Re: [PATCH v5 3/6] mm/gup: remove vmas parameter from get_user_pages_remote()

[Lorenzo Stoakes]

On Tue, May 16, 2023 at 11:49:19AM +0200, Anders Roxell wrote:
> On 2023-05-14 22:26, Lorenzo Stoakes wrote:
> > The only instances of get_user_pages_remote() invocations which used the
> > vmas parameter were for a single page which can instead simply look up the
> > VMA directly. In particular:-
> >
> > - __update_ref_ctr() looked up the VMA but did nothing with it so we simply
> >   remove it.
> >
> > - __access_remote_vm() was already using vma_lookup() when the original
> >   lookup failed so by doing the lookup directly this also de-duplicates the
> >   code.
> >
> > We are able to perform these VMA operations as we already hold the
> > mmap_lock in order to be able to call get_user_pages_remote().
> >
> > As part of this work we add get_user_page_vma_remote() which abstracts the
> > VMA lookup, error handling and decrementing the page reference count should
> > the VMA lookup fail.
> >
> > This forms part of a broader set of patches intended to eliminate the vmas
> > parameter altogether.
> >
> > Reviewed-by: Catalin Marinas <catalin.marinas@xxxxxxx> (for arm64)
> > Acked-by: David Hildenbrand <david@xxxxxxxxxx>
> > Reviewed-by: Janosch Frank <frankja@xxxxxxxxxxxxx> (for s390)
> > Signed-off-by: Lorenzo Stoakes <lstoakes@xxxxxxxxx>
> > ---
> >  arch/arm64/kernel/mte.c   | 17 +++++++++--------
> >  arch/s390/kvm/interrupt.c |  2 +-
> >  fs/exec.c                 |  2 +-
> >  include/linux/mm.h        | 34 +++++++++++++++++++++++++++++++---
> >  kernel/events/uprobes.c   | 13 +++++--------
> >  mm/gup.c                  | 12 ++++--------
> >  mm/memory.c               | 14 +++++++-------
> >  mm/rmap.c                 |  2 +-
> >  security/tomoyo/domain.c  |  2 +-
> >  virt/kvm/async_pf.c       |  3 +--
> >  10 files changed, 61 insertions(+), 40 deletions(-)
> >
>
> [...]
>
> > diff --git a/mm/memory.c b/mm/memory.c
> > index 146bb94764f8..63632a5eafc1 100644
> > --- a/mm/memory.c
> > +++ b/mm/memory.c
> > @@ -5590,7 +5590,6 @@ EXPORT_SYMBOL_GPL(generic_access_phys);
> >  int __access_remote_vm(struct mm_struct *mm, unsigned long addr, void *buf,
> >  		       int len, unsigned int gup_flags)
> >  {
> > -	struct vm_area_struct *vma;
> >  	void *old_buf = buf;
> >  	int write = gup_flags & FOLL_WRITE;
> >
> > @@ -5599,13 +5598,15 @@ int __access_remote_vm(struct mm_struct *mm, unsigned long addr, void *buf,
> >
> >  	/* ignore errors, just check how much was successfully transferred */
> >  	while (len) {
> > -		int bytes, ret, offset;
> > +		int bytes, offset;
> >  		void *maddr;
> > -		struct page *page = NULL;
> > +		struct vm_area_struct *vma;
> > +		struct page *page = get_user_page_vma_remote(mm, addr,
> > +							     gup_flags, &vma);
> > +
> > +		if (IS_ERR_OR_NULL(page)) {
> > +			int ret = 0;
>
> I see the warning below when building without CONFIG_HAVE_IOREMAP_PROT set.
>
> make --silent --keep-going --jobs=32 \
> O=/home/anders/.cache/tuxmake/builds/1244/build ARCH=arm \
> CROSS_COMPILE=arm-linux-gnueabihf- /home/anders/src/kernel/next/mm/memory.c: In function '__access_remote_vm':
> /home/anders/src/kernel/next/mm/memory.c:5608:29: warning: unused variable 'ret' [-Wunused-variable]
>  5608 |                         int ret = 0;
>       |                             ^~~
>

Ah damn, nice spot thanks!

>
> >
> > -		ret = get_user_pages_remote(mm, addr, 1,
> > -				gup_flags, &page, &vma, NULL);
> > -		if (ret <= 0) {
> >  #ifndef CONFIG_HAVE_IOREMAP_PROT
> >  			break;
> >  #else
> > @@ -5613,7 +5614,6 @@ int __access_remote_vm(struct mm_struct *mm, unsigned long addr, void *buf,
> >  			 * Check if this is a VM_IO | VM_PFNMAP VMA, which
> >  			 * we can access using slightly different code.
> >  			 */
> > -			vma = vma_lookup(mm, addr);
> >  			if (!vma)
> >  				break;
> >  			if (vma->vm_ops && vma->vm_ops->access)
>
> Cheers,
> Anders

I enclose a -fix patch for this below:-

----8<----