[PATCH 5.4 / 5.10] firmware: arm_scmi: Harden accesses to the reset domains

[Dominique Martinet]

From: Cristian Marussi <cristian.marussi at arm.com>

[ Upstream commit e9076ffbcaed5da6c182b144ef9f6e24554af268 ]

Accessing reset domains descriptors by the index upon the SCMI drivers
requests through the SCMI reset operations interface can potentially
lead to out-of-bound violations if the SCMI driver misbehave.

Add an internal consistency check before any such domains descriptors
accesses.

Link: https://lore.kernel.org/r/20220817172731.1185305-5-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi at arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla at arm.com>
Signed-off-by: Dominique Martinet <dominique.martinet at atmark-techno.com>
---
This is the backport I promised for CVE-2022-48655[1]
[1] https://lkml.kernel.org/r/Zj4t4q_w6gqzdvhz@codewreck.org

The 'pi' variable declaration context just changed a bit
(handle->reset_priv -> ph->get_priv(ph)) but the patch is
otherwise fine as is.
(I've also checked that num_domains is properly initialized at module
init time and this part of the code hasn't changed until 5.15, so it
should be safe to use this previously unused field)

This same patch applies cleanly to both 5.4.275 and 5.10.216.

Thanks!

 drivers/firmware/arm_scmi/reset.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/arm_scmi/reset.c b/drivers/firmware/arm_scmi/reset.c
index a981a22cfe89..b8388a3b9c06 100644
--- a/drivers/firmware/arm_scmi/reset.c
+++ b/drivers/firmware/arm_scmi/reset.c
@@ -149,8 +149,12 @@ static int scmi_domain_reset(const struct scmi_handle *handle, u32 domain,
 	struct scmi_xfer *t;
 	struct scmi_msg_reset_domain_reset *dom;
 	struct scmi_reset_info *pi = handle->reset_priv;
-	struct reset_dom_info *rdom = pi->dom_info + domain;
+	struct reset_dom_info *rdom;
 
+	if (domain >= pi->num_domains)
+		return -EINVAL;
+
+	rdom = pi->dom_info + domain;
 	if (rdom->async_reset)
 		flags |= ASYNCHRONOUS_RESET;
 
-- 
2.39.2