Trying add smt=off disabled cores to cpupool crash xen

2023-12-01 Thread René Winther Højgaard
If I set smt=off and try to configure cpupools with credit(1) as if all cores are available, I get the following crash.   The crash happens when I try to use xl cpupool-add-cpu on the disabled HT sibling cores. Hyper-threading is enabled in the firmware, and only disabled with smt=off. So

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-01 Thread Andrew Cooper
On 01/12/2023 7:59 pm, René Winther Højgaard wrote: > If I set smt=off and try to configure cpupools with credit(1) as if > all cores are available, I get the following crash.   > > The crash happens when I try to use xl cpupool-add-cpu on the disabled > HT sibling cores. > > Hyper-threading is ena

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-03 Thread Juergen Gross
On 01.12.23 21:12, Andrew Cooper wrote: On 01/12/2023 7:59 pm, René Winther Højgaard wrote: If I set smt=off and try to configure cpupools with credit(1) as if all cores are available, I get the following crash. The crash happens when I try to use xl cpupool-add-cpu on the disabled HT sibling c

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-04 Thread Jan Beulich
On 01.12.2023 21:12, Andrew Cooper wrote: > On 01/12/2023 7:59 pm, René Winther Højgaard wrote: >> If I set smt=off and try to configure cpupools with credit(1) as if >> all cores are available, I get the following crash.   >> >> The crash happens when I try to use xl cpupool-add-cpu on the disable

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-04 Thread Juergen Gross
On 04.12.23 10:15, Jan Beulich wrote: On 01.12.2023 21:12, Andrew Cooper wrote: On 01/12/2023 7:59 pm, René Winther Højgaard wrote: If I set smt=off and try to configure cpupools with credit(1) as if all cores are available, I get the following crash. The crash happens when I try to use xl cpu

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-04 Thread Jan Beulich
On 04.12.2023 11:02, Juergen Gross wrote: > On 04.12.23 10:15, Jan Beulich wrote: >> On 01.12.2023 21:12, Andrew Cooper wrote: >>> On 01/12/2023 7:59 pm, René Winther Højgaard wrote: If I set smt=off and try to configure cpupools with credit(1) as if all cores are available, I get the fol

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-04 Thread Juergen Gross
On 04.12.23 11:13, Jan Beulich wrote: On 04.12.2023 11:02, Juergen Gross wrote: On 04.12.23 10:15, Jan Beulich wrote: On 01.12.2023 21:12, Andrew Cooper wrote: On 01/12/2023 7:59 pm, René Winther Højgaard wrote: If I set smt=off and try to configure cpupools with credit(1) as if all cores are

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-04 Thread René Winther Højgaard
Hi Juergen, Sorry for the late reply. Here are the commands I execute, it is 'xl cpupool-cpu-add pcores 4-15' that crash the system. xl cpupool-cpu-remove Pool-0 4-31 xl cpupool-create name=\"ecores\" sched=\"credit\" xl cpupool-cpu-add ecores 16-31 xl cpupool-create name=\"pcores\" sched=\"cr

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-04 Thread Juergen Gross
On 04.12.23 18:40, René Winther Højgaard wrote: Hi Juergen, Sorry for the late reply. Here are the commands I execute, it is 'xl cpupool-cpu-add pcores 4-15' that crash the system. xl cpupool-cpu-remove Pool-0 4-31 xl cpupool-create name=\"ecores\" sched=\"credit\" xl cpupool-cpu-add ecores 1

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-05 Thread René Winther Højgaard
You are right about sched-gran=core being the issue. I don't know if this is relevant, but my CPU shouldn't be able to use sched-gran=core it's asymmetric. smt=on with sched-gran=core gives me a warning that it's falling back to sched-gran=cpu, I tested smt=off with sched-gran=cpu and it works.

Re: Trying add smt=off disabled cores to cpupool crash xen

2023-12-05 Thread Jan Beulich
On 05.12.2023 18:24, René Winther Højgaard wrote: > You are right about sched-gran=core being the issue. > > I don't know if this is relevant, but my CPU shouldn't be able to use > sched-gran=core it's asymmetric. > > smt=on with sched-gran=core gives me a warning that it's falling back to > sc