From: Bjorn Helgaas bhelgaas@google.com
[ Upstream commit d08c30d7a0d1826f771f16cde32bd86e48401791 ]
This reverts commit ec411e02b7a2e785a4ed9ed283207cd14f48699d.
Patrick reported that this commit broke hybrid graphics on a ThinkPad X1 Extreme 2nd with Intel UHD Graphics 630 and NVIDIA GeForce GTX 1650 Mobile:
nouveau 0000:01:00.0: fifo: PBDMA0: 01000000 [] ch 0 [00ff992000 DRM] subc 0 mthd 0008 data 00000000
Karol reported that this commit broke Nouveau firmware loading on a Lenovo P1G2 with Intel UHD Graphics 630 and NVIDIA TU117GLM [Quadro T1000 Mobile]:
nouveau 0000:01:00.0: acr: AHESASC binary failed
In both cases, reverting ec411e02b7a2 solved the problem. Unfortunately, this revert will reintroduce the "Thunderbolt bridges take long time to resume from D3cold" problem: https://bugzilla.kernel.org/show_bug.cgi?id=206837
Link: https://lore.kernel.org/r/CAErSpo5sTeK_my1dEhWp7aHD0xOp87+oHYWkTjbL7ALgDbXo-... Link: https://lore.kernel.org/r/CACO55tsAEa5GXw5oeJPG=mcn+qxNvspXreJYWDJGZBy5v82JD... Link: https://bugzilla.kernel.org/show_bug.cgi?id=208597 Reported-by: Patrick Volkerding volkerdi@gmail.com Reported-by: Karol Herbst kherbst@redhat.com Fixes: ec411e02b7a2 ("PCI/PM: Assume ports without DLL Link Active train links in 100 ms") Signed-off-by: Bjorn Helgaas bhelgaas@google.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/pci/pci.c | 30 +++++++++--------------------- 1 file changed, 9 insertions(+), 21 deletions(-)
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index d4758518a97bd..a4efc7e0061f5 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -4662,8 +4662,7 @@ static int pci_pm_reset(struct pci_dev *dev, int probe) * pcie_wait_for_link_delay - Wait until link is active or inactive * @pdev: Bridge device * @active: waiting for active or inactive? - * @delay: Delay to wait after link has become active (in ms). Specify %0 - * for no delay. + * @delay: Delay to wait after link has become active (in ms) * * Use this to wait till link becomes active or inactive. */ @@ -4704,7 +4703,7 @@ static bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active, msleep(10); timeout -= 10; } - if (active && ret && delay) + if (active && ret) msleep(delay); else if (ret != active) pci_info(pdev, "Data Link Layer Link Active not %s in 1000 msec\n", @@ -4825,28 +4824,17 @@ void pci_bridge_wait_for_secondary_bus(struct pci_dev *dev) if (!pcie_downstream_port(dev)) return;
- /* - * Per PCIe r5.0, sec 6.6.1, for downstream ports that support - * speeds > 5 GT/s, we must wait for link training to complete - * before the mandatory delay. - * - * We can only tell when link training completes via DLL Link - * Active, which is required for downstream ports that support - * speeds > 5 GT/s (sec 7.5.3.6). Unfortunately some common - * devices do not implement Link Active reporting even when it's - * required, so we'll check for that directly instead of checking - * the supported link speed. We assume devices without Link Active - * reporting can train in 100 ms regardless of speed. - */ - if (dev->link_active_reporting) { - pci_dbg(dev, "waiting for link to train\n"); - if (!pcie_wait_for_link_delay(dev, true, 0)) { + if (pcie_get_speed_cap(dev) <= PCIE_SPEED_5_0GT) { + pci_dbg(dev, "waiting %d ms for downstream link\n", delay); + msleep(delay); + } else { + pci_dbg(dev, "waiting %d ms for downstream link, after activation\n", + delay); + if (!pcie_wait_for_link_delay(dev, true, delay)) { /* Did not train, no need to wait any further */ return; } } - pci_dbg(child, "waiting %d ms to become accessible\n", delay); - msleep(delay);
if (!pci_device_is_present(child)) { pci_dbg(child, "waiting additional %d ms to become accessible\n", delay);