[opensuse-kernel] kernel-kvmsmall missing uinput module.
Hi kernel hackers, I tried the kernel-kvmsmall in a libvirt/KVM vm with current tumbleweed. Unfortunately, resizing the screen did not work. Usually I just have to call "xrandr -s 0" after resizing and the X server will adapt. Investigating Xorg.log differences, I found that spice-vdagent seemed to be not working. I then built and inserted the uinput module for this kernel, logged out and back in again and now resizing the VM window works again. Would it be possible to enable uinput for kernel-kvmsmall? Best regards, seife -- Stefan Seyfried "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Richard Feynman -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
18.01.2020 18:13, Stefan Seyfried пишет:
Hi kernel hackers,
I tried the kernel-kvmsmall in a libvirt/KVM vm with current tumbleweed.
Thanks for the hint, I was not aware. Indeed, it is quite attractive. But why it recommends kernel-firmware? We sure do not need any firmware inside QEMU VM, do not we? -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Sat, 18 Jan 2020 16:13:51 +0100, Stefan Seyfried wrote:
Hi kernel hackers,
I tried the kernel-kvmsmall in a libvirt/KVM vm with current tumbleweed.
Unfortunately, resizing the screen did not work. Usually I just have to call "xrandr -s 0" after resizing and the X server will adapt.
Investigating Xorg.log differences, I found that spice-vdagent seemed to be not working.
I then built and inserted the uinput module for this kernel, logged out and back in again and now resizing the VM window works again.
Would it be possible to enable uinput for kernel-kvmsmall?
Sounds like a reasonable addition. Care to create a bugzilla entry? thanks, Takashi -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Sat, 18 Jan 2020 16:51:23 +0100, Andrei Borzenkov wrote:
18.01.2020 18:13, Stefan Seyfried пишет:
Hi kernel hackers,
I tried the kernel-kvmsmall in a libvirt/KVM vm with current tumbleweed.
Thanks for the hint, I was not aware. Indeed, it is quite attractive.
But why it recommends kernel-firmware? We sure do not need any firmware inside QEMU VM, do not we?
Yes, it looks superfluous. Again, care to open a bugzilla entry? thanks, Takashi -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 18. 01. 20, 16:13, Stefan Seyfried wrote:
Hi kernel hackers,
I tried the kernel-kvmsmall in a libvirt/KVM vm with current tumbleweed.
Unfortunately, resizing the screen did not work. Usually I just have to call "xrandr -s 0" after resizing and the X server will adapt.
Investigating Xorg.log differences, I found that spice-vdagent seemed to be not working.
I then built and inserted the uinput module for this kernel, logged out and back in again and now resizing the VM window works again.
Would it be possible to enable uinput for kernel-kvmsmall?
Hi, sure, could you create a bug to track this? This diff is doing the job on x86_64 (other archs seem *not* to disable CONFIG_INPUT_MISC/CONFIG_INPUT_UINPUT): diff --git a/config/x86_64/kvmsmall b/config/x86_64/kvmsmall index 60ed0e50aaa..2a9a4a8e53c 100644 --- a/config/x86_64/kvmsmall +++ b/config/x86_64/kvmsmall @@ -302,12 +302,28 @@ CONFIG_HW_RANDOM_VIRTIO=y # CONFIG_IGC is not set # CONFIG_IIO is not set # CONFIG_INFINIBAND is not set +# CONFIG_INPUT_AD714X is not set +# CONFIG_INPUT_ADXL34X is not set +# CONFIG_INPUT_ATI_REMOTE2 is not set +# CONFIG_INPUT_ATLAS_BTNS is not set +# CONFIG_INPUT_BMA150 is not set +# CONFIG_INPUT_CM109 is not set +# CONFIG_INPUT_CMA3000 is not set +# CONFIG_INPUT_DRV2665_HAPTICS is not set +# CONFIG_INPUT_DRV2667_HAPTICS is not set +# CONFIG_INPUT_IDEAPAD_SLIDEBAR is not set # CONFIG_INPUT_JOYSTICK is not set -# CONFIG_INPUT_MISC is not set +# CONFIG_INPUT_KEYSPAN_REMOTE is not set +# CONFIG_INPUT_KXTJ9 is not set +# CONFIG_INPUT_MMA8450 is not set # CONFIG_INPUT_MOUSEDEV_PSAUX is not set +# CONFIG_INPUT_PCF8574 is not set +# CONFIG_INPUT_PCSPKR is not set +# CONFIG_INPUT_POWERMATE is not set # CONFIG_INPUT_SPARSEKMAP is not set # CONFIG_INPUT_TABLET is not set # CONFIG_INPUT_TOUCHSCREEN is not set +# CONFIG_INPUT_YEALINK is not set # CONFIG_INT340X_THERMAL is not set # CONFIG_INTEL_IDMA64 is not set # CONFIG_INTEL_IOATDMA is not set thanks, -- js suse labs -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Hi Jiri, Am 20.01.20 um 11:07 schrieb Jiri Slaby:
On 18. 01. 20, 16:13, Stefan Seyfried wrote:
Would it be possible to enable uinput for kernel-kvmsmall?
Hi,
sure, could you create a bug to track this?
Of course. https://bugzilla.opensuse.org/show_bug.cgi?id=1161305
This diff is doing the job on x86_64 (other archs seem *not* to disable> CONFIG_INPUT_MISC/CONFIG_INPUT_UINPUT):
I can't check easily right now (the test VM is at home and I'm at work), but my diff (after enabling UINPUT with "make menuconfig O=/tmp/build") was a bit bigger, especially as yours is missing "CONFIG_INPUT_UINPUT=m" completely :-) Thanks, Stefan
diff --git a/config/x86_64/kvmsmall b/config/x86_64/kvmsmall index 60ed0e50aaa..2a9a4a8e53c 100644 --- a/config/x86_64/kvmsmall +++ b/config/x86_64/kvmsmall @@ -302,12 +302,28 @@ CONFIG_HW_RANDOM_VIRTIO=y # CONFIG_IGC is not set # CONFIG_IIO is not set # CONFIG_INFINIBAND is not set +# CONFIG_INPUT_AD714X is not set +# CONFIG_INPUT_ADXL34X is not set +# CONFIG_INPUT_ATI_REMOTE2 is not set +# CONFIG_INPUT_ATLAS_BTNS is not set +# CONFIG_INPUT_BMA150 is not set +# CONFIG_INPUT_CM109 is not set +# CONFIG_INPUT_CMA3000 is not set +# CONFIG_INPUT_DRV2665_HAPTICS is not set +# CONFIG_INPUT_DRV2667_HAPTICS is not set +# CONFIG_INPUT_IDEAPAD_SLIDEBAR is not set # CONFIG_INPUT_JOYSTICK is not set -# CONFIG_INPUT_MISC is not set +# CONFIG_INPUT_KEYSPAN_REMOTE is not set +# CONFIG_INPUT_KXTJ9 is not set +# CONFIG_INPUT_MMA8450 is not set # CONFIG_INPUT_MOUSEDEV_PSAUX is not set +# CONFIG_INPUT_PCF8574 is not set +# CONFIG_INPUT_PCSPKR is not set +# CONFIG_INPUT_POWERMATE is not set # CONFIG_INPUT_SPARSEKMAP is not set # CONFIG_INPUT_TABLET is not set # CONFIG_INPUT_TOUCHSCREEN is not set +# CONFIG_INPUT_YEALINK is not set # CONFIG_INT340X_THERMAL is not set # CONFIG_INTEL_IDMA64 is not set # CONFIG_INTEL_IOATDMA is not set
thanks, -- Stefan Seyfried
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Richard Feynman -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On 20. 01. 20, 11:23, Stefan Seyfried wrote:
This diff is doing the job on x86_64 (other archs seem *not* to disable> CONFIG_INPUT_MISC/CONFIG_INPUT_UINPUT):
I can't check easily right now (the test VM is at home and I'm at work), but my diff (after enabling UINPUT with "make menuconfig O=/tmp/build") was a bit bigger, especially as yours is missing "CONFIG_INPUT_UINPUT=m" completely :-)
Hi, that's intentional: CONFIG_INPUT_UINPUT depends on CONFIG_INPUT_MISC. We switch the latter from =n to inherited (from -default) =y here. So with CONFIG_INPUT_MISC=y now, CONFIG_INPUT_UINPUT appears, but as -default has =m for that option, -kvmsmall inherits that and we don't need to specify it in -kvmsmall. You have to look at the merged configs during build to actually see the result :). thanks, -- js suse labs -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Jan 20, 2020 at 11:23:35AM +0100, Stefan Seyfried wrote:
Hi Jiri,
Am 20.01.20 um 11:07 schrieb Jiri Slaby:
On 18. 01. 20, 16:13, Stefan Seyfried wrote:
Would it be possible to enable uinput for kernel-kvmsmall?
Hi,
sure, could you create a bug to track this?
Of course. https://bugzilla.opensuse.org/show_bug.cgi?id=1161305
This diff is doing the job on x86_64 (other archs seem *not* to disable> CONFIG_INPUT_MISC/CONFIG_INPUT_UINPUT):
I can't check easily right now (the test VM is at home and I'm at work), but my diff (after enabling UINPUT with "make menuconfig O=/tmp/build") was a bit bigger, especially as yours is missing "CONFIG_INPUT_UINPUT=m" completely :-) It's a fragment config recording only differences from default config. uinput is enabled in default config so it won't appear in kvmsmall config.
Thanks Michal -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Am 20.01.20 um 13:41 schrieb Michal Suchánek:
It's a fragment config recording only differences from default config. uinput is enabled in default config so it won't appear in kvmsmall config.
I see, it's black magic, But I trust you to get it right :-) Thanks, this will make kernel-kvmsmall much more useful for the occasional "home user", who wants to have a graphical interface to his VMs and still save some space. -- Stefan Seyfried "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Richard Feynman -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Jan 20, 2020 at 11:01:11AM +0100, Takashi Iwai wrote:
On Sat, 18 Jan 2020 16:51:23 +0100, Andrei Borzenkov wrote:
18.01.2020 18:13, Stefan Seyfried пишет:
Hi kernel hackers,
I tried the kernel-kvmsmall in a libvirt/KVM vm with current tumbleweed.
Thanks for the hint, I was not aware. Indeed, it is quite attractive.
But why it recommends kernel-firmware? We sure do not need any firmware inside QEMU VM, do not we?
Yes, it looks superfluous. Again, care to open a bugzilla entry?
Do we need firmware in kernel-azure? Thanks Michal -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
20.01.2020 13:01, Takashi Iwai пишет:
On Sat, 18 Jan 2020 16:51:23 +0100, Andrei Borzenkov wrote:
18.01.2020 18:13, Stefan Seyfried пишет:
Hi kernel hackers,
I tried the kernel-kvmsmall in a libvirt/KVM vm with current tumbleweed.
Thanks for the hint, I was not aware. Indeed, it is quite attractive.
But why it recommends kernel-firmware? We sure do not need any firmware inside QEMU VM, do not we?
Yes, it looks superfluous. Again, care to open a bugzilla entry?
https://bugzilla.opensuse.org/show_bug.cgi?id=1161360 -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Jan 20, 2020 at 01:46:15PM +0100, Stefan Seyfried wrote:
Am 20.01.20 um 13:41 schrieb Michal Suchánek:
It's a fragment config recording only differences from default config. uinput is enabled in default config so it won't appear in kvmsmall config.
I see, it's black magic, But I trust you to get it right :-)
Thanks, this will make kernel-kvmsmall much more useful for the occasional "home user", who wants to have a graphical interface to his VMs and still save some space.
One word of warning before this gets out of hand and -kvmsmall flavor enters the dark path of abuse like kernel-default-base subpackage before it: the -kvmsmall flavor was *not* designed for "occasional home user" or, actually, any user. Its sole purpose is to serve as a flavor which builds as quickly as possible so that developers could quickly test their core fixes in a VM (or bisect regressions). Originally, we didn't even plan to build and publish it in BuildService. In other words, the -kvmsmall flavor is not intended for practical use and therefore "it would make it more useful for practical use" is not a valid justification for adding more stuff to it. Michal -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Hi Michal, Am 20.01.20 um 20:06 schrieb Michal Kubecek:
One word of warning before this gets out of hand and -kvmsmall flavor enters the dark path of abuse like kernel-default-base subpackage before it: the -kvmsmall flavor was *not* designed for "occasional home user" or, actually, any user. Its sole purpose is to serve as a flavor which builds as quickly as possible so that developers could quickly test their core fixes in a VM (or bisect regressions). Originally, we didn't even plan to build and publish it in BuildService.
Good point. This should probably be somewhere in the package %description. Oh! It is in the description. Which -- of course -- I did not read before ;-) I probably just judged from the name that it might be suitable for KVM use. I thought of it as "something like kernel-xen or kernel-ec2, but for KVM".
In other words, the -kvmsmall flavor is not intended for practical use and therefore "it would make it more useful for practical use" is not a valid justification for adding more stuff to it.
And actually, probably due to compressed modules(?), the size difference (168MB vs 92MB) in Factory is much less significant than it is in Leap 15.1 (where -default is 311MB vs -kvmsmall 117MB), so it is probably not as useful (size-wise) as I thought it would be. Thanks, seife -- Stefan Seyfried "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Richard Feynman -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Jan 20, 2020 at 08:06:54PM +0100, Michal Kubecek wrote:
On Mon, Jan 20, 2020 at 01:46:15PM +0100, Stefan Seyfried wrote:
Am 20.01.20 um 13:41 schrieb Michal Suchánek:
It's a fragment config recording only differences from default config. uinput is enabled in default config so it won't appear in kvmsmall config.
I see, it's black magic, But I trust you to get it right :-)
Thanks, this will make kernel-kvmsmall much more useful for the occasional "home user", who wants to have a graphical interface to his VMs and still save some space.
One word of warning before this gets out of hand and -kvmsmall flavor enters the dark path of abuse like kernel-default-base subpackage before it: the -kvmsmall flavor was *not* designed for "occasional home user" or, actually, any user. Its sole purpose is to serve as a flavor which builds as quickly as possible so that developers could quickly test their core fixes in a VM (or bisect regressions). Originally, we didn't even plan to build and publish it in BuildService.
In other words, the -kvmsmall flavor is not intended for practical use and therefore "it would make it more useful for practical use" is not a valid justification for adding more stuff to it.
That said, on x86 kvmsmall includes input support anyway (ie for power button and the virtual keyboard) so adding uinput is not something that deviates from the initial purpose. On powerpc input is not otherwise needed so adding it to support running virtual desktop is kind of overkill. Thanks Michal -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
participants (6)
-
Andrei Borzenkov -
Jiri Slaby -
Michal Kubecek -
Michal Suchánek -
Stefan Seyfried -
Takashi Iwai