lkml.org 
[lkml]   [2021]   [Jun]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH] PCI/APCI: Move acpi_pci_osc_support() check to negotiation phase
On Thu, Jun 03, 2021 at 02:48:14PM +0200, Joerg Roedel wrote:
> From: Joerg Roedel <jroedel@suse.de>

I like this patch a lot and I plan to apply it because you've managed
to simplify the nasty _OSC path a little bit. But I'm confused about
the justification.

> The acpi_pci_osc_support() does an _OSC query with _OSC supported set
> to what the OS supports but a zero _OSC control value. This is
> problematic on some platforms where the firmware allows to configure
> whether DPC is under OS or Firmware control.
>
> When DPC is configured to be under OS control these platforms will
> issue a warning in the firmware log that the OS does not support DPC.

My understanding is that DPC is under platform control until the OS
requests it via _OSC(Request, Control & OSC_PCI_EXPRESS_DPC_CONTROL)
and the platform grants it. And after the OS is granted control of
DPC, it must preserve OSC_PCI_EXPRESS_DPC_CONTROL in all subsequent
_OSC calls (i.e., there is no way for the OS to relinquish DPC
control).

So what does it mean for "DPC to be under OS control, but the OS does
_OSC(Query, Control=0)"? That doesn't sound like a legal sequence:
the OS has already been granted DPC control, but it failed to preserve
OSC_PCI_EXPRESS_DPC_CONTROL?

If instead you mean that the OS has *not* been granted DPC control,
but does _OSC(Query, SUPPORT=x, CONTROL=0), I think that means the OS
is telling the platform what it supports but not requesting anything.
That sounds legal to me, so if firmware complains about it, I would
say it's a firmware problem.

> Avoid an _OSC query with _OSC control set to zero by moving the
> supported check into the acpi_pci_osc_control_set() path. This is
> still early enough to fail as nothing before that depends on the
> results of acpi_pci_osc_support().
>
> As a result the acpi_pci_osc_support() function can be removed and
> acpi_pci_query_osc() be simplified because it no longer called with a
> NULL pointer for *control.

So I think we should do this, but not because it avoids a firmware
warning, which looks like a firmware bug to me. We should do it just
because it simplifies this ugly code.

But please help me out if I'm misunderstanding something above. I'm
never confident that I really understand _OSC.

> Signed-off-by: Joerg Roedel <jroedel@suse.de>
> ---
> drivers/acpi/pci_root.c | 50 ++++++++++++++++-------------------------
> 1 file changed, 19 insertions(+), 31 deletions(-)
>
> diff --git a/drivers/acpi/pci_root.c b/drivers/acpi/pci_root.c
> index dcd593766a64..530ecf4970b1 100644
> --- a/drivers/acpi/pci_root.c
> +++ b/drivers/acpi/pci_root.c
> @@ -199,16 +199,11 @@ static acpi_status acpi_pci_query_osc(struct acpi_pci_root *root,
>
> support &= OSC_PCI_SUPPORT_MASKS;
> support |= root->osc_support_set;
> + *control &= OSC_PCI_CONTROL_MASKS;

Unrelated to *this* patch, but I don't understand the point of
OSC_PCI_SUPPORT_MASKS and OSC_PCI_CONTROL_MASKS. These are all
internal static functions and it looks like pointless work to apply
masks here and in acpi_pci_osc_control_set().

I'm happy to make this change, but if you do it, please make it a
separate patch for bisection purposes.

> capbuf[OSC_QUERY_DWORD] = OSC_QUERY_ENABLE;
> capbuf[OSC_SUPPORT_DWORD] = support;
> - if (control) {
> - *control &= OSC_PCI_CONTROL_MASKS;
> - capbuf[OSC_CONTROL_DWORD] = *control | root->osc_control_set;
> - } else {
> - /* Run _OSC query only with existing controls. */
> - capbuf[OSC_CONTROL_DWORD] = root->osc_control_set;
> - }
> + capbuf[OSC_CONTROL_DWORD] = *control | root->osc_control_set;
>
> status = acpi_pci_run_osc(root->device->handle, capbuf, &result);
> if (ACPI_SUCCESS(status)) {

We can also drop the "if (control)" check inside the ACPI_SUCCESS()
block, can't we?

> @@ -219,11 +214,6 @@ static acpi_status acpi_pci_query_osc(struct acpi_pci_root *root,
> return status;
> }
>
> -static acpi_status acpi_pci_osc_support(struct acpi_pci_root *root, u32 flags)
> -{
> - return acpi_pci_query_osc(root, flags, NULL);
> -}
> -
> struct acpi_pci_root *acpi_pci_find_root(acpi_handle handle)
> {
> struct acpi_pci_root *root;
> @@ -346,7 +336,8 @@ EXPORT_SYMBOL_GPL(acpi_get_pci_dev);
> * _OSC bits the BIOS has granted control of, but its contents are meaningless
> * on failure.
> **/
> -static acpi_status acpi_pci_osc_control_set(acpi_handle handle, u32 *mask, u32 req)
> +static acpi_status acpi_pci_osc_control_set(acpi_handle handle, u32
> + *mask, u32 req, u32 support)
> {
> struct acpi_pci_root *root;
> acpi_status status;
> @@ -370,7 +361,7 @@ static acpi_status acpi_pci_osc_control_set(acpi_handle handle, u32 *mask, u32 r
>
> /* Need to check the available controls bits before requesting them. */
> while (*mask) {
> - status = acpi_pci_query_osc(root, root->osc_support_set, mask);
> + status = acpi_pci_query_osc(root, support, mask);
> if (ACPI_FAILURE(status))
> return status;
> if (ctrl == *mask)
> @@ -433,18 +424,6 @@ static void negotiate_os_control(struct acpi_pci_root *root, int *no_aspm,
> support |= OSC_PCI_EDR_SUPPORT;
>
> decode_osc_support(root, "OS supports", support);
> - status = acpi_pci_osc_support(root, support);
> - if (ACPI_FAILURE(status)) {
> - *no_aspm = 1;
> -
> - /* _OSC is optional for PCI host bridges */
> - if ((status == AE_NOT_FOUND) && !is_pcie)
> - return;
> -
> - dev_info(&device->dev, "_OSC: platform retains control of PCIe features (%s)\n",
> - acpi_format_exception(status));
> - return;
> - }
>
> if (pcie_ports_disabled) {
> dev_info(&device->dev, "PCIe port services disabled; not requesting _OSC control\n");

Also not related to this patch, but it seems pointless to compute and
decode "support" above when we're not going to use _OSC at all. I
think the "pcie_ports_disabled" test should be the very first thing in
this function (I'm assuming the "pcie_ports=compat" command line
argument *should* apply even on x86_apple_machine, which it doesn't
today).

Again, I'm happy to do this if it makes sense to you.

> @@ -483,7 +462,8 @@ static void negotiate_os_control(struct acpi_pci_root *root, int *no_aspm,
>
> requested = control;
> status = acpi_pci_osc_control_set(handle, &control,
> - OSC_PCI_EXPRESS_CAPABILITY_CONTROL);
> + OSC_PCI_EXPRESS_CAPABILITY_CONTROL,
> + support);
> if (ACPI_SUCCESS(status)) {
> decode_osc_control(root, "OS now controls", control);
> if (acpi_gbl_FADT.boot_flags & ACPI_FADT_NO_ASPM) {
> @@ -496,10 +476,8 @@ static void negotiate_os_control(struct acpi_pci_root *root, int *no_aspm,
> *no_aspm = 1;
> }
> } else {
> - decode_osc_control(root, "OS requested", requested);
> - decode_osc_control(root, "platform willing to grant", control);
> - dev_info(&device->dev, "_OSC: platform retains control of PCIe features (%s)\n",
> - acpi_format_exception(status));
> + /* Platform wants to control PCIe features */

Or _OSC just failed because of an OS or firmware defect ;)

> + root->osc_support_set = 0;
>
> /*
> * We want to disable ASPM here, but aspm_disabled
> @@ -509,6 +487,16 @@ static void negotiate_os_control(struct acpi_pci_root *root, int *no_aspm,
> * root scan.
> */
> *no_aspm = 1;
> +
> + /* _OSC is optional for PCI host bridges */
> + if ((status == AE_NOT_FOUND) && !is_pcie)
> + return;
> +
> + decode_osc_control(root, "OS requested", requested);
> + decode_osc_control(root, "platform willing to grant", control);
> + dev_info(&device->dev, "_OSC: platform retains control of PCIe features (%s)\n",
> + acpi_format_exception(status));
> +
> }
> }
>
> --
> 2.31.1
>

\
 
 \ /
  Last update: 2021-06-03 22:52    [W:0.104 / U:0.928 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site