Skip to contentSkip to conversation

bcachefs: Refactor memcpy into direct assignment

PATCH In review Kees Cook 2 participants 3 messages Oct 18, 2023 – Oct 19, 2023

lore
Share & tools
lore

Kees Cook

author
#0 lore raw

PATCH v3 1/1 bcachefs: Refactor memcpy into direct assignment

The memcpy() in bch2_bkey_append_ptr() is operating on an embedded fake flexible array which looks to the compiler like it has 0 size. This causes W=1 builds to emit warnings due to -Wstringop-overflow:

In file included from include/linux/string.h:254,
from include/linux/bitmap.h:11,
from include/linux/cpumask.h:12,
from include/linux/smp.h:13,
from include/linux/lockdep.h:14,
from include/linux/radix-tree.h:14,
from include/linux/backing-dev-defs.h:6,
from fs/bcachefs/bcachefs.h:182:
fs/bcachefs/extents.c: In function 'bch2_bkey_append_ptr':
include/linux/fortify-string.h:57:33: warning: writing 8 bytes into a region of size 0 [-Wstringop-overflow=]

      57 | #define __underlying_memcpy     __builtin_memcpy
         |                                 ^
   include/linux/fortify-string.h:648:9: note: in expansion of macro '__underlying_memcpy'
     648 |         __underlying_##op(p, q, __fortify_size);                        \
         |         ^~~~~~~~~~~~~
   include/linux/fortify-string.h:693:26: note: in expansion of macro '__fortify_memcpy_chk'
     693 | #define memcpy(p, q, s)  __fortify_memcpy_chk(p, q, s,                  \
         |                          ^~~~~~~~~~~~~~~~~~~~
   fs/bcachefs/extents.c:235:17: note: in expansion of macro 'memcpy'
     235 |                 memcpy((void *) &k->v + bkey_val_bytes(&k->k),
         |                 ^~~~~~
   fs/bcachefs/bcachefs_format.h:287:33: note: destination object 'v' of size 0
     287 |                 struct bch_val  v;
         |                                 ^

Avoid making any structure changes and just replace the u64 copy into a direct assignment, side-stepping the entire problem.

---
v3 - replace memcpy with assignment
v2 - https://lore.kernel.org/all/[email protected]
v1 - https://lore.kernel.org/r/[email protected]

---
 fs/bcachefs/extents.h | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
fs/bcachefs/extents.h +4 -4
index 7ee8d031bb6c..8c09c527fc4f 100644
@@ -632,6 +632,8 @@ void bch2_bkey_extent_entry_drop(struct bkey_i *, union bch_extent_entry *);
 
 static inline void bch2_bkey_append_ptr(struct bkey_i *k, struct bch_extent_ptr ptr)
 {
+	struct bch_extent_ptr *dest;
+
 	EBUG_ON(bch2_bkey_has_device(bkey_i_to_s(k), ptr.dev));
 
 	switch (k->k.type) {
@@ -641,10 +643,8 @@ static inline void bch2_bkey_append_ptr(struct bkey_i *k, struct bch_extent_ptr
 		EBUG_ON(bkey_val_u64s(&k->k) >= BKEY_EXTENT_VAL_U64s_MAX);
 
 		ptr.type = 1 << BCH_EXTENT_ENTRY_ptr;
-
-		memcpy((void *) &k->v + bkey_val_bytes(&k->k),
-		       &ptr,
-		       sizeof(ptr));
+		dest = (struct bch_extent_ptr *)((void *) &k->v + bkey_val_bytes(&k->k));
+		*dest = ptr;
 		k->k.u64s++;
 		break;
 	default:
Signature · 1 line
2.34.1
#1 lore raw
Quoted 29 lines · Kees Cook

The memcpy() in bch2_bkey_append_ptr() is operating on an embedded fake flexible array which looks to the compiler like it has 0 size. This causes W=1 builds to emit warnings due to -Wstringop-overflow:

In file included from include/linux/string.h:254,
from include/linux/bitmap.h:11, from include/linux/cpumask.h:12, from include/linux/smp.h:13, from include/linux/lockdep.h:14, from include/linux/radix-tree.h:14, from include/linux/backing-dev-defs.h:6, from fs/bcachefs/bcachefs.h:182:
fs/bcachefs/extents.c: In function 'bch2_bkey_append_ptr':

   include/linux/fortify-string.h:57:33: warning: writing 8 bytes into a region of size 0 [-Wstringop-overflow=]
      57 | #define __underlying_memcpy     __builtin_memcpy
         |                                 ^
   include/linux/fortify-string.h:648:9: note: in expansion of macro '__underlying_memcpy'
     648 |         __underlying_##op(p, q, __fortify_size);                        \
         |         ^~~~~~~~~~~~~
   include/linux/fortify-string.h:693:26: note: in expansion of macro '__fortify_memcpy_chk'
     693 | #define memcpy(p, q, s)  __fortify_memcpy_chk(p, q, s,                  \
         |                          ^~~~~~~~~~~~~~~~~~~~
   fs/bcachefs/extents.c:235:17: note: in expansion of macro 'memcpy'
     235 |                 memcpy((void *) &k->v + bkey_val_bytes(&k->k),
         |                 ^~~~~~
   fs/bcachefs/bcachefs_format.h:287:33: note: destination object 'v' of size 0
     287 |                 struct bch_val  v;
         |                                 ^
Quoted 2 lines

Avoid making any structure changes and just replace the u64 copy into a direct assignment, side-stepping the entire problem.

This does make me wonder about the usefulness of the fortify source stuff if it can be sidestepped this way, but hey, I'll take it :)

Pulled it into the testing branch, https://evilpiepirate.org/~testdashboard/ci?branch=bcachefs-testing

author
#2 lore raw
Quoted 30 lines · Kent Overstreet
Quoted 29 lines · Kees Cook

The memcpy() in bch2_bkey_append_ptr() is operating on an embedded fake flexible array which looks to the compiler like it has 0 size. This causes W=1 builds to emit warnings due to -Wstringop-overflow:

In file included from include/linux/string.h:254,
from include/linux/bitmap.h:11, from include/linux/cpumask.h:12, from include/linux/smp.h:13, from include/linux/lockdep.h:14, from include/linux/radix-tree.h:14, from include/linux/backing-dev-defs.h:6, from fs/bcachefs/bcachefs.h:182:
fs/bcachefs/extents.c: In function 'bch2_bkey_append_ptr':

   include/linux/fortify-string.h:57:33: warning: writing 8 bytes into a region of size 0 [-Wstringop-overflow=]
      57 | #define __underlying_memcpy     __builtin_memcpy
         |                                 ^
   include/linux/fortify-string.h:648:9: note: in expansion of macro '__underlying_memcpy'
     648 |         __underlying_##op(p, q, __fortify_size);                        \
         |         ^~~~~~~~~~~~~
   include/linux/fortify-string.h:693:26: note: in expansion of macro '__fortify_memcpy_chk'
     693 | #define memcpy(p, q, s)  __fortify_memcpy_chk(p, q, s,                  \
         |                          ^~~~~~~~~~~~~~~~~~~~
   fs/bcachefs/extents.c:235:17: note: in expansion of macro 'memcpy'
     235 |                 memcpy((void *) &k->v + bkey_val_bytes(&k->k),
         |                 ^~~~~~
   fs/bcachefs/bcachefs_format.h:287:33: note: destination object 'v' of size 0
     287 |                 struct bch_val  v;
         |                                 ^
Quoted 5 lines
Quoted 2 lines

Avoid making any structure changes and just replace the u64 copy into a direct assignment, side-stepping the entire problem.

This does make me wonder about the usefulness of the fortify source stuff if it can be sidestepped this way, but hey, I'll take it :)

Well, the "weird" cases like this are the ones that get attention. All the places it's working more cleanly are very effectively stomping real bugs.

Quoted 1 line

Pulled it into the testing branch, https://evilpiepirate.org/~testdashboard/ci?branch=bcachefs-testing

Thanks!

-Kees

Signature · 1 line
Kees Cook