Re: [PATCH v8 03/13] mm: introduce AS_NO_DIRECT_MAP

[Nikita Kalyazin <kalyazin@amazon.com>]

On 05/12/2025 18:35, John Hubbard wrote:
> On 12/5/25 8:58 AM, Kalyazin, Nikita wrote:
>> From: Patrick Roy <patrick.roy@linux.dev>
>>
>> Add AS_NO_DIRECT_MAP for mappings where direct map entries of folios are
>> set to not present. Currently, mappings that match this description are
>> secretmem mappings (memfd_secret()). Later, some guest_memfd
>> configurations will also fall into this category.
>>
>> Reject this new type of mappings in all locations that currently reject
>> secretmem mappings, on the assumption that if secretmem mappings are
>> rejected somewhere, it is precisely because of an inability to deal with
>> folios without direct map entries, and then make memfd_secret() use
>> AS_NO_DIRECT_MAP on its address_space to drop its special
>> vma_is_secretmem()/secretmem_mapping() checks.
>>
>> This drops a optimization in gup_fast_folio_allowed() where
>> secretmem_mapping() was only called if CONFIG_SECRETMEM=y. secretmem is
>> enabled by default since commit b758fe6df50d ("mm/secretmem: make it on
>> by default"), so the secretmem check did not actually end up elided in
>> most cases anymore anyway.
> 
> The above paragraph can be part (most) of a commit description for
> a separate patch for the gup aspects. Let's split it out that way,
> because it is a distinct change, and the diffs can stand alone
> from this patch here.

Split the patches in the v9, thanks.

> 
> thanks,
> -- 
> John Hubbard
> 
>>
>> Use a new flag instead of overloading AS_INACCESSIBLE (which is already
>> set by guest_memfd) because not all guest_memfd mappings will end up
>> being direct map removed (e.g. in pKVM setups, parts of guest_memfd that
>> can be mapped to userspace should also be GUP-able, and generally not
>> have restrictions on who can access it).
>>
>> Acked-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
>> Acked-by: David Hildenbrand (Red Hat)" <david@kernel.org>
>> Signed-off-by: Patrick Roy <patrick.roy@linux.dev>
>> Signed-off-by: Nikita Kalyazin <kalyazin@amazon.com>
>> ---
>>   include/linux/pagemap.h   | 16 ++++++++++++++++
>>   include/linux/secretmem.h | 18 ------------------
>>   lib/buildid.c             |  4 ++--
>>   mm/gup.c                  | 19 +++++--------------
>>   mm/mlock.c                |  2 +-
>>   mm/secretmem.c            |  8 ++------
>>   6 files changed, 26 insertions(+), 41 deletions(-)
>>
>> diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
>> index a17fabbc0269..d51e0c0404e2 100644
>> --- a/include/linux/pagemap.h
>> +++ b/include/linux/pagemap.h
>> @@ -213,6 +213,7 @@ enum mapping_flags {
>>       AS_WRITEBACK_MAY_DEADLOCK_ON_RECLAIM = 9,
>>       AS_KERNEL_FILE = 10,    /* mapping for a fake kernel file that 
>> shouldn't
>>                                  account usage to user cgroups */
>> +     AS_NO_DIRECT_MAP = 11,  /* Folios in the mapping are not in the 
>> direct map */
>>       /* Bits 16-25 are used for FOLIO_ORDER */
>>       AS_FOLIO_ORDER_BITS = 5,
>>       AS_FOLIO_ORDER_MIN = 16,
>> @@ -348,6 +349,21 @@ static inline bool 
>> mapping_writeback_may_deadlock_on_reclaim(const struct addres
>>       return test_bit(AS_WRITEBACK_MAY_DEADLOCK_ON_RECLAIM, &mapping- 
>> >flags);
>>   }
>>
>> +static inline void mapping_set_no_direct_map(struct address_space 
>> *mapping)
>> +{
>> +     set_bit(AS_NO_DIRECT_MAP, &mapping->flags);
>> +}
>> +
>> +static inline bool mapping_no_direct_map(const struct address_space 
>> *mapping)
>> +{
>> +     return test_bit(AS_NO_DIRECT_MAP, &mapping->flags);
>> +}
>> +
>> +static inline bool vma_has_no_direct_map(const struct vm_area_struct 
>> *vma)
>> +{
>> +     return vma->vm_file && mapping_no_direct_map(vma->vm_file- 
>> >f_mapping);
>> +}
>> +
>>   static inline gfp_t mapping_gfp_mask(const struct address_space 
>> *mapping)
>>   {
>>       return mapping->gfp_mask;
>> diff --git a/include/linux/secretmem.h b/include/linux/secretmem.h
>> index e918f96881f5..0ae1fb057b3d 100644
>> --- a/include/linux/secretmem.h
>> +++ b/include/linux/secretmem.h
>> @@ -4,28 +4,10 @@
>>
>>   #ifdef CONFIG_SECRETMEM
>>
>> -extern const struct address_space_operations secretmem_aops;
>> -
>> -static inline bool secretmem_mapping(struct address_space *mapping)
>> -{
>> -     return mapping->a_ops == &secretmem_aops;
>> -}
>> -
>> -bool vma_is_secretmem(struct vm_area_struct *vma);
>>   bool secretmem_active(void);
>>
>>   #else
>>
>> -static inline bool vma_is_secretmem(struct vm_area_struct *vma)
>> -{
>> -     return false;
>> -}
>> -
>> -static inline bool secretmem_mapping(struct address_space *mapping)
>> -{
>> -     return false;
>> -}
>> -
>>   static inline bool secretmem_active(void)
>>   {
>>       return false;
>> diff --git a/lib/buildid.c b/lib/buildid.c
>> index c4b0f376fb34..89e567954284 100644
>> --- a/lib/buildid.c
>> +++ b/lib/buildid.c
>> @@ -65,8 +65,8 @@ static int freader_get_folio(struct freader *r, 
>> loff_t file_off)
>>
>>       freader_put_folio(r);
>>
>> -     /* reject secretmem folios created with memfd_secret() */
>> -     if (secretmem_mapping(r->file->f_mapping))
>> +     /* reject folios without direct map entries (e.g. from 
>> memfd_secret() or guest_memfd()) */
>> +     if (mapping_no_direct_map(r->file->f_mapping))
>>               return -EFAULT;
>>
>>       r->folio = filemap_get_folio(r->file->f_mapping, file_off >> 
>> PAGE_SHIFT);
>> diff --git a/mm/gup.c b/mm/gup.c
>> index a8ba5112e4d0..719e4dbecc09 100644
>> --- a/mm/gup.c
>> +++ b/mm/gup.c
>> @@ -11,7 +11,6 @@
>>   #include <linux/rmap.h>
>>   #include <linux/swap.h>
>>   #include <linux/swapops.h>
>> -#include <linux/secretmem.h>
>>
>>   #include <linux/sched/signal.h>
>>   #include <linux/rwsem.h>
>> @@ -1216,7 +1215,7 @@ static int check_vma_flags(struct vm_area_struct 
>> *vma, unsigned long gup_flags)
>>       if ((gup_flags & FOLL_SPLIT_PMD) && is_vm_hugetlb_page(vma))
>>               return -EOPNOTSUPP;
>>
>> -     if (vma_is_secretmem(vma))
>> +     if (vma_has_no_direct_map(vma))
>>               return -EFAULT;
>>
>>       if (write) {
>> @@ -2724,7 +2723,7 @@ EXPORT_SYMBOL(get_user_pages_unlocked);
>>    * This call assumes the caller has pinned the folio, that the 
>> lowest page table
>>    * level still points to this folio, and that interrupts have been 
>> disabled.
>>    *
>> - * GUP-fast must reject all secretmem folios.
>> + * GUP-fast must reject all folios without direct map entries (such 
>> as secretmem).
>>    *
>>    * Writing to pinned file-backed dirty tracked folios is inherently 
>> problematic
>>    * (see comment describing the writable_file_mapping_allowed() 
>> function). We
>> @@ -2739,7 +2738,6 @@ static bool gup_fast_folio_allowed(struct folio 
>> *folio, unsigned int flags)
>>   {
>>       bool reject_file_backed = false;
>>       struct address_space *mapping;
>> -     bool check_secretmem = false;
>>       unsigned long mapping_flags;
>>
>>       /*
>> @@ -2751,18 +2749,10 @@ static bool gup_fast_folio_allowed(struct 
>> folio *folio, unsigned int flags)
>>               reject_file_backed = true;
>>
>>       /* We hold a folio reference, so we can safely access folio 
>> fields. */
>> -
>> -     /* secretmem folios are always order-0 folios. */
>> -     if (IS_ENABLED(CONFIG_SECRETMEM) && !folio_test_large(folio))
>> -             check_secretmem = true;
>> -
>> -     if (!reject_file_backed && !check_secretmem)
>> -             return true;
>> -
>>       if (WARN_ON_ONCE(folio_test_slab(folio)))
>>               return false;
>>
>> -     /* hugetlb neither requires dirty-tracking nor can be secretmem. */
>> +     /* hugetlb neither requires dirty-tracking nor can be without 
>> direct map. */
>>       if (folio_test_hugetlb(folio))
>>               return true;
>>
>> @@ -2800,8 +2790,9 @@ static bool gup_fast_folio_allowed(struct folio 
>> *folio, unsigned int flags)
>>        * At this point, we know the mapping is non-null and points to an
>>        * address_space object.
>>        */
>> -     if (check_secretmem && secretmem_mapping(mapping))
>> +     if (mapping_no_direct_map(mapping))
>>               return false;
>> +
>>       /* The only remaining allowed file system is shmem. */
>>       return !reject_file_backed || shmem_mapping(mapping);
>>   }
>> diff --git a/mm/mlock.c b/mm/mlock.c
>> index bb0776f5ef7c..506452ee8cad 100644
>> --- a/mm/mlock.c
>> +++ b/mm/mlock.c
>> @@ -474,7 +474,7 @@ static int mlock_fixup(struct vma_iterator *vmi, 
>> struct vm_area_struct *vma,
>>
>>       if (newflags == oldflags || (oldflags & VM_SPECIAL) ||
>>           is_vm_hugetlb_page(vma) || vma == get_gate_vma(current->mm) ||
>> -         vma_is_dax(vma) || vma_is_secretmem(vma) || (oldflags & 
>> VM_DROPPABLE))
>> +         vma_is_dax(vma) || vma_has_no_direct_map(vma) || (oldflags & 
>> VM_DROPPABLE))
>>               /* don't set VM_LOCKED or VM_LOCKONFAULT and don't count */
>>               goto out;
>>
>> diff --git a/mm/secretmem.c b/mm/secretmem.c
>> index b59350daffe3..c1eafe29c3e8 100644
>> --- a/mm/secretmem.c
>> +++ b/mm/secretmem.c
>> @@ -134,11 +134,6 @@ static int secretmem_mmap_prepare(struct 
>> vm_area_desc *desc)
>>       return 0;
>>   }
>>
>> -bool vma_is_secretmem(struct vm_area_struct *vma)
>> -{
>> -     return vma->vm_ops == &secretmem_vm_ops;
>> -}
>> -
>>   static const struct file_operations secretmem_fops = {
>>       .release        = secretmem_release,
>>       .mmap_prepare   = secretmem_mmap_prepare,
>> @@ -156,7 +151,7 @@ static void secretmem_free_folio(struct folio *folio)
>>       folio_zero_segment(folio, 0, folio_size(folio));
>>   }
>>
>> -const struct address_space_operations secretmem_aops = {
>> +static const struct address_space_operations secretmem_aops = {
>>       .dirty_folio    = noop_dirty_folio,
>>       .free_folio     = secretmem_free_folio,
>>       .migrate_folio  = secretmem_migrate_folio,
>> @@ -205,6 +200,7 @@ static struct file *secretmem_file_create(unsigned 
>> long flags)
>>
>>       mapping_set_gfp_mask(inode->i_mapping, GFP_HIGHUSER);
>>       mapping_set_unevictable(inode->i_mapping);
>> +     mapping_set_no_direct_map(inode->i_mapping);
>>
>>       inode->i_op = &secretmem_iops;
>>       inode->i_mapping->a_ops = &secretmem_aops;
> 
>