On Wed, Feb 21, 2018 at 09:28:41PM +0200, Andy Shevchenko wrote:
> On Wed, Feb 21, 2018 at 9:23 PM, Alexey Dobriyan <[email protected]> wrote:
> > /proc/*/cmdline is not different from /proc/*/environ as it accesses
> > target task's memory (and can access the very same region of memory)
> > but it doesn't go through ptrace_may_access() and thus doesn't go through
> > LSM.
>
> > +static int __mem_open(struct inode *inode, struct file *file, unsigned int
> > mode)
> > +{
> > + struct mm_struct *mm = proc_mem_open(inode, mode);
> > +
> > + if (IS_ERR(mm))
> > + return PTR_ERR(mm);
>
> So, is it possible to have it NULL?..
I haven't looked closely, but if kernel thread is accesses then yes.
Regardless, patch only moves function so that code compiles, untangling
this little mess is separate adventure.
> > +static int mem_release(struct inode *inode, struct file *file)
> > +{
> > + struct mm_struct *mm = file->private_data;
>
> > + if (mm)
>
> ...or I don't get this check.
>
> > + mmdrop(mm);
> > + return 0;
> > +}
It should trigger if kernel thread is accessed.
