Skip to contentSkip to conversation

irqchip/riscv-imsic: Add kernel parameter to disable IPIs

PATCH Superseded Anup Patel 1 participant 2 messages Jul 2, 2025

Subsystems (4)

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Anup Patel

author
#0 lore raw

PATCH v2 1/1 irqchip/riscv-imsic: Add kernel parameter to disable IPIs

When injecting IPIs to a set of harts, the IMSIC IPI support will do a separate MMIO write to the SETIPNUM_LE register of each target hart. This means on a platform where IMSIC is trap-n-emulated, there will be N MMIO traps when injecting IPI to N target harts hence IMSIC IPIs will be slow on such platform compared to the SBI IPI extension.

Unfortunately, there is no DT, ACPI, or any other way of discovering whether the underlying IMSIC is trap-n-emulated. Using MMIO write to the SETIPNUM_LE register for injecting IPI is purely a software choice in the IMSIC driver hence add a kernel parameter to allow users disable IMSIC IPIs on platforms with trap-n-emulated IMSIC.

---
 Documentation/admin-guide/kernel-parameters.txt |  7 +++++++
 drivers/irqchip/irq-riscv-imsic-early.c         | 12 ++++++++++++
 2 files changed, 19 insertions(+)
Documentation/admin-guide/kernel-parameters.txt +7 -0
index f1f2c0874da9..7f0e12d0d260 100644
@@ -2538,6 +2538,13 @@
 			requires the kernel to be built with
 			CONFIG_ARM64_PSEUDO_NMI.
 
+	irqchip.riscv_imsic_noipi
+			[RISC-V,EARLY]
+			Force the kernel to not use IMSIC software injected MSIs
+			as IPIs. Intended for system where IMSIC is trap-n-emulated,
+			and thus want to reduce MMIO traps when triggering IPIs
+			to multiple harts.
+
 	irqfixup	[HW]
 			When an interrupt is not handled search all handlers
 			for it. Intended to get systems with badly broken
drivers/irqchip/irq-riscv-imsic-early.c +12 -0
index 1dbc41d7fe80..c6fba92dd5a9 100644
@@ -9,6 +9,7 @@
 #include <linux/cpu.h>
 #include <linux/export.h>
 #include <linux/interrupt.h>
+#include <linux/init.h>
 #include <linux/io.h>
 #include <linux/irq.h>
 #include <linux/irqchip.h>
@@ -22,6 +23,14 @@
 #include "irq-riscv-imsic-state.h"
 
 static int imsic_parent_irq;
+static bool imsic_noipi;
+
+static int __init imsic_noipi_cfg(char *buf)
+{
+	imsic_noipi = true;
+	return 0;
+}
+early_param("irqchip.riscv_imsic_noipi", imsic_noipi_cfg);
 
 #ifdef CONFIG_SMP
 static void imsic_ipi_send(unsigned int cpu)
@@ -47,6 +56,9 @@ static int __init imsic_ipi_domain_init(void)
 {
 	int virq;
 
+	if (imsic_noipi)
+		return 0;
+
 	/* Create IMSIC IPI multiplexing */
 	virq = ipi_mux_create(IMSIC_NR_IPI, imsic_ipi_send);
 	if (virq <= 0)
Signature · 1 line
2.43.0
author
#1 lore raw
Quoted 65 lines · Anup Patel

When injecting IPIs to a set of harts, the IMSIC IPI support will do a separate MMIO write to the SETIPNUM_LE register of each target hart. This means on a platform where IMSIC is trap-n-emulated, there will be N MMIO traps when injecting IPI to N target harts hence IMSIC IPIs will be slow on such platform compared to the SBI IPI extension.

Unfortunately, there is no DT, ACPI, or any other way of discovering whether the underlying IMSIC is trap-n-emulated. Using MMIO write to the SETIPNUM_LE register for injecting IPI is purely a software choice in the IMSIC driver hence add a kernel parameter to allow users disable IMSIC IPIs on platforms with trap-n-emulated IMSIC.

---
 Documentation/admin-guide/kernel-parameters.txt |  7 +++++++
 drivers/irqchip/irq-riscv-imsic-early.c         | 12 ++++++++++++
 2 files changed, 19 insertions(+)
Documentation/admin-guide/kernel-parameters.txt +7 -0
index f1f2c0874da9..7f0e12d0d260 100644
@@ -2538,6 +2538,13 @@
                        requires the kernel to be built with
                        CONFIG_ARM64_PSEUDO_NMI.
 
+       irqchip.riscv_imsic_noipi
+                       [RISC-V,EARLY]
+                       Force the kernel to not use IMSIC software injected MSIs
+                       as IPIs. Intended for system where IMSIC is trap-n-emulated,
+                       and thus want to reduce MMIO traps when triggering IPIs
+                       to multiple harts.
+
        irqfixup        [HW]
                        When an interrupt is not handled search all handlers
                        for it. Intended to get systems with badly broken
drivers/irqchip/irq-riscv-imsic-early.c +9 -0
index 1dbc41d7fe80..c6fba92dd5a9 100644
@@ -9,6 +9,7 @@
 #include <linux/cpu.h>
 #include <linux/export.h>
 #include <linux/interrupt.h>
+#include <linux/init.h>
 #include <linux/io.h>
 #include <linux/irq.h>
 #include <linux/irqchip.h>
@@ -22,6 +23,14 @@
 #include "irq-riscv-imsic-state.h"
 
 static int imsic_parent_irq;
+static bool imsic_noipi;
+
+static int __init imsic_noipi_cfg(char *buf)
+{
+       imsic_noipi = true;
+       return 0;
+}
+early_param("irqchip.riscv_imsic_noipi", imsic_noipi_cfg);
 
 #ifdef CONFIG_SMP
 static void imsic_ipi_send(unsigned int cpu)

drivers/irqchip/irq-riscv-imsic-early.c static int __init imsic_ipi_domain_init(void) · line 61

Quoted 6 lines
drivers/irqchip/irq-riscv-imsic-early.c +3 -0
@@ -47,6 +56,9 @@ static int __init imsic_ipi_domain_init(void)
 {
        int virq;
 
+       if (imsic_noipi)
+               return 0;
+

We can skip enabling/disabling IMSIC_IPI_ID in imsic_ipi_starting_cpu() and imsic_ipi_dying_cpu() when imsic_noipi is set.

In addition to the above, we can also re-use the reserved IPI ID for devices.

I will quickly send v3. Apologies for the noise.

Regards,
Anup