linux/drivers/gpu/drm/amd/display, branch v6.18

drm/amd/display: Increase EDID read retries

2025-11-26T17:32:05Z

[WHY] When monitor is still booting EDID read can fail while DPCD read is successful. In this case no EDID data will be returned, and this could happen for a while. [HOW] Increase number of attempts to read EDID in dm_helpers_read_local_edid() to 25. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4672 Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Alex Hung Signed-off-by: Mario Limonciello (AMD) Signed-off-by: Alex Hung Tested-by: Dan Wheeler Signed-off-by: Alex Deucher (cherry picked from commit a76d6f2c76c3abac519ba753e2723e6ffe8e461c) Cc: stable@vger.kernel.org

drm/amd/display: Don't change brightness for disabled connectors

2025-11-26T17:31:26Z

[WHY] When a laptop lid is closed the connector is disabled but userspace can still try to change brightness. This doesn't work because the panel is turned off. It will eventually time out, but there is a lot of stutter along the way. [How] Iterate all connectors to check whether the matching one for the backlight index is enabled. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4675 Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Ray Wu Signed-off-by: Mario Limonciello (AMD) Signed-off-by: Alex Hung Signed-off-by: Alex Deucher (cherry picked from commit f6eeab30323d1174a4cc022e769d248fe8241304) Cc: stable@vger.kernel.org

drm/amd/display: Check NULL before accessing

2025-11-26T17:17:38Z

[WHAT] IGT kms_cursor_legacy's long-nonblocking-modeset-vs-cursor-atomic fails with NULL pointer dereference. This can be reproduced with both an eDP panel and a DP monitors connected. BUG: kernel NULL pointer dereference, address: 0000000000000000 #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page PGD 0 P4D 0 Oops: Oops: 0000 [#1] SMP NOPTI CPU: 13 UID: 0 PID: 2960 Comm: kms_cursor_lega Not tainted 6.16.0-99-custom #8 PREEMPT(voluntary) Hardware name: AMD ........ RIP: 0010:dc_stream_get_scanoutpos+0x34/0x130 [amdgpu] Code: 57 4d 89 c7 41 56 49 89 ce 41 55 49 89 d5 41 54 49 89 fc 53 48 83 ec 18 48 8b 87 a0 64 00 00 48 89 75 d0 48 c7 c6 e0 41 30 c2 <48> 8b 38 48 8b 9f 68 06 00 00 e8 8d d7 fd ff 31 c0 48 81 c3 e0 02 RSP: 0018:ffffd0f3c2bd7608 EFLAGS: 00010292 RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffd0f3c2bd7668 RDX: ffffd0f3c2bd7664 RSI: ffffffffc23041e0 RDI: ffff8b32494b8000 RBP: ffffd0f3c2bd7648 R08: ffffd0f3c2bd766c R09: ffffd0f3c2bd7760 R10: ffffd0f3c2bd7820 R11: 0000000000000000 R12: ffff8b32494b8000 R13: ffffd0f3c2bd7664 R14: ffffd0f3c2bd7668 R15: ffffd0f3c2bd766c FS: 000071f631b68700(0000) GS:ffff8b399f114000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000000 CR3: 00000001b8105000 CR4: 0000000000f50ef0 PKRU: 55555554 Call Trace: dm_crtc_get_scanoutpos+0xd7/0x180 [amdgpu] amdgpu_display_get_crtc_scanoutpos+0x86/0x1c0 [amdgpu] ? __pfx_amdgpu_crtc_get_scanout_position+0x10/0x10[amdgpu] amdgpu_crtc_get_scanout_position+0x27/0x50 [amdgpu] drm_crtc_vblank_helper_get_vblank_timestamp_internal+0xf7/0x400 drm_crtc_vblank_helper_get_vblank_timestamp+0x1c/0x30 drm_crtc_get_last_vbltimestamp+0x55/0x90 drm_crtc_next_vblank_start+0x45/0xa0 drm_atomic_helper_wait_for_fences+0x81/0x1f0 ... Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Aurabindo Pillai Signed-off-by: Alex Hung Signed-off-by: Alex Deucher (cherry picked from commit 621e55f1919640acab25383362b96e65f2baea3c) Cc: stable@vger.kernel.org

Revert "drm/amd/display: Move setup_stream_attribute"

2025-11-26T17:16:58Z

This reverts commit 2681bf4ae8d24df950138b8c9ea9c271cd62e414. This results in a blank screen on the HDMI port on some systems. Revert for now so as not to regress 6.18, can be addressed in 6.19 once the issue is root caused. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4652 Cc: Sunpeng.Li@amd.com Cc: ivan.lipski@amd.com Signed-off-by: Alex Deucher (cherry picked from commit d0e9de7a81503cdde37fb2d37f1d102f9e0f38fb)

drm/amd/display: Fix pbn to kbps Conversion

2025-11-18T17:08:51Z

[Why] Existing routine has two conversion sequence, pbn_to_kbps and kbps_to_pbn with margin. Non of those has without-margin calculation. kbps_to_pbn with margin conversion includes fec overhead which has already been included in pbn_div calculation with 0.994 factor considered. It is a double counted fec overhead factor that causes potential bw loss. [How] Add without-margin calculation. Fix fec overhead double counted issue. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3735 Reviewed-by: Aurabindo Pillai Signed-off-by: Fangzhi Zuo Signed-off-by: Ivan Lipski Tested-by: Dan Wheeler Signed-off-by: Alex Deucher (cherry picked from commit e0dec00f3d05e8c0eceaaebfdca217f8d10d380c) Cc: stable@vger.kernel.org

drm/amd/display: Clear the CUR_ENABLE register on DCN20 on DPP5

2025-11-18T17:07:50Z

[Why] On DCN20 & DCN30, the 6th DPP's & HUBP's are powered on permanently and cannot be power gated. Thus, when dpp_reset() is invoked for the DPP5, while it's still powered on, the cached cursor_state (dpp_base->pos.cur0_ctl.bits.cur0_enable) and the actual state (CUR0_ENABLE) bit are unsycned. This can cause a double cursor in full screen with non-native scaling. [How] Force disable cursor on DPP5 on plane powerdown for ASICs w/ 6 DPPs/HUBPs. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4673 Reviewed-by: Aric Cyr Signed-off-by: Ivan Lipski Tested-by: Dan Wheeler Signed-off-by: Alex Deucher (cherry picked from commit 79b3c037f972dcb13e325a8eabfb8da835764e15) Cc: stable@vger.kernel.org

drm/amd/display: Add an HPD filter for HDMI

2025-11-18T17:07:07Z

[Why] Some monitors perform rapid “autoscan” HPD re‑assertions right after a disconnect or powersaving mode enablement. These appear as a quick disconnect→reconnect with an identical EDID. Since Linux has no HDMI hotplug detection (HPD) filter, these quick reconnects are seen as hotplug events, which can unintentionally wake a system with DPMS off. An example: https://gitlab.freedesktop.org/drm/amd/-/issues/2876 Such 'fake reconnects' are considered when the interval between a disconnect and a connect is within 1500ms (experimentally chosen using several monitors), and the two connections have the same EDID. [How] Implement a time-based debounce mechanism: 1. On HDMI disconnect detection, instead of immediately processing the HPD event, save the current sink and schedule delayed work (default 1500ms) 2. If another HDMI disconnect HPD event arrives during the debounce period, it reschedules the pending work, ensuring only the final state is processed. 3. When the debounce timer expires, re-detect the display and compare the new sink with the cached one using EDID comparison. 4. If sinks match (same EDID), this was a spontaneous HPD toggle: - Update connector state internally - Skip hotplug event to prevent desktop rearrangement If sinks differ, this was a real display change: - Process normally with the hotplug event The debounce delay is configurable via module parameter 'hdmi_hpd_debounce_delay_ms'. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/2876 Reviewed-by: Sun peng (Leo) Li Signed-off-by: Ivan Lipski Tested-by: Dan Wheeler Signed-off-by: Alex Deucher (cherry picked from commit c918e75e1ed95be76f8e3156a411188f650fe03f)

drm/amd/display: Increase DPCD read retries

2025-11-18T17:06:34Z

[Why] Empirical measurement of some monitors that fail to read EDID while booting shows that the number of retries with a 30ms delay between tries is as high as 16. [How] Increase number of retries to 20. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4672 Reviewed-by: Alex Hung Signed-off-by: Mario Limonciello (AMD) Signed-off-by: Ivan Lipski Tested-by: Dan Wheeler Signed-off-by: Alex Deucher (cherry picked from commit ad1c59ad7cf74ec06e32fe2c330ac1e957222288) Cc: stable@vger.kernel.org

drm/amd/display: Move sleep into each retry for retrieve_link_cap()

2025-11-18T17:03:42Z

[Why] When a monitor is booting it's possible that it isn't ready to retrieve link caps and this can lead to an EDID read failure: ``` [drm:retrieve_link_cap [amdgpu]] *ERROR* retrieve_link_cap: Read receiver caps dpcd data failed. amdgpu 0000:c5:00.0: [drm] *ERROR* No EDID read. ``` [How] Rather than msleep once and try a few times, msleep each time. Should be no changes for existing working monitors, but should correct reading caps on a monitor that is slow to boot. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4672 Reviewed-by: Alex Hung Signed-off-by: Mario Limonciello (AMD) Signed-off-by: Ivan Lipski Tested-by: Dan Wheeler Signed-off-by: Alex Deucher (cherry picked from commit 669dca37b3348a447db04bbdcbb3def94d5997cc) Cc: stable@vger.kernel.org

drm/amd/display: Prevent Gating DTBCLK before It Is Properly Latched

2025-11-18T16:52:18Z

[why] 1. With allow_0_dtb_clk enabled, the time required to latch DTBCLK to 600 MHz depends on the SMU. If DTBCLK is not latched to 600 MHz before set_mode completes, gating DTBCLK causes the DP2 sink to lose its clock source. 2. The existing DTBCLK gating sequence ungates DTBCLK based on both pix_clk and ref_dtbclk, but gates DTBCLK when either pix_clk or ref_dtbclk is zero. pix_clk can be zero outside the set_mode sequence before DTBCLK is properly latched, which can lead to DTBCLK being gated by mistake. [how] Consider both pixel_clk and ref_dtbclk when determining when it is safe to gate DTBCLK; this is more accurate. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4701 Fixes: 5949e7c4890c ("drm/amd/display: Enable Dynamic DTBCLK Switch") Reviewed-by: Charlene Liu Reviewed-by: Aurabindo Pillai Signed-off-by: Fangzhi Zuo Signed-off-by: Roman Li Tested-by: Dan Wheeler Signed-off-by: Alex Deucher (cherry picked from commit d04eb0c402780ca037b62a6aecf23b863545ebca) Cc: stable@vger.kernel.org