Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 25 Apr 2023 13:49:37 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     "bibo, mao" <maobibo@loongson.cn>
Cc:     Frederic Weisbecker <frederic@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Huacai Chen <chenhuacai@kernel.org>,
        WANG Xuerui <kernel@xen0n.name>,
        "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>,
        Anna-Maria Behnsen <anna-maria@linutronix.de>,
        LKML <linux-kernel@vger.kernel.org>
Subject: Re: Loongson (and other $ARCHs?) idle VS timer enqueue
Message-ID: <20230425114937.GC1335080@hirez.programming.kicks-ass.net>
References: <ZEKDZEQmKExv0O7Q@lothringen>
 <87leil2r7v.ffs@tglx>
 <20230422081700.GB1214746@hirez.programming.kicks-ass.net>
 <ZEPteS82TbIhMQxe@lothringen>
 <20230422150409.GL1214746@hirez.programming.kicks-ass.net>
 <7d91fa2a-57c5-6c78-8e2d-7fbdd6a11cba@loongson.cn>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <7d91fa2a-57c5-6c78-8e2d-7fbdd6a11cba@loongson.cn>
Precedence: bulk

On Sun, Apr 23, 2023 at 09:52:49PM +0800, bibo, mao wrote:

> > @@ -40,6 +40,7 @@ SYM_FUNC_START(handle_vint)
> >   	ori	t0, t0, 0x1f
> >   	xori	t0, t0, 0x1f
> >   	bne	t0, t1, 1f
> > +	addi.d	t0, t0, 0x20
> It is more reasonable with this patch, this will jump out of idle function
> directly after interrupt returns. If so, can we remove checking
> _TIF_NEED_RESCHED in idle ASM function?
> 
> > +	move	t0, CSR_CRMD_IE
> > +	csrxchg	t0, t0, LOONGARCH_CSR_CRMD
> -   	LONG_L	t0, tp, TI_FLAGS
> +	nop
> >   	nop
> -	andi	t0, t0, _TIF_NEED_RESCHED
> -	bnez	t0, 1f
> +	nop
> +	nop
> >   	nop
> > -	nop
> > -	nop
> >   	idle	0

Would not something like the below be a more compact form?
That is; afaict there is no reason to keep it 32 bytes, we can easily go
16 and drop 4 nops.

Additionally, instead of truncating to the start, increase to the end by
doing:

	ip |= 0xf;
	ip++;

Also; I added a wee comment.


diff --git a/arch/loongarch/kernel/genex.S b/arch/loongarch/kernel/genex.S
index 44ff1ff64260..3c8a6bab98fe 100644
--- a/arch/loongarch/kernel/genex.S
+++ b/arch/loongarch/kernel/genex.S
@@ -18,27 +18,31 @@
 
 	.align	5
 SYM_FUNC_START(__arch_cpu_idle)
-	/* start of rollback region */
-	LONG_L	t0, tp, TI_FLAGS
-	nop
-	andi	t0, t0, _TIF_NEED_RESCHED
-	bnez	t0, 1f
-	nop
-	nop
-	nop
+	/* start of idle interrupt region */
+	move	t0, CSR_CRMD_IE
+	csrxchg	t0, t0, LOONGARCH_CSR_CRMD
+	/*
+	 * If an interrupt lands here; between enabling interrupts above and
+	 * going idle on the next instruction, we must *NOT* go idle since the
+	 * interrupt could have set TIF_NEED_RESCHED or caused an timer to need
+	 * reprogramming. Fall through -- see handle_vint() below -- and have
+	 * the idle loop take care of things.
+	 */
 	idle	0
-	/* end of rollback region */
-1:	jr	ra
+	nop
+	/* end of idle interrupt region */
+SYM_INNER_LBEL(__arch_cpu_idle_exit, SYM_L_LOCAL)
+	jr	ra
 SYM_FUNC_END(__arch_cpu_idle)
 
 SYM_FUNC_START(handle_vint)
 	BACKUP_T0T1
 	SAVE_ALL
-	la_abs	t1, __arch_cpu_idle
+	la_abs	t1, __arch_cpu_idle_exit
 	LONG_L	t0, sp, PT_ERA
-	/* 32 byte rollback region */
-	ori	t0, t0, 0x1f
-	xori	t0, t0, 0x1f
+	/* 16 byte idle interrupt region */
+	ori	t0, t0, 0x0f
+	addi.d	t0, t0, 1
 	bne	t0, t1, 1f
 	LONG_S	t0, sp, PT_ERA
 1:	move	a0, sp
diff --git a/arch/loongarch/kernel/idle.c b/arch/loongarch/kernel/idle.c
index 0b5dd2faeb90..5ba72d229920 100644
--- a/arch/loongarch/kernel/idle.c
+++ b/arch/loongarch/kernel/idle.c
@@ -11,7 +11,6 @@
 
 void __cpuidle arch_cpu_idle(void)
 {
-	raw_local_irq_enable();
 	__arch_cpu_idle(); /* idle instruction needs irq enabled */
 	raw_local_irq_disable();
 }