ARCv2: SMP: Emulate IPI to self using software triggered interrupt

commit bb143f8 upstream.

ARConnect/MCIP Inter-Core-Interrupt module can't send interrupt to
local core. So use core intc capability to trigger software
interrupt to self, using an unsued IRQ #21.

This showed up as csd deadlock with LTP trace_sched on a dual core
system. This test acts as scheduler fuzzer, triggering all sorts of
schedulting activity. Trouble starts with IPI to self, which doesn't get
delivered (effectively lost due to H/w capability), but the msg intended
to be sent remain enqueued in per-cpu @ipi_data.

All subsequent IPIs to this core from other cores get elided due to the
IPI coalescing optimization in ipi_send_msg_one() where a pending msg
implies an IPI already sent and assumes other core is yet to ack it.
After the elided IPI, other core simply goes into csd_lock_wait()
but never comes out as this core never sees the interrupt.

Fixes STAR 9001008624

Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Author: vineetgarc

arch/arc/include/asm/irqflags-arcv2.h

@@ -22,6 +22,7 @@
 #define AUX_IRQ_CTRL		0x00E
 #define AUX_IRQ_ACT		0x043	/* Active Intr across all levels */
 #define AUX_IRQ_LVL_PEND	0x200	/* Pending Intr across all levels */
+#define AUX_IRQ_HINT		0x201	/* For generating Soft Interrupts */
 #define AUX_IRQ_PRIORITY	0x206
 #define ICAUSE			0x40a
 #define AUX_IRQ_SELECT		0x40b
@@ -112,6 +113,16 @@ static inline int arch_irqs_disabled(void)
 	return arch_irqs_disabled_flags(arch_local_save_flags());
 }
 
+static inline void arc_softirq_trigger(int irq)
+{
+	write_aux_reg(AUX_IRQ_HINT, irq);
+}
+
+static inline void arc_softirq_clear(int irq)
+{
+	write_aux_reg(AUX_IRQ_HINT, 0);
+}
+
 #else
 
 .macro IRQ_DISABLE  scratch

arch/arc/kernel/entry-arcv2.S

@@ -45,11 +45,12 @@ VECTOR	reserved		; Reserved slots
 VECTOR	handle_interrupt	; (16) Timer0
 VECTOR	handle_interrupt	; unused (Timer1)
 VECTOR	handle_interrupt	; unused (WDT)
-VECTOR	handle_interrupt	; (19) ICI (inter core interrupt)
-VECTOR	handle_interrupt
-VECTOR	handle_interrupt
-VECTOR	handle_interrupt
-VECTOR	handle_interrupt	; (23) End of fixed IRQs
+VECTOR	handle_interrupt	; (19) Inter core Interrupt (IPI)
+VECTOR	handle_interrupt	; (20) perf Interrupt
+VECTOR	handle_interrupt	; (21) Software Triggered Intr (Self IPI)
+VECTOR	handle_interrupt	; unused
+VECTOR	handle_interrupt	; (23) unused
+# End of fixed IRQs
 
 .rept CONFIG_ARC_NUMBER_OF_INTERRUPTS - 8
 	VECTOR	handle_interrupt

arch/arc/kernel/mcip.c

@@ -11,9 +11,12 @@
 #include <linux/smp.h>
 #include <linux/irq.h>
 #include <linux/spinlock.h>
+#include <asm/irqflags-arcv2.h>
 #include <asm/mcip.h>
 #include <asm/setup.h>
 
+#define SOFTIRQ_IRQ	21
+
 static char smp_cpuinfo_buf[128];
 static int idu_detected;
 
@@ -22,13 +25,20 @@ static DEFINE_RAW_SPINLOCK(mcip_lock);
 static void mcip_setup_per_cpu(int cpu)
 {
 	smp_ipi_irq_setup(cpu, IPI_IRQ);
+	smp_ipi_irq_setup(cpu, SOFTIRQ_IRQ);
 }
 
 static void mcip_ipi_send(int cpu)
 {
 	unsigned long flags;
 	int ipi_was_pending;
 
+	/* ARConnect can only send IPI to others */
+	if (unlikely(cpu == raw_smp_processor_id())) {
+		arc_softirq_trigger(SOFTIRQ_IRQ);
+		return;
+	}
+
 	/*
 	 * NOTE: We must spin here if the other cpu hasn't yet
 	 * serviced a previous message. This can burn lots
@@ -63,6 +73,11 @@ static void mcip_ipi_clear(int irq)
 	unsigned long flags;
 	unsigned int __maybe_unused copy;
 
+	if (unlikely(irq == SOFTIRQ_IRQ)) {
+		arc_softirq_clear(irq);
+		return;
+	}
+
 	raw_spin_lock_irqsave(&mcip_lock, flags);
 
 	/* Who sent the IPI */