[Bug 285993] nvme device breakage in 14.2 STABLE n270867-25df691800f0

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=285993

--- Comment #20 from John Baldwin <jhb@FreeBSD.org> ---
(In reply to Jim Long from comment #17)
In this case, the PCI-e registers start at offset 0x90:

    cap 10[90] = PCI-Express 2 root port max data 128(256) ARI disabled

The SLOT_CAP register is at a relative offset of 0x14 from there, so can you
please do:

pciconf -r pcib10 0xa4

You can also perhaps try this (fine to do on top of your existing patches):

diff --git a/sys/dev/pci/pci_pci.c b/sys/dev/pci/pci_pci.c
index 5e71a376604b..27aea3ebd2ed 100644
--- a/sys/dev/pci/pci_pci.c
+++ b/sys/dev/pci/pci_pci.c
@@ -1294,9 +1294,12 @@ pcib_setup_hotplug(struct pcib_softc *sc)

        sc->pcie_link_sta = pcie_read_config(dev, PCIER_LINK_STA, 2);
        sc->pcie_slot_sta = pcie_read_config(dev, PCIER_SLOT_STA, 2);
+       device_printf(dev, "initial SLOT_STA %#x\n", sc->pcie_slot_sta);

        /* Clear any events previously pending. */
        pcie_write_config(dev, PCIER_SLOT_STA, sc->pcie_slot_sta, 2);
+       sc->pcie_slot_sta = pcie_read_config(dev, PCIER_SLOT_STA, 2);
+       device_printf(dev, "cleared SLOT_STA %#x\n", sc->pcie_slot_sta);

        /* Enable HotPlug events. */
        mask = PCIEM_SLOT_CTL_DLLSCE | PCIEM_SLOT_CTL_HPIE |

The one difference here beyond the printf's is that it is explicitly discarding
any pending events from POST.

-- 
You are receiving this mail because:
You are the assignee for the bug.