[PATCH] x86: Add check for number of available vectors before CPU down [v6]

[PATCH] x86: Add check for number of available vectors before CPU down [v6]

[Prarit Bhargava]

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=64791

When a cpu is downed on a system, the irqs on the cpu are assigned to
other cpus.  It is possible, however, that when a cpu is downed there
aren't enough free vectors on the remaining cpus to account for the
vectors from the cpu that is being downed.

This results in an interesting "overflow" condition where irqs are
"assigned" to a CPU but are not handled.

For example, when downing cpus on a 1-64 logical processor system:

<snip>
[  232.021745] smpboot: CPU 61 is now offline
[  238.480275] smpboot: CPU 62 is now offline
[  245.991080] ------------[ cut here ]------------
[  245.996270] WARNING: CPU: 0 PID: 0 at net/sched/sch_generic.c:264 dev_watchdog+0x246/0x250()
[  246.005688] NETDEV WATCHDOG: p786p1 (ixgbe): transmit queue 0 timed out
[  246.013070] Modules linked in: lockd sunrpc iTCO_wdt iTCO_vendor_support sb_edac ixgbe microcode e1000e pcspkr joydev edac_core lpc_ich ioatdma ptp mdio mfd_core i2c_i801 dca pps_core i2c_core wmi acpi_cpufreq isci libsas scsi_transport_sas
[  246.037633] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.12.0+ #14
[  246.044451] Hardware name: Intel Corporation S4600LH ........../SVRBD-ROW_T, BIOS SE5C600.86B.01.08.0003.022620131521 02/26/2013
[  246.057371]  0000000000000009 ffff88081fa03d40 ffffffff8164fbf6 ffff88081fa0ee48
[  246.065728]  ffff88081fa03d90 ffff88081fa03d80 ffffffff81054ecc ffff88081fa13040
[  246.074073]  0000000000000000 ffff88200cce0000 0000000000000040 0000000000000000
[  246.082430] Call Trace:
[  246.085174]  <IRQ>  [<ffffffff8164fbf6>] dump_stack+0x46/0x58
[  246.091633]  [<ffffffff81054ecc>] warn_slowpath_common+0x8c/0xc0
[  246.098352]  [<ffffffff81054fb6>] warn_slowpath_fmt+0x46/0x50
[  246.104786]  [<ffffffff815710d6>] dev_watchdog+0x246/0x250
[  246.110923]  [<ffffffff81570e90>] ? dev_deactivate_queue.constprop.31+0x80/0x80
[  246.119097]  [<ffffffff8106092a>] call_timer_fn+0x3a/0x110
[  246.125224]  [<ffffffff8106280f>] ? update_process_times+0x6f/0x80
[  246.132137]  [<ffffffff81570e90>] ? dev_deactivate_queue.constprop.31+0x80/0x80
[  246.140308]  [<ffffffff81061db0>] run_timer_softirq+0x1f0/0x2a0
[  246.146933]  [<ffffffff81059a80>] __do_softirq+0xe0/0x220
[  246.152976]  [<ffffffff8165fedc>] call_softirq+0x1c/0x30
[  246.158920]  [<ffffffff810045f5>] do_softirq+0x55/0x90
[  246.164670]  [<ffffffff81059d35>] irq_exit+0xa5/0xb0
[  246.170227]  [<ffffffff8166062a>] smp_apic_timer_interrupt+0x4a/0x60
[  246.177324]  [<ffffffff8165f40a>] apic_timer_interrupt+0x6a/0x70
[  246.184041]  <EOI>  [<ffffffff81505a1b>] ? cpuidle_enter_state+0x5b/0xe0
[  246.191559]  [<ffffffff81505a17>] ? cpuidle_enter_state+0x57/0xe0
[  246.198374]  [<ffffffff81505b5d>] cpuidle_idle_call+0xbd/0x200
[  246.204900]  [<ffffffff8100b7ae>] arch_cpu_idle+0xe/0x30
[  246.210846]  [<ffffffff810a47b0>] cpu_startup_entry+0xd0/0x250
[  246.217371]  [<ffffffff81646b47>] rest_init+0x77/0x80
[  246.223028]  [<ffffffff81d09e8e>] start_kernel+0x3ee/0x3fb
[  246.229165]  [<ffffffff81d0989f>] ? repair_env_string+0x5e/0x5e
[  246.235787]  [<ffffffff81d095a5>] x86_64_start_reservations+0x2a/0x2c
[  246.242990]  [<ffffffff81d0969f>] x86_64_start_kernel+0xf8/0xfc
[  246.249610] ---[ end trace fb74fdef54d79039 ]---
[  246.254807] ixgbe 0000:c2:00.0 p786p1: initiating reset due to tx timeout
[  246.262489] ixgbe 0000:c2:00.0 p786p1: Reset adapter
Last login: Mon Nov 11 08:35:14 from 10.18.17.119
[root@(none) ~]# [  246.792676] ixgbe 0000:c2:00.0 p786p1: detected SFP+: 5
[  249.231598] ixgbe 0000:c2:00.0 p786p1: NIC Link is Up 10 Gbps, Flow Control: RX/TX
[  246.792676] ixgbe 0000:c2:00.0 p786p1: detected SFP+: 5
[  249.231598] ixgbe 0000:c2:00.0 p786p1: NIC Link is Up 10 Gbps, Flow Control: RX/TX

(last lines keep repeating.  ixgbe driver is dead until module reload.)

If the downed cpu has more vectors than are free on the remaining cpus on the
system, it is possible that some vectors are "orphaned" even though they are
assigned to a cpu.  In this case, since the ixgbe driver had a watchdog, the
watchdog fired and notified that something was wrong.

This patch adds a function, check_vectors(), to compare the number of vectors
on the CPU going down and compares it to the number of vectors available on
the system.  If there aren't enough vectors for the CPU to go down, an
error is returned and propogated back to userspace.

v2: Do not need to look at percpu irqs
v3: Need to check affinity to prevent counting of MSIs in IOAPIC Lowest
    Priority Mode
v4: Additional changes suggested by Gong Chen.
v5/v6: Updated comment text

Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Reviewed-by: Gong Chen <gong.chen@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Seiji Aguchi <seiji.aguchi@hds.com>
Cc: Yang Zhang <yang.z.zhang@Intel.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Janet Morgan <janet.morgan@intel.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Ruiv Wang <ruiv.wang@gmail.com>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Cc: x86@kernel.org
---
 arch/x86/include/asm/irq.h |    1 +
 arch/x86/kernel/irq.c      |   78 ++++++++++++++++++++++++++++++++++++++++++++
 arch/x86/kernel/smpboot.c  |    6 ++++
 3 files changed, 85 insertions(+)

diff --git a/arch/x86/include/asm/irq.h b/arch/x86/include/asm/irq.h
index 0ea10f27..cb6cfcd 100644
--- a/arch/x86/include/asm/irq.h
+++ b/arch/x86/include/asm/irq.h
@@ -25,6 +25,7 @@ extern void irq_ctx_init(int cpu);
 
 #ifdef CONFIG_HOTPLUG_CPU
 #include <linux/cpumask.h>
+extern int check_irq_vectors_for_cpu_disable(void);
 extern void fixup_irqs(void);
 extern void irq_force_complete_move(int);
 #endif
diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c
index 22d0687..b23fdc8 100644
--- a/arch/x86/kernel/irq.c
+++ b/arch/x86/kernel/irq.c
@@ -262,6 +262,84 @@ __visible void smp_trace_x86_platform_ipi(struct pt_regs *regs)
 EXPORT_SYMBOL_GPL(vector_used_by_percpu_irq);
 
 #ifdef CONFIG_HOTPLUG_CPU
+/*
+ * This cpu is going to be removed and its vectors migrated to the remaining
+ * online cpus.  Check to see if there are enough vectors in the remaining cpus.
+ */
+int check_irq_vectors_for_cpu_disable(void)
+{
+	int irq, cpu;
+	unsigned int this_cpu, vector, this_count, count;
+	struct irq_desc *desc;
+	struct irq_data *data;
+	struct cpumask affinity_new, online_new;
+
+	this_cpu = smp_processor_id();
+	cpumask_copy(&online_new, cpu_online_mask);
+	cpu_clear(this_cpu, online_new);
+
+	this_count = 0;
+	for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) {
+		irq = __this_cpu_read(vector_irq[vector]);
+		if (irq >= 0) {
+			desc = irq_to_desc(irq);
+			data = irq_desc_get_irq_data(desc);
+			cpumask_copy(&affinity_new, data->affinity);
+			cpu_clear(this_cpu, affinity_new);
+			/*
+			 * The check below determines if this irq requires
+			 * an empty vector_irq[irq] entry on an online
+			 * cpu.
+			 *
+			 * The code only counts active non-percpu irqs, and
+			 * those irqs that are not linked to on an online cpu.
+			 * The first test is trivial, the second is not.
+			 *
+			 * The second test takes into account the
+			 * account that a single irq may be mapped to multiple
+			 * cpu's vector_irq[] (for example IOAPIC cluster
+			 * mode).  In this case we have two
+			 * possibilities:
+			 *
+			 * 1) the resulting affinity mask is empty; that is
+			 * this the down'd cpu is the last cpu in the irq's
+			 * affinity mask, and
+			 *
+			 * 2) the resulting affinity mask is no longer
+			 * a subset of the online cpus but the affinity
+			 * mask is not zero; that is the down'd cpu is the
+			 * last online cpu in a user set affinity mask.
+			 *
+			 * In both possibilities, we need to remap the irq
+			 * to a new vector_irq[].
+			 *
+			 */
+			if (irq_has_action(irq) && !irqd_is_per_cpu(data) &&
+			    (cpumask_empty(&affinity_new) ||
+			     !cpumask_subset(&affinity_new, &online_new)))
+				this_count++;
+		}
+	}
+
+	count = 0;
+	for_each_online_cpu(cpu) {
+		if (cpu == this_cpu)
+			continue;
+		for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS;
+		     vector++) {
+			if (per_cpu(vector_irq, cpu)[vector] < 0)
+				count++;
+		}
+	}
+
+	if (count < this_count) {
+		pr_warn("CPU %d disable failed: CPU has %u vectors assigned and there are only %u available.\n",
+			this_cpu, this_count, count);
+		return -ERANGE;
+	}
+	return 0;
+}
+
 /* A cpu has been removed from cpu_online_mask.  Reset irq affinities. */
 void fixup_irqs(void)
 {
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 85dc05a..391ea52 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -1312,6 +1312,12 @@ void cpu_disable_common(void)
 
 int native_cpu_disable(void)
 {
+	int ret;
+
+	ret = check_irq_vectors_for_cpu_disable();
+	if (ret)
+		return ret;
+
 	clear_local_APIC();
 
 	cpu_disable_common();
-- 
1.7.9.3

RE: [PATCH] x86: Add check for number of available vectors before CPU down [v6]

[Luck, Tony]

+	for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) {
+		irq = __this_cpu_read(vector_irq[vector]);
+		if (irq >= 0) {
+			desc = irq_to_desc(irq);
+			data = irq_desc_get_irq_data(desc);
+			cpumask_copy(&affinity_new, data->affinity);
+			cpu_clear(this_cpu, affinity_new);
+			/*
+			 * The check below determines if this irq requires
+			 * an empty vector_irq[irq] entry on an online
+			 * cpu.
+			 *
+			 * The code only counts active non-percpu irqs, and
+			 * those irqs that are not linked to on an online cpu.
+			 * The first test is trivial, the second is not.
+			 *
+			 * The second test takes into account the
+			 * account that a single irq may be mapped to multiple
+			 * cpu's vector_irq[] (for example IOAPIC cluster
+			 * mode).  In this case we have two
+			 * possibilities:
+			 *
+			 * 1) the resulting affinity mask is empty; that is
+			 * this the down'd cpu is the last cpu in the irq's
+			 * affinity mask, and
Code says "||" - so I think comment should say "or".
+			 *
+			 * 2) the resulting affinity mask is no longer
+			 * a subset of the online cpus but the affinity
+			 * mask is not zero; that is the down'd cpu is the
+			 * last online cpu in a user set affinity mask.
+			 *
+			 * In both possibilities, we need to remap the irq
+			 * to a new vector_irq[].
+			 *
+			 */
+			if (irq_has_action(irq) && !irqd_is_per_cpu(data) &&
+			    (cpumask_empty(&affinity_new) ||
+			     !cpumask_subset(&affinity_new, &online_new)))
+				this_count++;
+		}

That's an impressive 6:1 ratio of lines-of-comment to lines-of-code!

Perhaps it would be less scary if the test were broken up into the easy/obvious part
and the one that has taken all these revisions to work out?  E.g.

			/* no need to count inactive or per-cpu irqs */
			if (!irq_has_action(irq) || irqd_is_per_cpu(data))
				continue;

			/*
			  * We need to look for a new home for this irq if:
				... paste in 1), 2) from above here ... (but s/and/or/ to match code)
			 */
			if (cpumask_empty(&affinity_new) ||
			    !cpumask_subset(&affinity_new, &online_new))
				this_count++;

-Tony

Re: [PATCH] x86: Add check for number of available vectors before CPU down [v6]

[Prarit Bhargava]

On 01/07/2014 12:54 PM, Luck, Tony wrote:
> +	for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) {
> +		irq = __this_cpu_read(vector_irq[vector]);
> +		if (irq >= 0) {
> +			desc = irq_to_desc(irq);
> +			data = irq_desc_get_irq_data(desc);
> +			cpumask_copy(&affinity_new, data->affinity);
> +			cpu_clear(this_cpu, affinity_new);
> +			/*
> +			 * The check below determines if this irq requires
> +			 * an empty vector_irq[irq] entry on an online
> +			 * cpu.
> +			 *
> +			 * The code only counts active non-percpu irqs, and
> +			 * those irqs that are not linked to on an online cpu.
> +			 * The first test is trivial, the second is not.
> +			 *
> +			 * The second test takes into account the
> +			 * account that a single irq may be mapped to multiple
> +			 * cpu's vector_irq[] (for example IOAPIC cluster
> +			 * mode).  In this case we have two
> +			 * possibilities:
> +			 *
> +			 * 1) the resulting affinity mask is empty; that is
> +			 * this the down'd cpu is the last cpu in the irq's
> +			 * affinity mask, and
> Code says "||" - so I think comment should say "or".
> +			 *
> +			 * 2) the resulting affinity mask is no longer
> +			 * a subset of the online cpus but the affinity
> +			 * mask is not zero; that is the down'd cpu is the
> +			 * last online cpu in a user set affinity mask.
> +			 *
> +			 * In both possibilities, we need to remap the irq
> +			 * to a new vector_irq[].
> +			 *
> +			 */
> +			if (irq_has_action(irq) && !irqd_is_per_cpu(data) &&
> +			    (cpumask_empty(&affinity_new) ||
> +			     !cpumask_subset(&affinity_new, &online_new)))
> +				this_count++;
> +		}
> 
> That's an impressive 6:1 ratio of lines-of-comment to lines-of-code!

Heh -- I couldn't decide if I should keep it all together in one comment or
divide it up.  I guess it does look less scary if I divide it up.  So how about
(sorry for the cut-and-paste)


        for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) {
                irq = __this_cpu_read(vector_irq[vector]);
                if (irq >= 0) {
                        desc = irq_to_desc(irq);
                        data = irq_desc_get_irq_data(desc);
                        cpumask_copy(&affinity_new, data->affinity);
                        cpu_clear(this_cpu, affinity_new);

                        /* Do not count inactive or per-cpu irqs. */
                        if (!irq_has_action(irq) || irqd_is_per_cpu(data))
                                continue;

                        /*
                         * A single irq may be mapped to multiple
                         * cpu's vector_irq[] (for example IOAPIC cluster
                         * mode).  In this case we have two
                         * possibilities:
                         *
                         * 1) the resulting affinity mask is empty; that is
                         * this the down'd cpu is the last cpu in the irq's
                         * affinity mask, or
                         *
                         * 2) the resulting affinity mask is no longer
                         * a subset of the online cpus but the affinity
                         * mask is not zero; that is the down'd cpu is the
                         * last online cpu in a user set affinity mask.
                         */
                        if (cpumask_empty(&affinity_new) ||
                            !cpumask_subset(&affinity_new, &online_new))
                                this_count++;
                }
        }


Everyone okay with that?

P.