Re: [PATCH v5 04/44] x86: asm: instrument usercopy in get_user() and put_user()

[Yu Zhao]

On Tue, Aug 30, 2022 at 5:00 PM Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Tue, 30 Aug 2022 16:25:24 -0600 Yu Zhao <yuzhao@xxxxxxxxxx> wrote:
>
> > On Tue, Aug 30, 2022 at 4:05 PM Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > On Tue, 30 Aug 2022 16:23:44 +0200 Alexander Potapenko <glider@xxxxxxxxxx> wrote:
> > >
> > > > >                  from init/do_mounts.c:2:
> > > > > ./include/linux/page-flags.h: In function ‘page_fixed_fake_head’:
> > > > > ./include/linux/page-flags.h:226:36: error: invalid use of undefined type ‘const struct page’
> > > > >   226 |             test_bit(PG_head, &page->flags)) {
> > > > >       |                                    ^~
> > > > > ./include/linux/bitops.h:50:44: note: in definition of macro ‘bitop’
> > > > >    50 |           __builtin_constant_p((uintptr_t)(addr) != (uintptr_t)NULL) && \
> > > > >       |                                            ^~~~
> > > > > ./include/linux/page-flags.h:226:13: note: in expansion of macro ‘test_bit’
> > > > >   226 |             test_bit(PG_head, &page->flags)) {
> > > > >       |             ^~~~~~~~
> > > > > ...
> > > >
> > > > Gotcha, this is a circular dependency: mm_types.h -> sched.h ->
> > > > kmsan.h -> gfp.h -> mmzone.h -> page-flags.h -> mm_types.h, where the
> > > > inclusion of sched.h into mm_types.h was only introduced in "mm:
> > > > multi-gen LRU: support page table walks" - that's why the problem was
> > > > missing in other trees.
> > >
> > > Ah, thanks for digging that out.
> > >
> > > Yu, that inclusion is regrettable.
> >
> > Sorry for the trouble -- it's also superfluous because we don't call
> > lru_gen_use_mm() when switching to the kernel.
> >
> > I've queued the following for now.
>
> Well, the rest of us want it too.
>
> > --- a/include/linux/mm_types.h
> > +++ b/include/linux/mm_types.h
> > @@ -3,7 +3,6 @@
> >  #define _LINUX_MM_TYPES_H
> >
> >  #include <linux/mm_types_task.h>
> > -#include <linux/sched.h>
> >
> >  #include <linux/auxvec.h>
> >  #include <linux/kref.h>
> > @@ -742,8 +741,7 @@ static inline void lru_gen_init_mm(struct mm_struct *mm)
> >
> >  static inline void lru_gen_use_mm(struct mm_struct *mm)
> >  {
> > -       if (!(current->flags & PF_KTHREAD))
> > -               WRITE_ONCE(mm->lru_gen.bitmap, -1);
> > +       WRITE_ONCE(mm->lru_gen.bitmap, -1);
> >  }
>
> Doesn't apply.  I did:
>
> --- a/include/linux/mm_types.h~mm-multi-gen-lru-support-page-table-walks-fix
> +++ a/include/linux/mm_types.h
> @@ -3,7 +3,6 @@
>  #define _LINUX_MM_TYPES_H
>
>  #include <linux/mm_types_task.h>
> -#include <linux/sched.h>
>
>  #include <linux/auxvec.h>
>  #include <linux/kref.h>
> @@ -742,11 +741,7 @@ static inline void lru_gen_init_mm(struc
>
>  static inline void lru_gen_use_mm(struct mm_struct *mm)
>  {
> -       /* unlikely but not a bug when racing with lru_gen_migrate_mm() */
> -       VM_WARN_ON_ONCE(list_empty(&mm->lru_gen.list));

Yes. I got a report that somebody tripped over this "unlikely"
condition (and ascertained it's not a bug). So I deleted this part as
well.

Will refresh the series around rc5. Thanks.