On Mon, Aug 19, 2024 at 02:47:54PM GMT, Lizhi Xu wrote: > When journal v2 entry nr overflow, it will cause the value of ja->nr to > be incorrect, this will result in the allocated memory to ja->buckets > being too small, leading to out of bounds access. > > Reported-by: [email protected] > Closes: https://syzkaller.appspot.com/bug?extid=47ecc948aadfb2ab3efc > Signed-off-by: Lizhi Xu <[email protected]> > --- > fs/bcachefs/journal.c | 12 +++++++++++- > 1 file changed, 11 insertions(+), 1 deletion(-) > > diff --git a/fs/bcachefs/journal.c b/fs/bcachefs/journal.c > index 13669dd0e375..d4fb5a23b3f6 100644 > --- a/fs/bcachefs/journal.c > +++ b/fs/bcachefs/journal.c > @@ -1307,8 +1307,18 @@ int bch2_dev_journal_init(struct bch_dev *ca, struct > bch_sb *sb) > if (journal_buckets_v2) { > unsigned nr = > bch2_sb_field_journal_v2_nr_entries(journal_buckets_v2); > > - for (unsigned i = 0; i < nr; i++) > + for (unsigned i = 0; i < nr; i++) { > ja->nr += le64_to_cpu(journal_buckets_v2->d[i].nr); > + if (le64_to_cpu(journal_buckets_v2->d[i].nr) > > UINT_MAX) { > + struct bch_fs *c = ca->fs; > + struct printbuf buf = PRINTBUF; > + prt_printf(&buf, "journal v2 entry d[%u].nr %lu > overflow!\n", i, > + > le64_to_cpu(journal_buckets_v2->d[i].nr)); > + bch_info(c, "%s", buf.buf); > + printbuf_exit(&buf); > + return -BCH_ERR_ENOMEM_dev_journal_init; > + } > + }
this should be done in the validation code
