qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: David Gibson <david@gibson.dropbear.id.au>
To: Laurent Vivier <lvivier@redhat.com>
Cc: peter.maydell@linaro.org, Juan Quintela <quintela@redhat.com>,
	groug@kaod.org, qemu-devel@nongnu.org, qemu-ppc@nongnu.org,
	clg@kaod.org, "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Subject: Re: [PULL 06/28] spapr: Set VSMT to smp_threads by default
Date: Fri, 8 Nov 2019 14:26:10 +0000	[thread overview]
Message-ID: <20191108142610.GM2461@umbus.Home> (raw)
In-Reply-To: <9aebac7e-8ea3-0c16-30ca-251bcddd7699@redhat.com>

[-- Attachment #1: Type: text/plain, Size: 4462 bytes --]

On Fri, Nov 08, 2019 at 02:11:03PM +0100, Laurent Vivier wrote:
> On 24/10/2019 10:17, David Gibson wrote:
> > From: Greg Kurz <groug@kaod.org>
> > 
> > Support for setting VSMT is available in KVM since linux-4.13. Most distros
> > that support KVM on POWER already have it. It thus seem reasonable enough
> > to have the default machine to set VSMT to smp_threads.
> > 
> > This brings contiguous VCPU ids and thus brings their upper bound down to
> > the machine's max_cpus. This is especially useful for XIVE KVM devices,
> > which may thus allocate only one VP descriptor per VCPU.
> > 
> > Signed-off-by: Greg Kurz <groug@kaod.org>
> > Message-Id: <157010411885.246126.12610015369068227139.stgit@bahia.lan>
> > Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> > ---
> >  hw/ppc/spapr.c         | 7 ++++++-
> >  include/hw/ppc/spapr.h | 1 +
> >  2 files changed, 7 insertions(+), 1 deletion(-)
> > 
> > diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
> > index 4eb97d3a9b..428b834f30 100644
> > --- a/hw/ppc/spapr.c
> > +++ b/hw/ppc/spapr.c
> > @@ -2496,6 +2496,7 @@ static CPUArchId *spapr_find_cpu_slot(MachineState *ms, uint32_t id, int *idx)
> >  static void spapr_set_vsmt_mode(SpaprMachineState *spapr, Error **errp)
> >  {
> >      MachineState *ms = MACHINE(spapr);
> > +    SpaprMachineClass *smc = SPAPR_MACHINE_GET_CLASS(spapr);
> >      Error *local_err = NULL;
> >      bool vsmt_user = !!spapr->vsmt;
> >      int kvm_smt = kvmppc_smt_threads();
> > @@ -2522,7 +2523,7 @@ static void spapr_set_vsmt_mode(SpaprMachineState *spapr, Error **errp)
> >              goto out;
> >          }
> >          /* In this case, spapr->vsmt has been set by the command line */
> > -    } else {
> > +    } else if (!smc->smp_threads_vsmt) {
> >          /*
> >           * Default VSMT value is tricky, because we need it to be as
> >           * consistent as possible (for migration), but this requires
> > @@ -2531,6 +2532,8 @@ static void spapr_set_vsmt_mode(SpaprMachineState *spapr, Error **errp)
> >           * overwhelmingly common case in production systems.
> >           */
> >          spapr->vsmt = MAX(8, smp_threads);
> > +    } else {
> > +        spapr->vsmt = smp_threads;
> >      }
> >  
> >      /* KVM: If necessary, set the SMT mode: */
> > @@ -4438,6 +4441,7 @@ static void spapr_machine_class_init(ObjectClass *oc, void *data)
> >      smc->irq = &spapr_irq_dual;
> >      smc->dr_phb_enabled = true;
> >      smc->linux_pci_probe = true;
> > +    smc->smp_threads_vsmt = true;
> >  }
> >  
> >  static const TypeInfo spapr_machine_info = {
> > @@ -4505,6 +4509,7 @@ static void spapr_machine_4_1_class_options(MachineClass *mc)
> >  
> >      spapr_machine_4_2_class_options(mc);
> >      smc->linux_pci_probe = false;
> > +    smc->smp_threads_vsmt = false;
> >      compat_props_add(mc->compat_props, hw_compat_4_1, hw_compat_4_1_len);
> >      compat_props_add(mc->compat_props, compat, G_N_ELEMENTS(compat));
> >  }
> > diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
> > index cbd1a4c9f3..2009eb64f9 100644
> > --- a/include/hw/ppc/spapr.h
> > +++ b/include/hw/ppc/spapr.h
> > @@ -122,6 +122,7 @@ struct SpaprMachineClass {
> >      bool broken_host_serial_model; /* present real host info to the guest */
> >      bool pre_4_1_migration; /* don't migrate hpt-max-page-size */
> >      bool linux_pci_probe;
> > +    bool smp_threads_vsmt; /* set VSMT to smp_threads by default */
> >  
> >      void (*phb_placement)(SpaprMachineState *spapr, uint32_t index,
> >                            uint64_t *buid, hwaddr *pio, 
> > 
> 
> This patch breaks tests/migration-test on P8 host with kernel older than
> 4.3 because it tries by default to set the VSMT to 1.
> 
> qemu-system-ppc64: Failed to set KVM's VSMT mode to 1 (errno -22)
> On PPC, a VM with 1 threads/core on a host with 8 threads/core requires
> the use of VSMT mode 1.
> This KVM seems to be too old to support VSMT.
> 
> As this is clearly intentional, is there a way to fix migration-test?

Hrm.  I believe the argument for this was that the broken kernels were
old enough we didn't care.  What platform are you testing on where
you're hitting this?

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

  reply	other threads:[~2019-11-08 15:03 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-10-24  8:17 [PULL 00/28] ppc-for-4.2 queue 20191024 David Gibson
2019-10-24  8:17 ` [PULL 01/28] xive: Make some device types not user creatable David Gibson
2019-10-24  8:17 ` [PULL 02/28] xics: " David Gibson
2019-10-24  8:17 ` [PULL 03/28] target/ppc: Fix for optimized vsl/vsr instructions David Gibson
2019-10-24  8:17 ` [PULL 04/28] ppc/pnv: Improve trigger data definition David Gibson
2019-10-24  8:17 ` [PULL 05/28] ppc/pnv: Use address_space_stq_be() when triggering an interrupt from PSI David Gibson
2019-10-24  8:17 ` [PULL 06/28] spapr: Set VSMT to smp_threads by default David Gibson
2019-11-08 13:11   ` Laurent Vivier
2019-11-08 14:26     ` David Gibson [this message]
2019-11-08 15:34       ` Laurent Vivier
2019-11-08 15:42         ` Laurent Vivier
2019-10-24  8:17 ` [PULL 07/28] spapr, xics, xive: Introduce SpaprInterruptController QOM interface David Gibson
2019-10-24  8:17 ` [PULL 08/28] spapr, xics, xive: Move cpu_intc_create from SpaprIrq to SpaprInterruptController David Gibson
2019-10-24  8:17 ` [PULL 09/28] spapr, xics, xive: Move irq claim and free " David Gibson
2019-10-24  8:17 ` [PULL 10/28] spapr: Formalize notion of active interrupt controller David Gibson
2019-10-24  8:17 ` [PULL 11/28] spapr, xics, xive: Move set_irq from SpaprIrq to SpaprInterruptController David Gibson
2019-10-24  8:17 ` [PULL 12/28] spapr, xics, xive: Move print_info " David Gibson
2019-10-24  8:17 ` [PULL 13/28] spapr, xics, xive: Move dt_populate " David Gibson
2019-10-24  8:17 ` [PULL 14/28] spapr, xics, xive: Match signatures for XICS and XIVE KVM connect routines David Gibson
2019-10-24  8:18 ` [PULL 15/28] spapr: Remove SpaprIrq::init_kvm hook David Gibson
2019-10-24  8:18 ` [PULL 16/28] spapr, xics, xive: Move SpaprIrq::reset hook logic into activate/deactivate David Gibson
2019-10-24  8:18 ` [PULL 17/28] spapr, xics, xive: Move SpaprIrq::post_load hook to backends David Gibson
2019-10-24  8:18 ` [PULL 18/28] spapr: Remove SpaprIrq::nr_msis David Gibson
2019-10-24  8:18 ` [PULL 19/28] spapr: Move SpaprIrq::nr_xirqs to SpaprMachineClass David Gibson
2019-10-24  8:18 ` [PULL 20/28] pseries: Update SLOF firmware image David Gibson
2019-10-24  8:18 ` [PULL 21/28] spapr: Don't request to unplug the same core twice David Gibson
2019-10-24  8:18 ` [PULL 22/28] spapr: move CPU reset after presenter creation David Gibson
2019-10-24  8:18 ` [PULL 23/28] spapr_cpu_core: Implement DeviceClass::reset David Gibson
2019-10-24  8:18 ` [PULL 24/28] ppc/pnv: Introduce a PnvCore reset handler David Gibson
2019-10-24  8:18 ` [PULL 25/28] ppc/pnv: Add a PnvChip pointer to PnvCore David Gibson
2019-10-24  8:18 ` [PULL 26/28] ppc: Reset the interrupt presenter from the CPU reset handler David Gibson
2019-10-24  8:18 ` [PULL 27/28] ppc/pnv: Fix naming of routines realizing the CPUs David Gibson
2019-10-24  8:18 ` [PULL 28/28] spapr/xive: Set the OS CAM line at reset David Gibson
2019-10-24 16:09 ` [PULL 00/28] ppc-for-4.2 queue 20191024 Peter Maydell

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191108142610.GM2461@umbus.Home \
    --to=david@gibson.dropbear.id.au \
    --cc=clg@kaod.org \
    --cc=dgilbert@redhat.com \
    --cc=groug@kaod.org \
    --cc=lvivier@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-ppc@nongnu.org \
    --cc=quintela@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).