The family and device comparisons were using memcmp(), but this could
lead to Out-of-bounds reads when the length was larger than the
buffers being compared. Since these appear to always be NUL-terminated
strings, just use strcmp() instead.
This was found with Clang under LTO:
[ 92.405851][ T1] kernel BUG at lib/string_helpers.c:980!
...
[ 92.409141][ T1] RIP: 0010:fortify_panic (fbdev.c:?)
...
[ 92.410056][ T1] ni_assign_device_routes (fbdev.c:?)
[ 92.410056][ T1] ? unittest_enter (fbdev.c:?)
[ 92.410056][ T1] ni_routes_unittest (ni_routes_test.c:?)
[ 92.410056][ T1] ? unittest_enter (fbdev.c:?)
[ 92.410056][ T1] __initstub__kmod_ni_routes_test__505_604_ni_routes_unittest6 (fbdev.c:?)
[ 92.410056][ T1] do_one_initcall (fbdev.c:?)
Cc: Ian Abbott <[email protected]>
Cc: H Hartley Sweeten <[email protected]>
Cc: Spencer E. Olson <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Lee Jones <[email protected]>
Reported-by: kernel test robot <[email protected]>
Link: https://lore.kernel.org/lkml/20220210072821.GD4074@xsang-OptiPlex-9020
Fixes: 4bb90c87abbe ("staging: comedi: add interface to ni routing table information")
Signed-off-by: Kees Cook <[email protected]>
---
drivers/comedi/drivers/ni_routes.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/drivers/comedi/drivers/ni_routes.c b/drivers/comedi/drivers/ni_routes.c
index f24eeb464eba..295a3a9ee0c9 100644
--- a/drivers/comedi/drivers/ni_routes.c
+++ b/drivers/comedi/drivers/ni_routes.c
@@ -56,8 +56,7 @@ static const u8 *ni_find_route_values(const char *device_family)
int i;
for (i = 0; ni_all_route_values[i]; ++i) {
- if (memcmp(ni_all_route_values[i]->family, device_family,
- strnlen(device_family, 30)) == 0) {
+ if (!strcmp(ni_all_route_values[i]->family, device_family)) {
rv = &ni_all_route_values[i]->register_values[0][0];
break;
}
@@ -75,8 +74,7 @@ ni_find_valid_routes(const char *board_name)
int i;
for (i = 0; ni_device_routes_list[i]; ++i) {
- if (memcmp(ni_device_routes_list[i]->device, board_name,
- strnlen(board_name, 30)) == 0) {
+ if (!strcmp(ni_device_routes_list[i]->device, board_name)) {
dr = ni_device_routes_list[i];
break;
}
--
2.30.2
On 15/02/2022 17:10, Kees Cook wrote:
> The family and device comparisons were using memcmp(), but this could
> lead to Out-of-bounds reads when the length was larger than the
> buffers being compared. Since these appear to always be NUL-terminated
> strings, just use strcmp() instead.
>
> This was found with Clang under LTO:
>
> [ 92.405851][ T1] kernel BUG at lib/string_helpers.c:980!
> ...
> [ 92.409141][ T1] RIP: 0010:fortify_panic (fbdev.c:?)
> ...
> [ 92.410056][ T1] ni_assign_device_routes (fbdev.c:?)
> [ 92.410056][ T1] ? unittest_enter (fbdev.c:?)
> [ 92.410056][ T1] ni_routes_unittest (ni_routes_test.c:?)
> [ 92.410056][ T1] ? unittest_enter (fbdev.c:?)
> [ 92.410056][ T1] __initstub__kmod_ni_routes_test__505_604_ni_routes_unittest6 (fbdev.c:?)
> [ 92.410056][ T1] do_one_initcall (fbdev.c:?)
>
> Cc: Ian Abbott <[email protected]>
> Cc: H Hartley Sweeten <[email protected]>
> Cc: Spencer E. Olson <[email protected]>
> Cc: Greg Kroah-Hartman <[email protected]>
> Cc: Masahiro Yamada <[email protected]>
> Cc: Lee Jones <[email protected]>
> Reported-by: kernel test robot <[email protected]>
> Link: https://lore.kernel.org/lkml/20220210072821.GD4074@xsang-OptiPlex-9020
> Fixes: 4bb90c87abbe ("staging: comedi: add interface to ni routing table information")
> Signed-off-by: Kees Cook <[email protected]>
> ---
> drivers/comedi/drivers/ni_routes.c | 6 ++----
> 1 file changed, 2 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/comedi/drivers/ni_routes.c b/drivers/comedi/drivers/ni_routes.c
> index f24eeb464eba..295a3a9ee0c9 100644
> --- a/drivers/comedi/drivers/ni_routes.c
> +++ b/drivers/comedi/drivers/ni_routes.c
> @@ -56,8 +56,7 @@ static const u8 *ni_find_route_values(const char *device_family)
> int i;
>
> for (i = 0; ni_all_route_values[i]; ++i) {
> - if (memcmp(ni_all_route_values[i]->family, device_family,
> - strnlen(device_family, 30)) == 0) {
> + if (!strcmp(ni_all_route_values[i]->family, device_family)) {
> rv = &ni_all_route_values[i]->register_values[0][0];
> break;
> }
> @@ -75,8 +74,7 @@ ni_find_valid_routes(const char *board_name)
> int i;
>
> for (i = 0; ni_device_routes_list[i]; ++i) {
> - if (memcmp(ni_device_routes_list[i]->device, board_name,
> - strnlen(board_name, 30)) == 0) {
> + if (!strcmp(ni_device_routes_list[i]->device, board_name)) {
> dr = ni_device_routes_list[i];
> break;
> }
Looks good, thanks! I'm not sure why the tests used memcmp() like that.
Indeed, all the strings are statically allocated and null-terminated.
Reviewed-by: Ian Abbott <[email protected]>
--
-=( Ian Abbott <[email protected]> || MEV Ltd. is a company )=-
-=( registered in England & Wales. Regd. number: 02862268. )=-
-=( Regd. addr.: S11 & 12 Building 67, Europa Business Park, )=-
-=( Bird Hall Lane, STOCKPORT, SK3 0XA, UK. || http://www.mev.co.uk )=-
On Wed, Feb 16, 2022 at 8:03 AM Ian Abbott <[email protected]> wrote:
>
> On 15/02/2022 17:10, Kees Cook wrote:
> > The family and device comparisons were using memcmp(), but this could
> > lead to Out-of-bounds reads when the length was larger than the
> > buffers being compared. Since these appear to always be NUL-terminated
> > strings, just use strcmp() instead.
> >
> > This was found with Clang under LTO:
> >
> > [ 92.405851][ T1] kernel BUG at lib/string_helpers.c:980!
> > ...
> > [ 92.409141][ T1] RIP: 0010:fortify_panic (fbdev.c:?)
> > ...
> > [ 92.410056][ T1] ni_assign_device_routes (fbdev.c:?)
> > [ 92.410056][ T1] ? unittest_enter (fbdev.c:?)
> > [ 92.410056][ T1] ni_routes_unittest (ni_routes_test.c:?)
> > [ 92.410056][ T1] ? unittest_enter (fbdev.c:?)
> > [ 92.410056][ T1] __initstub__kmod_ni_routes_test__505_604_ni_routes_unittest6 (fbdev.c:?)
> > [ 92.410056][ T1] do_one_initcall (fbdev.c:?)
> >
> > Cc: Ian Abbott <[email protected]>
> > Cc: H Hartley Sweeten <[email protected]>
> > Cc: Spencer E. Olson <[email protected]>
> > Cc: Greg Kroah-Hartman <[email protected]>
> > Cc: Masahiro Yamada <[email protected]>
> > Cc: Lee Jones <[email protected]>
> > Reported-by: kernel test robot <[email protected]>
> > Link: https://lore.kernel.org/lkml/20220210072821.GD4074@xsang-OptiPlex-9020
> > Fixes: 4bb90c87abbe ("staging: comedi: add interface to ni routing table information")
> > Signed-off-by: Kees Cook <[email protected]>
> > ---
> > drivers/comedi/drivers/ni_routes.c | 6 ++----
> > 1 file changed, 2 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/comedi/drivers/ni_routes.c b/drivers/comedi/drivers/ni_routes.c
> > index f24eeb464eba..295a3a9ee0c9 100644
> > --- a/drivers/comedi/drivers/ni_routes.c
> > +++ b/drivers/comedi/drivers/ni_routes.c
> > @@ -56,8 +56,7 @@ static const u8 *ni_find_route_values(const char *device_family)
> > int i;
> >
> > for (i = 0; ni_all_route_values[i]; ++i) {
> > - if (memcmp(ni_all_route_values[i]->family, device_family,
> > - strnlen(device_family, 30)) == 0) {
> > + if (!strcmp(ni_all_route_values[i]->family, device_family)) {
> > rv = &ni_all_route_values[i]->register_values[0][0];
> > break;
> > }
> > @@ -75,8 +74,7 @@ ni_find_valid_routes(const char *board_name)
> > int i;
> >
> > for (i = 0; ni_device_routes_list[i]; ++i) {
> > - if (memcmp(ni_device_routes_list[i]->device, board_name,
> > - strnlen(board_name, 30)) == 0) {
> > + if (!strcmp(ni_device_routes_list[i]->device, board_name)) {
> > dr = ni_device_routes_list[i];
> > break;
> > }
>
> Looks good, thanks! I'm not sure why the tests used memcmp() like that.
> Indeed, all the strings are statically allocated and null-terminated.
>
> Reviewed-by: Ian Abbott <[email protected]>
>
> --
> -=( Ian Abbott <[email protected]> || MEV Ltd. is a company )=-
> -=( registered in England & Wales. Regd. number: 02862268. )=-
> -=( Regd. addr.: S11 & 12 Building 67, Europa Business Park, )=-
> -=( Bird Hall Lane, STOCKPORT, SK3 0XA, UK. || http://www.mev.co.uk )=-
I was probably just being paranoid when I wrote that.
Spencer