| Message ID | 20220602065544.2552771-1-nathan@nathanrossi.com |
|---|---|
| State | New |
| Headers | show |
| Series | PCI/ASPM: Wait for data link active after retraining | expand |
On Thu, 2 Jun 2022 at 16:55, Nathan Rossi <nathan@nathanrossi.com> wrote: > > From: Nathan Rossi <nathan.rossi@digi.com> > > When retraining the link either the child or the parent device may have > the data link layer state machine of the respective devices move out of > the active state despite the physical link training being completed. > Depending on how long is takes for the devices to return to the active > state, the device may not be ready and any further reads/writes to the > device can fail. > > This issue is present with the pci-mvebu controller paired with a device > supporting ASPM but without advertising the Slot Clock, where during > boot the pcie_aspm_cap_init call would cause common clocks to be made > consistent and then retrain the link. However the data link layer would > not be active before any device initialization (e.g. ASPM capability > queries, BAR configuration) causing improper configuration of the device > without error. > > To ensure the child device is accessible, after the link retraining use > pcie_wait_for_link to perform the associated state checks and any needed > delays. > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > --- Just pinging this patch, are there any comments or feedback for this change? Thanks, Nathan > drivers/pci/pcie/aspm.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > index a96b7424c9..4b8a1810be 100644 > --- a/drivers/pci/pcie/aspm.c > +++ b/drivers/pci/pcie/aspm.c > @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) > reg16 &= ~PCI_EXP_LNKCTL_CCC; > pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); > > - if (pcie_retrain_link(link)) > + /* Retrain link and then wait for the link to become active */ > + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) > return; > > /* Training failed. Restore common clock configurations */ > --- > 2.36.1
Hello! On Thursday 02 June 2022 06:55:44 Nathan Rossi wrote: > From: Nathan Rossi <nathan.rossi@digi.com> > > When retraining the link either the child or the parent device may have > the data link layer state machine of the respective devices move out of > the active state despite the physical link training being completed. > Depending on how long is takes for the devices to return to the active > state, the device may not be ready and any further reads/writes to the > device can fail. > > This issue is present with the pci-mvebu controller paired with a device > supporting ASPM but without advertising the Slot Clock, where during > boot the pcie_aspm_cap_init call would cause common clocks to be made > consistent and then retrain the link. However the data link layer would > not be active before any device initialization (e.g. ASPM capability > queries, BAR configuration) causing improper configuration of the device > without error. There is the known issue in marvell pcie controllers. They completely drop the link for PCIe GEN1 cards when Target Link Speed (Link Control2) in Root Port is configured to 5.0 GT/s or higher value and OS issues Retrain Link (Link Control). I think the proper way should be to workaround root of this issue by programming Target Link Speed in Link Control2 register to required value, instead of hacking couple of other places which are just implication of that issue... I can reproduce it for example with Qualcomm Atheros ath9k/ath10k wifi cards which have another issue that they go into "broken" state when in-band reset (e.g. pcie hot reset or pcie link down) is issues multiple times without longer delay. These two bugs (first in marvell pcie controller and second in wifi card) cause that setting kernel ASPM cause disappearing card from bus until cpu/board reset (or pcie warm reset; if board supports it at runtime without going to POR). I guess you are just observing result of this issue here. > To ensure the child device is accessible, after the link retraining use > pcie_wait_for_link to perform the associated state checks and any needed > delays. > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > --- > drivers/pci/pcie/aspm.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > index a96b7424c9..4b8a1810be 100644 > --- a/drivers/pci/pcie/aspm.c > +++ b/drivers/pci/pcie/aspm.c > @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) > reg16 &= ~PCI_EXP_LNKCTL_CCC; > pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); > > - if (pcie_retrain_link(link)) > + /* Retrain link and then wait for the link to become active */ > + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) > return; > > /* Training failed. Restore common clock configurations */ > --- > 2.36.1
On Mon, 3 Oct 2022 at 03:56, Pali Rohár <pali@kernel.org> wrote: > > Hello! > > On Thursday 02 June 2022 06:55:44 Nathan Rossi wrote: > > From: Nathan Rossi <nathan.rossi@digi.com> > > > > When retraining the link either the child or the parent device may have > > the data link layer state machine of the respective devices move out of > > the active state despite the physical link training being completed. > > Depending on how long is takes for the devices to return to the active > > state, the device may not be ready and any further reads/writes to the > > device can fail. > > > > This issue is present with the pci-mvebu controller paired with a device > > supporting ASPM but without advertising the Slot Clock, where during > > boot the pcie_aspm_cap_init call would cause common clocks to be made > > consistent and then retrain the link. However the data link layer would > > not be active before any device initialization (e.g. ASPM capability > > queries, BAR configuration) causing improper configuration of the device > > without error. > > There is the known issue in marvell pcie controllers. They completely > drop the link for PCIe GEN1 cards when Target Link Speed (Link Control2) > in Root Port is configured to 5.0 GT/s or higher value and OS issues > Retrain Link (Link Control). In the configuration we are having issues with, the downstream device is indeed a 2.5GT downstream, and the upstream is configured to support 2.5GT and 5GT (Armada 385). So it does make sense that this known issue would apply. I tested setting the target speed to 2.5GT within mvebu_pcie_setup_hw before the retraining occurs, and it does resolve the retraining delay/link drop. So this issue is indeed the problem we are having. Is this behaviour mentioned in any errata? > > I think the proper way should be to workaround root of this issue by > programming Target Link Speed in Link Control2 register to required > value, instead of hacking couple of other places which are just > implication of that issue... By programming the Target Link Speed, are you referring to programming the value like other controller drivers do with dtb configuration? Relying on this would be problematic for our design (mixed downstream link speed variants). Does it make more sense to set the Target Link Speed when the retrain bit is being set in Link Control (e.g. in the mvebu_pci_bridge_emul_pcie_conf_write function) essentially preventing the retraining from causing the link to drop only when the issue is expected to present itself (Link Status at 2.5GT)? Thanks, Nathan > > I can reproduce it for example with Qualcomm Atheros ath9k/ath10k wifi > cards which have another issue that they go into "broken" state when > in-band reset (e.g. pcie hot reset or pcie link down) is issues multiple > times without longer delay. > > These two bugs (first in marvell pcie controller and second in wifi > card) cause that setting kernel ASPM cause disappearing card from bus > until cpu/board reset (or pcie warm reset; if board supports it at > runtime without going to POR). > > I guess you are just observing result of this issue here. > > > To ensure the child device is accessible, after the link retraining use > > pcie_wait_for_link to perform the associated state checks and any needed > > delays. > > > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > > --- > > drivers/pci/pcie/aspm.c | 3 ++- > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > > index a96b7424c9..4b8a1810be 100644 > > --- a/drivers/pci/pcie/aspm.c > > +++ b/drivers/pci/pcie/aspm.c > > @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) > > reg16 &= ~PCI_EXP_LNKCTL_CCC; > > pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); > > > > - if (pcie_retrain_link(link)) > > + /* Retrain link and then wait for the link to become active */ > > + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) > > return; > > > > /* Training failed. Restore common clock configurations */ > > --- > > 2.36.1
On Wednesday 02 November 2022 18:31:04 Nathan Rossi wrote: > On Mon, 3 Oct 2022 at 03:56, Pali Rohár <pali@kernel.org> wrote: > > > > Hello! > > > > On Thursday 02 June 2022 06:55:44 Nathan Rossi wrote: > > > From: Nathan Rossi <nathan.rossi@digi.com> > > > > > > When retraining the link either the child or the parent device may have > > > the data link layer state machine of the respective devices move out of > > > the active state despite the physical link training being completed. > > > Depending on how long is takes for the devices to return to the active > > > state, the device may not be ready and any further reads/writes to the > > > device can fail. > > > > > > This issue is present with the pci-mvebu controller paired with a device > > > supporting ASPM but without advertising the Slot Clock, where during > > > boot the pcie_aspm_cap_init call would cause common clocks to be made > > > consistent and then retrain the link. However the data link layer would > > > not be active before any device initialization (e.g. ASPM capability > > > queries, BAR configuration) causing improper configuration of the device > > > without error. > > > > There is the known issue in marvell pcie controllers. They completely > > drop the link for PCIe GEN1 cards when Target Link Speed (Link Control2) > > in Root Port is configured to 5.0 GT/s or higher value and OS issues > > Retrain Link (Link Control). > > In the configuration we are having issues with, the downstream device > is indeed a 2.5GT downstream, and the upstream is configured to > support 2.5GT and 5GT (Armada 385). So it does make sense that this > known issue would apply. I tested setting the target speed to 2.5GT > within mvebu_pcie_setup_hw before the retraining occurs, and it does > resolve the retraining delay/link drop. So this issue is indeed the > problem we are having. I observed exactly same behavior as you described, on Armada 385 and similar also on Armada 3720. > Is this behaviour mentioned in any errata? I have not documented this behavior in any document. > > > > I think the proper way should be to workaround root of this issue by > > programming Target Link Speed in Link Control2 register to required > > value, instead of hacking couple of other places which are just > > implication of that issue... > > By programming the Target Link Speed, are you referring to programming > the value like other controller drivers do with dtb configuration? No, I mean programming Link Control 2 register in PCIe capabilities section of PCIe Root Port device (the one which is implemented throw emulated bridge kernel code). It should be programmed dynamically after detecting type of upstream device to the PCIe Root Port (other end of the serdes link). > Relying on this would be problematic for our design (mixed downstream > link speed variants). Yes, due to this reason it should be rather implemented dynamically in kernel code instead of "hacking" DTS files as it does not really fix it. > Does it make more sense to set the Target Link > Speed when the retrain bit is being set in Link Control (e.g. in the > mvebu_pci_bridge_emul_pcie_conf_write function) essentially preventing > the retraining from causing the link to drop only when the issue is > expected to present itself (Link Status at 2.5GT)? > > Thanks, > Nathan > > > > > I can reproduce it for example with Qualcomm Atheros ath9k/ath10k wifi > > cards which have another issue that they go into "broken" state when > > in-band reset (e.g. pcie hot reset or pcie link down) is issues multiple > > times without longer delay. > > > > These two bugs (first in marvell pcie controller and second in wifi > > card) cause that setting kernel ASPM cause disappearing card from bus > > until cpu/board reset (or pcie warm reset; if board supports it at > > runtime without going to POR). > > > > I guess you are just observing result of this issue here. > > > > > To ensure the child device is accessible, after the link retraining use > > > pcie_wait_for_link to perform the associated state checks and any needed > > > delays. > > > > > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > > > --- > > > drivers/pci/pcie/aspm.c | 3 ++- > > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > > > index a96b7424c9..4b8a1810be 100644 > > > --- a/drivers/pci/pcie/aspm.c > > > +++ b/drivers/pci/pcie/aspm.c > > > @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) > > > reg16 &= ~PCI_EXP_LNKCTL_CCC; > > > pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); > > > > > > - if (pcie_retrain_link(link)) > > > + /* Retrain link and then wait for the link to become active */ > > > + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) > > > return; > > > > > > /* Training failed. Restore common clock configurations */ > > > --- > > > 2.36.1
On Thu, Jun 02, 2022 at 06:55:44AM +0000, Nathan Rossi wrote: > From: Nathan Rossi <nathan.rossi@digi.com> > > When retraining the link either the child or the parent device may have > the data link layer state machine of the respective devices move out of > the active state despite the physical link training being completed. > Depending on how long is takes for the devices to return to the active > state, the device may not be ready and any further reads/writes to the > device can fail. > > This issue is present with the pci-mvebu controller paired with a device > supporting ASPM but without advertising the Slot Clock, where during > boot the pcie_aspm_cap_init call would cause common clocks to be made > consistent and then retrain the link. However the data link layer would > not be active before any device initialization (e.g. ASPM capability > queries, BAR configuration) causing improper configuration of the device > without error. > > To ensure the child device is accessible, after the link retraining use > pcie_wait_for_link to perform the associated state checks and any needed > delays. > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > --- > drivers/pci/pcie/aspm.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > index a96b7424c9..4b8a1810be 100644 > --- a/drivers/pci/pcie/aspm.c > +++ b/drivers/pci/pcie/aspm.c > @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) > reg16 &= ~PCI_EXP_LNKCTL_CCC; > pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); > > - if (pcie_retrain_link(link)) > + /* Retrain link and then wait for the link to become active */ > + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) pcie_retrain_link() waits for PCI_EXP_LNKSTA_LT (Link Training) to be cleared, which means the LTSSM has exited the Configuration/Recovery state. pcie_wait_for_link() waits for PCI_EXP_LNKSTA_DLLLA (Data Link Layer Link Active) to be set, which means the link is in DL_Active. I don't see an explicit procedure in the spec for determining when a link retrain is complete, but from PCIe r6.0, sec 6.2.11 (DPC): After software releases the Downstream Port from DPC, the Port’s LTSSM must transition to the Detect state, where the Link will attempt to retrain. Software can use Data Link Layer State Changed interrupts, DL_ACTIVE ERR_COR signaling, or both, to signal when the Link reaches the DL_Active state again. and sec 6.6: On the completion of Link Training (entering the DL_Active state, see Section 3.2), a component must be able to receive and process TLPs and DLLPs. The only use mentioned in the spec for the Link Training bit is the implementation note in sec 7.5.3.7 about avoiding race conditions when using the Retrain Link bit, where software should poll Link Training until it returns to zero before setting the Retrain Link bit to change link parameters. And I think you're absolutely right that what we *want* here is the data link layer DL_Active state, not just the link layer L0 state. This all makes me think that checking the Link Training bit might be the wrong thing to begin with. Of course, the Data Link Layer Link Active bit wasn't added until PCIe r1.1, and even now it's optional. Without it, I don't know if there's a way to make sure the link is in DL_Active. Maybe pcie_retrain_link() should wait for Data Link Layer Link Active if it is supported, and use the existing behavior of waiting for Link Training to be cleared otherwise? > return; > > /* Training failed. Restore common clock configurations */ > --- > 2.36.1
[+cc Maciej for similar retrain issue] On Tue, Nov 08, 2022 at 04:29:44PM -0600, Bjorn Helgaas wrote: > On Thu, Jun 02, 2022 at 06:55:44AM +0000, Nathan Rossi wrote: > > From: Nathan Rossi <nathan.rossi@digi.com> > > > > When retraining the link either the child or the parent device may have > > the data link layer state machine of the respective devices move out of > > the active state despite the physical link training being completed. > > Depending on how long is takes for the devices to return to the active > > state, the device may not be ready and any further reads/writes to the > > device can fail. > > > > This issue is present with the pci-mvebu controller paired with a device > > supporting ASPM but without advertising the Slot Clock, where during > > boot the pcie_aspm_cap_init call would cause common clocks to be made > > consistent and then retrain the link. However the data link layer would > > not be active before any device initialization (e.g. ASPM capability > > queries, BAR configuration) causing improper configuration of the device > > without error. > > > > To ensure the child device is accessible, after the link retraining use > > pcie_wait_for_link to perform the associated state checks and any needed > > delays. > > > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > > --- > > drivers/pci/pcie/aspm.c | 3 ++- > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > > index a96b7424c9..4b8a1810be 100644 > > --- a/drivers/pci/pcie/aspm.c > > +++ b/drivers/pci/pcie/aspm.c > > @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) > > reg16 &= ~PCI_EXP_LNKCTL_CCC; > > pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); > > > > - if (pcie_retrain_link(link)) > > + /* Retrain link and then wait for the link to become active */ > > + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) > > pcie_retrain_link() waits for PCI_EXP_LNKSTA_LT (Link Training) to be > cleared, which means the LTSSM has exited the Configuration/Recovery > state. pcie_wait_for_link() waits for PCI_EXP_LNKSTA_DLLLA (Data Link > Layer Link Active) to be set, which means the link is in DL_Active. > > I don't see an explicit procedure in the spec for determining when > a link retrain is complete, but from PCIe r6.0, sec 6.2.11 (DPC): > > After software releases the Downstream Port from DPC, the Port’s > LTSSM must transition to the Detect state, where the Link will > attempt to retrain. Software can use Data Link Layer State Changed > interrupts, DL_ACTIVE ERR_COR signaling, or both, to signal when the > Link reaches the DL_Active state again. > > and sec 6.6: > > On the completion of Link Training (entering the DL_Active state, > see Section 3.2), a component must be able to receive and process > TLPs and DLLPs. > > The only use mentioned in the spec for the Link Training bit is the > implementation note in sec 7.5.3.7 about avoiding race conditions when > using the Retrain Link bit, where software should poll Link Training > until it returns to zero before setting the Retrain Link bit to change > link parameters. > > And I think you're absolutely right that what we *want* here is the > data link layer DL_Active state, not just the link layer L0 state. > > This all makes me think that checking the Link Training bit might be > the wrong thing to begin with. > > Of course, the Data Link Layer Link Active bit wasn't added until PCIe > r1.1, and even now it's optional. Without it, I don't know if there's > a way to make sure the link is in DL_Active. > > Maybe pcie_retrain_link() should wait for Data Link Layer Link Active > if it is supported, and use the existing behavior of waiting for Link > Training to be cleared otherwise? Nathan, I meant to cc you on this thread, which is doing something very similar. Just FYI: https://lore.kernel.org/all/alpine.DEB.2.21.2209130050380.60554@angie.orcam.me.uk/
On Thu, 10 Nov 2022 at 03:34, Bjorn Helgaas <helgaas@kernel.org> wrote: > > [+cc Maciej for similar retrain issue] > > On Tue, Nov 08, 2022 at 04:29:44PM -0600, Bjorn Helgaas wrote: > > On Thu, Jun 02, 2022 at 06:55:44AM +0000, Nathan Rossi wrote: > > > From: Nathan Rossi <nathan.rossi@digi.com> > > > > > > When retraining the link either the child or the parent device may have > > > the data link layer state machine of the respective devices move out of > > > the active state despite the physical link training being completed. > > > Depending on how long is takes for the devices to return to the active > > > state, the device may not be ready and any further reads/writes to the > > > device can fail. > > > > > > This issue is present with the pci-mvebu controller paired with a device > > > supporting ASPM but without advertising the Slot Clock, where during > > > boot the pcie_aspm_cap_init call would cause common clocks to be made > > > consistent and then retrain the link. However the data link layer would > > > not be active before any device initialization (e.g. ASPM capability > > > queries, BAR configuration) causing improper configuration of the device > > > without error. > > > > > > To ensure the child device is accessible, after the link retraining use > > > pcie_wait_for_link to perform the associated state checks and any needed > > > delays. > > > > > > Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> > > > --- > > > drivers/pci/pcie/aspm.c | 3 ++- > > > 1 file changed, 2 insertions(+), 1 deletion(-) > > > > > > diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c > > > index a96b7424c9..4b8a1810be 100644 > > > --- a/drivers/pci/pcie/aspm.c > > > +++ b/drivers/pci/pcie/aspm.c > > > @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) > > > reg16 &= ~PCI_EXP_LNKCTL_CCC; > > > pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); > > > > > > - if (pcie_retrain_link(link)) > > > + /* Retrain link and then wait for the link to become active */ > > > + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) > > > > pcie_retrain_link() waits for PCI_EXP_LNKSTA_LT (Link Training) to be > > cleared, which means the LTSSM has exited the Configuration/Recovery > > state. pcie_wait_for_link() waits for PCI_EXP_LNKSTA_DLLLA (Data Link > > Layer Link Active) to be set, which means the link is in DL_Active. > > > > I don't see an explicit procedure in the spec for determining when > > a link retrain is complete, but from PCIe r6.0, sec 6.2.11 (DPC): > > > > After software releases the Downstream Port from DPC, the Port’s > > LTSSM must transition to the Detect state, where the Link will > > attempt to retrain. Software can use Data Link Layer State Changed > > interrupts, DL_ACTIVE ERR_COR signaling, or both, to signal when the > > Link reaches the DL_Active state again. > > > > and sec 6.6: > > > > On the completion of Link Training (entering the DL_Active state, > > see Section 3.2), a component must be able to receive and process > > TLPs and DLLPs. > > > > The only use mentioned in the spec for the Link Training bit is the > > implementation note in sec 7.5.3.7 about avoiding race conditions when > > using the Retrain Link bit, where software should poll Link Training > > until it returns to zero before setting the Retrain Link bit to change > > link parameters. > > > > And I think you're absolutely right that what we *want* here is the > > data link layer DL_Active state, not just the link layer L0 state. > > > > This all makes me think that checking the Link Training bit might be > > the wrong thing to begin with. > > > > Of course, the Data Link Layer Link Active bit wasn't added until PCIe > > r1.1, and even now it's optional. Without it, I don't know if there's > > a way to make sure the link is in DL_Active. My understanding is there is no way to check for the DL_Active state on these devices. Which is why pcie_wait_for_link_delay uses a fixed delay in that case. > > > > Maybe pcie_retrain_link() should wait for Data Link Layer Link Active > > if it is supported, and use the existing behavior of waiting for Link > > Training to be cleared otherwise? I think it still makes sense for pcie_retrain_link to wait for the Link Training bit, since it is typical for the retraining to never go to a LTSSM state where the LinkUp is 0 (e.g. Recovery -> Configuration), thus the Data Link Layer will stay in DL_Active. But it is still important to wait for the link training to complete in that case, but also because the training may not cause the Data Link state to change immediately. However since DLLLA reporting is optional, it is probably ideal if the pcie_retrain_link only calls pcie_wait_for_link if reporting is available? This would avoid the fixed delay of 1s upon link retraining for devices without reporting, which is unnecessary in the majority of cases. Thanks, Nathan > > Nathan, I meant to cc you on this thread, which is doing something > very similar. Just FYI: > > https://lore.kernel.org/all/alpine.DEB.2.21.2209130050380.60554@angie.orcam.me.uk/
Hello, On 14/11/22 09:29, Nathan Rossi wrote: > On Thu, 10 Nov 2022 at 03:34, Bjorn Helgaas <helgaas@kernel.org> wrote: >> >> [+cc Maciej for similar retrain issue] >> >> On Tue, Nov 08, 2022 at 04:29:44PM -0600, Bjorn Helgaas wrote: >>> On Thu, Jun 02, 2022 at 06:55:44AM +0000, Nathan Rossi wrote: >>>> From: Nathan Rossi <nathan.rossi@digi.com> >>>> >>>> When retraining the link either the child or the parent device may have >>>> the data link layer state machine of the respective devices move out of >>>> the active state despite the physical link training being completed. >>>> Depending on how long is takes for the devices to return to the active >>>> state, the device may not be ready and any further reads/writes to the >>>> device can fail. >>>> >>>> This issue is present with the pci-mvebu controller paired with a device >>>> supporting ASPM but without advertising the Slot Clock, where during >>>> boot the pcie_aspm_cap_init call would cause common clocks to be made >>>> consistent and then retrain the link. However the data link layer would >>>> not be active before any device initialization (e.g. ASPM capability >>>> queries, BAR configuration) causing improper configuration of the device >>>> without error. >>>> >>>> To ensure the child device is accessible, after the link retraining use >>>> pcie_wait_for_link to perform the associated state checks and any needed >>>> delays. >>>> >>>> Signed-off-by: Nathan Rossi <nathan.rossi@digi.com> >>>> --- >>>> drivers/pci/pcie/aspm.c | 3 ++- >>>> 1 file changed, 2 insertions(+), 1 deletion(-) >>>> >>>> diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c >>>> index a96b7424c9..4b8a1810be 100644 >>>> --- a/drivers/pci/pcie/aspm.c >>>> +++ b/drivers/pci/pcie/aspm.c >>>> @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) >>>> reg16 &= ~PCI_EXP_LNKCTL_CCC; >>>> pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); >>>> >>>> - if (pcie_retrain_link(link)) >>>> + /* Retrain link and then wait for the link to become active */ >>>> + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) >>> >>> pcie_retrain_link() waits for PCI_EXP_LNKSTA_LT (Link Training) to be >>> cleared, which means the LTSSM has exited the Configuration/Recovery >>> state. pcie_wait_for_link() waits for PCI_EXP_LNKSTA_DLLLA (Data Link >>> Layer Link Active) to be set, which means the link is in DL_Active. >>> >>> I don't see an explicit procedure in the spec for determining when >>> a link retrain is complete, but from PCIe r6.0, sec 6.2.11 (DPC): >>> >>> After software releases the Downstream Port from DPC, the Port’s >>> LTSSM must transition to the Detect state, where the Link will >>> attempt to retrain. Software can use Data Link Layer State Changed >>> interrupts, DL_ACTIVE ERR_COR signaling, or both, to signal when the >>> Link reaches the DL_Active state again. >>> >>> and sec 6.6: >>> >>> On the completion of Link Training (entering the DL_Active state, >>> see Section 3.2), a component must be able to receive and process >>> TLPs and DLLPs. >>> >>> The only use mentioned in the spec for the Link Training bit is the >>> implementation note in sec 7.5.3.7 about avoiding race conditions when >>> using the Retrain Link bit, where software should poll Link Training >>> until it returns to zero before setting the Retrain Link bit to change >>> link parameters. >>> >>> And I think you're absolutely right that what we *want* here is the >>> data link layer DL_Active state, not just the link layer L0 state. >>> >>> This all makes me think that checking the Link Training bit might be >>> the wrong thing to begin with. >>> >>> Of course, the Data Link Layer Link Active bit wasn't added until PCIe >>> r1.1, and even now it's optional. Without it, I don't know if there's >>> a way to make sure the link is in DL_Active. > > My understanding is there is no way to check for the DL_Active state > on these devices. Which is why pcie_wait_for_link_delay uses a fixed > delay in that case. In TI's J721E SoC, the PCIe controller doesn't report the Data Link State. The bit is hardwired to zero. However, the status of the Data Link is reported via a set of user specific registers, which aren't a part of the standard PCIe registers. With this, I was able to observe that despite the PCI_EXP_LNKSTA_LT bit being cleared, the Data Link initialization was still ongoing. This leads to a race condition if it isn't ensured that the Data Link initialization is complete before the PCI core attempts to obtain the configuration space addresses via the map_bus call. Based on the scheduling, it is possible for the kernel to crash, and I was able to observe this in my setup. With this patch applied, the issue is fixed. Tested-by: Siddharth Vadapalli <s-vadapalli@ti.com> Regards, Siddharth.
diff --git a/drivers/pci/pcie/aspm.c b/drivers/pci/pcie/aspm.c index a96b7424c9..4b8a1810be 100644 --- a/drivers/pci/pcie/aspm.c +++ b/drivers/pci/pcie/aspm.c @@ -288,7 +288,8 @@ static void pcie_aspm_configure_common_clock(struct pcie_link_state *link) reg16 &= ~PCI_EXP_LNKCTL_CCC; pcie_capability_write_word(parent, PCI_EXP_LNKCTL, reg16); - if (pcie_retrain_link(link)) + /* Retrain link and then wait for the link to become active */ + if (pcie_retrain_link(link) && pcie_wait_for_link(parent, true)) return; /* Training failed. Restore common clock configurations */