Skip to contentSkip to conversation

mm/memfd_luo: reject memfds whose page count exceeds UINT_MAX

PATCH In review David Carlier 3 participants 8 messages Apr 23 – May 4

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 mm/memfd_luo: reject memfds whose page count exceeds UINT_MAX

memfd_luo_preserve_folios() declares max_folios as unsigned int and computes it from the inode size, then passes it to memfd_pin_folios() which itself caps max_folios at unsigned int. For files whose base-page count exceeds UINT_MAX (larger than 16 TiB with 4 KiB pages), the assignment truncates silently: only a prefix of the file gets pinned and preserved, while memfd_luo_preserve() still records the full inode size in ser->size. On retrieve the inode is restored to the full size but only the preserved prefix repopulates the page cache, so the tail comes back as holes and user data is silently lost across the live update.

Reject such files at preserve time with -EFBIG rather than chunk the pin loop, which would also require enlarging the preserved folios array well beyond what is practical.

Fixes b3749f174d68 ("mm: memfd_luo: allow preserving memfd")
---
 mm/memfd_luo.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
mm/memfd_luo.c +13 -2
index b02b503c750d..f41d11053b7d 100644
@@ -259,7 +259,7 @@ static int memfd_luo_preserve(struct liveupdate_file_op_args *args)
 	struct inode *inode = file_inode(args->file);
 	struct memfd_luo_folio_ser *folios_ser;
 	struct memfd_luo_ser *ser;
-	u64 nr_folios;
+	u64 nr_folios, inode_size;
 	int err = 0, seals;
 
 	inode_lock(inode);
@@ -285,7 +285,18 @@ static int memfd_luo_preserve(struct liveupdate_file_op_args *args)
 	}
 
 	ser->pos = args->file->f_pos;
-	ser->size = i_size_read(inode);
+	inode_size = i_size_read(inode);
+
+	/*
+	 * memfd_pin_folios() caps at UINT_MAX folios; refuse larger
+	 * files to avoid silently preserving only a prefix.
+	 */
+	if (DIV_ROUND_UP_ULL(inode_size, PAGE_SIZE) > UINT_MAX) {
+		err = -EFBIG;
+		goto err_free_ser;
+	}
+
+	ser->size = inode_size;
 	ser->seals = seals;
 
 	err = memfd_luo_preserve_folios(args->file, &ser->folios,
Signature · 1 line
2.53.0
#1 lore raw

PATCH 2/2 mm/memfd_luo: document preservation of file seals

Commit 8a552d68a86e ("mm: memfd_luo: preserve file seals") started preserving file seals across live update and restoring them via memfd_add_seals() on retrieve, but the DOC header was not updated and still listed seals under "Non-Preserved Properties" as being unsealed on restore.

Move the Seals entry to the "Preserved Properties" section and describe the actual behavior, including the MEMFD_LUO_ALL_SEALS restriction that both preserve and retrieve enforce.

Fixes 8a552d68a86e ("mm: memfd_luo: preserve file seals")
---
 mm/memfd_luo.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
mm/memfd_luo.c +5 -4
index f41d11053b7d..c5250f68c096 100644
@@ -50,6 +50,11 @@
  *   memfds are always opened with ``O_RDWR`` and ``O_LARGEFILE``. This property
  *   is maintained.
  *
+ * Seals
+ *   File seals set on the memfd are preserved and re-applied on restore.
+ *   Only seals known to this LUO version (see ``MEMFD_LUO_ALL_SEALS``) may
+ *   be present; preservation fails with ``-EOPNOTSUPP`` otherwise.
+ *
  * Non-Preserved Properties
  * ========================
  *
@@ -61,10 +66,6 @@
  *   A memfd can be created with the ``MFD_CLOEXEC`` flag that sets the
  *   ``FD_CLOEXEC`` on the file. This flag is not preserved and must be set
  *   again after restore via ``fcntl()``.
- *
- * Seals
- *   File seals are not preserved. The file is unsealed on restore and if
- *   needed, must be sealed again via ``fcntl()``.
  */
 
 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Signature · 1 line
2.53.0
maintainer
#4 lore raw
In reply to David Carlier
Quoted 34 lines · David Carlier

memfd_luo_preserve_folios() declares max_folios as unsigned int and computes it from the inode size, then passes it to memfd_pin_folios() which itself caps max_folios at unsigned int. For files whose base-page count exceeds UINT_MAX (larger than 16 TiB with 4 KiB pages), the assignment truncates silently: only a prefix of the file gets pinned and preserved, while memfd_luo_preserve() still records the full inode size in ser->size. On retrieve the inode is restored to the full size but only the preserved prefix repopulates the page cache, so the tail comes back as holes and user data is silently lost across the live update.

Reject such files at preserve time with -EFBIG rather than chunk the pin loop, which would also require enlarging the preserved folios array well beyond what is practical.

Fixes b3749f174d68 ("mm: memfd_luo: allow preserving memfd")
---
 mm/memfd_luo.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
mm/memfd_luo.c +1 -1
index b02b503c750d..f41d11053b7d 100644
@@ -259,7 +259,7 @@ static int memfd_luo_preserve(struct liveupdate_file_op_args *args)
 	struct inode *inode = file_inode(args->file);
 	struct memfd_luo_folio_ser *folios_ser;
 	struct memfd_luo_ser *ser;
-	u64 nr_folios;
+	u64 nr_folios, inode_size;
 	int err = 0, seals;
 
 	inode_lock(inode);

mm/memfd_luo.c static int memfd_luo_preserve(struct liveupdate_file_op_args *args) · line 293

Quoted 10 lines
mm/memfd_luo.c +6 -1
@@ -285,7 +285,18 @@ static int memfd_luo_preserve(struct liveupdate_file_op_args *args)
 	}
 
 	ser->pos = args->file->f_pos;
-	ser->size = i_size_read(inode);
+	inode_size = i_size_read(inode);
+
+	/*
+	 * memfd_pin_folios() caps at UINT_MAX folios; refuse larger
+	 * files to avoid silently preserving only a prefix.
+	 */

I think, the fix should be first done at memfd_pin_folios() to change max_folios to 'long' or 'unsigned long', and then just updated memfd_luo.c to match.

Pasha

Quoted 12 lines
mm/memfd_luo.c +6 -0
+	if (DIV_ROUND_UP_ULL(inode_size, PAGE_SIZE) > UINT_MAX) {
+		err = -EFBIG;
+		goto err_free_ser;
+	}
+
+	ser->size = inode_size;
 	ser->seals = seals;
 
 	err = memfd_luo_preserve_folios(args->file, &ser->folios,
Signature · 1 line
2.53.0
#5 lore raw
Quoted 1 line · Pasha Tatashin

I think, the fix should be first done at memfd_pin_folios() to

change

Quoted 2 lines

max_folios to 'long' or 'unsigned long', and then just updated memfd_luo.c to match.

Even with memfd_pin_folios() widened, memfd_luo_preserve_folios()
still
kvmalloc_objs()s the folios array up front -- at UINT_MAX entries
that's
already ~32 GiB of pointer array, which won't realistically succeed.
So
the guard here still makes sense, just on allocation grounds rather than interface truncation.

Happy to widen memfd_pin_folios() as a separate patch, but I'd keep
the
-EFBIG check regardless. Or did you have a different shape in mind?

Cheers.

maintainer Reviewed-by
#6 lore raw
Quoted 15 lines · David CARLIER
Quoted 1 line · Pasha Tatashin

I think, the fix should be first done at memfd_pin_folios() to

change

Quoted 2 lines

max_folios to 'long' or 'unsigned long', and then just updated memfd_luo.c to match.

Even with memfd_pin_folios() widened, memfd_luo_preserve_folios()
still
kvmalloc_objs()s the folios array up front -- at UINT_MAX entries
that's
already ~32 GiB of pointer array, which won't realistically succeed.
So
the guard here still makes sense, just on allocation grounds rather than interface truncation.

Quoted 3 lines

Happy to widen memfd_pin_folios() as a separate patch, but I'd keep
the
-EFBIG check regardless. Or did you have a different shape in mind?

Ah, Good point about kvmalloc_objs(), I am not against this, even though it is a very theoretical issue.

maintainer Reviewed-by on v1 2/2
#2 lore raw

Re: [PATCH 2/2] mm/memfd_luo: document preservation of file seals

In reply to David Carlier
Quoted 6 lines · David Carlier

Commit 8a552d68a86e ("mm: memfd_luo: preserve file seals") started preserving file seals across live update and restoring them via memfd_add_seals() on retrieve, but the DOC header was not updated and still listed seals under "Non-Preserved Properties" as being unsealed on restore.

Quoted 6 lines

Move the Seals entry to the "Preserved Properties" section and describe the actual behavior, including the MEMFD_LUO_ALL_SEALS restriction that both preserve and retrieve enforce.

Fixes 8a552d68a86e ("mm: memfd_luo: preserve file seals")
maintainer Reviewed-by
#7 lore raw
In reply to Pasha Tatashin
Quoted 16 lines · Pasha Tatashin
Quoted 15 lines · David CARLIER
Quoted 1 line · Pasha Tatashin

I think, the fix should be first done at memfd_pin_folios() to

change

Quoted 2 lines

max_folios to 'long' or 'unsigned long', and then just updated memfd_luo.c to match.

Even with memfd_pin_folios() widened, memfd_luo_preserve_folios()
still
kvmalloc_objs()s the folios array up front -- at UINT_MAX entries
that's
already ~32 GiB of pointer array, which won't realistically succeed.
So
the guard here still makes sense, just on allocation grounds rather than interface truncation.

Quoted 6 lines
Quoted 3 lines

Happy to widen memfd_pin_folios() as a separate patch, but I'd keep
the
-EFBIG check regardless. Or did you have a different shape in mind?

Ah, Good point about kvmalloc_objs(), I am not against this, even though it is a very theoretical issue.

Yeah. I plan to get rid of the pinning anyway and do this via the page cache directly (might be easier said than done though), so this should be fine for now I think. I suspect we might run into a lot more problems if we get a memfd like this anyway.

Quoted 2 lines
Signature · 2 lines
Regards,
Pratyush Yadav
maintainer Reviewed-by on v1 2/2
#3 lore raw

Re: [PATCH 2/2] mm/memfd_luo: document preservation of file seals

In reply to David Carlier
Quoted 6 lines · David Carlier

Commit 8a552d68a86e ("mm: memfd_luo: preserve file seals") started preserving file seals across live update and restoring them via memfd_add_seals() on retrieve, but the DOC header was not updated and still listed seals under "Non-Preserved Properties" as being unsealed on restore.

Quoted 6 lines

Move the Seals entry to the "Preserved Properties" section and describe the actual behavior, including the MEMFD_LUO_ALL_SEALS restriction that both preserve and retrieve enforce.

Fixes 8a552d68a86e ("mm: memfd_luo: preserve file seals")

Thanks for the fix!

[...]

Signature · 2 lines
Regards,
Pratyush Yadav