Re: [PATCH 2/2] x86/mm: replace a goto by merging two if clause

[Wei Yang]

On Mon, Sep 23, 2019 at 11:22:31AM +0200, Peter Zijlstra wrote:
>On Thu, Sep 19, 2019 at 10:08:44AM +0800, Wei Yang wrote:
>> There is only one place to use good_area jump, which could be reduced by
>> merging the following two if clause.
>> 
>> Signed-off-by: Wei Yang 
>> ---
>>  arch/x86/mm/fault.c | 11 +--
>>  1 file changed, 5 insertions(+), 6 deletions(-)
>> 
>> diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
>> index 9d18b73b5f77..72ce6c69e195 100644
>> --- a/arch/x86/mm/fault.c
>> +++ b/arch/x86/mm/fault.c
>> @@ -1390,18 +1390,17 @@ void do_user_addr_fault(struct pt_regs *regs,
>>  vma = find_vma(mm, address);
>>  if (unlikely(!vma))
>>  goto bad_area;
>> -if (likely(vma->vm_start <= address))
>> -goto good_area;
>> -if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)))
>> -goto bad_area;
>> -if (unlikely(expand_stack(vma, address)))
>> +if (likely(vma->vm_start <= address)) {
>> +/* good area, do nothing */
>> +} else if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)) ||
>> +   unlikely(expand_stack(vma, address))) {
>>  goto bad_area;
>> +}
>>  
>>  /*
>>   * Ok, we have a good vm_area for this memory access, so
>>   * we can handle it..
>>   */
>> -good_area:
>>  if (unlikely(access_error(hw_error_code, vma))) {
>>  bad_area_access_error(regs, hw_error_code, address, vma);
>>  return;
>
>I find the old code far easier to read... is there any actual reason to
>do this?

Hi, Peter,

Do you have some comment for the Patch 1?

-- 
Wei Yang
Help you, Help me

Re: [PATCH 2/2] x86/mm: replace a goto by merging two if clause

[Wei Yang]

On Mon, Sep 23, 2019 at 11:22:31AM +0200, Peter Zijlstra wrote:
>On Thu, Sep 19, 2019 at 10:08:44AM +0800, Wei Yang wrote:
>> There is only one place to use good_area jump, which could be reduced by
>> merging the following two if clause.
>> 
>> Signed-off-by: Wei Yang 
>> ---
>>  arch/x86/mm/fault.c | 11 +--
>>  1 file changed, 5 insertions(+), 6 deletions(-)
>> 
>> diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
>> index 9d18b73b5f77..72ce6c69e195 100644
>> --- a/arch/x86/mm/fault.c
>> +++ b/arch/x86/mm/fault.c
>> @@ -1390,18 +1390,17 @@ void do_user_addr_fault(struct pt_regs *regs,
>>  vma = find_vma(mm, address);
>>  if (unlikely(!vma))
>>  goto bad_area;
>> -if (likely(vma->vm_start <= address))
>> -goto good_area;
>> -if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)))
>> -goto bad_area;
>> -if (unlikely(expand_stack(vma, address)))
>> +if (likely(vma->vm_start <= address)) {
>> +/* good area, do nothing */
>> +} else if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)) ||
>> +   unlikely(expand_stack(vma, address))) {
>>  goto bad_area;
>> +}
>>  
>>  /*
>>   * Ok, we have a good vm_area for this memory access, so
>>   * we can handle it..
>>   */
>> -good_area:
>>  if (unlikely(access_error(hw_error_code, vma))) {
>>  bad_area_access_error(regs, hw_error_code, address, vma);
>>  return;
>
>I find the old code far easier to read... is there any actual reason to
>do this?

No, just want to make it easy to read.

-- 
Wei Yang
Help you, Help me

Re: [PATCH 2/2] x86/mm: replace a goto by merging two if clause

[Peter Zijlstra]

On Thu, Sep 19, 2019 at 10:08:44AM +0800, Wei Yang wrote:
> There is only one place to use good_area jump, which could be reduced by
> merging the following two if clause.
> 
> Signed-off-by: Wei Yang 
> ---
>  arch/x86/mm/fault.c | 11 +--
>  1 file changed, 5 insertions(+), 6 deletions(-)
> 
> diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
> index 9d18b73b5f77..72ce6c69e195 100644
> --- a/arch/x86/mm/fault.c
> +++ b/arch/x86/mm/fault.c
> @@ -1390,18 +1390,17 @@ void do_user_addr_fault(struct pt_regs *regs,
>   vma = find_vma(mm, address);
>   if (unlikely(!vma))
>   goto bad_area;
> - if (likely(vma->vm_start <= address))
> - goto good_area;
> - if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)))
> - goto bad_area;
> - if (unlikely(expand_stack(vma, address)))
> + if (likely(vma->vm_start <= address)) {
> + /* good area, do nothing */
> + } else if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)) ||
> +unlikely(expand_stack(vma, address))) {
>   goto bad_area;
> + }
>  
>   /*
>* Ok, we have a good vm_area for this memory access, so
>* we can handle it..
>*/
> -good_area:
>   if (unlikely(access_error(hw_error_code, vma))) {
>   bad_area_access_error(regs, hw_error_code, address, vma);
>   return;

I find the old code far easier to read... is there any actual reason to
do this?

[PATCH 2/2] x86/mm: replace a goto by merging two if clause

[Wei Yang]

There is only one place to use good_area jump, which could be reduced by
merging the following two if clause.

Signed-off-by: Wei Yang 
---
 arch/x86/mm/fault.c | 11 +--
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 9d18b73b5f77..72ce6c69e195 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -1390,18 +1390,17 @@ void do_user_addr_fault(struct pt_regs *regs,
vma = find_vma(mm, address);
if (unlikely(!vma))
goto bad_area;
-   if (likely(vma->vm_start <= address))
-   goto good_area;
-   if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)))
-   goto bad_area;
-   if (unlikely(expand_stack(vma, address)))
+   if (likely(vma->vm_start <= address)) {
+   /* good area, do nothing */
+   } else if (unlikely(!(vma->vm_flags & VM_GROWSDOWN)) ||
+  unlikely(expand_stack(vma, address))) {
goto bad_area;
+   }
 
/*
 * Ok, we have a good vm_area for this memory access, so
 * we can handle it..
 */
-good_area:
if (unlikely(access_error(hw_error_code, vma))) {
bad_area_access_error(regs, hw_error_code, address, vma);
return;
-- 
2.17.1