On 2024/6/4 10:01, Kent Overstreet wrote:
On Tue, Jun 04, 2024 at 09:22:51AM +0800, Hongbo Li wrote:
I found the following issue on:
HEAD commit: 587f7f5a4765a76f9ea05159963d14f8967cb45b(cmd_list_journal: open
with noexcl, so we can work on mounted filesystems)
git tree: master
git repo: https://github.com/koverstreet/bcachefs-tools.git
util-linux: 2.37.2
IMPORTANT: If someone fix this issue, please kindly add the following tag to
the commit:
Reported-by: Hongbo Li <[email protected]>
[Tue Jun 4 09:16:44 2024] bcachefs (loop1): mounting version 1.9:
disk_accounting_v2
[Tue Jun 4 09:16:44 2024] bcachefs (loop1): initializing new filesystem
[Tue Jun 4 09:16:44 2024] bcachefs (loop1): going read-write
[Tue Jun 4 09:16:44 2024] bcachefs (loop1): marking superblocks
[Tue Jun 4 09:16:44 2024] ------------[ cut here ]------------
[Tue Jun 4 09:16:44 2024] memcpy: detected field-spanning write (size 4) of
single field "r" at fs/bcachefs/disk_accounting.c:185 (size 3)
[Tue Jun 4 09:16:44 2024] WARNING: CPU: 6 PID: 11 at
Thanks, this just needs an unsafe_memcpy() annotation
commit 1c39379a4cddbf8d68c588ac9df7149840b9e4ff
Author: Kent Overstreet <[email protected]>
Date: Mon Jun 3 22:01:25 2024 -0400
fixup! bcachefs: Disk space accounting rewrite
diff --git a/fs/bcachefs/disk_accounting.c b/fs/bcachefs/disk_accounting.c
index 48d68db286a3..f118d5614e3a 100644
--- a/fs/bcachefs/disk_accounting.c
+++ b/fs/bcachefs/disk_accounting.c
@@ -182,7 +182,9 @@ static inline bool accounting_to_replicas(struct
bch_replicas_entry_v1 *r, struc
switch (acc_k.type) {
case BCH_DISK_ACCOUNTING_replicas:
- memcpy(r, &acc_k.replicas,
replicas_entry_bytes(&acc_k.replicas));
+ unsafe_memcpy(r, &acc_k.replicas,
+ replicas_entry_bytes(&acc_k.replicas),
+ "variable length struct");
I have tested it on this patch. It Looks good! Feel free to add:
Reported-by: Hongbo Li <[email protected]> and Tested-by: Hongbo Li
<[email protected]>
Thanks,
Hongbo
return true;
default:
return false;
