Re: [libvirt] [RFC 12/42] pc: initialize legacy hotplug only for 2.6 and older machine types
On Thu, May 12, 2016 at 12:29:55PM +0200, Igor Mammedov wrote:
On Thu, 12 May 2016 10:07:03 +0300 "Michael S. Tsirkin" <mst@redhat.com> wrote:
On Wed, May 11, 2016 at 08:28:59PM -0300, Eduardo Habkost wrote:
On Thu, May 12, 2016 at 12:51:55AM +0300, Michael S. Tsirkin wrote:
On Wed, May 11, 2016 at 03:50:39PM +0200, Igor Mammedov wrote:
On Tue, 10 May 2016 17:24:14 -0300 Eduardo Habkost <ehabkost@redhat.com> wrote:
On Mon, May 02, 2016 at 02:33:21PM +0200, Igor Mammedov wrote: > on old machine types CPU hotplug was uncondtionally > enabled since it was introduced, consuming IO ports > and providing AML regardles of whether it was actually > in use or not. Keep it so for 2.6 and older machines. > > New machine types will have an option to turn CPU > hotplug on if it's needed while by default it stays > disabled not consuming extra RAM/IO resources. > > Signed-off-by: Igor Mammedov <imammedo@redhat.com>
What if people are using "-machine pc -smp N,max_cpus=M"? Shouldn't we at least warning about missing CPU hotplug support when using just "max_cpus" with no "cpu-hotplug=on" with pc-2.7 and newer? Yep, I'll add it on next respin. Would hard error better than warning?
Most people don't need cpu hotplug, attempts to hotplug fail, should be enough.
People who don't need CPU hotplug shouldn't be using the max_cpus option.
I agree.
I believe we should at least warn people (early, during initialization) that their configuration don't make sense anymore.
I think we should try to keep old command line working if we can. this patch won't break old command line for old machine types, but for new machine type users would need to fix it and be explicit if they want cpu-hotplug.
I don't think we should do that, unless users already had time to update their scripts and libvirt had time to implement code supporting the new method. I believe libvirt (and people's scripts) use maxcpus only when they want CPU hotplug, so making max_cpus > smp_cpus enable CPU hotplug implicitly would probably solve the compatibility issue. If we want to deprecate the use of maxcpus to enable CPU hotplug, then we can make it print a warning for a few releases, so people have time to update their code. -- Eduardo
On Thu, May 12, 2016 at 07:52:23 -0300, Eduardo Habkost wrote:
I don't think we should do that, unless users already had time to update their scripts and libvirt had time to implement code supporting the new method.
I believe libvirt (and people's scripts) use maxcpus only when they want CPU hotplug, so making max_cpus > smp_cpus enable CPU hotplug implicitly would probably solve the compatibility issue.
Libvirt uses maxcpus only if the configuration explicitly has less active cpus than the maximum number. This option would be the best IMO.
If we want to deprecate the use of maxcpus to enable CPU hotplug, then we can make it print a warning for a few releases, so people have time to update their code.
At that point libvirt also needs a way to detect that the new argument is supported by qemu, so we can start passing it on the command line basically every time we now pass 'maxcpus'. The warning will get most probably ignored by people using libvirt as the stdout/err of qemu is not visible to them. Peter
On Thu, 12 May 2016 13:04:23 +0200 Peter Krempa <pkrempa@redhat.com> wrote:
On Thu, May 12, 2016 at 07:52:23 -0300, Eduardo Habkost wrote:
I don't think we should do that, unless users already had time to update their scripts and libvirt had time to implement code supporting the new method.
I believe libvirt (and people's scripts) use maxcpus only when they want CPU hotplug, so making max_cpus > smp_cpus enable CPU hotplug implicitly would probably solve the compatibility issue.
Libvirt uses maxcpus only if the configuration explicitly has less active cpus than the maximum number. This option would be the best IMO.
If we want to deprecate the use of maxcpus to enable CPU hotplug, then we can make it print a warning for a few releases, so people have time to update their code.
At that point libvirt also needs a way to detect that the new argument is supported by qemu, so we can start passing it on the command line basically every time we now pass 'maxcpus'.
The warning will get most probably ignored by people using libvirt as the stdout/err of qemu is not visible to them. Ok, to make things less complicated I'll drop machine.cpu-hotplug and leave it always enabled as it used to be and as Michael suggested.
I'll drop following patches: 12, 13, 14, 20, 23 and respin series
Peter
On Thu, May 12, 2016 at 15:20:51 +0200, Igor Mammedov wrote:
On Thu, 12 May 2016 13:04:23 +0200 Peter Krempa <pkrempa@redhat.com> wrote:
On Thu, May 12, 2016 at 07:52:23 -0300, Eduardo Habkost wrote:
I don't think we should do that, unless users already had time to update their scripts and libvirt had time to implement code supporting the new method.
I believe libvirt (and people's scripts) use maxcpus only when they want CPU hotplug, so making max_cpus > smp_cpus enable CPU hotplug implicitly would probably solve the compatibility issue.
Libvirt uses maxcpus only if the configuration explicitly has less active cpus than the maximum number. This option would be the best IMO.
If we want to deprecate the use of maxcpus to enable CPU hotplug, then we can make it print a warning for a few releases, so people have time to update their code.
At that point libvirt also needs a way to detect that the new argument is supported by qemu, so we can start passing it on the command line basically every time we now pass 'maxcpus'.
The warning will get most probably ignored by people using libvirt as the stdout/err of qemu is not visible to them. Ok, to make things less complicated I'll drop machine.cpu-hotplug and leave it always enabled as it used to be and as Michael suggested.
I'll drop following patches: 12, 13, 14, 20, 23 and respin series
I actually don't mind disabling it. But I'd be glad if it was based on the max_cpus value as Eduardo suggested. Peter
On Thu, 12 May 2016 15:27:04 +0200 Peter Krempa <pkrempa@redhat.com> wrote:
On Thu, May 12, 2016 at 15:20:51 +0200, Igor Mammedov wrote:
On Thu, 12 May 2016 13:04:23 +0200 Peter Krempa <pkrempa@redhat.com> wrote:
On Thu, May 12, 2016 at 07:52:23 -0300, Eduardo Habkost wrote:
I don't think we should do that, unless users already had time to update their scripts and libvirt had time to implement code supporting the new method.
I believe libvirt (and people's scripts) use maxcpus only when they want CPU hotplug, so making max_cpus > smp_cpus enable CPU hotplug implicitly would probably solve the compatibility issue.
Libvirt uses maxcpus only if the configuration explicitly has less active cpus than the maximum number. This option would be the best IMO.
If we want to deprecate the use of maxcpus to enable CPU hotplug, then we can make it print a warning for a few releases, so people have time to update their code.
At that point libvirt also needs a way to detect that the new argument is supported by qemu, so we can start passing it on the command line basically every time we now pass 'maxcpus'.
The warning will get most probably ignored by people using libvirt as the stdout/err of qemu is not visible to them. Ok, to make things less complicated I'll drop machine.cpu-hotplug and leave it always enabled as it used to be and as Michael suggested.
I'll drop following patches: 12, 13, 14, 20, 23 and respin series
I actually don't mind disabling it. But I'd be glad if it was based on the max_cpus value as Eduardo suggested. I've already dropped it, and simplifies series quite a bit. So lets continue without disabling for now. Later we can consider disabling it if we agree on conditions when it should happen.
smp_cpus < max_cpus is not a good enough I think:
On Wed, 11 May 2016 15:50:39 +0200 Igor Mammedov <imammedo@redhat.com> wrote:
On Tue, 10 May 2016 17:24:14 -0300 Eduardo Habkost <ehabkost@redhat.com> wrote: [...]
Should max_cpus > smp_cpus automatically set cpu-hotplug=on? I'd prefer dumb explicit feature enablement, as it doesn't put any assumptions on other options and QEMU + mgmt don't have to maintain logic for implicit rules that might enable it.
and if I didn't manage to push 'device_add x86cpu' in 2.7 time frame, guess work gets a bit confusing with current cpu-add semantic, consider current:
SRC-QEMU -smp 1,maxcpus=2 cpu-add 1 DST-QEMU -smp 2,maxcpus=2
vs would be device_add:
SRC-QEMU -smp 1,maxcpus=2 device_add cpu DST-QEMU -smp 1,maxcpus=2 -device cpu
so instead of qemu/users guessing, I suggest to make it explictly enabled to get feature (which is mostly optional) or cleanly fail qemu start if confusing options are specified with a clear error message.
participants (3)
-
Eduardo Habkost -
Igor Mammedov -
Peter Krempa