Re: [PATCH 1/6] RISC-V: simplify software interrupt / IPI code

From: Palmer Dabbelt
Date: Wed Jul 25 2018 - 17:48:00 EST


On Wed, 25 Jul 2018 02:36:44 PDT (-0700), Christoph Hellwig wrote:
Rename handle_ipi to riscv_software_interrupt, drop the unused return
value and provide a stub for the !SMP build. This allows simplifying
the upcoming interrupt controller driver by not providing a wrapper
for it.

Signed-off-by: Christoph Hellwig <hch@xxxxxx>
---
arch/riscv/include/asm/smp.h | 13 +++++++++++--
arch/riscv/kernel/smp.c | 6 ++----
2 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/arch/riscv/include/asm/smp.h b/arch/riscv/include/asm/smp.h
index 85e4220839b0..80ecb957fe9f 100644
--- a/arch/riscv/include/asm/smp.h
+++ b/arch/riscv/include/asm/smp.h
@@ -44,8 +44,17 @@ void arch_send_call_function_single_ipi(int cpu);
*/
#define raw_smp_processor_id() (*((int*)((char*)get_current() + TASK_TI_CPU)))

-/* Interprocessor interrupt handler */
-irqreturn_t handle_ipi(void);
+/* Software interrupt handler */
+void riscv_software_interrupt(void);
+
+#else /* CONFIG_SMP */
+
+/*
+ * We currently only use software interrupts to pass inter-processor
+ * interrupts, so if a non-SMP system gets a software interrupt then we
+ * don't know what to do.
+ */
+#define riscv_software_interrupt() WARN_ON()

#endif /* CONFIG_SMP */

diff --git a/arch/riscv/kernel/smp.c b/arch/riscv/kernel/smp.c
index 6d3962435720..906fe21ea21b 100644
--- a/arch/riscv/kernel/smp.c
+++ b/arch/riscv/kernel/smp.c
@@ -45,7 +45,7 @@ int setup_profiling_timer(unsigned int multiplier)
return -EINVAL;
}

-irqreturn_t handle_ipi(void)
+void riscv_software_interrupt(void)
{
unsigned long *pending_ipis = &ipi_data[smp_processor_id()].bits;

@@ -60,7 +60,7 @@ irqreturn_t handle_ipi(void)

ops = xchg(pending_ipis, 0);
if (ops == 0)
- return IRQ_HANDLED;
+ return;

if (ops & (1 << IPI_RESCHEDULE))
scheduler_ipi();
@@ -73,8 +73,6 @@ irqreturn_t handle_ipi(void)
/* Order data access and bit testing. */
mb();
}
-
- return IRQ_HANDLED;
}

static void

Acked-by: Palmer Dabbelt <palmer@xxxxxxxxxx>

I think it's probably easier to have the whole patch set go through the IRQ
tree, so I won't put these in the RISC-V tree unless someone says something.