1. Patchew
  2. linux
  3. [PATCH v4 0/5] Remove onstack cpumask var for sparc
  4. View patch

[PATCH v4 1/5] sparc/srmmu: Remove on-stack cpumask var

Dawei Li posted 5 patches 1 year, 9 months ago
[PATCH v4 1/5] sparc/srmmu: Remove on-stack cpumask var
Posted by Dawei Li 1 year, 9 months ago 
In general it's preferable to avoid placing cpumasks on the stack, as
for large values of NR_CPUS these can consume significant amounts of
stack space and make stack overflows more likely.

Use cpumask_any_but() to avoid the need for a temporary cpumask on
the stack and simplify code.

Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Dawei Li <dawei.li@shingroup.cn>
---
 arch/sparc/mm/srmmu.c | 40 ++++++++++++----------------------------
 1 file changed, 12 insertions(+), 28 deletions(-)

diff --git a/arch/sparc/mm/srmmu.c b/arch/sparc/mm/srmmu.c
index 852085ada368..9df51a62333d 100644
--- a/arch/sparc/mm/srmmu.c
+++ b/arch/sparc/mm/srmmu.c
@@ -1653,13 +1653,15 @@ static void smp_flush_tlb_all(void)
 	local_ops->tlb_all();
 }
 
+static bool any_other_mm_cpus(struct mm_struct *mm)
+{
+	return cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids;
+}
+
 static void smp_flush_cache_mm(struct mm_struct *mm)
 {
 	if (mm->context != NO_CONTEXT) {
-		cpumask_t cpu_mask;
-		cpumask_copy(&cpu_mask, mm_cpumask(mm));
-		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
-		if (!cpumask_empty(&cpu_mask))
+		if (any_other_mm_cpus(mm))
 			xc1(local_ops->cache_mm, (unsigned long)mm);
 		local_ops->cache_mm(mm);
 	}
@@ -1668,10 +1670,7 @@ static void smp_flush_cache_mm(struct mm_struct *mm)
 static void smp_flush_tlb_mm(struct mm_struct *mm)
 {
 	if (mm->context != NO_CONTEXT) {
-		cpumask_t cpu_mask;
-		cpumask_copy(&cpu_mask, mm_cpumask(mm));
-		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
-		if (!cpumask_empty(&cpu_mask)) {
+		if (any_other_mm_cpus(mm)) {
 			xc1(local_ops->tlb_mm, (unsigned long)mm);
 			if (atomic_read(&mm->mm_users) == 1 && current->active_mm == mm)
 				cpumask_copy(mm_cpumask(mm),
@@ -1688,10 +1687,7 @@ static void smp_flush_cache_range(struct vm_area_struct *vma,
 	struct mm_struct *mm = vma->vm_mm;
 
 	if (mm->context != NO_CONTEXT) {
-		cpumask_t cpu_mask;
-		cpumask_copy(&cpu_mask, mm_cpumask(mm));
-		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
-		if (!cpumask_empty(&cpu_mask))
+		if (any_other_mm_cpus(mm))
 			xc3(local_ops->cache_range, (unsigned long)vma, start,
 			    end);
 		local_ops->cache_range(vma, start, end);
@@ -1705,10 +1701,7 @@ static void smp_flush_tlb_range(struct vm_area_struct *vma,
 	struct mm_struct *mm = vma->vm_mm;
 
 	if (mm->context != NO_CONTEXT) {
-		cpumask_t cpu_mask;
-		cpumask_copy(&cpu_mask, mm_cpumask(mm));
-		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
-		if (!cpumask_empty(&cpu_mask))
+		if (any_other_mm_cpus(mm))
 			xc3(local_ops->tlb_range, (unsigned long)vma, start,
 			    end);
 		local_ops->tlb_range(vma, start, end);
@@ -1720,10 +1713,7 @@ static void smp_flush_cache_page(struct vm_area_struct *vma, unsigned long page)
 	struct mm_struct *mm = vma->vm_mm;
 
 	if (mm->context != NO_CONTEXT) {
-		cpumask_t cpu_mask;
-		cpumask_copy(&cpu_mask, mm_cpumask(mm));
-		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
-		if (!cpumask_empty(&cpu_mask))
+		if (any_other_mm_cpus(mm))
 			xc2(local_ops->cache_page, (unsigned long)vma, page);
 		local_ops->cache_page(vma, page);
 	}
@@ -1734,10 +1724,7 @@ static void smp_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
 	struct mm_struct *mm = vma->vm_mm;
 
 	if (mm->context != NO_CONTEXT) {
-		cpumask_t cpu_mask;
-		cpumask_copy(&cpu_mask, mm_cpumask(mm));
-		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
-		if (!cpumask_empty(&cpu_mask))
+		if (any_other_mm_cpus(mm))
 			xc2(local_ops->tlb_page, (unsigned long)vma, page);
 		local_ops->tlb_page(vma, page);
 	}
@@ -1759,10 +1746,7 @@ static void smp_flush_page_to_ram(unsigned long page)
 
 static void smp_flush_sig_insns(struct mm_struct *mm, unsigned long insn_addr)
 {
-	cpumask_t cpu_mask;
-	cpumask_copy(&cpu_mask, mm_cpumask(mm));
-	cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
-	if (!cpumask_empty(&cpu_mask))
+	if (any_other_mm_cpus(mm))
 		xc2(local_ops->sig_insns, (unsigned long)mm, insn_addr);
 	local_ops->sig_insns(mm, insn_addr);
 }
-- 
2.27.0
Re: [PATCH v4 1/5] sparc/srmmu: Remove on-stack cpumask var
Posted by Andreas Larsson 1 year, 9 months ago 
On 2024-04-24 04:55, Dawei Li wrote:
> In general it's preferable to avoid placing cpumasks on the stack, as
> for large values of NR_CPUS these can consume significant amounts of
> stack space and make stack overflows more likely.
> 
> Use cpumask_any_but() to avoid the need for a temporary cpumask on
> the stack and simplify code.
> 
> Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
> Signed-off-by: Dawei Li <dawei.li@shingroup.cn>
> ---
>  arch/sparc/mm/srmmu.c | 40 ++++++++++++----------------------------
>  1 file changed, 12 insertions(+), 28 deletions(-)
> 
> diff --git a/arch/sparc/mm/srmmu.c b/arch/sparc/mm/srmmu.c
> index 852085ada368..9df51a62333d 100644
> --- a/arch/sparc/mm/srmmu.c
> +++ b/arch/sparc/mm/srmmu.c
> @@ -1653,13 +1653,15 @@ static void smp_flush_tlb_all(void)
>  	local_ops->tlb_all();
>  }
>  
> +static bool any_other_mm_cpus(struct mm_struct *mm)
> +{
> +	return cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids;
> +}
> +
>  static void smp_flush_cache_mm(struct mm_struct *mm)
>  {
>  	if (mm->context != NO_CONTEXT) {
> -		cpumask_t cpu_mask;
> -		cpumask_copy(&cpu_mask, mm_cpumask(mm));
> -		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
> -		if (!cpumask_empty(&cpu_mask))
> +		if (any_other_mm_cpus(mm))
>  			xc1(local_ops->cache_mm, (unsigned long)mm);
>  		local_ops->cache_mm(mm);
>  	}
> @@ -1668,10 +1670,7 @@ static void smp_flush_cache_mm(struct mm_struct *mm)
>  static void smp_flush_tlb_mm(struct mm_struct *mm)
>  {
>  	if (mm->context != NO_CONTEXT) {
> -		cpumask_t cpu_mask;
> -		cpumask_copy(&cpu_mask, mm_cpumask(mm));
> -		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
> -		if (!cpumask_empty(&cpu_mask)) {
> +		if (any_other_mm_cpus(mm)) {
>  			xc1(local_ops->tlb_mm, (unsigned long)mm);
>  			if (atomic_read(&mm->mm_users) == 1 && current->active_mm == mm)
>  				cpumask_copy(mm_cpumask(mm),
> @@ -1688,10 +1687,7 @@ static void smp_flush_cache_range(struct vm_area_struct *vma,
>  	struct mm_struct *mm = vma->vm_mm;
>  
>  	if (mm->context != NO_CONTEXT) {
> -		cpumask_t cpu_mask;
> -		cpumask_copy(&cpu_mask, mm_cpumask(mm));
> -		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
> -		if (!cpumask_empty(&cpu_mask))
> +		if (any_other_mm_cpus(mm))
>  			xc3(local_ops->cache_range, (unsigned long)vma, start,
>  			    end);
>  		local_ops->cache_range(vma, start, end);
> @@ -1705,10 +1701,7 @@ static void smp_flush_tlb_range(struct vm_area_struct *vma,
>  	struct mm_struct *mm = vma->vm_mm;
>  
>  	if (mm->context != NO_CONTEXT) {
> -		cpumask_t cpu_mask;
> -		cpumask_copy(&cpu_mask, mm_cpumask(mm));
> -		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
> -		if (!cpumask_empty(&cpu_mask))
> +		if (any_other_mm_cpus(mm))
>  			xc3(local_ops->tlb_range, (unsigned long)vma, start,
>  			    end);
>  		local_ops->tlb_range(vma, start, end);
> @@ -1720,10 +1713,7 @@ static void smp_flush_cache_page(struct vm_area_struct *vma, unsigned long page)
>  	struct mm_struct *mm = vma->vm_mm;
>  
>  	if (mm->context != NO_CONTEXT) {
> -		cpumask_t cpu_mask;
> -		cpumask_copy(&cpu_mask, mm_cpumask(mm));
> -		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
> -		if (!cpumask_empty(&cpu_mask))
> +		if (any_other_mm_cpus(mm))
>  			xc2(local_ops->cache_page, (unsigned long)vma, page);
>  		local_ops->cache_page(vma, page);
>  	}
> @@ -1734,10 +1724,7 @@ static void smp_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
>  	struct mm_struct *mm = vma->vm_mm;
>  
>  	if (mm->context != NO_CONTEXT) {
> -		cpumask_t cpu_mask;
> -		cpumask_copy(&cpu_mask, mm_cpumask(mm));
> -		cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
> -		if (!cpumask_empty(&cpu_mask))
> +		if (any_other_mm_cpus(mm))
>  			xc2(local_ops->tlb_page, (unsigned long)vma, page);
>  		local_ops->tlb_page(vma, page);
>  	}
> @@ -1759,10 +1746,7 @@ static void smp_flush_page_to_ram(unsigned long page)
>  
>  static void smp_flush_sig_insns(struct mm_struct *mm, unsigned long insn_addr)
>  {
> -	cpumask_t cpu_mask;
> -	cpumask_copy(&cpu_mask, mm_cpumask(mm));
> -	cpumask_clear_cpu(smp_processor_id(), &cpu_mask);
> -	if (!cpumask_empty(&cpu_mask))
> +	if (any_other_mm_cpus(mm))
>  		xc2(local_ops->sig_insns, (unsigned long)mm, insn_addr);
>  	local_ops->sig_insns(mm, insn_addr);
>  }

Reviewed-by: Andreas Larsson <andreas@gaisler.com>
Tested-by: Andreas Larsson <andreas@gaisler.com>

Picking this up to my for-next.

Thanks,
Andreas

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.