linux-pci.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [RESEND PATCH v1] PCI: pcie_bus_config can be set at build time
@ 2020-09-08 16:32 Jim Quinlan
  2020-09-10  2:25 ` Bjorn Helgaas
  0 siblings, 1 reply; 4+ messages in thread
From: Jim Quinlan @ 2020-09-08 16:32 UTC (permalink / raw)
  To: linux-pci, bcm-kernel-feedback-list, james.quinlan
  Cc: Bjorn Helgaas, open list

The Kconfig is modified so that the pcie_bus_config setting can be done at
build time in the same manner as the CONFIG_PCIEASPM_XXXX choice.  The
pci_bus_config setting may still be overridden by the bootline param.

Signed-off-by: Jim Quinlan <james.quinlan@broadcom.com>
---
 drivers/pci/Kconfig | 40 ++++++++++++++++++++++++++++++++++++++++
 drivers/pci/pci.c   | 12 ++++++++++++
 2 files changed, 52 insertions(+)

diff --git a/drivers/pci/Kconfig b/drivers/pci/Kconfig
index 4bef5c2bae9f..efe69b0d9f7f 100644
--- a/drivers/pci/Kconfig
+++ b/drivers/pci/Kconfig
@@ -187,6 +187,46 @@ config PCI_HYPERV
 	  The PCI device frontend driver allows the kernel to import arbitrary
 	  PCI devices from a PCI backend to support PCI driver domains.
 
+choice
+	prompt "PCIE default bus config setting"
+	default PCIE_BUS_DEFAULT
+	depends on PCI
+	help
+	  One of the following choices will set the pci_bus_config at
+	  compile time.  This will still be overridden by the appropriate
+	  pci bootline parameter.
+
+config PCIE_BUS_TUNE_OFF
+	bool "Tune Off"
+	depends on PCI
+	help
+	  Use the BIOS defaults; doesn't touch MPS at all.
+
+config PCIE_BUS_DEFAULT
+	bool "Default"
+	depends on PCI
+	help
+	  Ensure MPS matches upstream bridge.
+
+config PCIE_BUS_SAFE
+	bool "Safe"
+	depends on PCI
+	help
+	  Use largest MPS boot-time devices support.
+
+config PCIE_BUS_PERFORMANCE
+	bool "Performance"
+	depends on PCI
+	help
+	  Use MPS and MRRS for best performance.
+
+config PCIE_BUS_PEER2PEER
+	bool "Peer2peer"
+	depends on PCI
+	help
+	  Set MPS = 128 for all devices.
+endchoice
+
 source "drivers/pci/hotplug/Kconfig"
 source "drivers/pci/controller/Kconfig"
 source "drivers/pci/endpoint/Kconfig"
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index e39c5499770f..dfb52ed4a931 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -101,7 +101,19 @@ unsigned long pci_hotplug_mmio_pref_size = DEFAULT_HOTPLUG_MMIO_PREF_SIZE;
 #define DEFAULT_HOTPLUG_BUS_SIZE	1
 unsigned long pci_hotplug_bus_size = DEFAULT_HOTPLUG_BUS_SIZE;
 
+
+/* PCIE bus config, can be overridden by bootline param */
+#ifdef CONFIG_PCIE_BUS_TUNE_OFF
+enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_TUNE_OFF;
+#elif defined CONFIG_PCIE_BUS_SAFE
+enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_SAFE;
+#elif defined CONFIG_PCIE_BUS_PERFORMANCE
+enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_PERFORMANCE;
+#elif defined CONFIG_PCIE_BUS_PEER2PEER
+enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_PEER2PEER;
+#else
 enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_DEFAULT;
+#endif
 
 /*
  * The default CLS is used if arch didn't set CLS explicitly and not
-- 
2.17.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [RESEND PATCH v1] PCI: pcie_bus_config can be set at build time
  2020-09-08 16:32 [RESEND PATCH v1] PCI: pcie_bus_config can be set at build time Jim Quinlan
@ 2020-09-10  2:25 ` Bjorn Helgaas
  2020-09-10 12:57   ` Jim Quinlan
  0 siblings, 1 reply; 4+ messages in thread
From: Bjorn Helgaas @ 2020-09-10  2:25 UTC (permalink / raw)
  To: Jim Quinlan
  Cc: linux-pci, bcm-kernel-feedback-list, Bjorn Helgaas, linux-kernel

On Tue, Sep 08, 2020 at 12:32:48PM -0400, Jim Quinlan wrote:
> The Kconfig is modified so that the pcie_bus_config setting can be done at
> build time in the same manner as the CONFIG_PCIEASPM_XXXX choice.  The
> pci_bus_config setting may still be overridden by the bootline param.

I guess...  I really hate these build-time config settings for both
ASPM and MPS/MRRS.  But Linux just isn't smart or flexible enough to
do the right thing at run-time, so I guess we're kind of stuck.

I guess you have systems where you need a different default?

It'd be nice if we could put a little more detail in the Kconfig to
help users choose the correct one.  "Ensure MPS matches upstream
bridge" is *accurate*, but it doesn't really tell me why I would
choose this rather than a different one.

Maybe we could mention the corresponding command-line parameters,
e.g., "This is the same as booting with 'pci=pcie_bus_tune_off'"?

> Signed-off-by: Jim Quinlan <james.quinlan@broadcom.com>
> ---
>  drivers/pci/Kconfig | 40 ++++++++++++++++++++++++++++++++++++++++
>  drivers/pci/pci.c   | 12 ++++++++++++
>  2 files changed, 52 insertions(+)
> 
> diff --git a/drivers/pci/Kconfig b/drivers/pci/Kconfig
> index 4bef5c2bae9f..efe69b0d9f7f 100644
> --- a/drivers/pci/Kconfig
> +++ b/drivers/pci/Kconfig
> @@ -187,6 +187,46 @@ config PCI_HYPERV
>  	  The PCI device frontend driver allows the kernel to import arbitrary
>  	  PCI devices from a PCI backend to support PCI driver domains.
>  
> +choice
> +	prompt "PCIE default bus config setting"
> +	default PCIE_BUS_DEFAULT
> +	depends on PCI
> +	help
> +	  One of the following choices will set the pci_bus_config at
> +	  compile time.  This will still be overridden by the appropriate
> +	  pci bootline parameter.
> +
> +config PCIE_BUS_TUNE_OFF
> +	bool "Tune Off"
> +	depends on PCI
> +	help
> +	  Use the BIOS defaults; doesn't touch MPS at all.
> +
> +config PCIE_BUS_DEFAULT
> +	bool "Default"
> +	depends on PCI
> +	help
> +	  Ensure MPS matches upstream bridge.
> +
> +config PCIE_BUS_SAFE
> +	bool "Safe"
> +	depends on PCI
> +	help
> +	  Use largest MPS boot-time devices support.
> +
> +config PCIE_BUS_PERFORMANCE
> +	bool "Performance"
> +	depends on PCI
> +	help
> +	  Use MPS and MRRS for best performance.
> +
> +config PCIE_BUS_PEER2PEER
> +	bool "Peer2peer"
> +	depends on PCI
> +	help
> +	  Set MPS = 128 for all devices.
> +endchoice
> +
>  source "drivers/pci/hotplug/Kconfig"
>  source "drivers/pci/controller/Kconfig"
>  source "drivers/pci/endpoint/Kconfig"
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index e39c5499770f..dfb52ed4a931 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -101,7 +101,19 @@ unsigned long pci_hotplug_mmio_pref_size = DEFAULT_HOTPLUG_MMIO_PREF_SIZE;
>  #define DEFAULT_HOTPLUG_BUS_SIZE	1
>  unsigned long pci_hotplug_bus_size = DEFAULT_HOTPLUG_BUS_SIZE;
>  
> +
> +/* PCIE bus config, can be overridden by bootline param */
> +#ifdef CONFIG_PCIE_BUS_TUNE_OFF
> +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_TUNE_OFF;
> +#elif defined CONFIG_PCIE_BUS_SAFE
> +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_SAFE;
> +#elif defined CONFIG_PCIE_BUS_PERFORMANCE
> +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_PERFORMANCE;
> +#elif defined CONFIG_PCIE_BUS_PEER2PEER
> +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_PEER2PEER;
> +#else
>  enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_DEFAULT;
> +#endif
>  
>  /*
>   * The default CLS is used if arch didn't set CLS explicitly and not
> -- 
> 2.17.1
> 

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [RESEND PATCH v1] PCI: pcie_bus_config can be set at build time
  2020-09-10  2:25 ` Bjorn Helgaas
@ 2020-09-10 12:57   ` Jim Quinlan
  2020-09-10 13:31     ` Bjorn Helgaas
  0 siblings, 1 reply; 4+ messages in thread
From: Jim Quinlan @ 2020-09-10 12:57 UTC (permalink / raw)
  To: Bjorn Helgaas
  Cc: open list:PCI NATIVE HOST BRIDGE AND ENDPOINT DRIVERS,
	maintainer:BROADCOM BCM7XXX ARM ARCHITECTURE, Bjorn Helgaas,
	open list

[-- Attachment #1: Type: text/plain, Size: 4502 bytes --]

Hi Bjorn,

On Wed, Sep 9, 2020 at 10:25 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
>
> On Tue, Sep 08, 2020 at 12:32:48PM -0400, Jim Quinlan wrote:
> > The Kconfig is modified so that the pcie_bus_config setting can be done at
> > build time in the same manner as the CONFIG_PCIEASPM_XXXX choice.  The
> > pci_bus_config setting may still be overridden by the bootline param.
>
> I guess...  I really hate these build-time config settings for both
> ASPM and MPS/MRRS.  But Linux just isn't smart or flexible enough to
> do the right thing at run-time, so I guess we're kind of stuck.
>
> I guess you have systems where you need a different default?
Yes, we've been shipping our kernel with the DEFAULT and since we do
not have FW it is not configured optimally.  Some customers have
noticed and I tell them to put 'pci=pcie_bus_safe' on their bootline
but I'd rather have this setting work for all customers as it yields
the option we want.

>
> It'd be nice if we could put a little more detail in the Kconfig to
> help users choose the correct one.  "Ensure MPS matches upstream
> bridge" is *accurate*, but it doesn't really tell me why I would
> choose this rather than a different one.
IIRC I just copied the comments that were in the bootline settings.
I'm concerned about there being the same comment in two places; sooner
or later someone will update one place and not the other.

>
> Maybe we could mention the corresponding command-line parameters,
> e.g., "This is the same as booting with 'pci=pcie_bus_tune_off'"?
Will do and resubmit.

Thanks,
Jim Quinlan
Broadcom STB
>
> > Signed-off-by: Jim Quinlan <james.quinlan@broadcom.com>
> > ---
> >  drivers/pci/Kconfig | 40 ++++++++++++++++++++++++++++++++++++++++
> >  drivers/pci/pci.c   | 12 ++++++++++++
> >  2 files changed, 52 insertions(+)
> >
> > diff --git a/drivers/pci/Kconfig b/drivers/pci/Kconfig
> > index 4bef5c2bae9f..efe69b0d9f7f 100644
> > --- a/drivers/pci/Kconfig
> > +++ b/drivers/pci/Kconfig
> > @@ -187,6 +187,46 @@ config PCI_HYPERV
> >         The PCI device frontend driver allows the kernel to import arbitrary
> >         PCI devices from a PCI backend to support PCI driver domains.
> >
> > +choice
> > +     prompt "PCIE default bus config setting"
> > +     default PCIE_BUS_DEFAULT
> > +     depends on PCI
> > +     help
> > +       One of the following choices will set the pci_bus_config at
> > +       compile time.  This will still be overridden by the appropriate
> > +       pci bootline parameter.
> > +
> > +config PCIE_BUS_TUNE_OFF
> > +     bool "Tune Off"
> > +     depends on PCI
> > +     help
> > +       Use the BIOS defaults; doesn't touch MPS at all.
> > +
> > +config PCIE_BUS_DEFAULT
> > +     bool "Default"
> > +     depends on PCI
> > +     help
> > +       Ensure MPS matches upstream bridge.
> > +
> > +config PCIE_BUS_SAFE
> > +     bool "Safe"
> > +     depends on PCI
> > +     help
> > +       Use largest MPS boot-time devices support.
> > +
> > +config PCIE_BUS_PERFORMANCE
> > +     bool "Performance"
> > +     depends on PCI
> > +     help
> > +       Use MPS and MRRS for best performance.
> > +
> > +config PCIE_BUS_PEER2PEER
> > +     bool "Peer2peer"
> > +     depends on PCI
> > +     help
> > +       Set MPS = 128 for all devices.
> > +endchoice
> > +
> >  source "drivers/pci/hotplug/Kconfig"
> >  source "drivers/pci/controller/Kconfig"
> >  source "drivers/pci/endpoint/Kconfig"
> > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> > index e39c5499770f..dfb52ed4a931 100644
> > --- a/drivers/pci/pci.c
> > +++ b/drivers/pci/pci.c
> > @@ -101,7 +101,19 @@ unsigned long pci_hotplug_mmio_pref_size = DEFAULT_HOTPLUG_MMIO_PREF_SIZE;
> >  #define DEFAULT_HOTPLUG_BUS_SIZE     1
> >  unsigned long pci_hotplug_bus_size = DEFAULT_HOTPLUG_BUS_SIZE;
> >
> > +
> > +/* PCIE bus config, can be overridden by bootline param */
> > +#ifdef CONFIG_PCIE_BUS_TUNE_OFF
> > +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_TUNE_OFF;
> > +#elif defined CONFIG_PCIE_BUS_SAFE
> > +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_SAFE;
> > +#elif defined CONFIG_PCIE_BUS_PERFORMANCE
> > +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_PERFORMANCE;
> > +#elif defined CONFIG_PCIE_BUS_PEER2PEER
> > +enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_PEER2PEER;
> > +#else
> >  enum pcie_bus_config_types pcie_bus_config = PCIE_BUS_DEFAULT;
> > +#endif
> >
> >  /*
> >   * The default CLS is used if arch didn't set CLS explicitly and not
> > --
> > 2.17.1
> >

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4167 bytes --]

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [RESEND PATCH v1] PCI: pcie_bus_config can be set at build time
  2020-09-10 12:57   ` Jim Quinlan
@ 2020-09-10 13:31     ` Bjorn Helgaas
  0 siblings, 0 replies; 4+ messages in thread
From: Bjorn Helgaas @ 2020-09-10 13:31 UTC (permalink / raw)
  To: Jim Quinlan
  Cc: open list:PCI NATIVE HOST BRIDGE AND ENDPOINT DRIVERS,
	maintainer:BROADCOM BCM7XXX ARM ARCHITECTURE, Bjorn Helgaas,
	open list

On Thu, Sep 10, 2020 at 08:57:10AM -0400, Jim Quinlan wrote:
> Hi Bjorn,
> 
> On Wed, Sep 9, 2020 at 10:25 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
> >
> > On Tue, Sep 08, 2020 at 12:32:48PM -0400, Jim Quinlan wrote:
> > > The Kconfig is modified so that the pcie_bus_config setting can be done at
> > > build time in the same manner as the CONFIG_PCIEASPM_XXXX choice.  The
> > > pci_bus_config setting may still be overridden by the bootline param.
> >
> > I guess...  I really hate these build-time config settings for both
> > ASPM and MPS/MRRS.  But Linux just isn't smart or flexible enough to
> > do the right thing at run-time, so I guess we're kind of stuck.
> >
> > I guess you have systems where you need a different default?
>
> Yes, we've been shipping our kernel with the DEFAULT and since we do
> not have FW it is not configured optimally.  Some customers have
> noticed and I tell them to put 'pci=pcie_bus_safe' on their bootline
> but I'd rather have this setting work for all customers as it yields
> the option we want.

I'm guessing you probably don't have any hotplug slots.  Seems like we
ought to be able to recognize that and pick pcie_bus_safe
automatically.  Someday.

Maybe that's part of the description: if you have a closed system with
no possibility of adding new devices, we can use the largest MPS
that's supported by all devices, i.e., pcie_bus_safe.

> > It'd be nice if we could put a little more detail in the Kconfig to
> > help users choose the correct one.  "Ensure MPS matches upstream
> > bridge" is *accurate*, but it doesn't really tell me why I would
> > choose this rather than a different one.
>
> IIRC I just copied the comments that were in the bootline settings.
> I'm concerned about there being the same comment in two places; sooner
> or later someone will update one place and not the other.

True.  It'd be nice if we at least had *one* place with a useful
description.  I don't think we have any today.

Bjorn

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2020-09-10 21:55 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-09-08 16:32 [RESEND PATCH v1] PCI: pcie_bus_config can be set at build time Jim Quinlan
2020-09-10  2:25 ` Bjorn Helgaas
2020-09-10 12:57   ` Jim Quinlan
2020-09-10 13:31     ` Bjorn Helgaas

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).