Skip to contentSkip to conversation

Fix usage of irq affinity_hint

PATCH In review Haiyang Zhang 3 participants 11 messages Jan 26, 2023 – Jan 29, 2023

Subsystems (5)
lore
Share & tools
lore

Haiyang Zhang

author maintainer
#0 lore raw

cover Fix usage of irq affinity_hint

Fix 2 bugs of irq affinity_hint usage in the mana driver

Haiyang Zhang (2):
net: mana: Fix hint value before free irq
net: mana: Fix accessing freed irq affinity_hint

 .../net/ethernet/microsoft/mana/gdma_main.c   | 45 +++++++++++--------
 include/net/mana/gdma.h                       |  1 +
 2 files changed, 28 insertions(+), 18 deletions(-)
Signature · 1 line
2.25.1
author maintainer
#1 lore raw

PATCH 1/2 net: mana: Fix hint value before free irq

Need to clear affinity_hint before free_irq(), otherwise there is a one-time warning and stack trace during module unloading.

To fix this bug, set affinity_hint to NULL before free as required.

Fixes 71fa6887eeca ("net: mana: Assign interrupts to CPUs based on NUMA nodes")
---
 drivers/net/ethernet/microsoft/mana/gdma_main.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/ethernet/microsoft/mana/gdma_main.c +5 -0
index b144f2237748..3bae9d4c1f08 100644
@@ -1297,6 +1297,8 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 	for (j = i - 1; j >= 0; j--) {
 		irq = pci_irq_vector(pdev, j);
 		gic = &gc->irq_contexts[j];
+
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
@@ -1324,6 +1326,9 @@ static void mana_gd_remove_irqs(struct pci_dev *pdev)
 			continue;
 
 		gic = &gc->irq_contexts[i];
+
+		/* Need to clear the hint before free_irq */
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
Signature · 1 line
2.25.1
#2 lore raw

Re: [PATCH net, 1/2] net: mana: Fix hint value before free irq

Quoted 2 lines · Haiyang Zhang

Need to clear affinity_hint before free_irq(), otherwise there is a one-time warning and stack trace during module unloading.

Please add this warning and stack trace to the commit message.

Thanks

Quoted 36 lines

To fix this bug, set affinity_hint to NULL before free as required.

Fixes 71fa6887eeca ("net: mana: Assign interrupts to CPUs based on NUMA nodes")
---
 drivers/net/ethernet/microsoft/mana/gdma_main.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/ethernet/microsoft/mana/gdma_main.c +5 -0
index b144f2237748..3bae9d4c1f08 100644
@@ -1297,6 +1297,8 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 	for (j = i - 1; j >= 0; j--) {
 		irq = pci_irq_vector(pdev, j);
 		gic = &gc->irq_contexts[j];
+
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
@@ -1324,6 +1326,9 @@ static void mana_gd_remove_irqs(struct pci_dev *pdev)
 			continue;
 
 		gic = &gc->irq_contexts[i];
+
+		/* Need to clear the hint before free_irq */
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
Signature · 1 line
2.25.1
author maintainer
#3 lore raw

RE: [PATCH net, 1/2] net: mana: Fix hint value before free irq

Quoted 12 lines

-----Original Message-----
From: Leon Romanovsky <[email protected]>
Sent: Sunday, January 29, 2023 4:28 AM
To: Haiyang Zhang <[email protected]>
Cc: [email protected]; [email protected]; Dexuan Cui <[email protected]>; KY Srinivasan <[email protected]>; Paul Rosswurm <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [PATCH net, 1/2] net: mana: Fix hint value before free irq

Quoted 5 lines
Quoted 2 lines · Haiyang Zhang

Need to clear affinity_hint before free_irq(), otherwise there is a one-time warning and stack trace during module unloading.

Please add this warning and stack trace to the commit message.

Will do.

Thanks,
- Haiyang

#4 lore raw

RE: [PATCH net, 1/2] net: mana: Fix hint value before free irq

In reply to Haiyang Zhang

From: LKML haiyangz <[email protected]> On Behalf Of Haiyang Zhang Sent: Thursday, January 26, 2023 1:05 PM

Quoted 38 lines

Need to clear affinity_hint before free_irq(), otherwise there is a one-time warning and stack trace during module unloading.

To fix this bug, set affinity_hint to NULL before free as required.

Fixes 71fa6887eeca ("net: mana: Assign interrupts to CPUs based on NUMA nodes")
---
 drivers/net/ethernet/microsoft/mana/gdma_main.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/ethernet/microsoft/mana/gdma_main.c +5 -1
index b144f2237748..3bae9d4c1f08 100644
@@ -1297,6 +1297,8 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 	for (j = i - 1; j >= 0; j--) {
 		irq = pci_irq_vector(pdev, j);
 		gic = &gc->irq_contexts[j];
+
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
@@ -1324,6 +1326,9 @@ static void mana_gd_remove_irqs(struct pci_dev *pdev)
 			continue;
 
 		gic = &gc->irq_contexts[i];
+
+		/* Need to clear the hint before free_irq */
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
--
Quoted 1 line

2.25.1

I think this patch should be folded into the second patch of this series. While this patch makes the warning go away, it doesn't really solve any problems by itself. It just papers over the warning. My first reaction on seeing this patch is that the warning exists because the memory for the mask likely had been incorrectly managed, which is exactly the case. Since this patch doesn't really fix a problem on its own, I'd say merge it into the second patch.

That's just my $.02. If you really want to keep it separate, that's not a blocker for me.

Michael

author maintainer
#5 lore raw

RE: [PATCH net, 1/2] net: mana: Fix hint value before free irq

Quoted 65 lines

-----Original Message-----
From: Michael Kelley (LINUX) <[email protected]>
Sent: Sunday, January 29, 2023 9:26 AM
To: Haiyang Zhang <[email protected]>; [email protected]; [email protected]
Cc: Haiyang Zhang <[email protected]>; Dexuan Cui <[email protected]>; KY Srinivasan <[email protected]>; Paul Rosswurm <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: RE: [PATCH net, 1/2] net: mana: Fix hint value before free irq

From: LKML haiyangz <[email protected]> On Behalf Of Haiyang Zhang
Sent: Thursday, January 26, 2023 1:05 PM

Quoted 8 lines

Need to clear affinity_hint before free_irq(), otherwise there is a one-time warning and stack trace during module unloading.

To fix this bug, set affinity_hint to NULL before free as required.

Fixes 71fa6887eeca ("net: mana: Assign interrupts to CPUs based on NUMA

nodes")

drivers/net/ethernet/microsoft/mana/gdma_main.c static int mana_gd_setup_irqs(struct pci_dev

Quoted 11 lines
---
 drivers/net/ethernet/microsoft/mana/gdma_main.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/ethernet/microsoft/mana/gdma_main.c +0 -0
index b144f2237748..3bae9d4c1f08 100644
@@ -1297,6 +1297,8 @@ static int mana_gd_setup_irqs(struct pci_dev

*pdev)

drivers/net/ethernet/microsoft/mana/gdma_main.c static void mana_gd_remove_irqs(struct pci_dev

Quoted 9 lines
drivers/net/ethernet/microsoft/mana/gdma_main.c +2 -0
 	for (j = i - 1; j >= 0; j--) {
 		irq = pci_irq_vector(pdev, j);
 		gic = &gc->irq_contexts[j];
+
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
@@ -1324,6 +1326,9 @@ static void mana_gd_remove_irqs(struct pci_dev

*pdev)

Quoted 11 lines
drivers/net/ethernet/microsoft/mana/gdma_main.c +3 -1
 			continue;
 
 		gic = &gc->irq_contexts[i];
+
+		/* Need to clear the hint before free_irq */
+		irq_update_affinity_hint(irq, NULL);
 		free_irq(irq, gic);
 	}
 
--

2.25.1

I think this patch should be folded into the second patch of this series. While this patch makes the warning go away, it doesn't really solve any problems by itself. It just papers over the warning. My first reaction on seeing this patch is that the warning exists because the memory for the mask likely had been incorrectly managed, which is exactly the case. Since this patch doesn't really fix a problem on its own, I'd say merge it into the second patch.

Quoted 2 lines

That's just my $.02. If you really want to keep it separate, that's not a blocker for me.

Will do.

Thanks,
- Haiyang

author maintainer
#6 lore raw

PATCH 2/2 net: mana: Fix accessing freed irq affinity_hint

In reply to Haiyang Zhang

After calling irq_set_affinity_and_hint(), the cpumask pointer is saved in desc->affinity_hint, and will be used later when reading /proc/irq/<num>/affinity_hint. So the cpumask variable needs to be allocated per irq, and available until freeing the irq. Otherwise, we are accessing freed memory when reading the affinity_hint file.

To fix the bug, allocate the cpumask per irq, and free it just before freeing the irq.

Fixes 71fa6887eeca ("net: mana: Assign interrupts to CPUs based on NUMA nodes")
---
 .../net/ethernet/microsoft/mana/gdma_main.c   | 40 ++++++++++---------
 include/net/mana/gdma.h                       |  1 +
 2 files changed, 23 insertions(+), 18 deletions(-)
drivers/net/ethernet/microsoft/mana/gdma_main.c +22 -18
index 3bae9d4c1f08..37473ae3859c 100644
@@ -1219,7 +1219,6 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 	struct gdma_irq_context *gic;
 	unsigned int max_irqs;
 	u16 *cpus;
-	cpumask_var_t req_mask;
 	int nvec, irq;
 	int err, i = 0, j;
 
@@ -1240,25 +1239,26 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 		goto free_irq_vector;
 	}
 
-	if (!zalloc_cpumask_var(&req_mask, GFP_KERNEL)) {
-		err = -ENOMEM;
-		goto free_irq;
-	}
-
 	cpus = kcalloc(nvec, sizeof(*cpus), GFP_KERNEL);
 	if (!cpus) {
 		err = -ENOMEM;
-		goto free_mask;
+		goto free_gic;
 	}
 	for (i = 0; i < nvec; i++)
 		cpus[i] = cpumask_local_spread(i, gc->numa_node);
 
 	for (i = 0; i < nvec; i++) {
-		cpumask_set_cpu(cpus[i], req_mask);
 		gic = &gc->irq_contexts[i];
 		gic->handler = NULL;
 		gic->arg = NULL;
 
+		if (!zalloc_cpumask_var(&gic->cpu_hint, GFP_KERNEL)) {
+			err = -ENOMEM;
+			goto free_irq;
+		}
+
+		cpumask_set_cpu(cpus[i], gic->cpu_hint);
+
 		if (!i)
 			snprintf(gic->name, MANA_IRQ_NAME_SZ, "mana_hwc@pci:%s",
 				 pci_name(pdev));
@@ -1269,17 +1269,18 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 		irq = pci_irq_vector(pdev, i);
 		if (irq < 0) {
 			err = irq;
-			goto free_mask;
+			free_cpumask_var(gic->cpu_hint);
+			goto free_irq;
 		}
 
 		err = request_irq(irq, mana_gd_intr, 0, gic->name, gic);
-		if (err)
-			goto free_mask;
-		irq_set_affinity_and_hint(irq, req_mask);
-		cpumask_clear(req_mask);
+		if (err) {
+			free_cpumask_var(gic->cpu_hint);
+			goto free_irq;
+		}
+
+		irq_set_affinity_and_hint(irq, gic->cpu_hint);
 	}
-	free_cpumask_var(req_mask);
-	kfree(cpus);
 
 	err = mana_gd_alloc_res_map(nvec, &gc->msix_resource);
 	if (err)
@@ -1288,20 +1289,22 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 	gc->max_num_msix = nvec;
 	gc->num_msix_usable = nvec;
 
+	kfree(cpus);
 	return 0;
 
-free_mask:
-	free_cpumask_var(req_mask);
-	kfree(cpus);
 free_irq:
 	for (j = i - 1; j >= 0; j--) {
 		irq = pci_irq_vector(pdev, j);
 		gic = &gc->irq_contexts[j];
 
 		irq_update_affinity_hint(irq, NULL);
+		free_cpumask_var(gic->cpu_hint);
 		free_irq(irq, gic);
 	}
 
+	kfree(cpus);
+
+free_gic:
 	kfree(gc->irq_contexts);
 	gc->irq_contexts = NULL;
 free_irq_vector:
@@ -1329,6 +1332,7 @@ static void mana_gd_remove_irqs(struct pci_dev *pdev)
 
 		/* Need to clear the hint before free_irq */
 		irq_update_affinity_hint(irq, NULL);
+		free_cpumask_var(gic->cpu_hint);
 		free_irq(irq, gic);
 	}
 
include/net/mana/gdma.h +1 -0
index 56189e4252da..4dcafecbd89e 100644
@@ -342,6 +342,7 @@ struct gdma_irq_context {
 	void (*handler)(void *arg);
 	void *arg;
 	char name[MANA_IRQ_NAME_SZ];
+	cpumask_var_t cpu_hint;
 };
 
 struct gdma_context {
Signature · 1 line
2.25.1
Reviewed-by on v1 2/2
#7 lore raw

Re: [PATCH net, 2/2] net: mana: Fix accessing freed irq affinity_hint

Quoted 9 lines · Haiyang Zhang

After calling irq_set_affinity_and_hint(), the cpumask pointer is saved in desc->affinity_hint, and will be used later when reading /proc/irq/<num>/affinity_hint. So the cpumask variable needs to be allocated per irq, and available until freeing the irq. Otherwise, we are accessing freed memory when reading the affinity_hint file.

To fix the bug, allocate the cpumask per irq, and free it just before freeing the irq.

Quoted 8 lines
Fixes 71fa6887eeca ("net: mana: Assign interrupts to CPUs based on NUMA nodes")
---
 .../net/ethernet/microsoft/mana/gdma_main.c   | 40 ++++++++++---------
 include/net/mana/gdma.h                       |  1 +
 2 files changed, 23 insertions(+), 18 deletions(-)

Thanks,

#8 lore raw

RE: [PATCH net, 2/2] net: mana: Fix accessing freed irq affinity_hint

In reply to Haiyang Zhang

From: LKML haiyangz <[email protected]> On Behalf Of Haiyang Zhang Sent: Thursday, January 26, 2023 1:05 PM

Quoted 9 lines

After calling irq_set_affinity_and_hint(), the cpumask pointer is saved in desc->affinity_hint, and will be used later when reading /proc/irq/<num>/affinity_hint. So the cpumask variable needs to be allocated per irq, and available until freeing the irq. Otherwise, we are accessing freed memory when reading the affinity_hint file.

To fix the bug, allocate the cpumask per irq, and free it just before freeing the irq.

Since the cpumask being passed to irq_set_affinity_and_hint() always contains exactly one CPU, the code can be considerably simplified by using the pre-calculated and persistent masks available as cpumask_of(cpu). All allocation of cpumasks in this code goes away, and you can set the affinity_hint to NULL in the cleanup and remove paths without having to free any masks.

Michael

Quoted 131 lines
Fixes 71fa6887eeca ("net: mana: Assign interrupts to CPUs based on NUMA nodes")
---
 .../net/ethernet/microsoft/mana/gdma_main.c   | 40 ++++++++++---------
 include/net/mana/gdma.h                       |  1 +
 2 files changed, 23 insertions(+), 18 deletions(-)
drivers/net/ethernet/microsoft/mana/gdma_main.c +8 -8
index 3bae9d4c1f08..37473ae3859c 100644
@@ -1219,7 +1219,6 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 	struct gdma_irq_context *gic;
 	unsigned int max_irqs;
 	u16 *cpus;
-	cpumask_var_t req_mask;
 	int nvec, irq;
 	int err, i = 0, j;
 
@@ -1240,25 +1239,26 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 		goto free_irq_vector;
 	}
 
-	if (!zalloc_cpumask_var(&req_mask, GFP_KERNEL)) {
-		err = -ENOMEM;
-		goto free_irq;
-	}
-
 	cpus = kcalloc(nvec, sizeof(*cpus), GFP_KERNEL);
 	if (!cpus) {
 		err = -ENOMEM;
-		goto free_mask;
+		goto free_gic;
 	}
 	for (i = 0; i < nvec; i++)
 		cpus[i] = cpumask_local_spread(i, gc->numa_node);
 
 	for (i = 0; i < nvec; i++) {
-		cpumask_set_cpu(cpus[i], req_mask);
 		gic = &gc->irq_contexts[i];
 		gic->handler = NULL;
 		gic->arg = NULL;
 
+		if (!zalloc_cpumask_var(&gic->cpu_hint, GFP_KERNEL)) {
+			err = -ENOMEM;
+			goto free_irq;
+		}
+
+		cpumask_set_cpu(cpus[i], gic->cpu_hint);
+
 		if (!i)
 			snprintf(gic->name, MANA_IRQ_NAME_SZ,
 "mana_hwc@pci:%s",
drivers/net/ethernet/microsoft/mana/gdma_main.c +14 -10
 				 pci_name(pdev));
@@ -1269,17 +1269,18 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 		irq = pci_irq_vector(pdev, i);
 		if (irq < 0) {
 			err = irq;
-			goto free_mask;
+			free_cpumask_var(gic->cpu_hint);
+			goto free_irq;
 		}
 
 		err = request_irq(irq, mana_gd_intr, 0, gic->name, gic);
-		if (err)
-			goto free_mask;
-		irq_set_affinity_and_hint(irq, req_mask);
-		cpumask_clear(req_mask);
+		if (err) {
+			free_cpumask_var(gic->cpu_hint);
+			goto free_irq;
+		}
+
+		irq_set_affinity_and_hint(irq, gic->cpu_hint);
 	}
-	free_cpumask_var(req_mask);
-	kfree(cpus);
 
 	err = mana_gd_alloc_res_map(nvec, &gc->msix_resource);
 	if (err)
@@ -1288,20 +1289,22 @@ static int mana_gd_setup_irqs(struct pci_dev *pdev)
 	gc->max_num_msix = nvec;
 	gc->num_msix_usable = nvec;
 
+	kfree(cpus);
 	return 0;
 
-free_mask:
-	free_cpumask_var(req_mask);
-	kfree(cpus);
 free_irq:
 	for (j = i - 1; j >= 0; j--) {
 		irq = pci_irq_vector(pdev, j);
 		gic = &gc->irq_contexts[j];
 
 		irq_update_affinity_hint(irq, NULL);
+		free_cpumask_var(gic->cpu_hint);
 		free_irq(irq, gic);
 	}
 
+	kfree(cpus);
+
+free_gic:
 	kfree(gc->irq_contexts);
 	gc->irq_contexts = NULL;
 free_irq_vector:
@@ -1329,6 +1332,7 @@ static void mana_gd_remove_irqs(struct pci_dev *pdev)
 
 		/* Need to clear the hint before free_irq */
 		irq_update_affinity_hint(irq, NULL);
+		free_cpumask_var(gic->cpu_hint);
 		free_irq(irq, gic);
 	}
 
include/net/mana/gdma.h +1 -1
index 56189e4252da..4dcafecbd89e 100644
@@ -342,6 +342,7 @@ struct gdma_irq_context {
 	void (*handler)(void *arg);
 	void *arg;
 	char name[MANA_IRQ_NAME_SZ];
+	cpumask_var_t cpu_hint;
 };
 
 struct gdma_context {
--

2.25.1

author maintainer
#9 lore raw

RE: [PATCH net, 2/2] net: mana: Fix accessing freed irq affinity_hint

Quoted 25 lines

-----Original Message-----
From: Michael Kelley (LINUX) <[email protected]>
Sent: Sunday, January 29, 2023 9:27 AM
To: Haiyang Zhang <[email protected]>; [email protected]; [email protected]
Cc: Haiyang Zhang <[email protected]>; Dexuan Cui <[email protected]>; KY Srinivasan <[email protected]>; Paul Rosswurm <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: RE: [PATCH net, 2/2] net: mana: Fix accessing freed irq affinity_hint

From: LKML haiyangz <[email protected]> On Behalf Of Haiyang Zhang
Sent: Thursday, January 26, 2023 1:05 PM

Quoted 9 lines

After calling irq_set_affinity_and_hint(), the cpumask pointer is saved in desc->affinity_hint, and will be used later when reading /proc/irq/<num>/affinity_hint. So the cpumask variable needs to be allocated per irq, and available until freeing the irq. Otherwise, we are accessing freed memory when reading the affinity_hint file.

To fix the bug, allocate the cpumask per irq, and free it just before freeing the irq.

Quoted 7 lines

Since the cpumask being passed to irq_set_affinity_and_hint() always contains exactly one CPU, the code can be considerably simplified by using the pre-calculated and persistent masks available as cpumask_of(cpu). All allocation of cpumasks in this code goes away, and you can set the affinity_hint to NULL in the cleanup and remove paths without having to free any masks.

Great idea!
Will update the patch accordingly.

- Haiyang

author maintainer
#10 lore raw

RE: [PATCH net, 2/2] net: mana: Fix accessing freed irq affinity_hint

Quoted 49 lines

-----Original Message-----
From: Haiyang Zhang <[email protected]>
Sent: Sunday, January 29, 2023 1:51 PM
To: Michael Kelley (LINUX) <[email protected]>; linux- [email protected]; [email protected]
Cc: Dexuan Cui <[email protected]>; KY Srinivasan <[email protected]>; Paul Rosswurm <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: RE: [PATCH net, 2/2] net: mana: Fix accessing freed irq affinity_hint

Quoted 4 lines

-----Original Message-----
From: Michael Kelley (LINUX) <[email protected]>
Sent: Sunday, January 29, 2023 9:27 AM
To: Haiyang Zhang <[email protected]>; linux-

[email protected];

Quoted 3 lines

[email protected]
Cc: Haiyang Zhang <[email protected]>; Dexuan Cui <[email protected]>; KY Srinivasan <[email protected]>; Paul

Rosswurm

Quoted 2 lines

<[email protected]>; [email protected]; [email protected]; [email protected]; [email protected];

[email protected]

Quoted 3 lines

Subject: RE: [PATCH net, 2/2] net: mana: Fix accessing freed irq affinity_hint

From: LKML haiyangz <[email protected]> On Behalf Of Haiyang

Zhang

Quoted 11 lines

Sent: Thursday, January 26, 2023 1:05 PM

Quoted 9 lines

After calling irq_set_affinity_and_hint(), the cpumask pointer is saved in desc->affinity_hint, and will be used later when reading /proc/irq/<num>/affinity_hint. So the cpumask variable needs to be allocated per irq, and available until freeing the irq. Otherwise, we are accessing freed memory when reading the affinity_hint file.

To fix the bug, allocate the cpumask per irq, and free it just before freeing the irq.

Quoted 7 lines

Since the cpumask being passed to irq_set_affinity_and_hint() always contains exactly one CPU, the code can be considerably simplified by using the pre-calculated and persistent masks available as cpumask_of(cpu). All allocation of cpumasks in this code goes away, and you can set the affinity_hint to NULL in the cleanup and remove paths without having to free any masks.

Quoted 2 lines

Great idea!
Will update the patch accordingly.

Also, I saw this alloc isn't necessary either:

	cpus = kcalloc(nvec, sizeof(*cpus), GFP_KERNEL);

We can simply use the return from cpumask_local_spread() without saving all cpu numbers in a tmp array.

I will clean this up too :)

Thanks,
- Haiyang