Skip to contentSkip to conversation

drm/amd/display: fix division by zero in get_estimated_bw()

PATCH In review Hari Mishal 3 participants 3 messages Aug 22 – Aug 28

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/amd/display: fix division by zero in get_estimated_bw()

get_estimated_bw() divides by link->dpia_bw_alloc_config.bw_granularity, which is zeroed by reset_bw_alloc_struct() and only populated once DP_TUNNELING_BW_ALLOC_CAP_CHANGED has been handled.

link_dp_dpia_handle_bw_alloc_status(), the DPCD interrupt handler, calls get_estimated_bw() whenever DP_TUNNELING_ESTIMATED_BW_CHANGED is set, independently of whether DP_TUNNELING_BW_ALLOC_CAP_CHANGED has ever fired for that link. A connected USB4/DPIA tunneling device that reports an estimated-bandwidth change before ever reporting a capability change drives a division by zero in this IRQ path.

link_dpia_send_bw_alloc_request() already guards the same bw_granularity division; add the identical guard here rather than introducing a new pattern.

Fixes 8e5cfe547bf3 ("drm/amd/display: upstream link_dp_dpia_bw.c")
Assisted-by gkh_clanker_t1000
---
 .../gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c  | 5 +++++
 1 file changed, 5 insertions(+)
drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c +5 -0
index dd854d992692..f43fc4b78a8d 100644
@@ -103,6 +103,11 @@ static int get_estimated_bw(struct dc_link *link)
 {
 	uint8_t bw_estimated_bw = 0;
 
+	if (link->dpia_bw_alloc_config.bw_granularity == 0) {
+		DC_LOG_ERROR("%s: BW granularity is zero!\n", __func__);
+		return 0;
+	}
+
 	core_link_read_dpcd(
 			link,
 			ESTIMATED_BW,
Signature · 1 line
2.43.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Uncapped request_bw is assigned to allocated_bw, causing an oversized value that leads to an integer underflow during bandwidth validation. --

--- Patch [1]: [PATCH] drm/amd/display: fix division by zero in get_estimated_bw() --- commit d3870b3b2da2897f934f4033da62b969b6443914
Author: Hari Mishal <[email protected]>

drm/amd/display: fix division by zero in get_estimated_bw()

This commit fixes a division by zero in get_estimated_bw() by returning 0 if bw_granularity is zero. This prevents a crash when a connected DPIA device reports an estimated bandwidth change before reporting a capability change.

Quoted 3 lines

link_dpia_send_bw_alloc_request() already guards the same bw_granularity division; add the identical guard here rather than introducing a new pattern.

[Severity: High]
This is a pre-existing issue, but since the commit message references link_dpia_send_bw_alloc_request(), I noticed a potential integer underflow issue involving allocated_bw.

In link_dpia_send_bw_alloc_request(), the uncapped request_bw variable is assigned to allocated_bw:

drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c:link_dpia_send_bw_alloc_request() {
    ...
		temp = req_bw * link->dpia_bw_alloc_config.bw_granularity;
		request_reg_val = temp / Kbps_TO_Gbps;
		if (temp % Kbps_TO_Gbps)
			++request_reg_val;
	}

	link->dpia_bw_alloc_config.allocated_bw = request_bw;
    ...
}

Later, in link_dpia_validate_dp_tunnel_bandwidth(), remaining_bw is calculated using unsigned 32-bit subtraction:

drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c:link_dpia_validate_dp_tunnel_bandwidth() {
    ...
			if (router_sets[j].cm_id == dp_tunnel_settings->cm_id) {
				uint32_t remaining_bw =
					dp_tunnel_settings->estimated_bw - dp_tunnel_settings->allocated_bw;
    ...
}

If a connected DPIA display requests more bandwidth than estimated, allocated_bw will be strictly greater than estimated_bw.

Could this cause the subtraction to underflow to an extremely large positive value (~4.29 billion), thereby bypassing the subsequent capacity checks (required_bw > total_bw) and allowing illegal configurations that oversubscribe the DP tunnel?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Reviewed-by on v1 1/1
#2 lore raw
In reply to Hari Mishal

Weekly promotion tests found no regression.

This patch is

Quoted 42 lines · Hari Mishal

[Some people who received this message don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

get_estimated_bw() divides by link->dpia_bw_alloc_config.bw_granularity, which is zeroed by reset_bw_alloc_struct() and only populated once DP_TUNNELING_BW_ALLOC_CAP_CHANGED has been handled.

link_dp_dpia_handle_bw_alloc_status(), the DPCD interrupt handler, calls get_estimated_bw() whenever DP_TUNNELING_ESTIMATED_BW_CHANGED is set, independently of whether DP_TUNNELING_BW_ALLOC_CAP_CHANGED has ever fired for that link. A connected USB4/DPIA tunneling device that reports an estimated-bandwidth change before ever reporting a capability change drives a division by zero in this IRQ path.

link_dpia_send_bw_alloc_request() already guards the same bw_granularity division; add the identical guard here rather than introducing a new pattern.

Fixes 8e5cfe547bf3 ("drm/amd/display: upstream link_dp_dpia_bw.c")
Assisted-by gkh_clanker_t1000
---
  .../gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c  | 5 +++++
  1 file changed, 5 insertions(+)
drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c +5 -1
index dd854d992692..f43fc4b78a8d 100644
@@ -103,6 +103,11 @@ static int get_estimated_bw(struct dc_link *link)
  {
         uint8_t bw_estimated_bw = 0;
 
+       if (link->dpia_bw_alloc_config.bw_granularity == 0) {
+               DC_LOG_ERROR("%s: BW granularity is zero!\n", __func__);
+               return 0;
+       }
+
         core_link_read_dpcd(
                         link,
                         ESTIMATED_BW,
--

2.43.0