Re: [PATCH v2 4/9] PCI/PM: Rework changing power states of PCI devices

From: Mika Westerberg
Date: Tue Apr 12 2022 - 08:28:15 EST

Next message: Ondřej Jirman: "Re: [PATCH 3/5] Input: pinephone-keyboard - Build in the default keymap"
Previous message: Serge Semin: "Re: [PATCH 04/11] dt-bindings: spi: Add compatible for Pensando Elba SoC"
In reply to: Rafael J. Wysocki: "Re: [PATCH v2 4/9] PCI/PM: Rework changing power states of PCI devices"
Next in thread: Rafael J. Wysocki: "[PATCH v2 3/9] PCI/PM: Rearrange pci_update_current_state()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi,

On Tue, Apr 12, 2022 at 01:31:57PM +0200, Rafael J. Wysocki wrote:
> On Tue, Apr 12, 2022 at 1:17 PM Mika Westerberg
> <mika.westerberg@xxxxxxxxxxxxxxx> wrote:
> >
> > On Mon, Apr 11, 2022 at 04:25:12PM +0200, Rafael J. Wysocki wrote:
> > > From: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
> > >
> > > There are some issues related to changing power states of PCI
> > > devices, mostly related to carrying out unnecessary actions in some
> > > places, and the code is generally hard to follow.
> > >
> > > 1. pci_power_up() has two callers, pci_set_power_state() and
> > > pci_pm_default_resume_early(). The latter updates the current
> > > power state of the device right after calling pci_power_up()
> > > and it restores the entire config space of the device right
> > > after that, so pci_power_up() itself need not read the
> > > PCI_PM_CTRL register or restore the BARs after programming the
> > > device into D0 in that case.
> > >
> > > 2. It is generally hard to get a clear view of the pci_power_up()
> > > code flow, especially in some corner cases, due to all of the
> > > involved PCI_PM_CTRL register reads and writes occurring in
> > > pci_platform_power_transition() and in pci_raw_set_power_state(),
> > > some of which are redundant.
> > >
> > > 3. The transitions from low-power states to D0 and the other way
> > > around are unnecessarily tangled in pci_raw_set_power_state()
> > > which causes it to use a redundant local variable and makes it
> > > rather hard to follow.
> > >
> > > To address the above shortcomings, make the following changes:
> > >
> > > a. Remove the code handling transitions into D0
> >
> > Should this be D3?
>
> No. Transitions into D0 will be handled by pci_power_up() directly,
> so they need not be handled by pci_raw_set_power_state().

OK.

> > > from pci_raw_set_power_state() and rename it as
> > > pci_set_low_power_state().
> > >
> > > b. Add the code handling transitions into D0 directly
> > > to pci_power_up() and to a new wrapper function
> > > pci_set_full_power_state() calling it internally that is
> > > only used in pci_set_power_state().
> > >
> > > c. Make pci_power_up() avoid redundant PCI_PM_CTRL register reads
> > > and make it work in the same way for transitions from any
> > > low-power states (transitions from D1 and D2 are handled
> > > slightly differently before the change).
> > >
> > > d. Put the restoration of the BARs and the PCI_PM_CTRL
> > > register read confirming the power state change into
> > > pci_set_full_power_state() to avoid doing that in
> > > pci_pm_default_resume_early() unnecessarily.
> > >
> > > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
> > > ---
> > >
> > > v1 -> v2:
> > > * Do not add a redundant check to pci_set_low_power_state().
> > >
> > > ---
> > > drivers/pci/pci.c | 154 +++++++++++++++++++++++++++++++++++-------------------
> > > 1 file changed, 101 insertions(+), 53 deletions(-)
> > >
> > > Index: linux-pm/drivers/pci/pci.c
> > > ===================================================================
> > > --- linux-pm.orig/drivers/pci/pci.c
> > > +++ linux-pm/drivers/pci/pci.c
> > > @@ -1068,10 +1068,9 @@ static inline bool platform_pci_bridge_d
> > > }
> > >
> > > /**
> > > - * pci_raw_set_power_state - Use PCI PM registers to set the power state of
> > > - * given PCI device
> > > + * pci_set_low_power_state - Program the given device into a low-power state
> > > * @dev: PCI device to handle.
> > > - * @state: PCI power state (D0, D1, D2, D3hot) to put the device into.
> > > + * @state: PCI power state (D1, D2, D3hot) to put the device into.
> > > *
> > > * RETURN VALUE:
> > > * -EINVAL if the requested state is invalid.
> > > @@ -1080,10 +1079,9 @@ static inline bool platform_pci_bridge_d
> > > * 0 if device already is in the requested state.
> > > * 0 if device's power state has been successfully changed.
> > > */
> > > -static int pci_raw_set_power_state(struct pci_dev *dev, pci_power_t state)
> > > +static int pci_set_low_power_state(struct pci_dev *dev, pci_power_t state)
> > > {
> > > u16 pmcsr;
> > > - bool need_restore = false;
> > >
> > > /* Check if we're already there */
> > > if (dev->current_state == state)
> > > @@ -1092,7 +1090,7 @@ static int pci_raw_set_power_state(struc
> > > if (!dev->pm_cap)
> > > return -EIO;
> > >
> > > - if (state < PCI_D0 || state > PCI_D3hot)
> > > + if (state < PCI_D1 || state > PCI_D3hot)
> > > return -EINVAL;
> > >
> > > /*
> > > @@ -1101,8 +1099,7 @@ static int pci_raw_set_power_state(struc
> > > * we can go from D1 to D3, but we can't go directly from D3 to D1;
> > > * we'd have to go from D3 to D0, then to D1.
> > > */
> > > - if (state != PCI_D0 && dev->current_state <= PCI_D3cold
> > > - && dev->current_state > state) {
> > > + if (dev->current_state <= PCI_D3cold && dev->current_state > state) {
> > > pci_err(dev, "invalid power transition (from %s to %s)\n",
> > > pci_power_name(dev->current_state),
> > > pci_power_name(state));
> > > @@ -1122,29 +1119,8 @@ static int pci_raw_set_power_state(struc
> > > return -EIO;
> > > }
> > >
> > > - /*
> > > - * If we're (effectively) in D3, force entire word to 0.
> > > - * This doesn't affect PME_Status, disables PME_En, and
> > > - * sets PowerState to 0.
> > > - */
> > > - switch (dev->current_state) {
> > > - case PCI_D0:
> > > - case PCI_D1:
> > > - case PCI_D2:
> > > - pmcsr &= ~PCI_PM_CTRL_STATE_MASK;
> > > - pmcsr |= state;
> > > - break;
> > > - case PCI_D3hot:
> > > - case PCI_D3cold:
> > > - case PCI_UNKNOWN: /* Boot-up */
> > > - if ((pmcsr & PCI_PM_CTRL_STATE_MASK) == PCI_D3hot
> > > - && !(pmcsr & PCI_PM_CTRL_NO_SOFT_RESET))
> > > - need_restore = true;
> > > - fallthrough; /* force to D0 */
> > > - default:
> > > - pmcsr = 0;
> > > - break;
> > > - }
> > > + pmcsr &= ~PCI_PM_CTRL_STATE_MASK;
> > > + pmcsr |= state;
> > >
> > > /* Enter specified state */
> > > pci_write_config_word(dev, dev->pm_cap + PCI_PM_CTRL, pmcsr);
> > > @@ -1153,9 +1129,9 @@ static int pci_raw_set_power_state(struc
> > > * Mandatory power management transition delays; see PCI PM 1.1
> > > * 5.6.1 table 18
> > > */
> > > - if (state == PCI_D3hot || dev->current_state == PCI_D3hot)
> > > + if (state == PCI_D3hot)
> > > pci_dev_d3_sleep(dev);
> > > - else if (state == PCI_D2 || dev->current_state == PCI_D2)
> > > + else if (state == PCI_D2)
> > > udelay(PCI_PM_D2_DELAY);
> > >
> > > pci_read_config_word(dev, dev->pm_cap + PCI_PM_CTRL, &pmcsr);
> > > @@ -1165,22 +1141,6 @@ static int pci_raw_set_power_state(struc
> > > pci_power_name(dev->current_state),
> > > pci_power_name(state));
> > >
> > > - /*
> > > - * According to section 5.4.1 of the "PCI BUS POWER MANAGEMENT
> > > - * INTERFACE SPECIFICATION, REV. 1.2", a device transitioning
> > > - * from D3hot to D0 _may_ perform an internal reset, thereby
> > > - * going to "D0 Uninitialized" rather than "D0 Initialized".
> > > - * For example, at least some versions of the 3c905B and the
> > > - * 3c556B exhibit this behaviour.
> > > - *
> > > - * At least some laptop BIOSen (e.g. the Thinkpad T21) leave
> > > - * devices in a D3hot state at boot. Consequently, we need to
> > > - * restore at least the BARs so that the device will be
> > > - * accessible to its driver.
> > > - */
> > > - if (need_restore)
> > > - pci_restore_bars(dev);
> > > -
> > > if (dev->bus->self)
> > > pcie_aspm_pm_state_change(dev->bus->self);
> > >
> > > @@ -1312,8 +1272,54 @@ static int pci_dev_wait(struct pci_dev *
> > > */
> > > int pci_power_up(struct pci_dev *dev)
> > > {
> > > - pci_platform_power_transition(dev, PCI_D0);
> > > - return pci_raw_set_power_state(dev, PCI_D0);
> > > + int ret;
> > > +
> > > + ret = pci_platform_power_transition(dev, PCI_D0);
> > > + if (ret) {
> >
> > Here pci_platform_power_transition() returned an error so we go and read
> > back the PM_CTRL to check in which power state the device is in? Perhaps
> > add a comment here explaining why we need to do this?
>
> That's the comment below, but I gather it is insufficient as is.
> Please let me know if rephrasing it this way would help:
>
> "Since pci_platform_power_transition() has returned an error, the
> PCI_PM_CTRL register has not been read by it and the current power
> state of the device is unknown. Read the PCI_PM_CTRL register now and
> bail out if that fails."

Yes, that's better, thanks!

> And I've just realized that pm_cap should be checked here, because it
> is not guaranteed to be set.

Good point.

Next message: Ondřej Jirman: "Re: [PATCH 3/5] Input: pinephone-keyboard - Build in the default keymap"
Previous message: Serge Semin: "Re: [PATCH 04/11] dt-bindings: spi: Add compatible for Pensando Elba SoC"
In reply to: Rafael J. Wysocki: "Re: [PATCH v2 4/9] PCI/PM: Rework changing power states of PCI devices"
Next in thread: Rafael J. Wysocki: "[PATCH v2 3/9] PCI/PM: Rearrange pci_update_current_state()"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]