From: Stefan Roese sr@denx.de
commit 83dbf898a2d45289be875deb580e93050ba67529 upstream.
Masking all unused MSI-X entries is done to ensure that a crash kernel starts from a clean slate, which correponds to the reset state of the device as defined in the PCI-E specificion 3.0 and later:
Vector Control for MSI-X Table Entries --------------------------------------
"00: Mask bit: When this bit is set, the function is prohibited from sending a message using this MSI-X Table entry. ... This bit’s state after reset is 1 (entry is masked)."
A Marvell NVME device fails to deliver MSI interrupts after trying to enable MSI-X interrupts due to that masking. It seems to take the MSI-X mask bits into account even when MSI-X is disabled.
While not specification compliant, this can be cured by moving the masking into the success path, so that the MSI-X table entries stay in device reset state when the MSI-X setup fails.
[ tglx: Move it into the success path, add comment and amend changelog ]
Fixes: aa8092c1d1f1 ("PCI/MSI: Mask all unused MSI-X entries") Signed-off-by: Stefan Roese sr@denx.de Signed-off-by: Thomas Gleixner tglx@linutronix.de Cc: linux-pci@vger.kernel.org Cc: Bjorn Helgaas bhelgaas@google.com Cc: Michal Simek michal.simek@xilinx.com Cc: Marek Vasut marex@denx.de Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20211210161025.3287927-1-sr@denx.de Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org --- drivers/pci/msi.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-)
--- a/drivers/pci/msi.c +++ b/drivers/pci/msi.c @@ -827,9 +827,6 @@ static int msix_capability_init(struct p goto out_disable; }
- /* Ensure that all table entries are masked. */ - msix_mask_all(base, tsize); - ret = msix_setup_entries(dev, base, entries, nvec, affd); if (ret) goto out_disable; @@ -852,6 +849,16 @@ static int msix_capability_init(struct p /* Set MSI-X enabled bits and unmask the function */ pci_intx_for_msi(dev, 0); dev->msix_enabled = 1; + + /* + * Ensure that all table entries are masked to prevent + * stale entries from firing in a crash kernel. + * + * Done late to deal with a broken Marvell NVME device + * which takes the MSI-X mask bits into account even + * when MSI-X is disabled, which prevents MSI delivery. + */ + msix_mask_all(base, tsize); pci_msix_clear_and_set_ctrl(dev, PCI_MSIX_FLAGS_MASKALL, 0);
pcibios_free_irq(dev);