Message-ID: <af74b8d4f1f7072ffebc8a7f5cf392140da73dc7.camel@linux.intel.com>
Date: Tue, 30 Apr 2024 11:26:46 -0700
From: "David E. Box" <david.e.box@...ux.intel.com>
To: Jian-Hong Pan <jhp@...lessos.org>
Cc: Bjorn Helgaas <helgaas@...nel.org>, Johan Hovold <johan@...nel.org>, 
 Ilpo Järvinen <ilpo.jarvinen@...ux.intel.com>,
 Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...ux.intel.com>,
 Mika Westerberg <mika.westerberg@...ux.intel.com>, Damien Le Moal
 <dlemoal@...nel.org>,  Nirmal Patel <nirmal.patel@...ux.intel.com>,
 Jonathan Derrick <jonathan.derrick@...ux.dev>,  linux-pci@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH v5 4/4] PCI/ASPM: Fix L1.2 parameters when enable link
 state

On Tue, 2024-04-30 at 15:46 +0800, Jian-Hong Pan wrote:
> David E. Box <david.e.box@...ux.intel.com> 於 2024年4月27日 週六 上午8:03寫道：
> > 
> > Hi Jian-Hong,
> > 
> > On Wed, 2024-04-24 at 19:02 +0800, Jian-Hong Pan wrote:
> > > Currently, when enable link's L1.2 features with
> > > __pci_enable_link_state(),
> > > it configs the link directly without ensuring related L1.2 parameters,
> > > such
> > > as T_POWER_ON, Common_Mode_Restore_Time, and LTR_L1.2_THRESHOLD have been
> > > programmed.
> > > 
> > > This leads the link's L1.2 between PCIe Root Port and child device gets
> > > wrong configs when a caller tries to enabled it.
> > > 
> > > Here is a failed example on ASUS B1400CEAE with enabled VMD:
> > > 
> > > 10000:e0:06.0 PCI bridge: Intel Corporation 11th Gen Core Processor PCIe
> > > Controller (rev 01) (prog-if 00 [Normal decode])
> > >     ...
> > >     Capabilities: [200 v1] L1 PM Substates
> > >         L1SubCap: PCI-PM_L1.2+ PCI-PM_L1.1+ ASPM_L1.2+ ASPM_L1.1+
> > > L1_PM_Substates+
> > >                   PortCommonModeRestoreTime=45us PortTPowerOnTime=50us
> > >         L1SubCtl1: PCI-PM_L1.2- PCI-PM_L1.1- ASPM_L1.2+ ASPM_L1.1-
> > >                    T_CommonMode=45us LTR1.2_Threshold=101376ns
> > >         L1SubCtl2: T_PwrOn=50us
> > > 
> > > 10000:e1:00.0 Non-Volatile memory controller: Sandisk Corp WD Blue SN550
> > > NVMe
> > > SSD (rev 01) (prog-if 02 [NVM Express])
> > >     ...
> > >     Capabilities: [900 v1] L1 PM Substates
> > >         L1SubCap: PCI-PM_L1.2+ PCI-PM_L1.1- ASPM_L1.2+ ASPM_L1.1-
> > > L1_PM_Substates+
> > >                   PortCommonModeRestoreTime=32us PortTPowerOnTime=10us
> > >         L1SubCtl1: PCI-PM_L1.2- PCI-PM_L1.1- ASPM_L1.2+ ASPM_L1.1-
> > >                    T_CommonMode=0us LTR1.2_Threshold=0ns
> > >         L1SubCtl2: T_PwrOn=10us
> > > 
> > > According to "PCIe r6.0, sec 5.5.4", before enabling ASPM L1.2 on the PCIe
> > > Root Port and the child NVMe, they should be programmed with the same
> > > LTR1.2_Threshold value. However, they have different values in this case.
> > > 
> > > Invoke aspm_calc_l12_info() to program the L1.2 parameters properly before
> > > enable L1.2 bits of L1 PM Substates Control Register in
> > > __pci_enable_link_state().
> > > 
> > > Link: https://bugzilla.kernel.org/show_bug.cgi?id=218394
> > > Signed-off-by: Jian-Hong Pan <jhp@...lessos.org>
> > > ---
> > > v2:
> > > - Prepare the PCIe LTR parameters before enable L1 Substates
> > > 
> > > v3:
> > > - Only enable supported features for the L1 Substates part
> > > 
> > > v4:
> > > - Focus on fixing L1.2 parameters, instead of re-initializing whole L1SS
> > > 
> > > v5:
> > > - Fix typo and commit message
> > > - Split introducing aspm_get_l1ss_cap() to "PCI/ASPM: Introduce
> > >   aspm_get_l1ss_cap()"
> > > 
> > >  drivers/pci/pcie/aspm.c | 12 ++++++++++++
> > >  1 file changed, 12 insertions(+)
> > > 
> > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c
> > > index c55ac11faa73..553327dee991 100644
> > > --- a/drivers/pci/pcie/aspm.c
> > > +++ b/drivers/pci/pcie/aspm.c
> > > @@ -1402,6 +1402,8 @@ EXPORT_SYMBOL(pci_disable_link_state);
> > >  static int __pci_enable_link_state(struct pci_dev *pdev, int state, bool
> > > locked)
> > >  {
> > >         struct pcie_link_state *link = pcie_aspm_get_link(pdev);
> > > +       struct pci_dev *child = link->downstream, *parent = link->pdev;
> > > +       u32 parent_l1ss_cap, child_l1ss_cap;
> > > 
> > >         if (!link)
> > >                 return -EINVAL;
> > > @@ -1433,6 +1435,16 @@ static int __pci_enable_link_state(struct pci_dev
> > > *pdev, int state, bool locked)
> > >                 link->aspm_default |= ASPM_STATE_L1_1_PCIPM |
> > > ASPM_STATE_L1;
> > >         if (state & PCIE_LINK_STATE_L1_2_PCIPM)
> > >                 link->aspm_default |= ASPM_STATE_L1_2_PCIPM |
> > > ASPM_STATE_L1;
> > > +       /*
> > > +        * Ensure L1.2 parameters: Common_Mode_Restore_Times, T_POWER_ON
> > > and
> > > +        * LTR_L1.2_THRESHOLD are programmed properly before enable bits
> > > for
> > > +        * L1.2, per PCIe r6.0, sec 5.5.4.
> > > +        */
> > > +       if (state & link->aspm_capable & ASPM_STATE_L1_2_MASK) {
> > 
> > This is still mixing PCIE_LINK_STATE flags with ASPM_STATE flags.
> 
> Thanks for your review, but I notice some description in PCIe spec,
> 5.5.4 L1 PM Substates Configuration:
> "Prior to setting either or both of the enable bits for L1.2, the
> values for TPOWER_ON, Common_Mode_Restore_Time, and, if the ASPM L1.2
> Enable bit is to be Set, the LTR_L1.2_THRESHOLD (both Value and Scale
> fields) must be programmed." => I think this includes both "ASPM L1.2
> Enable" and "PCI-PM L1.2 Enable" bits.

That's fine. While the spec clearly calls out the ASPM L1.2 Enable bit here, I
see no harm in including PCI-PM L1.2 in that check. This is what the code
already does in aspm_l1ss_init().

The issue is the mixed used of two different types of flags that don't have the
same meaning. 'state' contains PCIE_LINK_STATE flags which are part of the
caller API to the pci_<enabled/disable>_link_state() functions. The ASPM_STATE
flags are used internally to aspm.c to track all states and their meaningful
combinations such as ASPM_STATE_L1_2_MASK which includes ASPM L1.2 and PCI-PM
L1.2. You should not do bit operations between them.

Also, you should not require that the timings be calculated only if L1_2 is
enabled. You should calculate the timings as long as it's capable. This is also
what aspm_l1ss_init() does.

The confusion might be over the fact that you are having
__pci_enable_link_state() call aspm_calc_l12_info(). This should have been
handled during initialization of the link in aspm_l1ss_init() and I'm not sure
why it didn't. Maybe it's because, for VMD, ASPM default state would have
started out all disabled and this somehow led to aspm_l1ss_init() not getting
called. But looking through the code I don't see it. It would be great if you
can confirm why they weren't calculated before.

David

> 
> Jain-Hong Pan
> 
> > 'state' should not even matter.
> > The timings should always be calculated and programmed as long
> > as L1_2 is capable. That way the timings are ready even if L1_2 isn't being
> > enabled now (in case the user enables it later).
> > 
> > David
> > 
> > > +               parent_l1ss_cap = aspm_get_l1ss_cap(parent);
> > > +               child_l1ss_cap = aspm_get_l1ss_cap(child);
> > > +               aspm_calc_l12_info(link, parent_l1ss_cap, child_l1ss_cap);
> > > +       }
> > >         pcie_config_aspm_link(link, policy_to_aspm_state(link));
> > > 
> > >         link->clkpm_default = (state & PCIE_LINK_STATE_CLKPM) ? 1 : 0;
> > 

Powered by blists - more mailing lists