* [PATCH v5] PCI/PTM: Enable PTM only if it advertises a role
@ 2025-11-12 7:46 Mika Westerberg
2025-11-12 23:42 ` Bjorn Helgaas
0 siblings, 1 reply; 2+ messages in thread
From: Mika Westerberg @ 2025-11-12 7:46 UTC (permalink / raw)
To: linux-pci; +Cc: Bjorn Helgaas, Lukas Wunner, Mika Westerberg
We have a Upstream Port (2b:00.0) that has following in the PTM capability:
Capabilities: [220 v1] Precision Time Measurement
PTMCap: Requester- Responder- Root-
Linux enables PTM for this without looking into what roles it actually
supports. Immediately after enabling PTM we start getting these:
pci 0000:2b:00.0: [8086:5786] type 01 class 0x060400 PCIe Switch Upstream Port
...
pci 0000:2b:00.0: PTM enabled, 4ns granularity
...
pcieport 0000:00:07.1: AER: Multiple Uncorrectable (Non-Fatal) error message received from 0000:00:07.1
pcieport 0000:00:07.1: PCIe Bus Error: severity=Uncorrectable (Non-Fatal), type=Transaction Layer, (Receiver ID)
pcieport 0000:00:07.1: device [8086:e44f] error status/mask=00200000/00000000
pcieport 0000:00:07.1: [21] ACSViol (First)
pcieport 0000:00:07.1: AER: TLP Header: 0x34000000 0x00000052 0x00000000 0x00000000
Fix this by enabling PTM only if any of the following conditions are
true (see more in PCIe r7.0 sec 6.21.1 figure 6-21):
- PCIe Endpoint that has PTM capability must to declare requester
capable
- PCIe Switch Upstream Port that has PTM capability must declare
at least responder capable
- PCIe Root Port must declare root port capable.
While there make the enabling happen for all in __pci_enable_ptm() instead
of enabling some in pci_ptm_init() and some in __pci_enable_ptm().
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
---
Hi,
I hope I did not make any stupid mistakes this time ;-) My testing still
passed: the Root Port that has ->ptm_root (and ->ptm_responder) PTM is
enabled and the Switch Upstream Port that does not have ->ptm_responder is
not enabled (and I don't see the flood of AER errors).
Previous versions can be seen:
v4: https://lore.kernel.org/linux-pci/20251111061048.681752-1-mika.westerberg@linux.intel.com/
v3: https://lore.kernel.org/linux-pci/20251030134606.3782352-1-mika.westerberg@linux.intel.com/
v2: https://lore.kernel.org/linux-pci/20251028060427.2163115-1-mika.westerberg@linux.intel.com/
v1: https://lore.kernel.org/linux-pci/20251021104833.3729120-1-mika.westerberg@linux.intel.com/
Changes from v4:
- Do not enable PTM automatically for all components (e.g keep the
existing behavior).
- Make the switch-case new lines consistent.
Changes from v3:
- Cache the responder and requester capability bits.
- Enable PTM only in __pci_enable_ptm().
- Update $subject and commit message.
- Since this is changed quite a lot, I dropped the Reviewed-by from Lukas
and also stable tag.
Changes from v2:
- Limit the check in __pci_enable_ptm() to Endpoints and Legacy
Endpoints.
- Added stable tags suggested by Lukas, and PCIe spec reference.
- Added Reviewed-by tag from Lukas (hope it is okay to keep).
Changes from v1:
- Limit Switch Upstream Port only to Responder, not both Requester and
Responder.
drivers/pci/pcie/ptm.c | 35 +++++++++++++++++++++++++++++++++++
include/linux/pci.h | 2 ++
2 files changed, 37 insertions(+)
diff --git a/drivers/pci/pcie/ptm.c b/drivers/pci/pcie/ptm.c
index 65e4b008be00..fb1f3d0d8448 100644
--- a/drivers/pci/pcie/ptm.c
+++ b/drivers/pci/pcie/ptm.c
@@ -81,6 +81,11 @@ void pci_ptm_init(struct pci_dev *dev)
dev->ptm_granularity = 0;
}
+ if (cap & PCI_PTM_CAP_RES)
+ dev->ptm_responder = 1;
+ if (cap & PCI_PTM_CAP_REQ)
+ dev->ptm_requester = 1;
+
if (pci_pcie_type(dev) == PCI_EXP_TYPE_ROOT_PORT ||
pci_pcie_type(dev) == PCI_EXP_TYPE_UPSTREAM)
pci_enable_ptm(dev, NULL);
@@ -144,6 +149,36 @@ static int __pci_enable_ptm(struct pci_dev *dev)
return -EINVAL;
}
+ switch (pci_pcie_type(dev)) {
+ case PCI_EXP_TYPE_ROOT_PORT:
+ /*
+ * Root Port must declare Root Capable if we want to enable
+ * PTM for it.
+ */
+ if (!dev->ptm_root)
+ return -EINVAL;
+ break;
+ case PCI_EXP_TYPE_UPSTREAM:
+ /*
+ * Switch Upstream Ports must at least declare Responder
+ * Capable if we want to enable PTM for it.
+ */
+ if (!dev->ptm_responder)
+ return -EINVAL;
+ break;
+ case PCI_EXP_TYPE_ENDPOINT:
+ case PCI_EXP_TYPE_LEG_END:
+ /*
+ * PCIe Endpoint must declare Requester Capable before we
+ * can enable PTM for it.
+ */
+ if (!dev->ptm_requester)
+ return -EINVAL;
+ break;
+ default:
+ return -EINVAL;
+ }
+
pci_read_config_dword(dev, ptm + PCI_PTM_CTRL, &ctrl);
ctrl |= PCI_PTM_CTRL_ENABLE;
diff --git a/include/linux/pci.h b/include/linux/pci.h
index d1fdf81fbe1e..d5018cb5c331 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -500,6 +500,8 @@ struct pci_dev {
#ifdef CONFIG_PCIE_PTM
u16 ptm_cap; /* PTM Capability */
unsigned int ptm_root:1;
+ unsigned int ptm_responder:1;
+ unsigned int ptm_requester:1;
unsigned int ptm_enabled:1;
u8 ptm_granularity;
#endif
--
2.50.1
^ permalink raw reply related [flat|nested] 2+ messages in thread* Re: [PATCH v5] PCI/PTM: Enable PTM only if it advertises a role
2025-11-12 7:46 [PATCH v5] PCI/PTM: Enable PTM only if it advertises a role Mika Westerberg
@ 2025-11-12 23:42 ` Bjorn Helgaas
0 siblings, 0 replies; 2+ messages in thread
From: Bjorn Helgaas @ 2025-11-12 23:42 UTC (permalink / raw)
To: Mika Westerberg; +Cc: linux-pci, Bjorn Helgaas, Lukas Wunner
On Wed, Nov 12, 2025 at 08:46:14AM +0100, Mika Westerberg wrote:
> We have a Upstream Port (2b:00.0) that has following in the PTM capability:
>
> Capabilities: [220 v1] Precision Time Measurement
> PTMCap: Requester- Responder- Root-
Sec 7.9.15.2 says "Switches supporting PTM must Set PTM Responder
Capable". To me that sounds like "Switches with a PTM Capability must
Set PTM Responder Capable". But I guess whoever designed this device
didn't think the same way.
> Linux enables PTM for this without looking into what roles it actually
> supports. Immediately after enabling PTM we start getting these:
>
> pci 0000:2b:00.0: [8086:5786] type 01 class 0x060400 PCIe Switch Upstream Port
> ...
> pci 0000:2b:00.0: PTM enabled, 4ns granularity
> ...
> pcieport 0000:00:07.1: AER: Multiple Uncorrectable (Non-Fatal) error message received from 0000:00:07.1
> pcieport 0000:00:07.1: PCIe Bus Error: severity=Uncorrectable (Non-Fatal), type=Transaction Layer, (Receiver ID)
> pcieport 0000:00:07.1: device [8086:e44f] error status/mask=00200000/00000000
> pcieport 0000:00:07.1: [21] ACSViol (First)
> pcieport 0000:00:07.1: AER: TLP Header: 0x34000000 0x00000052 0x00000000 0x00000000
>
> Fix this by enabling PTM only if any of the following conditions are
> true (see more in PCIe r7.0 sec 6.21.1 figure 6-21):
>
> - PCIe Endpoint that has PTM capability must to declare requester
> capable
> - PCIe Switch Upstream Port that has PTM capability must declare
> at least responder capable
> - PCIe Root Port must declare root port capable.
>
> While there make the enabling happen for all in __pci_enable_ptm() instead
> of enabling some in pci_ptm_init() and some in __pci_enable_ptm().
>
> Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Applied to pci/ptm for v6.19, thanks!
> ---
> Hi,
>
> I hope I did not make any stupid mistakes this time ;-) My testing still
> passed: the Root Port that has ->ptm_root (and ->ptm_responder) PTM is
> enabled and the Switch Upstream Port that does not have ->ptm_responder is
> not enabled (and I don't see the flood of AER errors).
>
> Previous versions can be seen:
>
> v4: https://lore.kernel.org/linux-pci/20251111061048.681752-1-mika.westerberg@linux.intel.com/
> v3: https://lore.kernel.org/linux-pci/20251030134606.3782352-1-mika.westerberg@linux.intel.com/
> v2: https://lore.kernel.org/linux-pci/20251028060427.2163115-1-mika.westerberg@linux.intel.com/
> v1: https://lore.kernel.org/linux-pci/20251021104833.3729120-1-mika.westerberg@linux.intel.com/
>
> Changes from v4:
>
> - Do not enable PTM automatically for all components (e.g keep the
> existing behavior).
> - Make the switch-case new lines consistent.
>
> Changes from v3:
>
> - Cache the responder and requester capability bits.
> - Enable PTM only in __pci_enable_ptm().
> - Update $subject and commit message.
> - Since this is changed quite a lot, I dropped the Reviewed-by from Lukas
> and also stable tag.
>
> Changes from v2:
>
> - Limit the check in __pci_enable_ptm() to Endpoints and Legacy
> Endpoints.
> - Added stable tags suggested by Lukas, and PCIe spec reference.
> - Added Reviewed-by tag from Lukas (hope it is okay to keep).
>
> Changes from v1:
>
> - Limit Switch Upstream Port only to Responder, not both Requester and
> Responder.
>
> drivers/pci/pcie/ptm.c | 35 +++++++++++++++++++++++++++++++++++
> include/linux/pci.h | 2 ++
> 2 files changed, 37 insertions(+)
>
> diff --git a/drivers/pci/pcie/ptm.c b/drivers/pci/pcie/ptm.c
> index 65e4b008be00..fb1f3d0d8448 100644
> --- a/drivers/pci/pcie/ptm.c
> +++ b/drivers/pci/pcie/ptm.c
> @@ -81,6 +81,11 @@ void pci_ptm_init(struct pci_dev *dev)
> dev->ptm_granularity = 0;
> }
>
> + if (cap & PCI_PTM_CAP_RES)
> + dev->ptm_responder = 1;
> + if (cap & PCI_PTM_CAP_REQ)
> + dev->ptm_requester = 1;
> +
> if (pci_pcie_type(dev) == PCI_EXP_TYPE_ROOT_PORT ||
> pci_pcie_type(dev) == PCI_EXP_TYPE_UPSTREAM)
> pci_enable_ptm(dev, NULL);
> @@ -144,6 +149,36 @@ static int __pci_enable_ptm(struct pci_dev *dev)
> return -EINVAL;
> }
>
> + switch (pci_pcie_type(dev)) {
> + case PCI_EXP_TYPE_ROOT_PORT:
> + /*
> + * Root Port must declare Root Capable if we want to enable
> + * PTM for it.
> + */
> + if (!dev->ptm_root)
> + return -EINVAL;
> + break;
> + case PCI_EXP_TYPE_UPSTREAM:
> + /*
> + * Switch Upstream Ports must at least declare Responder
> + * Capable if we want to enable PTM for it.
> + */
> + if (!dev->ptm_responder)
> + return -EINVAL;
> + break;
> + case PCI_EXP_TYPE_ENDPOINT:
> + case PCI_EXP_TYPE_LEG_END:
> + /*
> + * PCIe Endpoint must declare Requester Capable before we
> + * can enable PTM for it.
> + */
> + if (!dev->ptm_requester)
> + return -EINVAL;
> + break;
> + default:
> + return -EINVAL;
> + }
> +
> pci_read_config_dword(dev, ptm + PCI_PTM_CTRL, &ctrl);
>
> ctrl |= PCI_PTM_CTRL_ENABLE;
> diff --git a/include/linux/pci.h b/include/linux/pci.h
> index d1fdf81fbe1e..d5018cb5c331 100644
> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -500,6 +500,8 @@ struct pci_dev {
> #ifdef CONFIG_PCIE_PTM
> u16 ptm_cap; /* PTM Capability */
> unsigned int ptm_root:1;
> + unsigned int ptm_responder:1;
> + unsigned int ptm_requester:1;
> unsigned int ptm_enabled:1;
> u8 ptm_granularity;
> #endif
> --
> 2.50.1
>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2025-11-12 23:42 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-11-12 7:46 [PATCH v5] PCI/PTM: Enable PTM only if it advertises a role Mika Westerberg
2025-11-12 23:42 ` Bjorn Helgaas
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox