[PATCH -next v19 20/24] riscv: Add prctl controls for userspace vector management

Andy Chiu andy.chiu at sifive.com
Tue May 9 09:11:45 PDT 2023


On Tue, May 9, 2023 at 7:14 PM Heiko Stübner <heiko at sntech.de> wrote:
>
> Hi,
>
> need to poke this more, but one issue popped up at first compile.
>
> Am Dienstag, 9. Mai 2023, 12:30:29 CEST schrieb Andy Chiu:
> > This patch add two riscv-specific prctls, to allow usespace control the
> > use of vector unit:
> >
> >  * PR_RISCV_V_SET_CONTROL: control the permission to use Vector at next,
> >    or all following execve for a thread. Turning off a thread's Vector
> >    live is not possible since libraries may have registered ifunc that
> >    may execute Vector instructions.
> >  * PR_RISCV_V_GET_CONTROL: get the same permission setting for the
> >    current thread, and the setting for following execve(s).
> >
> > Signed-off-by: Andy Chiu <andy.chiu at sifive.com>
> > Reviewed-by: Greentime Hu <greentime.hu at sifive.com>
> > Reviewed-by: Vincent Chen <vincent.chen at sifive.com>
>
>
> > diff --git a/arch/riscv/kernel/vector.c b/arch/riscv/kernel/vector.c
> > index 960a343799c6..16ccb35625a9 100644
> > --- a/arch/riscv/kernel/vector.c
> > +++ b/arch/riscv/kernel/vector.c
> > @@ -9,6 +9,7 @@
> >  #include <linux/slab.h>
> >  #include <linux/sched.h>
> >  #include <linux/uaccess.h>
> > +#include <linux/prctl.h>
> >
> >  #include <asm/thread_info.h>
> >  #include <asm/processor.h>
> > @@ -19,6 +20,8 @@
> >  #include <asm/ptrace.h>
> >  #include <asm/bug.h>
> >
> > +static bool riscv_v_implicit_uacc = !IS_ENABLED(CONFIG_RISCV_V_DISABLE);
> > +
> >  unsigned long riscv_v_vsize __read_mostly;
> >  EXPORT_SYMBOL_GPL(riscv_v_vsize);
> >
> > @@ -91,11 +94,51 @@ static int riscv_v_thread_zalloc(void)
> >       return 0;
> >  }
> >
> > +#define VSTATE_CTRL_GET_CUR(x) ((x) & PR_RISCV_V_VSTATE_CTRL_CUR_MASK)
> > +#define VSTATE_CTRL_GET_NEXT(x) (((x) & PR_RISCV_V_VSTATE_CTRL_NEXT_MASK) >> 2)
> > +#define VSTATE_CTRL_MAKE_NEXT(x) (((x) << 2) & PR_RISCV_V_VSTATE_CTRL_NEXT_MASK)
> > +#define VSTATE_CTRL_GET_INHERIT(x) (!!((x) & PR_RISCV_V_VSTATE_CTRL_INHERIT))
> > +static inline int riscv_v_get_cur_ctrl(struct task_struct *tsk)
> > +{
> > +     return VSTATE_CTRL_GET_CUR(tsk->thread.vstate_ctrl);
> > +}
> > +
> > +static inline int riscv_v_get_next_ctrl(struct task_struct *tsk)
> > +{
> > +     return VSTATE_CTRL_GET_NEXT(tsk->thread.vstate_ctrl);
> > +}
> > +
> > +static inline bool riscv_v_test_ctrl_inherit(struct task_struct *tsk)
> > +{
> > +     return VSTATE_CTRL_GET_INHERIT(tsk->thread.vstate_ctrl);
> > +}
> > +
> > +static inline void riscv_v_set_ctrl(struct task_struct *tsk, int cur, int nxt,
> > +                                 bool inherit)
> > +{
> > +     unsigned long ctrl;
> > +
> > +     ctrl = cur & PR_RISCV_V_VSTATE_CTRL_CUR_MASK;
> > +     ctrl |= VSTATE_CTRL_MAKE_NEXT(nxt);
> > +     if (inherit)
> > +             ctrl |= PR_RISCV_V_VSTATE_CTRL_INHERIT;
> > +     tsk->thread.vstate_ctrl = ctrl;
> > +}
> > +
> > +bool riscv_v_user_allowed(void)
> > +{
> > +     return riscv_v_get_cur_ctrl(current) == PR_RISCV_V_VSTATE_CTRL_ON;
> > +}
>
> EXPORT_SYMBOL(riscv_v_user_allowed);

It's a shame. KVM is a builtin on my test platform so I missed this
obvious thing. Or, maybe we should make them inline functions or
macros at the header file because of the size. And because other
modules may rarely use them.

>
> kvm is allowed to be built as module, so you could end up with:
>
> ERROR: modpost: "riscv_v_user_allowed" [arch/riscv/kvm/kvm.ko] undefined!
> make[2]: *** [../scripts/Makefile.modpost:136: Module.symvers] Fehler 1
> make[1]: *** [/home/devel/hstuebner/00_git-repos/linux-riscv/Makefile:1978: modpost] Fehler 2
> make[1]: Verzeichnis „/home/devel/hstuebner/00_git-repos/linux-riscv/_build-riscv64“ wird verlassen
> make: *** [Makefile:226: __sub-make] Fehler 2
>
>
> Heiko
>
>
>

Thanks,
Andy



More information about the kvm-riscv mailing list