x86/ioapic: Generate RTE directly from parent irqchip's MSI message

The I/O-APIC generates an MSI cycle with address/data bits taken from its
Redirection Table Entry in some combination which used to make sense, but
now is just a bunch of bits which get passed through in some seemingly
arbitrary order.

Instead of making IRQ remapping drivers directly frob the I/OA-PIC RTE, let
them just do their job and generate an MSI message. The bit swizzling to
turn that MSI message into the I/O-APIC's RTE is the same in all cases,
since it's a function of the I/O-APIC hardware. The IRQ remappers have no
real need to get involved with that.

The only slight caveat is that the I/OAPIC is interpreting some of those
fields too, and it does want the 'vector' field to be unique to make EOI
work. The AMD IOMMU happens to put its IRTE index in the bits that the
I/O-APIC thinks are the vector field, and accommodates this requirement by
reserving the first 32 indices for the I/O-APIC.  The Intel IOMMU doesn't
actually use the bits that the I/O-APIC thinks are the vector field, so it
fills in the 'pin' value there instead.

[ tglx: Replaced the unreadably macro maze with the cleaned up RTE/msi_msg
  	bitfields and added commentry to explain the mapping magic ]

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20201024213535.443185-22-dwmw2@infradead.org

1 files changed, 72 insertions, 31 deletions

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 07e754131854..ea983da1a57f 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -48,6 +48,7 @@
 #include <linux/jiffies.h>	/* time_after() */
 #include <linux/slab.h>
 #include <linux/memblock.h>
+#include <linux/msi.h>
 
 #include <asm/irqdomain.h>
 #include <asm/io.h>
@@ -63,7 +64,6 @@
 #include <asm/setup.h>
 #include <asm/irq_remapping.h>
 #include <asm/hw_irq.h>
-
 #include <asm/apic.h>
 
 #define	for_each_ioapic(idx)		\
@@ -1848,21 +1848,58 @@ static void ioapic_ir_ack_level(struct irq_data *irq_data)
 	eoi_ioapic_pin(data->entry.vector, data);
 }
 
+/*
+ * The I/OAPIC is just a device for generating MSI messages from legacy
+ * interrupt pins. Various fields of the RTE translate into bits of the
+ * resulting MSI which had a historical meaning.
+ *
+ * With interrupt remapping, many of those bits have different meanings
+ * in the underlying MSI, but the way that the I/OAPIC transforms them
+ * from its RTE to the MSI message is the same. This function allows
+ * the parent IRQ domain to compose the MSI message, then takes the
+ * relevant bits to put them in the appropriate places in the RTE in
+ * order to generate that message when the IRQ happens.
+ *
+ * The setup here relies on a preconfigured route entry (is_level,
+ * active_low, masked) because the parent domain is merely composing the
+ * generic message routing information which is used for the MSI.
+ */
+static void ioapic_setup_msg_from_msi(struct irq_data *irq_data,
+				      struct IO_APIC_route_entry *entry)
+{
+	struct msi_msg msg;
+
+	/* Let the parent domain compose the MSI message */
+	irq_chip_compose_msi_msg(irq_data, &msg);
+
+	/*
+	 * - Real vector
+	 * - DMAR/IR: 8bit subhandle (ioapic.pin)
+	 * - AMD/IR:  8bit IRTE index
+	 */
+	entry->vector			= msg.arch_data.vector;
+	/* Delivery mode (for DMAR/IR all 0) */
+	entry->delivery_mode		= msg.arch_data.delivery_mode;
+	/* Destination mode or DMAR/IR index bit 15 */
+	entry->dest_mode_logical	= msg.arch_addr_lo.dest_mode_logical;
+	/* DMAR/IR: 1, 0 for all other modes */
+	entry->ir_format		= msg.arch_addr_lo.dmar_format;
+	/*
+	 * DMAR/IR: index bit 0-14.
+	 *
+	 * All other modes have bit 0-6 of dmar_index_0_14 cleared and the
+	 * topmost 8 bits are destination id bit 0-7 (entry::destid_0_7).
+	 */
+	entry->ir_index_0_14		= msg.arch_addr_lo.dmar_index_0_14;
+}
+
 static void ioapic_configure_entry(struct irq_data *irqd)
 {
 	struct mp_chip_data *mpd = irqd->chip_data;
-	struct irq_cfg *cfg = irqd_cfg(irqd);
 	struct irq_pin_list *entry;
 
-	/*
-	 * Only update when the parent is the vector domain, don't touch it
-	 * if the parent is the remapping domain. Check the installed
-	 * ioapic chip to verify that.
-	 */
-	if (irqd->chip == &ioapic_chip) {
-		mpd->entry.destid_0_7 = cfg->dest_apicid;
-		mpd->entry.vector = cfg->vector;
-	}
+	ioapic_setup_msg_from_msi(irqd, &mpd->entry);
+
 	for_each_irq_pin(entry, mpd->irq_2_pin)
 		__ioapic_write_entry(entry->apic, entry->pin, mpd->entry);
 }
@@ -2919,14 +2956,23 @@ static void mp_irqdomain_get_attr(u32 gsi, struct mp_chip_data *data,
 	}
 }
 
-static void mp_setup_entry(struct irq_cfg *cfg, struct mp_chip_data *data,
-			   struct IO_APIC_route_entry *entry)
+/*
+ * Configure the I/O-APIC specific fields in the routing entry.
+ *
+ * This is important to setup the I/O-APIC specific bits (is_level,
+ * active_low, masked) because the underlying parent domain will only
+ * provide the routing information and is oblivious of the I/O-APIC
+ * specific bits.
+ *
+ * The entry is just preconfigured at this point and not written into the
+ * RTE. This happens later during activation which will fill in the actual
+ * routing information.
+ */
+static void mp_preconfigure_entry(struct mp_chip_data *data)
 {
+	struct IO_APIC_route_entry *entry = &data->entry;
+
 	memset(entry, 0, sizeof(*entry));
-	entry->delivery_mode	 = apic->delivery_mode;
-	entry->dest_mode_logical = apic->dest_mode_logical;
-	entry->destid_0_7	 = cfg->dest_apicid;
-	entry->vector		 = cfg->vector;
 	entry->is_level		 = data->is_level;
 	entry->active_low	 = data->active_low;
 	/*
@@ -2939,11 +2985,10 @@ static void mp_setup_entry(struct irq_cfg *cfg, struct mp_chip_data *data,
 int mp_irqdomain_alloc(struct irq_domain *domain, unsigned int virq,
 		       unsigned int nr_irqs, void *arg)
 {
-	int ret, ioapic, pin;
-	struct irq_cfg *cfg;
-	struct irq_data *irq_data;
-	struct mp_chip_data *data;
 	struct irq_alloc_info *info = arg;
+	struct mp_chip_data *data;
+	struct irq_data *irq_data;
+	int ret, ioapic, pin;
 	unsigned long flags;
 
 	if (!info || nr_irqs > 1)
@@ -2961,7 +3006,6 @@ int mp_irqdomain_alloc(struct irq_domain *domain, unsigned int virq,
 	if (!data)
 		return -ENOMEM;
 
-	info->ioapic.entry = &data->entry;
 	ret = irq_domain_alloc_irqs_parent(domain, virq, nr_irqs, info);
 	if (ret < 0) {
 		kfree(data);
@@ -2975,23 +3019,20 @@ int mp_irqdomain_alloc(struct irq_domain *domain, unsigned int virq,
 	irq_data->chip_data = data;
 	mp_irqdomain_get_attr(mp_pin_to_gsi(ioapic, pin), data, info);
 
-	cfg = irqd_cfg(irq_data);
 	add_pin_to_irq_node(data, ioapic_alloc_attr_node(info), ioapic, pin);
 
-	local_irq_save(flags);
-	if (info->ioapic.entry)
-		mp_setup_entry(cfg, data, info->ioapic.entry);
+	mp_preconfigure_entry(data);
 	mp_register_handler(virq, data->is_level);
+
+	local_irq_save(flags);
 	if (virq < nr_legacy_irqs())
 		legacy_pic->mask(virq);
 	local_irq_restore(flags);
 
 	apic_printk(APIC_VERBOSE, KERN_DEBUG
-		    "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> IRQ %d Mode:%i Active:%i Dest:%d)\n",
-		    ioapic, mpc_ioapic_id(ioapic), pin, cfg->vector,
-		    virq, data->is_level, data->active_low,
-		    cfg->dest_apicid);
-
+		    "IOAPIC[%d]: Preconfigured routing entry (%d-%d -> IRQ %d Level:%i ActiveLow:%i)\n",
+		    ioapic, mpc_ioapic_id(ioapic), pin, virq,
+		    data->is_level, data->active_low);
 	return 0;
 }