Nouveau - Jun 2018 - no mouse cursor on nv50

Hi!
On v4.18-rc1, the mouse cursor is missing on my right monitor.
Card is G98 [GeForce 8400 GS Rev. 2].

I have two monitors: one small landscape 1280x1024 on DVI-I-1 left, and one
big 1600x1200 (1200x1600 portrait) on HDMI-1 right.  Curiously, the cursor
is missing not only with proper xrandr setup after logging in but even in
mirrored mode at the lightdm greeter[1].  How is this even possible if both
screens are supposed to be identical?

Bisected:
# bad: [ce397d215ccd07b8ae3f71db689aedb85d56ab40] Linux 4.18-rc1
# good: [29dcea88779c856c7dc92040a0c01233263101d4] Linux 4.17
git bisect start 'v4.18-rc1' 'v4.17'
# bad: [1c8c5a9d38f607c0b6fd12c91cbe1a4418762a21] Merge
git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
git bisect bad 1c8c5a9d38f607c0b6fd12c91cbe1a4418762a21
# bad: [135c5504a600ff9b06e321694fbcac78a9530cd4] Merge tag
'drm-next-2018-06-06-1' of git://anongit.freedesktop.org/drm/drm
git bisect bad 135c5504a600ff9b06e321694fbcac78a9530cd4
# good: [5231804cf9e584f3e7e763a0d6d2fffe011c1bce] Merge tag
'leds_for_4.18-rc1' of
git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski/linux-leds
git bisect good 5231804cf9e584f3e7e763a0d6d2fffe011c1bce
# good: [315852b422972e6ebb1dfddaadada09e46a2681a] drm: rcar-du: Fix build
failure
git bisect good 315852b422972e6ebb1dfddaadada09e46a2681a
# good: [ec064d3c6b40697fd72f4b1eeabbf293b7947a04] Merge tag
'driver-core-4.18-rc1' of
git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
git bisect good ec064d3c6b40697fd72f4b1eeabbf293b7947a04
# bad: [ce234ccc03cfee004e168a1ae4b9d0cfb1974a32] Merge tag
'drm/tegra/for-4.18-rc1' of git://anongit.freedesktop.org/tegra/linux
into drm-next
git bisect bad ce234ccc03cfee004e168a1ae4b9d0cfb1974a32
# bad: [d7c6e97a32329032ba7af1f53cab2767832fed77] drm/nouveau/kms/nv50-: modify
base allocation so the code can be split
git bisect bad d7c6e97a32329032ba7af1f53cab2767832fed77
# good: [0a5b97304b9e2cd07c78a399c5395d5fb0118341] drm/nouveau/gr/gf100-:
virtualise init_sked_hww_esr
git bisect good 0a5b97304b9e2cd07c78a399c5395d5fb0118341
# good: [18d17221dd58741a8590ba0a40a9ded82aa5d619] drm/nouveau/gr/gf100-:
virtualise r419e00
git bisect good 18d17221dd58741a8590ba0a40a9ded82aa5d619
# good: [e9d03335f604a1123b8de3103ce8e06db4ad777a] drm/nouveau/gr/gp100-: use
correct registers for zbc colour/depth setup
git bisect good e9d03335f604a1123b8de3103ce8e06db4ad777a
# good: [512fa0b8a398539c3c2db251f6c40da4ef065d09] drm/nouveau/drm/nv50-: remove
allocation of sw class
git bisect good 512fa0b8a398539c3c2db251f6c40da4ef065d09
# good: [62b290fc7b36e8fec2a370b946d7117c1899b6c1] drm/nouveau/kms/nv50-: fix
i2c-over-aux on anx9805
git bisect good 62b290fc7b36e8fec2a370b946d7117c1899b6c1
# bad: [a97c530eb968bad8d945d4f64fb550fa37a9d362] drm/nouveau/kms/nv50-: modify
overlay allocation so the code can be split
git bisect bad a97c530eb968bad8d945d4f64fb550fa37a9d362
# bad: [5bca1621c07c3ad37b5a4943450a892e18984df0] drm/nouveau/kms/nv50-: move fb
ctxdma tracking into windows
git bisect bad 5bca1621c07c3ad37b5a4943450a892e18984df0
# first bad commit: [5bca1621c07c3ad37b5a4943450a892e18984df0]
drm/nouveau/kms/nv50-: move fb ctxdma tracking into windows

Alas, the bad commit can't be easily reverted as there's a lot of
further
work on the file on question.


Meow!

[1]. It's possible to setup lightdm for proper layout even on the login
screen, but that gives only aesthetic bonus while possibly causing problems
when the layout changes.
-- 
⢀⣴⠾⠻⢶⣦⠀ There's an easy way to tell toy operating systems from real ones.
⣾⠁⢰⠒⠀⣿⡁ Just look at how their shipped fonts display U+1F52B, this makes
⢿⡄⠘⠷⠚⠋⠀ the intended audience obvious.  It's also interesting to see OSes
⠈⠳⣄⠀⠀⠀⠀ go back and forth wrt their intended target.

Lyude bisected a similar problem on a much newer GPU to this very same
change as well. [Sorry, no useful information beyond that, but thought
I'd make the connection.]

On Sun, Jun 17, 2018 at 5:46 PM, Adam Borowski <kilobyte at angband.pl>
wrote:
> Hi!
> On v4.18-rc1, the mouse cursor is missing on my right monitor.
> Card is G98 [GeForce 8400 GS Rev. 2].
>
> I have two monitors: one small landscape 1280x1024 on DVI-I-1 left, and one
> big 1600x1200 (1200x1600 portrait) on HDMI-1 right.  Curiously, the cursor
> is missing not only with proper xrandr setup after logging in but even in
> mirrored mode at the lightdm greeter[1].  How is this even possible if both
> screens are supposed to be identical?
>
> Bisected:
> # bad: [ce397d215ccd07b8ae3f71db689aedb85d56ab40] Linux 4.18-rc1
> # good: [29dcea88779c856c7dc92040a0c01233263101d4] Linux 4.17
> git bisect start 'v4.18-rc1' 'v4.17'
> # bad: [1c8c5a9d38f607c0b6fd12c91cbe1a4418762a21] Merge
git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
> git bisect bad 1c8c5a9d38f607c0b6fd12c91cbe1a4418762a21
> # bad: [135c5504a600ff9b06e321694fbcac78a9530cd4] Merge tag
'drm-next-2018-06-06-1' of git://anongit.freedesktop.org/drm/drm
> git bisect bad 135c5504a600ff9b06e321694fbcac78a9530cd4
> # good: [5231804cf9e584f3e7e763a0d6d2fffe011c1bce] Merge tag
'leds_for_4.18-rc1' of
git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski/linux-leds
> git bisect good 5231804cf9e584f3e7e763a0d6d2fffe011c1bce
> # good: [315852b422972e6ebb1dfddaadada09e46a2681a] drm: rcar-du: Fix build
failure
> git bisect good 315852b422972e6ebb1dfddaadada09e46a2681a
> # good: [ec064d3c6b40697fd72f4b1eeabbf293b7947a04] Merge tag
'driver-core-4.18-rc1' of
git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
> git bisect good ec064d3c6b40697fd72f4b1eeabbf293b7947a04
> # bad: [ce234ccc03cfee004e168a1ae4b9d0cfb1974a32] Merge tag
'drm/tegra/for-4.18-rc1' of git://anongit.freedesktop.org/tegra/linux
into drm-next
> git bisect bad ce234ccc03cfee004e168a1ae4b9d0cfb1974a32
> # bad: [d7c6e97a32329032ba7af1f53cab2767832fed77] drm/nouveau/kms/nv50-:
modify base allocation so the code can be split
> git bisect bad d7c6e97a32329032ba7af1f53cab2767832fed77
> # good: [0a5b97304b9e2cd07c78a399c5395d5fb0118341] drm/nouveau/gr/gf100-:
virtualise init_sked_hww_esr
> git bisect good 0a5b97304b9e2cd07c78a399c5395d5fb0118341
> # good: [18d17221dd58741a8590ba0a40a9ded82aa5d619] drm/nouveau/gr/gf100-:
virtualise r419e00
> git bisect good 18d17221dd58741a8590ba0a40a9ded82aa5d619
> # good: [e9d03335f604a1123b8de3103ce8e06db4ad777a] drm/nouveau/gr/gp100-:
use correct registers for zbc colour/depth setup
> git bisect good e9d03335f604a1123b8de3103ce8e06db4ad777a
> # good: [512fa0b8a398539c3c2db251f6c40da4ef065d09] drm/nouveau/drm/nv50-:
remove allocation of sw class
> git bisect good 512fa0b8a398539c3c2db251f6c40da4ef065d09
> # good: [62b290fc7b36e8fec2a370b946d7117c1899b6c1] drm/nouveau/kms/nv50-:
fix i2c-over-aux on anx9805
> git bisect good 62b290fc7b36e8fec2a370b946d7117c1899b6c1
> # bad: [a97c530eb968bad8d945d4f64fb550fa37a9d362] drm/nouveau/kms/nv50-:
modify overlay allocation so the code can be split
> git bisect bad a97c530eb968bad8d945d4f64fb550fa37a9d362
> # bad: [5bca1621c07c3ad37b5a4943450a892e18984df0] drm/nouveau/kms/nv50-:
move fb ctxdma tracking into windows
> git bisect bad 5bca1621c07c3ad37b5a4943450a892e18984df0
> # first bad commit: [5bca1621c07c3ad37b5a4943450a892e18984df0]
drm/nouveau/kms/nv50-: move fb ctxdma tracking into windows
>
> Alas, the bad commit can't be easily reverted as there's a lot of
further
> work on the file on question.
>
>
> Meow!
>
> [1]. It's possible to setup lightdm for proper layout even on the login
> screen, but that gives only aesthetic bonus while possibly causing problems
> when the layout changes.
> --
> ⢀⣴⠾⠻⢶⣦⠀ There's an easy way to tell toy operating systems from real
ones.
> ⣾⠁⢰⠒⠀⣿⡁ Just look at how their shipped fonts display U+1F52B, this makes
> ⢿⡄⠘⠷⠚⠋⠀ the intended audience obvious.  It's also interesting to see
OSes
> ⠈⠳⣄⠀⠀⠀⠀ go back and forth wrt their intended target.
> _______________________________________________
> Nouveau mailing list
> Nouveau at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/nouveau

On Mon, Jun 18, 2018 at 7:46 AM, Adam Borowski <kilobyte at angband.pl>
wrote:
> Hi!
> On v4.18-rc1, the mouse cursor is missing on my right monitor.
> Card is G98 [GeForce 8400 GS Rev. 2].
>
> I have two monitors: one small landscape 1280x1024 on DVI-I-1 left, and one
> big 1600x1200 (1200x1600 portrait) on HDMI-1 right.  Curiously, the cursor
> is missing not only with proper xrandr setup after logging in but even in
> mirrored mode at the lightdm greeter[1].  How is this even possible if both
> screens are supposed to be identical?
>
> Bisected:
> # bad: [ce397d215ccd07b8ae3f71db689aedb85d56ab40] Linux 4.18-rc1
> # good: [29dcea88779c856c7dc92040a0c01233263101d4] Linux 4.17
> git bisect start 'v4.18-rc1' 'v4.17'
> # bad: [1c8c5a9d38f607c0b6fd12c91cbe1a4418762a21] Merge
git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
> git bisect bad 1c8c5a9d38f607c0b6fd12c91cbe1a4418762a21
> # bad: [135c5504a600ff9b06e321694fbcac78a9530cd4] Merge tag
'drm-next-2018-06-06-1' of git://anongit.freedesktop.org/drm/drm
> git bisect bad 135c5504a600ff9b06e321694fbcac78a9530cd4
> # good: [5231804cf9e584f3e7e763a0d6d2fffe011c1bce] Merge tag
'leds_for_4.18-rc1' of
git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski/linux-leds
> git bisect good 5231804cf9e584f3e7e763a0d6d2fffe011c1bce
> # good: [315852b422972e6ebb1dfddaadada09e46a2681a] drm: rcar-du: Fix build
failure
> git bisect good 315852b422972e6ebb1dfddaadada09e46a2681a
> # good: [ec064d3c6b40697fd72f4b1eeabbf293b7947a04] Merge tag
'driver-core-4.18-rc1' of
git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
> git bisect good ec064d3c6b40697fd72f4b1eeabbf293b7947a04
> # bad: [ce234ccc03cfee004e168a1ae4b9d0cfb1974a32] Merge tag
'drm/tegra/for-4.18-rc1' of git://anongit.freedesktop.org/tegra/linux
into drm-next
> git bisect bad ce234ccc03cfee004e168a1ae4b9d0cfb1974a32
> # bad: [d7c6e97a32329032ba7af1f53cab2767832fed77] drm/nouveau/kms/nv50-:
modify base allocation so the code can be split
> git bisect bad d7c6e97a32329032ba7af1f53cab2767832fed77
> # good: [0a5b97304b9e2cd07c78a399c5395d5fb0118341] drm/nouveau/gr/gf100-:
virtualise init_sked_hww_esr
> git bisect good 0a5b97304b9e2cd07c78a399c5395d5fb0118341
> # good: [18d17221dd58741a8590ba0a40a9ded82aa5d619] drm/nouveau/gr/gf100-:
virtualise r419e00
> git bisect good 18d17221dd58741a8590ba0a40a9ded82aa5d619
> # good: [e9d03335f604a1123b8de3103ce8e06db4ad777a] drm/nouveau/gr/gp100-:
use correct registers for zbc colour/depth setup
> git bisect good e9d03335f604a1123b8de3103ce8e06db4ad777a
> # good: [512fa0b8a398539c3c2db251f6c40da4ef065d09] drm/nouveau/drm/nv50-:
remove allocation of sw class
> git bisect good 512fa0b8a398539c3c2db251f6c40da4ef065d09
> # good: [62b290fc7b36e8fec2a370b946d7117c1899b6c1] drm/nouveau/kms/nv50-:
fix i2c-over-aux on anx9805
> git bisect good 62b290fc7b36e8fec2a370b946d7117c1899b6c1
> # bad: [a97c530eb968bad8d945d4f64fb550fa37a9d362] drm/nouveau/kms/nv50-:
modify overlay allocation so the code can be split
> git bisect bad a97c530eb968bad8d945d4f64fb550fa37a9d362
> # bad: [5bca1621c07c3ad37b5a4943450a892e18984df0] drm/nouveau/kms/nv50-:
move fb ctxdma tracking into windows
> git bisect bad 5bca1621c07c3ad37b5a4943450a892e18984df0
> # first bad commit: [5bca1621c07c3ad37b5a4943450a892e18984df0]
drm/nouveau/kms/nv50-: move fb ctxdma tracking into windows
>
> Alas, the bad commit can't be easily reverted as there's a lot of
further
> work on the file on question.
Thanks for the report!  Can you give
https://github.com/skeggsb/linux/commit/fffdc521f313031b1745a264ec0d5c848c2475d1
a try and see if it helps the issue you're seeing?

I'm presuming gnome must either render the cursor in sw, or at lease
have some kind of fallback, as multi-head cursors are working fine
here, which is why I didn't notice the issue earlier.

Ben.
>
> Meow!
>
> [1]. It's possible to setup lightdm for proper layout even on the login
> screen, but that gives only aesthetic bonus while possibly causing problems
> when the layout changes.
> --
> ⢀⣴⠾⠻⢶⣦⠀ There's an easy way to tell toy operating systems from real
ones.
> ⣾⠁⢰⠒⠀⣿⡁ Just look at how their shipped fonts display U+1F52B, this makes
> ⢿⡄⠘⠷⠚⠋⠀ the intended audience obvious.  It's also interesting to see
OSes
> ⠈⠳⣄⠀⠀⠀⠀ go back and forth wrt their intended target.

On Mon, Jun 18, 2018 at 06:18:34PM +1000, Ben Skeggs
wrote:
> On Mon, Jun 18, 2018 at 7:46 AM, Adam Borowski <kilobyte at
angband.pl> wrote:
> > On v4.18-rc1, the mouse cursor is missing on my right monitor.
> > Card is G98 [GeForce 8400 GS Rev. 2].
> >
> > I have two monitors: one small landscape 1280x1024 on DVI-I-1 left,
and one
> > big 1600x1200 (1200x1600 portrait) on HDMI-1 right.  Curiously, the
cursor
> > is missing not only with proper xrandr setup after logging in but even
in
> > mirrored mode at the lightdm greeter[1].  How is this even possible if
both
> > screens are supposed to be identical?
> >
> > # first bad commit: [5bca1621c07c3ad37b5a4943450a892e18984df0]
drm/nouveau/kms/nv50-: move fb ctxdma tracking into windows
> >
> > Alas, the bad commit can't be easily reverted as there's a lot
of further
> > work on the file on question.
> 
> Thanks for the report!  Can you give
>
https://github.com/skeggsb/linux/commit/fffdc521f313031b1745a264ec0d5c848c2475d1
> a try and see if it helps the issue you're seeing?
Yay, it works!  Both for mirrored and extended w/ different orientation.


Meow.
-- 
⢀⣴⠾⠻⢶⣦⠀ There's an easy way to tell toy operating systems from real ones.
⣾⠁⢰⠒⠀⣿⡁ Just look at how their shipped fonts display U+1F52B, this makes
⢿⡄⠘⠷⠚⠋⠀ the intended audience obvious.  It's also interesting to see OSes
⠈⠳⣄⠀⠀⠀⠀ go back and forth wrt their intended target.