Thread (28 messages) 28 messages, 4 authors, 2025-02-19

Re: [PATCH v1 02/13] genirq/msi: Rename iommu_dma_compose_msi_msg() to msi_msg_set_addr()

From: Thomas Gleixner <hidden>
Date: 2025-02-13 12:11:52
Also in: linux-iommu, linux-kselftest, linux-patches, lkml

On Sat, Feb 08 2025 at 01:02, Nicolin Chen wrote:
From: Jason Gunthorpe <jgg@nvidia.com>

The new function is used to take in a u64 MSI address and store it in the
Which new function? The subject claims this is a rename. That's
confusing at best.
msi_msg. If the iommu has provided an alternative address then that is
replaced instead.

All callers have a tidy u64 already so this also consolidates the repeated
low/high code into a small helper.
Now I'm really confused. Something like:

    genirq/msi: Refactor iommu_dma_compose_msi_msg()

    The call sites of iommu_dma_compose_msi_msg() have the following
    pattern:

      1) Set the MSI message address to the non-translated address

      2) Invoke iommu_dma_compose_msi_msg() to apply IOVA translation if
         available.

    Rework and rename iommu_dma_compose_msi_msg() to handles both
    address set up variants in one consolidated helper.

Or something to that effect.
-/**
- * iommu_dma_compose_msi_msg() - Apply translation to an MSI message
- * @desc: MSI descriptor prepared by iommu_dma_prepare_msi()
- * @msg: MSI message containing target physical address
- */
Please keep and rework the documentation comment.
-static inline void iommu_dma_compose_msi_msg(struct msi_desc *desc,
-					     struct msi_msg *msg)
+static inline void msi_msg_set_addr(struct msi_desc *desc, struct msi_msg *msg,
+				    u64 msi_addr)
No line break required. You have 100 characters available.
quoted hunk ↗ jump to hunk
 {
 #ifdef CONFIG_IRQ_MSI_IOMMU
 	if (desc->iommu_msi_page_shift) {
@@ -314,11 +309,14 @@ static inline void iommu_dma_compose_msi_msg(struct msi_desc *desc,
 			       << desc->iommu_msi_page_shift;
 
 		msg->address_hi = upper_32_bits(msi_iova);
-		msg->address_lo = lower_32_bits(msi_iova) |
-				  (msg->address_lo &
-				   ((1 << desc->iommu_msi_page_shift) - 1));
+		msg->address_lo =
+			lower_32_bits(msi_iova) |
+			(msi_addr & ((1 << desc->iommu_msi_page_shift) - 1));
Please avoid unrelated random formatting changes. Especially in this
case this is even more non-sensical as you introduced the original
formatting in the previous patch. So fix it up there and make it:

		msg->address_lo = lower_32_bits(msi_iova) |
				  (msg->address_lo & ((1 << desc->iommu_msi_page_shift) - 1));

which is way more readable than this displaced variant above.
+		return;
 
-	iommu_dma_compose_msi_msg(irq_data_get_msi_desc(d), msg);
+	msg->data = its_get_event_id(d);
+	msi_msg_set_addr(irq_data_get_msi_desc(d), msg,
+			 its_dev->its->get_msi_base(its_dev));
No line break required here and at the other places.

Thanks,

        tglx
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help