Re: [RESEND PATCH] libfs: fix error cast of negative value in simple_attr_write()

[Andrew Morton]

On Wed, 11 Nov 2020 18:18:31 +0800 Yicong Yang <yangyicong@xxxxxxxxxxxxx> wrote:

> Hi,
> 
> Thanks for reviewing this.
> 
> 
> On 2020/11/11 3:18, Andrew Morton wrote:
> > On Tue, 10 Nov 2020 17:25:24 +0800 Yicong Yang <yangyicong@xxxxxxxxxxxxx> wrote:
> >
> >> The attr->set() receive a value of u64, but simple_strtoll() is used
> >> for doing the conversion. It will lead to the error cast if user inputs
> >> a negative value.
> >>
> >> Use kstrtoull() instead of simple_strtoll() to convert a string got
> >> from the user to an unsigned value. The former will return '-EINVAL' if
> >> it gets a negetive value, but the latter can't handle the situation
> >> correctly.
> >>
> >> ...
> >>
> >> --- a/fs/libfs.c
> >> +++ b/fs/libfs.c
> >> @@ -977,7 +977,9 @@ ssize_t simple_attr_write(struct file *file, const char __user *buf,
> >>  		goto out;
> >>  
> >>  	attr->set_buf[size] = '\0';
> >> -	val = simple_strtoll(attr->set_buf, NULL, 0);
> >> +	ret = kstrtoull(attr->set_buf, 0, &val);
> >> +	if (ret)
> >> +		goto out;
> >>  	ret = attr->set(attr->data, val);
> >>  	if (ret == 0)
> >>  		ret = len; /* on success, claim we got the whole input */
> > kstrtoull() takes an `unsigned long long *', but `val' is a u64.
> >
> > I think this probably works OK on all architectures (ie, no 64-bit
> > architectures are using `unsigned long' for u64).  But perhaps `val'
> > should have type `unsigned long long'?
> 
> the attr->set() takes 'val' as u64, so maybe we can stay it unchanged here
> if it works well.

Sure.  But the compiler will convert an unsigned long long into a u64
quite happily, regardless of how u64 was actually implemented.

However the compiler will not convert a `u64 *' into an `unsigned long
long *' if the underlying type of u64 happens to be `unsigned long'. 
It will warn.