[REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[John The Cooling Fan]

[-- Attachment #1: Type: text/plain, Size: 721 bytes --]

Noticed a regression with VRR after updating the kernel. My main
monitor's VRR stopped being recognized, I was no longer able to enable
it in my desktop environment settings and
/sys/kernel/debug/dri/0/DP-1/vrr_range showed a range of 0 to 0. I did
a bisect on the linux kernel git tree, and found that the commit
introducing this issue is 0159f88a99c9f5722dbe52ac42faf3446f371dea.

The monitor is connected via DisplayPort, no converters as far as I
know. The GPU in use is AMD Radeon RX 6700XT. There is another
monitor, connected over HDMI and it has VRR working in all the
versions I've tested. Attached the DisplayPort monitor's edid copied
from sysfs.

#regzbot introduced: 0159f88a99c9f5722dbe52ac42faf3446f371dea

[-- Attachment #2: main-edid.bin --]
[-- Type: application/octet-stream, Size: 256 bytes --]

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Mario Limonciello]

On 2/19/26 4:33 PM, John The Cooling Fan wrote:
> Noticed a regression with VRR after updating the kernel. My main
> monitor's VRR stopped being recognized, I was no longer able to enable
> it in my desktop environment settings and
> /sys/kernel/debug/dri/0/DP-1/vrr_range showed a range of 0 to 0. I did
> a bisect on the linux kernel git tree, and found that the commit
> introducing this issue is 0159f88a99c9f5722dbe52ac42faf3446f371dea.
> 
> The monitor is connected via DisplayPort, no converters as far as I
> know. The GPU in use is AMD Radeon RX 6700XT. There is another
> monitor, connected over HDMI and it has VRR working in all the
> versions I've tested. Attached the DisplayPort monitor's edid copied
> from sysfs.
> 
> #regzbot introduced: 0159f88a99c9f5722dbe52ac42faf3446f371dea

Can you please share your bisect log?  There was a regression from that 
commit but it was already fixed a long time back.

commit 7f2b5237e313 ("drm/amd/display: restore invalid MSA timing check 
for freesync")

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Ivan Sergeev]

[-- Attachment #1.1: Type: text/plain, Size: 426 bytes --]

On Fri Feb 20, 2026 at 5:26 AM MSK, Mario Limonciello wrote:
>
> Can you please share your bisect log?  There was a regression from that 
> commit but it was already fixed a long time back.
>
> commit 7f2b5237e313 ("drm/amd/display: restore invalid MSA timing check 
> for freesync")

Here's my bisect log. I've done the bisect using the v6.12.68 tag as
good and v6.18.8 tag as bad. I'll check the mentioned commit.

[-- Attachment #2: vrr-issue-bisect-deepened.log --]
[-- Type: text/x-log, Size: 5235 bytes --]

git bisect start
# status: waiting for both good and bad commits
# good: [90ecf7d9daa2d0167e20b2e9f6f52e01f47b8147] Linux 6.12.68
git bisect good 90ecf7d9daa2d0167e20b2e9f6f52e01f47b8147
# bad: [251aeb0f2f570db5290d0dc2f8ebf87247b00b85] Linux 6.13.6
git bisect bad 251aeb0f2f570db5290d0dc2f8ebf87247b00b85
# good: [adc218676eef25575469234709c2d87185ca223a] Linux 6.12
git bisect good adc218676eef25575469234709c2d87185ca223a
# bad: [5c71729ab92c7e710d48ed93043a2d1e35cc8d3c] net: pcs: xpcs: fix DW_VR_MII_DIG_CTRL1_2G5_EN bit being set for 1G SGMII w/o inband
git bisect bad 5c71729ab92c7e710d48ed93043a2d1e35cc8d3c
# bad: [4800575d8c0b2f354ab05ab1c4749e45e213bf73] Merge tag 'xfs-fixes-6.13-rc3' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
git bisect bad 4800575d8c0b2f354ab05ab1c4749e45e213bf73
# bad: [896d8946da97332d4dc80fa1937d8dd6b1c35ad4] Merge tag 'net-6.13-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
git bisect bad 896d8946da97332d4dc80fa1937d8dd6b1c35ad4
# bad: [31f1b55d5d7e531cd827419e5d71c19f24de161c] net :mana :Request a V2 response version for MANA_QUERY_GF_STAT
git bisect bad 31f1b55d5d7e531cd827419e5d71c19f24de161c
# good: [d1524d040b12e5bee1ee1d04ce50122a1ea35258] Merge branch 'net-fix-some-callers-of-copy_from_sockptr'
git bisect good d1524d040b12e5bee1ee1d04ce50122a1ea35258
# bad: [51ee075d698bb3e8df8287ac80447a18c723dd5a] Merge tag 'linux-can-fixes-for-6.13-20241202' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can
git bisect bad 51ee075d698bb3e8df8287ac80447a18c723dd5a
# bad: [f1cd565ce57760923d5e0fbd9e9914b415c0620a] Merge branch 'enetc-mqprio-fixes' Wei Fang sayus:
git bisect bad f1cd565ce57760923d5e0fbd9e9914b415c0620a
# good: [04f5cb48995d51deed0af71aaba1b8699511313f] Documentation: tls_offload: fix typos and grammar
git bisect good 04f5cb48995d51deed0af71aaba1b8699511313f
# bad: [d83ec4afd209a0619b32b867f45c845d4dddc2eb] Merge tag 'regulator-fix-v6.13-merge-window' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator
git bisect bad d83ec4afd209a0619b32b867f45c845d4dddc2eb
# good: [5262bcbb41d526f944a91a6a6c88dfb6fba3889b] dt-bindings: regulator: qcom-labibb-regulator: document the pmi8950 labibb regulator
git bisect good 5262bcbb41d526f944a91a6a6c88dfb6fba3889b
# good: [351f2bfe6362c663f45f5c6111f14365cfd094ab] regulator: core: Ignore unset max_uA constraints in current limit check
git bisect good 351f2bfe6362c663f45f5c6111f14365cfd094ab
# bad: [448ecd5771e255629bef0fb16c9b78c4bbd7bd56] Merge tag 'for-v6.13' of git://git.kernel.org/pub/scm/linux/kernel/git/sre/linux-power-supply
git bisect bad 448ecd5771e255629bef0fb16c9b78c4bbd7bd56
# good: [fcc79e1714e8c2b8e216dc3149812edd37884eef] Merge tag 'net-next-6.13' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next
git bisect good fcc79e1714e8c2b8e216dc3149812edd37884eef
# bad: [6f9baa9b92c2e4e28e385fa581f0511621db3f9a] Merge tag 'pm-6.13-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
git bisect bad 6f9baa9b92c2e4e28e385fa581f0511621db3f9a
# bad: [28eb75e178d389d325f1666e422bc13bbbb9804c] Merge tag 'drm-next-2024-11-21' of https://gitlab.freedesktop.org/drm/kernel
git bisect bad 28eb75e178d389d325f1666e422bc13bbbb9804c
# bad: [9bed494cd627cecab1114ce0416c9ddb3ad66064] Merge tag 'mediatek-drm-next-6.13' of https://git.kernel.org/pub/scm/linux/kernel/git/chunkuang.hu/linux into drm-next
git bisect bad 9bed494cd627cecab1114ce0416c9ddb3ad66064
# good: [c9ff14d0339a7838b71d9f196bd4244eeb6e2808] Merge tag 'drm-intel-gt-next-2024-10-23' of https://gitlab.freedesktop.org/drm/i915/kernel into drm-next
git bisect good c9ff14d0339a7838b71d9f196bd4244eeb6e2808
# bad: [30194b716bd8ff6215b4a49a088ede8de1a36d15] drm/amd/display: update fullscreen status to SPL
git bisect bad 30194b716bd8ff6215b4a49a088ede8de1a36d15
# good: [1845752b2f6a8acd17949c83e41c5aba4bd4e2bb] drm/amdgpu: Fix spelling mistake "initializtion" -> "initialization"
git bisect good 1845752b2f6a8acd17949c83e41c5aba4bd4e2bb
# bad: [f83fc3abd59c34fcc9f561db75e35bb01c07885a] drm/amdgpu: optimize fn gfx_v12_ring_insert_nop
git bisect bad f83fc3abd59c34fcc9f561db75e35bb01c07885a
# good: [82e2ccaa46811356f1daaac2127da79ecd2442d4] drm/amd/display: 3.2.304
git bisect good 82e2ccaa46811356f1daaac2127da79ecd2442d4
# bad: [04bdba46542c953e26f3ba4ac84dd65fbb4e7382] drm/amdgpu: Use drm_print_memory_stats helper from fdinfo
git bisect bad 04bdba46542c953e26f3ba4ac84dd65fbb4e7382
# bad: [2fe79508d9c393bb9931b0037c5ecaee09a8dc39] drm/amd/display: fix hibernate entry for DCN35+
git bisect bad 2fe79508d9c393bb9931b0037c5ecaee09a8dc39
# good: [9f293c4544f74a428275e6fbf9b2b62587ced7f0] drm/amd/display: switch to setting physical address directly
git bisect good 9f293c4544f74a428275e6fbf9b2b62587ced7f0
# bad: [0159f88a99c9f5722dbe52ac42faf3446f371dea] drm/amd/display: remove redundant freesync parser for DP
git bisect bad 0159f88a99c9f5722dbe52ac42faf3446f371dea
# good: [abb91c545a51feb9ea871593dd819c42644f25ad] drm/amd/display: always call connector_update when parsing freesync_caps
git bisect good abb91c545a51feb9ea871593dd819c42644f25ad
# first bad commit: [0159f88a99c9f5722dbe52ac42faf3446f371dea] drm/amd/display: remove redundant freesync parser for DP

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Ivan Sergeev]

Checked commit 7f2b5237e313, it still has the issue

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Ivan Sergeev]

On Fri Feb 20, 2026 at 5:38 AM MSK, Ivan Sergeev wrote:
>
> I've done the bisect using the v6.12.68 tag as good and v6.18.8 tag as bad

Sorry, those are not the tags I've used for the bisect. Before doing a
bisect in git I first checked versions from 6.18 to 6.13, and could
reproduce the issue in all of them, so after that I went for the git
bisect between 6.12.68 and 6.13.6.

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Mario Limonciello]

On 2/19/26 9:02 PM, Ivan Sergeev wrote:
> On Fri Feb 20, 2026 at 5:38 AM MSK, Ivan Sergeev wrote:
>>
>> I've done the bisect using the v6.12.68 tag as good and v6.18.8 tag as bad
> 
> Sorry, those are not the tags I've used for the bisect. Before doing a
> bisect in git I first checked versions from 6.18 to 6.13, and could
> reproduce the issue in all of them, so after that I went for the git
> bisect between 6.12.68 and 6.13.6.

I'm wondering if you actually have a second regression manifesting 
similarly.

Can you check out at your bad commit and apply just 7f2b5237e313 at the 
bad commit?  Confirm that fixes it.

If it does, then redo your bisect applying 7f2b5237e313 where applicable.

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Ivan Sergeev]

On Fri Feb 20, 2026 at 6:14 AM MSK, Mario Limonciello wrote:
>
> Can you check out at your bad commit and apply just 7f2b5237e313 at the 
> bad commit?  Confirm that fixes it.

Applied 7f2b5237e313 on top of 0159f88a99c9 and the issue persists.

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Melissa Wen]

[-- Attachment #1: Type: text/plain, Size: 445 bytes --]

Hi Ivan,

Thanks for reporting this issue and follow-up bisecting.
Can you check if the attached patch resolves the issue?

Thanks in advance,

Melissa Wen

On 20/02/2026 00:35, Ivan Sergeev wrote:
> On Fri Feb 20, 2026 at 6:14 AM MSK, Mario Limonciello wrote:
>> Can you check out at your bad commit and apply just 7f2b5237e313 at the
>> bad commit?  Confirm that fixes it.
> Applied 7f2b5237e313 on top of 0159f88a99c9 and the issue persists.

[-- Attachment #2: vrr-without-continuous-freq-support.patch --]
[-- Type: text/x-patch, Size: 957 bytes --]

From 5d5c4052fc64a63803f2f0acf7c2760b5b8b5218 Mon Sep 17 00:00:00 2001
From: Melissa Wen <mwen@igalia.com>
Date: Fri, 20 Feb 2026 19:38:43 -0300
Subject: [PATCH] drm/drm_edid: ignore continuous freq support for monitor range

Get the min and max refresh rate even if edid doesn't say the display
supports continuous frequencies.

Signed-off-by: Melissa Wen <mwen@igalia.com>
---
 drivers/gpu/drm/drm_edid.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c
index 26bb7710a462..17c32b33172e 100644
--- a/drivers/gpu/drm/drm_edid.c
+++ b/drivers/gpu/drm/drm_edid.c
@@ -6516,9 +6516,6 @@ static void drm_get_monitor_range(struct drm_connector *connector,
 	if (drm_edid->edid->revision < 4)
 		return;
 
-	if (!(drm_edid->edid->features & DRM_EDID_FEATURE_CONTINUOUS_FREQ))
-		return;
-
 	drm_for_each_detailed_block(drm_edid, get_monitor_range, &closure);
 
 	drm_dbg_kms(connector->dev,
-- 
2.50.1

Re: [REGRESSION] VRR not detected on a DisplayPort monitor using an AMD GPU

[Ivan Sergeev]

On Sat Feb 21, 2026 at 1:47 AM MSK, Melissa Wen wrote:
> Can you check if the attached patch resolves the issue?

The patch does resolve the issue!

There was a problem with my particular setup caused by the fact that my
greeter didn't have the VRR feature enabled, which made the desktop
environment take longer to start and the following error messages were
printed to the kernel log:

amdgpu 0000:2d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000028 SMN_C2PMSG_82:0x00000000
amdgpu 0000:2d:00.0: amdgpu: Failed to enable gfxoff!
amdgpu 0000:2d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000028 SMN_C2PMSG_82:0x00000000
amdgpu 0000:2d:00.0: amdgpu: Failed to set workload mask 0x00000001
amdgpu 0000:2d:00.0: amdgpu: (-62) failed to disable fullscreen 3D power profile mode

These errors do not appear without the patch, nor do they appear with
the patch if the greeter and the DE have VRR enabled (in my case, GDM
and GNOME Shell respectively). I have not tested the case of greeter
having VRR enabled, but disabled in DE.