fs/bcachefs/opts.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
Use opt->max-1 in bch2_opt_validate when option type is BCH_OPT_STR. When
option type is BCH_OPT_STR, real option's max is one less than the max
value stored in option structure.
Reported-by: syzbot+bee87a0c3291c06aa8c6@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=bee87a0c3291c06aa8c6
Fixes: 63c4b2545382 ("bcachefs: Better superblock opt validation")
Signed-off-by: Piotr Zalewski <pZ010001011111@proton.me>
---
fs/bcachefs/opts.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/fs/bcachefs/opts.c b/fs/bcachefs/opts.c
index 83f55cf99d46..bffcbe6a6fd0 100644
--- a/fs/bcachefs/opts.c
+++ b/fs/bcachefs/opts.c
@@ -290,6 +290,8 @@ static int bch2_mount_opt_lookup(const char *name)
int bch2_opt_validate(const struct bch_option *opt, u64 v, struct printbuf *err)
{
+ const u64 opt_max = opt->type == BCH_OPT_STR ? opt->max - 1 : opt->max;
+
if (v < opt->min) {
if (err)
prt_printf(err, "%s: too small (min %llu)",
@@ -297,10 +299,10 @@ int bch2_opt_validate(const struct bch_option *opt, u64 v, struct printbuf *err)
return -BCH_ERR_ERANGE_option_too_small;
}
- if (opt->max && v >= opt->max) {
+ if (opt->max && v >= opt_max) {
if (err)
prt_printf(err, "%s: too big (max %llu)",
- opt->attr.name, opt->max);
+ opt->attr.name, opt_max);
return -BCH_ERR_ERANGE_option_too_big;
}
--
2.47.0
Hi Kent, Did you see this? Best regards On Friday, November 1st, 2024 at 12:22 AM, Piotr Zalewski <pZ010001011111@proton.me> wrote: > Use opt->max-1 in bch2_opt_validate when option type is BCH_OPT_STR. When > > option type is BCH_OPT_STR, real option's max is one less than the max > value stored in option structure. > > Reported-by: syzbot+bee87a0c3291c06aa8c6@syzkaller.appspotmail.com > Closes: https://syzkaller.appspot.com/bug?extid=bee87a0c3291c06aa8c6 > Fixes: 63c4b2545382 ("bcachefs: Better superblock opt validation") > Signed-off-by: Piotr Zalewski pZ010001011111@proton.me > > --- > fs/bcachefs/opts.c | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/fs/bcachefs/opts.c b/fs/bcachefs/opts.c > index 83f55cf99d46..bffcbe6a6fd0 100644 > --- a/fs/bcachefs/opts.c > +++ b/fs/bcachefs/opts.c > @@ -290,6 +290,8 @@ static int bch2_mount_opt_lookup(const char *name) > > int bch2_opt_validate(const struct bch_option *opt, u64 v, struct printbuf *err) > { > + const u64 opt_max = opt->type == BCH_OPT_STR ? opt->max - 1 : opt->max; > > + > if (v < opt->min) { > > if (err) > prt_printf(err, "%s: too small (min %llu)", > @@ -297,10 +299,10 @@ int bch2_opt_validate(const struct bch_option *opt, u64 v, struct printbuf *err) > return -BCH_ERR_ERANGE_option_too_small; > } > > - if (opt->max && v >= opt->max) { > > + if (opt->max && v >= opt_max) { > > if (err) > prt_printf(err, "%s: too big (max %llu)", > - opt->attr.name, opt->max); > > + opt->attr.name, opt_max); > > return -BCH_ERR_ERANGE_option_too_big; > } > > -- > 2.47.0 >
On Wed, Nov 06, 2024 at 08:11:13AM +0000, Piotr Zalewski wrote: > Hi Kent, > > Did you see this? Whoops, I did miss it the first time. I think it'd be better to fix it in the OPT_STR() macro though.
Sent with Proton Mail secure email. On Wednesday, November 6th, 2024 at 6:52 PM, Kent Overstreet <kent.overstreet@linux.dev> wrote: > On Wed, Nov 06, 2024 at 08:11:13AM +0000, Piotr Zalewski wrote: > > > Hi Kent, > > > > Did you see this? > > > Whoops, I did miss it the first time. np > I think it'd be better to fix it in the OPT_STR() macro though. If changed in OPT_STR() macro it would also require a change in bch2_opt_to_text:434 (it also does -1 there). Best regards
On Wed, Nov 06, 2024 at 06:58:04PM +0000, Piotr Zalewski wrote: > > > > > > Sent with Proton Mail secure email. > > On Wednesday, November 6th, 2024 at 6:52 PM, Kent Overstreet <kent.overstreet@linux.dev> wrote: > > > On Wed, Nov 06, 2024 at 08:11:13AM +0000, Piotr Zalewski wrote: > > > > > Hi Kent, > > > > > > Did you see this? > > > > > > Whoops, I did miss it the first time. > > np > > > I think it'd be better to fix it in the OPT_STR() macro though. > > If changed in OPT_STR() macro it would also require a change in > bch2_opt_to_text:434 (it also does -1 there). *nod* I think we should do it that way then, we don't want opt->max meaning different things in different contexts if we can avoid it
© 2016 - 2024 Red Hat, Inc.