2013-09-22 18:17:35

by Felipe Pena

Subject: [PATCH 1/1] arch/parisc: mm: fix uninitialized variable usage

The FAULT_FLAG_WRITE flag has been set based on uninitialized variable

Signed-off-by: Felipe Pena <[email protected]>
---
arch/parisc/mm/fault.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/parisc/mm/fault.c b/arch/parisc/mm/fault.c
index d10d27a..6b38026 100644
--- a/arch/parisc/mm/fault.c
+++ b/arch/parisc/mm/fault.c
@@ -182,8 +182,6 @@ void do_page_fault(struct pt_regs *regs, unsigned long code,

if (user_mode(regs))
flags |= FAULT_FLAG_USER;
- if (acc_type & VM_WRITE)
- flags |= FAULT_FLAG_WRITE;
retry:
down_read(&mm->mmap_sem);
vma = find_vma_prev(mm, address, &prev_vma);
@@ -201,6 +199,9 @@ good_area:
if ((vma->vm_flags & acc_type) != acc_type)
goto bad_area;

+ if (acc_type & VM_WRITE)
+ flags |= FAULT_FLAG_WRITE;
+
/*
* If for any reason at all we couldn't handle the fault, make
* sure we exit gracefully rather than endlessly redo the
--
1.7.10.4

2013-09-22 22:58:17

by Johannes Weiner

Subject: Re: [PATCH 1/1] arch/parisc: mm: fix uninitialized variable usage

Hello Felipe,

On Sun, Sep 22, 2013 at 03:17:46PM -0300, Felipe Pena wrote:
> The FAULT_FLAG_WRITE flag has been set based on uninitialized variable

Oops, you are right.

> Signed-off-by: Felipe Pena <[email protected]>
> ---
> arch/parisc/mm/fault.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/arch/parisc/mm/fault.c b/arch/parisc/mm/fault.c
> index d10d27a..6b38026 100644
> --- a/arch/parisc/mm/fault.c
> +++ b/arch/parisc/mm/fault.c
> @@ -182,8 +182,6 @@ void do_page_fault(struct pt_regs *regs, unsigned long code,
>
> if (user_mode(regs))
> flags |= FAULT_FLAG_USER;
> - if (acc_type & VM_WRITE)
> - flags |= FAULT_FLAG_WRITE;
> retry:
> down_read(&mm->mmap_sem);
> vma = find_vma_prev(mm, address, &prev_vma);
> @@ -201,6 +199,9 @@ good_area:
> if ((vma->vm_flags & acc_type) != acc_type)
> goto bad_area;
>
> + if (acc_type & VM_WRITE)
> + flags |= FAULT_FLAG_WRITE;

Can acc_type actually change between between the first round and a
retry? Otherwise, it might make sense to pull this up and place it
next to the flag initialization instead of pulling one flag down.

2013-09-23 00:23:50

by Felipe Pena

Subject: Re: [PATCH 1/1] arch/parisc: mm: fix uninitialized variable usage

Hello Johannes,

On Sun, Sep 22, 2013 at 7:58 PM, Johannes Weiner <[email protected]> wrote:
> Hello Felipe,
>
> On Sun, Sep 22, 2013 at 03:17:46PM -0300, Felipe Pena wrote:
>> The FAULT_FLAG_WRITE flag has been set based on uninitialized variable
>
> Oops, you are right.
>
>> Signed-off-by: Felipe Pena <[email protected]>
>> ---
>> arch/parisc/mm/fault.c | 5 +++--
>> 1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/parisc/mm/fault.c b/arch/parisc/mm/fault.c
>> index d10d27a..6b38026 100644
>> --- a/arch/parisc/mm/fault.c
>> +++ b/arch/parisc/mm/fault.c
>> @@ -182,8 +182,6 @@ void do_page_fault(struct pt_regs *regs, unsigned long code,
>>
>> if (user_mode(regs))
>> flags |= FAULT_FLAG_USER;
>> - if (acc_type & VM_WRITE)
>> - flags |= FAULT_FLAG_WRITE;
>> retry:
>> down_read(&mm->mmap_sem);
>> vma = find_vma_prev(mm, address, &prev_vma);
>> @@ -201,6 +199,9 @@ good_area:
>> if ((vma->vm_flags & acc_type) != acc_type)
>> goto bad_area;
>>
>> + if (acc_type & VM_WRITE)
>> + flags |= FAULT_FLAG_WRITE;
>
> Can acc_type actually change between between the first round and a
> retry? Otherwise, it might make sense to pull this up and place it
> next to the flag initialization instead of pulling one flag down.

>From what I've analyzed, this make sense. I'll make the suggested
changes and send another patch.

Thanks.

--
Regards,
Felipe Pena