Re: [PATCH] MIPS: Fix strnlen_user access check

[Tiezhu Yang <yangtiezhu@xxxxxxxxxxx>]

On 04/12/2021 11:02 AM, Tiezhu Yang wrote:
> On 04/11/2021 07:04 PM, Jinyang He wrote:
> > Commit 04324f44cb69 ("MIPS: Remove get_fs/set_fs") brought a problem for
> > strnlen_user(). Jump out when checking access_ok() with condition that
> > (s + strlen(s)) < __UA_LIMIT <= (s + n). The old __strnlen_user_asm()
> > just checked (ua_limit & s) without checking (ua_limit & (s + n)).
> > Therefore, find strlen form s to __UA_LIMIT - 1 in that condition.
> >
> > Signed-off-by: Jinyang He <hejinyang@xxxxxxxxxxx>
> > ---
> >   arch/mips/include/asm/uaccess.h | 11 +++++++++--
> >   1 file changed, 9 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/mips/include/asm/uaccess.h 
> > b/arch/mips/include/asm/uaccess.h
> > index 91bc7fb..85ba0c8 100644
> > --- a/arch/mips/include/asm/uaccess.h
> > +++ b/arch/mips/include/asm/uaccess.h
> > @@ -630,8 +630,15 @@ static inline long strnlen_user(const char 
> > __user *s, long n)
> >   {
> >       long res;
> >   -    if (!access_ok(s, n))
> > -        return -0;
> > +    if (unlikely(n <= 0))
> > +        return 0;
> > +
> > +    if (!access_ok(s, n)) {
> > +        if (!access_ok(s, 0))
> > +            return 0;
> > +
> > +        n = __UA_LIMIT - (unsigned long)s - 1;
> > +    }
> >         might_fault();
> >       __asm__ __volatile__(
>
> The following simple changes are OK to fix this issue?
>
> diff --git a/arch/mips/include/asm/uaccess.h 
> b/arch/mips/include/asm/uaccess.h
> index 91bc7fb..eafc99b 100644
> --- a/arch/mips/include/asm/uaccess.h
> +++ b/arch/mips/include/asm/uaccess.h
> @@ -630,8 +630,8 @@ static inline long strnlen_user(const char __user 
> *s, long n)
>  {
>         long res;
>
> -       if (!access_ok(s, n))
> -               return -0;
> +       if (!access_ok(s, 1))
> +               return 0;
>
>         might_fault();
>         __asm__ __volatile__(
>
> Thanks,
> Tiezhu

Hi all,

Here is some detail info about background and analysis process,
I hope it is useful to understand this issue.

When update kernel with the latest mips-next, we can not login through a
graphical interface, this is because drm radeon GPU driver does not work,
we can not see the boot message "[drm] radeon kernel modesetting enabled."
through the serial console.

drivers/gpu/drm/radeon/radeon_drv.c
static int __init radeon_module_init(void)
{
        [...]
        DRM_INFO("radeon kernel modesetting enabled.\n");
        [...]
}

I use git bisect to find commit 04324f44cb69 ("MIPS: Remove get_fs/set_fs")
is the first bad commit:

 $ git bisect log
 git bisect start
 # good: [666c1fc90cd82184624d4cc5d124c66025f89a47] mips: bmips: 
bcm63268: populate device tree nodes
 git bisect good 666c1fc90cd82184624d4cc5d124c66025f89a47
 # bad: [e86e75596623e1ce5d784db8214687326712a8ae] MIPS: octeon: Add 
__raw_copy_[from|to|in]_user symbols
 git bisect bad e86e75596623e1ce5d784db8214687326712a8ae
 # good: [45deb5faeb9e02951361ceba5ffee721745661c3] MIPS: uaccess: 
Remove get_fs/set_fs call sites
 git bisect good 45deb5faeb9e02951361ceba5ffee721745661c3
 # bad: [5e65c52ec716af6e8f51dacdaeb4a4d872249af1] MIPS: Loongson64: 
Use _CACHE_UNCACHED instead of _CACHE_UNCACHED_ACCELERATED
 git bisect bad 5e65c52ec716af6e8f51dacdaeb4a4d872249af1
 # bad: [04324f44cb69a03fdc8f2ee52386a4fdf6a0043b] MIPS: Remove 
get_fs/set_fs
 git bisect bad 04324f44cb69a03fdc8f2ee52386a4fdf6a0043b
 # first bad commit: [04324f44cb69a03fdc8f2ee52386a4fdf6a0043b] MIPS: 
Remove get_fs/set_fs

I analysis and test the changes in the above first bad commit and find out
the following obvious difference which leads to the login issue.

arch/mips/include/asm/uaccess.h
static inline long strnlen_user(const char __user *s, long n)
{
        [...]
        if (!access_ok(s, n))
                return -0;
        [...]
}

Thanks,
Tiezhu