Re: [PATCH v4 08/10] mm/vma: improve and document __is_vma_write_locked()

[Lorenzo Stoakes <lorenzo.stoakes@oracle.com>]

On Mon, Jan 26, 2026 at 11:21:29AM -0800, Suren Baghdasaryan wrote:
> On Mon, Jan 26, 2026 at 8:29 AM Lorenzo Stoakes
> <lorenzo.stoakes@oracle.com> wrote:
> >
> > On Mon, Jan 26, 2026 at 12:30:04PM +0100, Vlastimil Babka wrote:
> > > On 1/23/26 21:12, Lorenzo Stoakes wrote:
> > > > We don't actually need to return an output parameter providing mm sequence
> > > > number, rather we can separate that out into another function -
> > > > __vma_raw_mm_seqnum() - and have any callers which need to obtain that
> > > > invoke that instead.
> > > >
> > > > The access to the raw sequence number requires that we hold the exclusive
> > > > mmap lock such that we know we can't race vma_end_write_all(), so move the
> > > > assert to __vma_raw_mm_seqnum() to make this requirement clear.
> > > >
> > > > Also while we're here, convert all of the VM_BUG_ON_VMA()'s to
> > > > VM_WARN_ON_ONCE_VMA()'s in line with the convention that we do not invoke
> > > > oopses when we can avoid it.
> > > >
> > > > Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> > >
> > > Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
>
> Sorry but I have one more comment below.
>
> Reviewed-by: Suren Baghdasaryan <surenb@google.com>

Thanks!

>
> >
> > Thanks!
> >
> > >
> > > Few nits:
> > >
> > > > ---
> > > >  include/linux/mmap_lock.h | 44 ++++++++++++++++++++++-----------------
> > > >  1 file changed, 25 insertions(+), 19 deletions(-)
> > > >
> > > > diff --git a/include/linux/mmap_lock.h b/include/linux/mmap_lock.h
> > > > index 678f90080fa6..23bde4bd5a85 100644
> > > > --- a/include/linux/mmap_lock.h
> > > > +++ b/include/linux/mmap_lock.h
> > > > @@ -258,17 +258,30 @@ static inline void vma_end_read(struct vm_area_struct *vma)
> > > >     vma_refcount_put(vma);
> > > >  }
> > > >
> > > > -/* WARNING! Can only be used if mmap_lock is expected to be write-locked */
> > > > -static inline bool __is_vma_write_locked(struct vm_area_struct *vma, unsigned int *mm_lock_seq)
> > > > +static inline unsigned int __vma_raw_mm_seqnum(struct vm_area_struct *vma)
>
> This function returns the mm->mm_lock_seq.sequence attribute of mm, so
> no real commection to VMA. IMO it's better to rename it into
> __raw_mm_lock_seqnum(const struct mm_struct *mm) and have the callers
> pass vma->vm_mm.

Sorry missed these comments.

We are only ever referencing in terms of VMA's, so I think it makes sense still
to pass a VMA ptr, even if it just references vma->vm_mm.

I think the name of the function also makes things pretty clear, given it's
called essentially 'VMA['s] raw mm seqnum'.

>
> > > >  {
> > > > +   const struct mm_struct *mm = vma->vm_mm;
> > > > +
> > > > +   /* We must hold an exclusive write lock for this access to be valid. */
> > > >     mmap_assert_write_locked(vma->vm_mm);
>
> If for some reason you need to keep this function VMA-centric, then in
> the above line please s/vma->vm_mm/mm

Ah yeah, maybe if there's a respin? As kinda trivial thing, if that makes sense
to you!

Cheers, Lorenzo