| Message ID | 20220327053139.2572-1-xiam0nd.tong@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | device: fix missing check on list iterator | expand |
On 3/26/22 22:31, Xiaomeng Tong wrote: > The bug is here: > lo = pstate->base.domain[domain->name]; > > The list iterator 'pstate' will point to a bogus position containing > HEAD if the list is empty or no element is found. This case should > be checked before any use of the iterator, otherwise it will lead > to a invalid memory access. > > To fix this bug, add an check. Use a new value 'iter' as the list > iterator, while use the old value 'pstate' as a dedicated variable > to point to the found element. > > Cc: stable@vger.kernel.org > Fixes: 9838366c1597d ("drm/nouveau/device: initial control object class, with pstate control methods") > Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com> > --- > drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c | 11 ++++++++--- > 1 file changed, 8 insertions(+), 3 deletions(-) > > diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c > index ce774579c89d..6b768635e8ba 100644 > --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c > +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c > @@ -72,7 +72,7 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) > } *args = data; > struct nvkm_clk *clk = ctrl->device->clk; > const struct nvkm_domain *domain; > - struct nvkm_pstate *pstate; > + struct nvkm_pstate *pstate = NULL, *iter; > struct nvkm_cstate *cstate; > int i = 0, j = -1; > u32 lo, hi; > @@ -103,11 +103,16 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) > return -EINVAL; > > if (args->v0.state != NVIF_CONTROL_PSTATE_ATTR_V0_STATE_CURRENT) { > - list_for_each_entry(pstate, &clk->states, head) { > - if (i++ == args->v0.state) > + list_for_each_entry(iter, &clk->states, head) { > + if (i++ == args->v0.state) { > + pstate = iter; Is iter and the assignment really necessary ? Unless I am missing something, list_for_each_entry() always assigns pos (pstate/iter), even if the list is empty. If nothing is found, pstate would be NULL at the end, so > break; > + } > } > > + if (!pstate) > + return -EINVAL; > + ... just this check should do to cover both the "not found" and "list empty" cases. Thanks, Guenter > lo = pstate->base.domain[domain->name]; > hi = lo; > list_for_each_entry(cstate, &pstate->list, head) {
On Sat, 26 Mar 2022 22:38:05 -0700, Guenter Roeck <linux@roeck-us.net> wrote: > > @@ -103,11 +103,16 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) > > return -EINVAL; > > > > if (args->v0.state != NVIF_CONTROL_PSTATE_ATTR_V0_STATE_CURRENT) { > > - list_for_each_entry(pstate, &clk->states, head) { > > - if (i++ == args->v0.state) > > + list_for_each_entry(iter, &clk->states, head) { > > + if (i++ == args->v0.state) { > > + pstate = iter; > > Is iter and the assignment really necessary ? Unless I am missing something, > list_for_each_entry() always assigns pos (pstate/iter), even if the list is > empty. If nothing is found, pstate would be NULL at the end, so the pstate will not be NULL at the end! so the assignment is necessary! #define list_for_each_entry(pos, head, member) \ for (pos = __container_of((head)->next, pos, member); \ &pos->member != (head); \ pos = __container_of(pos->member.next, pos, member)) -- Xiaomeng Tong
On 3/26/22 23:59, Xiaomeng Tong wrote: > On Sat, 26 Mar 2022 22:38:05 -0700, Guenter Roeck <linux@roeck-us.net> wrote: >>> @@ -103,11 +103,16 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) >>> return -EINVAL; >>> >>> if (args->v0.state != NVIF_CONTROL_PSTATE_ATTR_V0_STATE_CURRENT) { >>> - list_for_each_entry(pstate, &clk->states, head) { >>> - if (i++ == args->v0.state) >>> + list_for_each_entry(iter, &clk->states, head) { >>> + if (i++ == args->v0.state) { >>> + pstate = iter; >> >> Is iter and the assignment really necessary ? Unless I am missing something, >> list_for_each_entry() always assigns pos (pstate/iter), even if the list is >> empty. If nothing is found, pstate would be NULL at the end, so > > the pstate will not be NULL at the end! so the assignment is necessary! > #define list_for_each_entry(pos, head, member) \ > for (pos = __container_of((head)->next, pos, member); \ > &pos->member != (head); \ > pos = __container_of(pos->member.next, pos, member)) > Uuh, yes, you are correct. Sorry for the noise. Guenter
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c index ce774579c89d..6b768635e8ba 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c @@ -72,7 +72,7 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) } *args = data; struct nvkm_clk *clk = ctrl->device->clk; const struct nvkm_domain *domain; - struct nvkm_pstate *pstate; + struct nvkm_pstate *pstate = NULL, *iter; struct nvkm_cstate *cstate; int i = 0, j = -1; u32 lo, hi; @@ -103,11 +103,16 @@ nvkm_control_mthd_pstate_attr(struct nvkm_control *ctrl, void *data, u32 size) return -EINVAL; if (args->v0.state != NVIF_CONTROL_PSTATE_ATTR_V0_STATE_CURRENT) { - list_for_each_entry(pstate, &clk->states, head) { - if (i++ == args->v0.state) + list_for_each_entry(iter, &clk->states, head) { + if (i++ == args->v0.state) { + pstate = iter; break; + } } + if (!pstate) + return -EINVAL; + lo = pstate->base.domain[domain->name]; hi = lo; list_for_each_entry(cstate, &pstate->list, head) {
The bug is here: lo = pstate->base.domain[domain->name]; The list iterator 'pstate' will point to a bogus position containing HEAD if the list is empty or no element is found. This case should be checked before any use of the iterator, otherwise it will lead to a invalid memory access. To fix this bug, add an check. Use a new value 'iter' as the list iterator, while use the old value 'pstate' as a dedicated variable to point to the found element. Cc: stable@vger.kernel.org Fixes: 9838366c1597d ("drm/nouveau/device: initial control object class, with pstate control methods") Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com> --- drivers/gpu/drm/nouveau/nvkm/engine/device/ctrl.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-)