[tip: irq/urgent] PCI/MSI: Add an option to write MSIX ENTRY_DATA before any reads

From: tip-bot2 for Jonathan Currier
Date: Tue Apr 15 2025 - 02:36:48 EST


The following commit has been merged into the irq/urgent branch of tip:

Commit-ID: cf761e3dacc6ad5f65a4886d00da1f9681e6805a
Gitweb: https://git.kernel.org/tip/cf761e3dacc6ad5f65a4886d00da1f9681e6805a
Author: Jonathan Currier <dullfire@xxxxxxxxx>
AuthorDate: Sun, 17 Nov 2024 17:48:42 -06:00
Committer: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
CommitterDate: Tue, 15 Apr 2025 08:32:18 +02:00

PCI/MSI: Add an option to write MSIX ENTRY_DATA before any reads

Commit 7d5ec3d36123 ("PCI/MSI: Mask all unused MSI-X entries") introduced a
readl() from ENTRY_VECTOR_CTRL before the writel() to ENTRY_DATA.

This is correct, however some hardware, like the Sun Neptune chips, the NIU
module, will cause an error and/or fatal trap if any MSIX table entry is
read before the corresponding ENTRY_DATA field is written to.

Add an optional early writel() in msix_prepare_msi_desc().

Fixes: 7d5ec3d36123 ("PCI/MSI: Mask all unused MSI-X entries")
Signed-off-by: Jonathan Currier <dullfire@xxxxxxxxx>
Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx
Link: https://lore.kernel.org/all/20241117234843.19236-2-dullfire@xxxxxxxxx
---
drivers/pci/msi/msi.c | 3 +++
include/linux/pci.h | 2 ++
2 files changed, 5 insertions(+)

diff --git a/drivers/pci/msi/msi.c b/drivers/pci/msi/msi.c
index 6569ba3..8b88487 100644
--- a/drivers/pci/msi/msi.c
+++ b/drivers/pci/msi/msi.c
@@ -615,6 +615,9 @@ void msix_prepare_msi_desc(struct pci_dev *dev, struct msi_desc *desc)
void __iomem *addr = pci_msix_desc_addr(desc);

desc->pci.msi_attrib.can_mask = 1;
+ /* Workaround for SUN NIU insanity, which requires write before read */
+ if (dev->dev_flags & PCI_DEV_FLAGS_MSIX_TOUCH_ENTRY_DATA_FIRST)
+ writel(0, addr + PCI_MSIX_ENTRY_DATA);
desc->pci.msix_ctrl = readl(addr + PCI_MSIX_ENTRY_VECTOR_CTRL);
}
}
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 0e8e3fd..51e2bd6 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -245,6 +245,8 @@ enum pci_dev_flags {
PCI_DEV_FLAGS_NO_RELAXED_ORDERING = (__force pci_dev_flags_t) (1 << 11),
/* Device does honor MSI masking despite saying otherwise */
PCI_DEV_FLAGS_HAS_MSI_MASKING = (__force pci_dev_flags_t) (1 << 12),
+ /* Device requires write to PCI_MSIX_ENTRY_DATA before any MSIX reads */
+ PCI_DEV_FLAGS_MSIX_TOUCH_ENTRY_DATA_FIRST = (__force pci_dev_flags_t) (1 << 13),
};

enum pci_irq_reroute_variant {