Re: [PATCH] hugetlb: Add guard support

[David Hildenbrand (Red Hat)]

On 11/28/25 12:23, Fushuai Wang wrote:
> > >    include/linux/hugetlb.h | 8 ++++++++
> > >    1 file changed, 8 insertions(+)
> > >
> > > diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
> > > index 8e63e46b8e1f..f97cb2f123a3 100644
> > > --- a/include/linux/hugetlb.h
> > > +++ b/include/linux/hugetlb.h
> > > @@ -15,6 +15,7 @@
> > >    #include <linux/gfp.h>
> > >    #include <linux/userfaultfd_k.h>
> > >    #include <linux/nodemask.h>
> > > +#include <linux/cleanup.h>
> > >    
> > >    struct ctl_table;
> > >    struct user_struct;
> > > @@ -468,6 +469,13 @@ static inline void hugetlb_split(struct vm_area_struct *vma, unsigned long addr)
> > >    
> > >    #endif /* !CONFIG_HUGETLB_PAGE */
> > >    
> > > +DEFINE_GUARD(hugetlb_vma_read, struct vm_area_struct *,
> > > +	     hugetlb_vma_lock_read(_T), hugetlb_vma_unlock_read(_T))
> > > +
> > > +DEFINE_GUARD(hugetlb_vma_write, struct vm_area_struct *,
> > > +	     hugetlb_vma_lock_write(_T), hugetlb_vma_unlock_write(_T))
> > > +DEFINE_GUARD_COND(hugetlb_vma_write, _try, hugetlb_vma_trylock_write(_T))
> >
> > Do I understand correctly that the code you are introducing does not
> > have any users?
>
> Yes, it is not used now, but it can be used in future development to simplify
> the code. If there are complex hugetlb_vma locks in the current code, this can
> also be used later to optimize them.

Right, but we usually don't add dead code. So it should ideally be 
introduced with actual users.
--
Cheers

David