Bug 153451 - [bisected] display corruption issue on Mobility Radeon HD 3650 (RV635/M86) since kernel 4.7
Summary: [bisected] display corruption issue on Mobility Radeon HD 3650 (RV635/M86) si...
Status: RESOLVED PATCH_ALREADY_AVAILABLE
Alias: None
Product: Drivers
Classification: Unclassified
Component: Video(DRI - non Intel) (show other bugs)
Hardware: x86-64 Linux
: P1 high
Assignee: drivers_video-dri
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2016-08-19 22:22 UTC by Denis Ollier
Modified: 2016-08-19 22:37 UTC (History)
1 user (show)

See Also:
Kernel Version: 4.7 and above
Subsystem:
Regression: No
Bisected commit-id:


Attachments
Photo of the corruption issue (poor quality but the image is ugly anyway) (3.84 MB, image/jpeg)
2016-08-19 22:22 UTC, Denis Ollier
Details

Description Denis Ollier 2016-08-19 22:22:00 UTC
Created attachment 229491 [details]
Photo of the corruption issue (poor quality but the image is ugly anyway)

After upgrading the kernel package to version 4.7 on ArchLinux, display output gets corrupted as soon as the radeon module is loaded on my laptop.

As a result, the display output is *ugly*: flickering, bluring, wrong color depth... (see attachment)

Involved hardware:
- ID_MODEL_FROM_DATABASE=RV635/M86 [Mobility Radeon HD 3650]
- ID_PCI_CLASS_FROM_DATABASE=Display controller
- ID_PCI_INTERFACE_FROM_DATABASE=VGA controller
- ID_PCI_SUBCLASS_FROM_DATABASE=VGA compatible controller
- ID_VENDOR_FROM_DATABASE=Advanced Micro Devices, Inc. [AMD/ATI]
- DRIVER=radeon
- MODALIAS=pci:v00001002d00009591sv00001631sd0000C223bc03sc00i00
- PCI_CLASS=30000
- PCI_ID=1002:9591
- PCI_SLOT_NAME=0000:01:00.0
- PCI_SUBSYS_ID=1631:C223
- SUBSYSTEM=pci

Running git-bisect points out the following commit as the culprit:

commit 9ef8537e68941d858924a3eacee5a1945767cbab
Author: Christian König <christian.koenig@amd.com>
Date:   Mon Jun 13 16:09:53 2016 +0200

    drm/radeon: don't use fractional dividers on RS[78]80 if SS is enabled
    
    Seems to cause problems for some older hardware. Kudos to Thom Kouwenhoven
    for working a lot with the PLLs and figuring this out.
    
    Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
    Signed-off-by: Christian König <christian.koenig@amd.com>
    Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/radeon/atombios_crtc.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c
index 2e216e2..259cd6e 100644
--- a/drivers/gpu/drm/radeon/atombios_crtc.c
+++ b/drivers/gpu/drm/radeon/atombios_crtc.c
@@ -589,7 +589,8 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc,
        if (ASIC_IS_DCE41(rdev) || ASIC_IS_DCE61(rdev) || ASIC_IS_DCE8(rdev))
            radeon_crtc->pll_flags |= RADEON_PLL_USE_FRAC_FB_DIV;
        /* use frac fb div on RS780/RS880 */
-       if ((rdev->family == CHIP_RS780) || (rdev->family == CHIP_RS880))
+       if (((rdev->family == CHIP_RS780) || (rdev->family == CHIP_RS880))
+           && !radeon_crtc->ss_enabled)
            radeon_crtc->pll_flags |= RADEON_PLL_USE_FRAC_FB_DIV;
        if (ASIC_IS_DCE32(rdev) && mode->clock > 165000)
            radeon_crtc->pll_flags |= RADEON_PLL_USE_FRAC_FB_DIV;
@@ -626,7 +627,7 @@ static u32 atombios_adjust_pll(struct drm_crtc *crtc,
            if (radeon_crtc->ss.refdiv) {
                radeon_crtc->pll_flags |= RADEON_PLL_USE_REF_DIV;
                radeon_crtc->pll_reference_div = radeon_crtc->ss.refdiv;
-               if (ASIC_IS_AVIVO(rdev))
+               if (rdev->family >= CHIP_RV770)
                    radeon_crtc->pll_flags |= RADEON_PLL_USE_FRAC_FB_DIV;
            }
        }

The change has two hunks:
- reverting only the first one doesn't change anything,
- reverting only the second one fix the corruption issue.
Comment 1 Alex Deucher 2016-08-19 22:24:19 UTC
Duplicate of:
https://bugs.freedesktop.org/show_bug.cgi?id=97099
See that bug for the fix which is already on it's way upstream.
Comment 2 Denis Ollier 2016-08-19 22:37:18 UTC
My bad, thanks for the link.

Note You need to log in before you can comment on or make changes to this bug.