Skip to contentSkip to conversation

bcachefs: byte order swap bch_alloc_v4.fragmentation_lru field

PATCH New Brian Foster 1 participant 2 messages Oct 9, 2023

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 bcachefs: byte order swap bch_alloc_v4.fragmentation_lru field

A simple test to populate a filesystem on one CPU architecture and fsck on an arch of the opposite byte order produces errors related to the fragmentation LRU. This occurs because the 64-bit fragmentation_lru field is not byte-order swapped when reads detect that the on-disk/bset key values were written in opposite byte-order of the current CPU.

Update the bch2_alloc_v4 swab callback to handle fragmentation_lru as is done for other multi-byte fields. This doesn't affect existing filesystems when accessed by CPUs of the same endianness because the ->swab() callback is only called when the bset flags indicate an endianness mismatch between the CPU and on-disk data.

---

Pushed to my CI branch as well (on top of the writeback fix I sent a few days or so ago...):

https://evilpiepirate.org/~testdashboard/ci?branch=bfoster

Brian

 fs/bcachefs/alloc_background.c | 1 +
 1 file changed, 1 insertion(+)
fs/bcachefs/alloc_background.c +1 -0
index d1c323fd3f0b..26b19ccad89b 100644
@@ -324,6 +324,7 @@ void bch2_alloc_v4_swab(struct bkey_s k)
 	a->io_time[1]		= swab64(a->io_time[1]);
 	a->stripe		= swab32(a->stripe);
 	a->nr_external_backpointers = swab32(a->nr_external_backpointers);
+	a->fragmentation_lru	= swab64(a->fragmentation_lru);
 
 	bps = alloc_v4_backpointers(a);
 	for (bp = bps; bp < bps + BCH_ALLOC_V4_NR_BACKPOINTERS(a); bp++) {
Signature · 1 line
2.41.0
#1 lore raw

PATCH 2/2 bcachefs: use swab40 for bch_backpointer.bucket_offset bitfield

The bucket_offset field of bch_backpointer is a 40-bit bitfield, but the bch2_backpointer_swab() helper uses swab32. This leads to inconsistency when an on-disk fs is accessed from an opposite endian machine.

As it turns out, we already have an internal swab40() helper that is used from the bch_alloc_v4 swab callback. Lift it into the backpointers header file and use it consistently in both places.

---

Here's another one I just tacked on that seems to address the backpointer related problems when switching CPU byte order...

Brian

 fs/bcachefs/alloc_background.c | 9 ---------
 fs/bcachefs/backpointers.c     | 2 +-
 fs/bcachefs/backpointers.h     | 9 +++++++++
 3 files changed, 10 insertions(+), 10 deletions(-)
fs/bcachefs/alloc_background.c +0 -9
index 26b19ccad89b..6380de7cd90f 100644
@@ -302,15 +302,6 @@ int bch2_alloc_v4_invalid(const struct bch_fs *c, struct bkey_s_c k,
 	return 0;
 }
 
-static inline u64 swab40(u64 x)
-{
-	return (((x & 0x00000000ffULL) << 32)|
-		((x & 0x000000ff00ULL) << 16)|
-		((x & 0x0000ff0000ULL) >>  0)|
-		((x & 0x00ff000000ULL) >> 16)|
-		((x & 0xff00000000ULL) >> 32));
-}
-
 void bch2_alloc_v4_swab(struct bkey_s k)
 {
 	struct bch_alloc_v4 *a = bkey_s_to_alloc_v4(k).v;
fs/bcachefs/backpointers.c +1 -1
index cc856150a948..b9955f937657 100644
@@ -76,7 +76,7 @@ void bch2_backpointer_swab(struct bkey_s k)
 {
 	struct bkey_s_backpointer bp = bkey_s_to_backpointer(k);
 
-	bp.v->bucket_offset	= swab32(bp.v->bucket_offset);
+	bp.v->bucket_offset	= swab40(bp.v->bucket_offset);
 	bp.v->bucket_len	= swab32(bp.v->bucket_len);
 	bch2_bpos_swab(&bp.v->pos);
 }
fs/bcachefs/backpointers.h +9 -0
index 547e0617602a..b0006e656f3a 100644
@@ -7,6 +7,15 @@
 #include "buckets.h"
 #include "super.h"
 
+static inline u64 swab40(u64 x)
+{
+	return (((x & 0x00000000ffULL) << 32)|
+		((x & 0x000000ff00ULL) << 16)|
+		((x & 0x0000ff0000ULL) >>  0)|
+		((x & 0x00ff000000ULL) >> 16)|
+		((x & 0xff00000000ULL) >> 32));
+}
+
 int bch2_backpointer_invalid(const struct bch_fs *, struct bkey_s_c k,
 			     enum bkey_invalid_flags, struct printbuf *);
 void bch2_backpointer_to_text(struct printbuf *, const struct bch_backpointer *);
Signature · 1 line
2.41.0