Re: [PATCH] fs/proc/task_mmu.c: fix clear_refs_write() input sanity check

["Li, Ming Chun" <macli@brc.ubc.ca>]

On Wed, 12 Aug 2009, David Rientjes wrote:

> On Wed, 12 Aug 2009, Vincent Li wrote:
> 
> > diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> > index 2079969..026b532 100644
> > --- a/fs/proc/task_mmu.c
> > +++ b/fs/proc/task_mmu.c
> > @@ -494,18 +494,17 @@ static ssize_t clear_refs_write(struct file *file, const char __user *buf,
> >  	char buffer[PROC_NUMBUF], *end;
> >  	struct mm_struct *mm;
> >  	struct vm_area_struct *vma;
> > -	int type;
> > +	long res;
> >  
> 
> What does `res' mean?  Why can't it just be left as `type'?

res mean the result parsed from userspace input, I can keep it as name 
'type'.

> 
> >  	memset(buffer, 0, sizeof(buffer));
> >  	if (count > sizeof(buffer) - 1)
> >  		count = sizeof(buffer) - 1;
> >  	if (copy_from_user(buffer, buf, count))
> >  		return -EFAULT;
> > -	type = simple_strtol(buffer, &end, 0);
> > -	if (type < CLEAR_REFS_ALL || type > CLEAR_REFS_MAPPED)
> > +	if (strict_strtol(strstrip(buffer), 10, &res)
> > +		return -EINVAL;
> 
> Please test your patch submissions (which would require that the above 
> actually compiles).
> 

I did get compiling error and warnings, I will resend the patch with 
varible name 'type' left and no compiling errors.


Vincent Li
Biomedical Research Center
University of British Columbia