்
Manivannan Sadhasivam July 25, 2023, 10:03 a.m. UTC | #8
On Tue, Jul 25, 2023 at 02:51:10PM +0530, Vidya Sagar wrote:
> 
> 
> On 7/25/2023 2:00 PM, Manivannan Sadhasivam wrote:
> > External email: Use caution opening links or attachments
> > 
> > 
> > On Tue, Jul 25, 2023 at 01:49:35PM +0530, Vidya Sagar wrote:
> > > 
> > > 
> > > On 7/25/2023 1:21 PM, Manivannan Sadhasivam wrote:
> > > > External email: Use caution opening links or attachments
> > > > 
> > > > 
> > > > On Thu, Jun 08, 2023 at 03:06:52PM +0530, Vidya Sagar wrote:
> > > > > This reverts commit 4fb8e46c1bc4 ("PCI: tegra194: Enable
> > > > > support for 256 Byte payload")
> > > > > 
> > > > > Consider a PCIe hierarchy with a PCIe switch and a device connected
> > > > > downstream of the switch that has support for MPS which is the minimum
> > > > > in the hierarchy, and root port programmed with an MPS in its DevCtl
> > > > > register that is greater than the minimum. In this scenario, the default
> > > > > bus configuration of the kernel i.e. "PCIE_BUS_DEFAULT" doesn't
> > > > > configure the MPS settings in the hierarchy correctly resulting in the
> > > > > device with support for minimum MPS in the hierarchy receiving the TLPs
> > > > > of size more than that. Although this can be addresed by appending
> > > > > "pci=pcie_bus_safe" to the kernel command line, it doesn't seem to be a
> > > > > good idea to always have this commandline argument even for the basic
> > > > > functionality to work.
> > > > > Reverting commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256
> > > > > Byte payload") avoids this requirement and ensures that the basic
> > > > > functionality of the devices irrespective of the hierarchy and the MPS of
> > > > > the devices in the hierarchy.
> > > > > To reap the benefits of having support for higher MPS, optionally, one can
> > > > > always append the kernel command line with "pci=pcie_bus_perf".
> > > > > 
> > > > > Fixes: 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte payload")
> > > > > Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
> > > > 
> > > > I know that this patch is merged. But I happen to test a similar change on Qcom
> > > > platform during a patch review and found that the PCI core changes MPS to 128
> > > > when a 128byte supported device is found:
> > > > 
> > > > [    3.174290] pci 0000:01:00.0: Upstream bridge's Max Payload Size set to 128 (was 256, max 128)
> > > > [    3.186538] pci 0000:01:00.0: Max Payload Size set to 128 (was 128, max 128)
> > > > 
> > > > This was just randomly tested on a platform whose Root port DEVCAP was 128, but
> > > > it shouldn't matter. And I didn't change the default bus configuration.
> > > > 
> > > > Wondering how you ended up facing issues with it.
> > > 
> > > If the endpiont device that has support only for 128byte MPS is connected
> > > directly to the root port, then, I agree that the PCIe sub-system takes care
> > > of changing the MPS to the common lowest MPS, but if the endpoint device is
> > > connected behind a PCIe switch, then the PCIe subsystem doesn't configure
> > > the MPS properly.
> > > 
> > 
> > Ah, I missed the fact that your issue only happens with a PCIe switch. But
> > shouldn't this be fixed in the PCI core instead?
> > 
> > I mean, PCI core should use 128byte in your case for Root port unless it is not
> > allowed in the spec (which I doubt).
> well, if the RP's DevCtl is set to 256MPS by default, then, the core
> wouldn't do it automatically unless either 'pcie_bus_safe' or
> 'pcie_bus_perf' is passed.
> 

That's what I'm referring to. The default configuration shouldn't cause Root
port to send TLPs with unsupported payload. Moreover, this is not the case for
immediate children. So definitely the PCI core should exhibit the same behavior
for all downstream devices. 

- Mani

> 
> > 
> > - Mani
> > 
> > > -Vidya Sagar
> > > 
> > > > 
> > > > - Mani
> > > > 
> > > > > ---
> > > > >    drivers/pci/controller/dwc/pcie-tegra194.c | 13 -------------
> > > > >    1 file changed, 13 deletions(-)
> > > > > 
> > > > > diff --git a/drivers/pci/controller/dwc/pcie-tegra194.c b/drivers/pci/controller/dwc/pcie-tegra194.c
> > > > > index 4fdadc7b045f..877d81b13334 100644
> > > > > --- a/drivers/pci/controller/dwc/pcie-tegra194.c
> > > > > +++ b/drivers/pci/controller/dwc/pcie-tegra194.c
> > > > > @@ -892,7 +892,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
> > > > >         struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
> > > > >         struct tegra_pcie_dw *pcie = to_tegra_pcie(pci);
> > > > >         u32 val;
> > > > > -     u16 val_16;
> > > > > 
> > > > >         pp->bridge->ops = &tegra_pci_ops;
> > > > > 
> > > > > @@ -900,11 +899,6 @@ static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
> > > > >                 pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
> > > > >                                                               PCI_CAP_ID_EXP);
> > > > > 
> > > > > -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
> > > > > -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
> > > > > -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
> > > > > -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
> > > > > -
> > > > >         val = dw_pcie_readl_dbi(pci, PCI_IO_BASE);
> > > > >         val &= ~(IO_BASE_IO_DECODE | IO_BASE_IO_DECODE_BIT8);
> > > > >         dw_pcie_writel_dbi(pci, PCI_IO_BASE, val);
> > > > > @@ -1756,7 +1750,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
> > > > >         struct device *dev = pcie->dev;
> > > > >         u32 val;
> > > > >         int ret;
> > > > > -     u16 val_16;
> > > > > 
> > > > >         if (pcie->ep_state == EP_STATE_ENABLED)
> > > > >                 return;
> > > > > @@ -1887,11 +1880,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
> > > > >         pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
> > > > >                                                       PCI_CAP_ID_EXP);
> > > > > 
> > > > > -     val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
> > > > > -     val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
> > > > > -     val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
> > > > > -     dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
> > > > > -
> > > > >         /* Clear Slot Clock Configuration bit if SRNS configuration */
> > > > >         if (pcie->enable_srns) {
> > > > >                 val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base +
> > > > > @@ -1900,7 +1888,6 @@ static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
> > > > >                 dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_LNKSTA,
> > > > >                                    val_16);
> > > > >         }
> > > > > -
> > > > >         clk_set_rate(pcie->core_clk, GEN4_CORE_CLK_FREQ);
> > > > > 
> > > > >         val = (ep->msi_mem_phys & MSIX_ADDR_MATCH_LOW_OFF_MASK);
> > > > > --
> > > > > 2.25.1
> > > > > 
> > > > 
> > > > --
> > > > மணிவண்ணன் சதாசிவம்
> > 
> > --
> > மணிவண்ணன் சதாசிவம்
diff mbox series

Patch

diff --git a/drivers/pci/controller/dwc/pcie-tegra194.c b/drivers/pci/controller/dwc/pcie-tegra194.c
index 4fdadc7b045f..877d81b13334 100644
--- a/drivers/pci/controller/dwc/pcie-tegra194.c
+++ b/drivers/pci/controller/dwc/pcie-tegra194.c
@@ -892,7 +892,6 @@  static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
 	struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
 	struct tegra_pcie_dw *pcie = to_tegra_pcie(pci);
 	u32 val;
-	u16 val_16;
 
 	pp->bridge->ops = &tegra_pci_ops;
 
@@ -900,11 +899,6 @@  static int tegra_pcie_dw_host_init(struct dw_pcie_rp *pp)
 		pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
 							      PCI_CAP_ID_EXP);
 
-	val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
-	val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
-	val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
-	dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
-
 	val = dw_pcie_readl_dbi(pci, PCI_IO_BASE);
 	val &= ~(IO_BASE_IO_DECODE | IO_BASE_IO_DECODE_BIT8);
 	dw_pcie_writel_dbi(pci, PCI_IO_BASE, val);
@@ -1756,7 +1750,6 @@  static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
 	struct device *dev = pcie->dev;
 	u32 val;
 	int ret;
-	u16 val_16;
 
 	if (pcie->ep_state == EP_STATE_ENABLED)
 		return;
@@ -1887,11 +1880,6 @@  static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
 	pcie->pcie_cap_base = dw_pcie_find_capability(&pcie->pci,
 						      PCI_CAP_ID_EXP);
 
-	val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL);
-	val_16 &= ~PCI_EXP_DEVCTL_PAYLOAD;
-	val_16 |= PCI_EXP_DEVCTL_PAYLOAD_256B;
-	dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_DEVCTL, val_16);
-
 	/* Clear Slot Clock Configuration bit if SRNS configuration */
 	if (pcie->enable_srns) {
 		val_16 = dw_pcie_readw_dbi(pci, pcie->pcie_cap_base +
@@ -1900,7 +1888,6 @@  static void pex_ep_event_pex_rst_deassert(struct tegra_pcie_dw *pcie)
 		dw_pcie_writew_dbi(pci, pcie->pcie_cap_base + PCI_EXP_LNKSTA,
 				   val_16);
 	}
-
 	clk_set_rate(pcie->core_clk, GEN4_CORE_CLK_FREQ);
 
 	val = (ep->msi_mem_phys & MSIX_ADDR_MATCH_LOW_OFF_MASK);