On 04.09.2025 05:19, Damon Ding wrote: > On 9/1/2025 6:25 PM, Marek Szyprowski wrote: >> On 01.09.2025 05:41, Damon Ding wrote: >>> On 8/29/2025 4:23 PM, Marek Szyprowski wrote: >>>> On 29.08.2025 10:08, Damon Ding wrote: >>>>> On 8/20/2025 5:20 AM, Marek Szyprowski wrote: >>>>>> On 15.08.2025 04:59, Damon Ding wrote: >>>>>>> On 2025/8/15 5:16, Marek Szyprowski wrote: >>>>>>>> On 14.08.2025 16:33, Marek Szyprowski wrote: >>>>>>>>> On 14.08.2025 12:47, Damon Ding wrote: >>>>>>>>>> PATCH 1 is a small format optimization for struct >>>>>>>>>> analogid_dp_device. >>>>>>>>>> PATCH 2 is to perform mode setting in >>>>>>>>>> &drm_bridge_funcs.atomic_enable. >>>>>>>>>> PATCH 3-6 are preparations for apply drm_bridge_connector >>>>>>>>>> helper. >>>>>>>>>> PATCH 7 is to apply the drm_bridge_connector helper. >>>>>>>>>> PATCH 8-10 are to move the panel/bridge parsing to the Analogix >>>>>>>>>> side. >>>>>>>>>> PATCH 11-12 are preparations for apply panel_bridge helper. >>>>>>>>>> PATCH 13 is to apply the panel_bridge helper. >>>>>>>>> ... >>>>>> >>>>> >>>>> Could you please provide the related DTS file for the test? I will >>>>> also try to find out the reason for this unexpected issue. ;-) >>>> >>>> Unfortunately I didn't find enough time to debug this further. The >>>> above >>>> log is from Samsung Snow Chromebook, >>>> arch/arm/boot/dts/samsung/exynos5250-snow.dts >>>> >>>> >>> >>> I compare the differences in the following display path before and >>> after this patch series: >>> >>> exynos_dp -> nxp-ptn3460 -> panel "auo,b116xw03" >>> >>> The issue is likely caused by the &drm_connector_funcs.detect() >>> related logic. Before this patch series, the nxp-ptn3460 connector is >>> always connector_status_connected because there is not available >>> &drm_connector_funcs.detect(). After it, the DRM_BRIDGE_OP_DETECT flag >>> make the connection status depend on analogix_dp_bridge_detect(). >>> >>> Could you please add the following patches additionally and try again? >>> (Not the final solution, just validation) >>> >>> diff --git a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c >>> b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c >>> index a93ff8f0a468..355911c47354 100644 >>> --- a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c >>> +++ b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c >>> @@ -1491,9 +1491,11 @@ int analogix_dp_bind(struct analogix_dp_device >>> *dp, struct drm_device *drm_dev) >>> } >>> } >>> >>> - bridge->ops = DRM_BRIDGE_OP_DETECT | >>> - DRM_BRIDGE_OP_EDID | >>> + bridge->ops = DRM_BRIDGE_OP_EDID | >>> DRM_BRIDGE_OP_MODES; >>> + if (drm_bridge_is_panel(dp->plat_data->next_bridge)) >>> + bridge->ops |= DRM_BRIDGE_OP_DETECT; >>> + >>> bridge->of_node = dp->dev->of_node; >>> bridge->type = DRM_MODE_CONNECTOR_eDP; >>> ret = devm_drm_bridge_add(dp->dev, &dp->bridge); >> >> It is better. Now the display panel is detected and reported to >> userspace, but it looks that something is not properly initialized, >> because there is garbage instead of the proper picture. >> > > I simulated the display path mentioned above on my RK3588S EVB1 Board. > To my slight surprise, it displayed normally with the reported > connector type DRM_MODE_CONNECTOR_LVDS. ;-) > > The modifications included: > 1.Synchronized the Analogix DP ralated DT configurations with Samsung > Snow Chromebook. > 2.Skipped the I2C transfers and GPIO operations in nxp-ptn3460 driver. > 3.Set the EDID to that of eDP Panel LP079QX1-SP0V forcibly. > > Additionally, I added debug logs to verify whether the functions in > ptn3460_bridge_funcs were actually called. > > Did you encounter any unexpected logs during your investigation? I'd > like to perform additional tests on this issue. :-)
Okay, I've finally went to the office and tested manually all 3 Chromebook boards witch use exynos-dp based display hardware. Previously I only did the remote tests and observed result on a webcam, which was not directed directly at the tested displays, so I only saw the glare from the display panel. The results are as follows: 1. Snow (arch/arm/boot/dts/samsung/exynos5250-snow.dts) - exynos-dp -> nxp-ptn3460 1366x768 lvds panel - works fine with the above mentioned change. 2. Peach-Pit (arch/arm/boot/dts/samsung/exynos5420-peach-pit.dts) - exynos-dp -> parade,ps8625 -> auo,b116xw03 1366x768 lvds panel - displays garbage, this was the only board I previously was able to see partially on the webcam. 3. Peach-Pi (arch/arm/boot/dts/samsung/exynos5800-peach-pi.dts) - exynos-dp -> auo,b133htn01 1920x1080 edp panel - also displays garbage. Then I found why both Peach boards displays garbage on boot - the framebuffer emulation is initialized for 1024x786 resolution, which is not handled by those panels. This is a bit strange, because the connector implemented by the panel reports proper native mode to drm and userspace. 'modetest -c' shows the right resolution. Also when I run 'modetest -s' with the panel's native resolution then the test pattern is correctly displayed on all three boards. Then I've played a bit with the analogix code and it looks that this 1024x768 mode is some kind default mode which comes from analogix_dp_bridge_edid_read() function, which has been introduced by this patchset. When I removed DRM_BRIDGE_OP_EDID flag from bridge->ops, then I got it finally working again properly on all three boards. I hope this helps fixing this issue. Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland
