Re: [PATCH RFC V6 02/24] hw/core, qemu-options.hx: Introduce 'disabledcpus' SMP parameter

[Igor Mammedov <imammedo@redhat.com>]

On Thu, 9 Oct 2025 11:28:40 +0000
Miguel Luis <miguel.luis@oracle.com> wrote:

> Hi Salil,
> 
> > On 1 Oct 2025, at 01:01, salil.mehta@opnsrc.net wrote:
> > 
> > From: Salil Mehta <salil.mehta@huawei.com>
> > 
> > Add support for a new SMP configuration parameter, 'disabledcpus', which
> > specifies the number of additional CPUs that are present in the virtual
> > machine but administratively disabled at boot. These CPUs are visible in
> > firmware (e.g. ACPI tables) yet unavailable to the guest until explicitly
> > enabled via QMP/HMP, or via the 'device_set' API (introduced in later
> > patches).
> > 
> > This feature is intended for architectures that lack native CPU hotplug
> > support but can change the administrative power state of present CPUs.
> > It allows simulating CPU hot-add–like scenarios while all CPUs remain
> > physically present in the topology at boot time.
> > 
> > Note: ARM is the first architecture to support this concept.
> > 
> > Changes include:
> > - Extend CpuTopology with a 'disabledcpus' field.
> > - Update machine_parse_smp_config() to account for disabled CPUs when
> >   computing 'cpus' and 'maxcpus'.
> > - Update SMPConfiguration in QAPI to accept 'disabledcpus'.
> > - Extend -smp option documentation to describe 'disabledcpus' usage and
> >   behavior.
> >   
> 
> Specifying a new parameter for the user seems unnecessary when the system could
> infer the number of present and disabled from (maxcpus - cpus) and those this
> patch calls "disabledcpus" could be obtained this way.
> 
> Naming is hard although it is of my opinion that we shouldn't be
> calling 'disabledcpus' here; I understand that gets carried by previous
> administrative power state meanings but machine-smp level being at a different
> abstraction level the administrative power state concept could be
> decoupled from machine-smp realm.


> My suggestion would be calling those cpus 'inactive' and not carry previous
> patch's nomenclature.
> 
> CPUs in 'inactive' state are still present in the virtual machine although this
> pre-condition may require post actions like being explicitly 'enabled'/active via
> [QH]MP.
> 
> Overall, I believe the above should be all it takes to simplify acommodation of
> CPUs not to be brought online at boot time within this patch's context.

See my reply to cover letter, we shouldn't touch -smp at all.
And make -device/device_add work with virt/arm cpus instead,
that solves a number of issues that this and -deviceset patches introduce. 

> Thanks
> Miguel
> 
> 
> > Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
> > ---
> > hw/core/machine-smp.c | 24 +++++++-----
> > include/hw/boards.h   |  2 +
> > qapi/machine.json     |  3 ++
> > qemu-options.hx       | 86 +++++++++++++++++++++++++++++++++----------
> > system/vl.c           |  3 ++
> > 5 files changed, 89 insertions(+), 29 deletions(-)
> > 
> > diff --git a/hw/core/machine-smp.c b/hw/core/machine-smp.c
> > index 0be0ac044c..c1a09fdc3f 100644
> > --- a/hw/core/machine-smp.c
> > +++ b/hw/core/machine-smp.c
> > @@ -87,6 +87,7 @@ void machine_parse_smp_config(MachineState *ms,
> > {
> >     MachineClass *mc = MACHINE_GET_CLASS(ms);
> >     unsigned cpus    = config->has_cpus ? config->cpus : 0;
> > +    unsigned disabledcpus = config->has_disabledcpus ? config->disabledcpus : 0;
> >     unsigned drawers = config->has_drawers ? config->drawers : 0;
> >     unsigned books   = config->has_books ? config->books : 0;
> >     unsigned sockets = config->has_sockets ? config->sockets : 0;
> > @@ -166,8 +167,13 @@ void machine_parse_smp_config(MachineState *ms,
> >         sockets = sockets > 0 ? sockets : 1;
> >         cores = cores > 0 ? cores : 1;
> >         threads = threads > 0 ? threads : 1;
> > +
> > +        maxcpus = drawers * books * sockets * dies * clusters *
> > +                    modules * cores * threads;
> > +        cpus = maxcpus - disabledcpus;
> >     } else {
> > -        maxcpus = maxcpus > 0 ? maxcpus : cpus;
> > +        maxcpus = maxcpus > 0 ? maxcpus : cpus + disabledcpus;
> > +        cpus = cpus > 0 ? cpus : maxcpus - disabledcpus;
> > 
> >         if (mc->smp_props.prefer_sockets) {
> >             /* prefer sockets over cores before 6.2 */
> > @@ -207,12 +213,8 @@ void machine_parse_smp_config(MachineState *ms,
> >         }
> >     }
> > 
> > -    total_cpus = drawers * books * sockets * dies *
> > -                 clusters * modules * cores * threads;
> > -    maxcpus = maxcpus > 0 ? maxcpus : total_cpus;
> > -    cpus = cpus > 0 ? cpus : maxcpus;
> > -
> >     ms->smp.cpus = cpus;
> > +    ms->smp.disabledcpus = disabledcpus;
> >     ms->smp.drawers = drawers;
> >     ms->smp.books = books;
> >     ms->smp.sockets = sockets;
> > @@ -226,6 +228,8 @@ void machine_parse_smp_config(MachineState *ms,
> >     mc->smp_props.has_clusters = config->has_clusters;
> > 
> >     /* sanity-check of the computed topology */
> > +    total_cpus = maxcpus = drawers * books * sockets * dies * clusters *
> > +                modules * cores * threads;
> >     if (total_cpus != maxcpus) {
> >         g_autofree char *topo_msg = cpu_hierarchy_to_string(ms);
> >         error_setg(errp, "Invalid CPU topology: "
> > @@ -235,12 +239,12 @@ void machine_parse_smp_config(MachineState *ms,
> >         return;
> >     }
> > 
> > -    if (maxcpus < cpus) {
> > +    if (maxcpus < (cpus + disabledcpus)) {
> >         g_autofree char *topo_msg = cpu_hierarchy_to_string(ms);
> >         error_setg(errp, "Invalid CPU topology: "
> > -                   "maxcpus must be equal to or greater than smp: "
> > -                   "%s == maxcpus (%u) < smp_cpus (%u)",
> > -                   topo_msg, maxcpus, cpus);
> > +                   "maxcpus must be equal to or greater than smp[+disabledcpus]:"
> > +                   "%s == maxcpus (%u) < smp_cpus (%u) [+ offline cpus (%u)]",
> > +                   topo_msg, maxcpus, cpus, disabledcpus);
> >         return;
> >     }
> > 
> > diff --git a/include/hw/boards.h b/include/hw/boards.h
> > index f94713e6e2..2b182d7817 100644
> > --- a/include/hw/boards.h
> > +++ b/include/hw/boards.h
> > @@ -361,6 +361,7 @@ typedef struct DeviceMemoryState {
> > /**
> >  * CpuTopology:
> >  * @cpus: the number of present logical processors on the machine
> > + * @disabledcpus: the number additional present but admin disabled cpus
> >  * @drawers: the number of drawers on the machine
> >  * @books: the number of books in one drawer
> >  * @sockets: the number of sockets in one book
> > @@ -373,6 +374,7 @@ typedef struct DeviceMemoryState {
> >  */
> > typedef struct CpuTopology {
> >     unsigned int cpus;
> > +    unsigned int disabledcpus;
> >     unsigned int drawers;
> >     unsigned int books;
> >     unsigned int sockets;
> > diff --git a/qapi/machine.json b/qapi/machine.json
> > index 038eab281c..e45740da33 100644
> > --- a/qapi/machine.json
> > +++ b/qapi/machine.json
> > @@ -1634,6 +1634,8 @@
> > #
> > # @cpus: number of virtual CPUs in the virtual machine
> > #
> > +# @disabledcpus: number of additional present but disabled(or offline) CPUs
> > +#
> > # @maxcpus: maximum number of hotpluggable virtual CPUs in the virtual
> > #     machine
> > #
> > @@ -1657,6 +1659,7 @@
> > ##
> > { 'struct': 'SMPConfiguration', 'data': {
> >      '*cpus': 'int',
> > +     '*disabledcpus': 'int',
> >      '*drawers': 'int',
> >      '*books': 'int',
> >      '*sockets': 'int',
> > diff --git a/qemu-options.hx b/qemu-options.hx
> > index ab23f14d21..83ccde341b 100644
> > --- a/qemu-options.hx
> > +++ b/qemu-options.hx
> > @@ -326,12 +326,15 @@ SRST
> > ERST
> > 
> > DEF("smp", HAS_ARG, QEMU_OPTION_smp,
> > -    "-smp [[cpus=]n][,maxcpus=maxcpus][,drawers=drawers][,books=books][,sockets=sockets]\n"
> > -    "               [,dies=dies][,clusters=clusters][,modules=modules][,cores=cores]\n"
> > -    "               [,threads=threads]\n"
> > -    "                set the number of initial CPUs to 'n' [default=1]\n"
> > -    "                maxcpus= maximum number of total CPUs, including\n"
> > -    "                offline CPUs for hotplug, etc\n"
> > +    "-smp [[cpus=]n][,disabledcpus=disabledcpus][,maxcpus=maxcpus][,drawers=drawers][,books=books]\n"
> > +    "               [,sockets=sockets][,dies=dies][,clusters=clusters][,modules=modules]\n"
> > +    "               [,cores=cores][,threads=threads]\n"
> > +    "                set the initial number of CPUs present and\n"
> > +    "                  administratively enabled at boot time to 'n' [default=1]\n"
> > +    "                disabledcpus= number of present but administratively\n"
> > +    "                  disabled CPUs (unavailable to the guest at boot)\n"
> > +    "                maxcpus= maximum total CPUs (present + hotpluggable)\n"
> > +    "                  on machines without CPU hotplug, defaults to n + disabledcpus\n"
> >     "                drawers= number of drawers on the machine board\n"
> >     "                books= number of books in one drawer\n"
> >     "                sockets= number of sockets in one book\n"
> > @@ -351,22 +354,49 @@ DEF("smp", HAS_ARG, QEMU_OPTION_smp,
> >     "      For a particular machine type board, an expected CPU topology hierarchy\n"
> >     "      can be defined through the supported sub-option. Unsupported parameters\n"
> >     "      can also be provided in addition to the sub-option, but their values\n"
> > -    "      must be set as 1 in the purpose of correct parsing.\n",
> > +    "      must be set as 1 in the purpose of correct parsing.\n"
> > +    "                                                          \n"
> > +    "      Administratively disabled CPUs: Some machine types do not support vCPU\n"
> > +    "      hotplug but their CPUs can be marked disabled (powered off) and kept\n"
> > +    "      unavailable to the guest. Later, such CPUs can be enabled via QMP/HMP\n"
> > +    "      (e.g., 'device_set ... admin-state=enable'). This is similar to hotplug,\n"
> > +    "      except all disabled CPUs are already present at boot. Useful on\n"
> > +    "      architectures that lack architectural CPU hotplug.\n",
> >     QEMU_ARCH_ALL)
> > SRST
> > -``-smp [[cpus=]n][,maxcpus=maxcpus][,drawers=drawers][,books=books][,sockets=sockets][,dies=dies][,clusters=clusters][,modules=modules][,cores=cores][,threads=threads]``
> > -    Simulate a SMP system with '\ ``n``\ ' CPUs initially present on
> > -    the machine type board. On boards supporting CPU hotplug, the optional
> > -    '\ ``maxcpus``\ ' parameter can be set to enable further CPUs to be
> > -    added at runtime. When both parameters are omitted, the maximum number
> > +``-smp [[cpus=]n][,disabledcpus=disabledcpus][,maxcpus=maxcpus][,drawers=drawers][,books=books][,sockets=sockets][,dies=dies][,clusters=clusters][,modules=modules][,cores=cores][,threads=threads]``
> > +    Simulate a SMP system with '\ ``n``\ ' CPUs initially present & enabled on
> > +    the machine type board. Furthermore, on architectures that support changing
> > +    the administrative power state of CPUs, optional '\ ``disabledcpus``\ '
> > +    parameter specifies *additional* CPUs that are present in firmware (e.g.,
> > +    ACPI) but are administratively disabled (i.e., not usable by the guest at
> > +    boot time).
> > +
> > +    This is different from CPU hotplug where additional CPUs are not even
> > +    present in the system description. Administratively disabled CPUs appear in
> > +    ACPI tables i.e. are provisioned, but cannot be used until explicitly
> > +    enabled via QMP/HMP or the deviceset API.
> > +
> > +    On boards supporting CPU hotplug, the optional '\ ``maxcpus``\ ' parameter
> > +    can be set to enable further CPUs to be added at runtime. When both
> > +    '\ ``n``\ ' & '\ ``maxcpus``\ ' parameters are omitted, the maximum number
> >     of CPUs will be calculated from the provided topology members and the
> > -    initial CPU count will match the maximum number. When only one of them
> > -    is given then the omitted one will be set to its counterpart's value.
> > -    Both parameters may be specified, but the maximum number of CPUs must
> > -    be equal to or greater than the initial CPU count. Product of the
> > -    CPU topology hierarchy must be equal to the maximum number of CPUs.
> > -    Both parameters are subject to an upper limit that is determined by
> > -    the specific machine type chosen.
> > +    initial CPU count will match the maximum number. When only one of them is
> > +    given then the omitted one will be set to its counterpart's value. Both
> > +    parameters may be specified, but the maximum number of CPUs must be equal
> > +    to or greater than the initial CPU count. Product of the CPU topology
> > +    hierarchy must be equal to the maximum number of CPUs. Both parameters are
> > +    subject to an upper limit that is determined by the specific machine type
> > +    chosen. Boards that support administratively disabled CPUs but do *not*
> > +    support CPU hotplug derive the maximum number of CPUs implicitly:
> > +    '\ ``maxcpus``\ ' is treated as '\ ``n + disabledcpus``\ ' (the total CPUs
> > +    present in firmware). If '\ ``maxcpus``\ ' is provided, it must equal
> > +    '\ ``n + disabledcpus``\ '. The topology product must equal this derived
> > +    maximum as well.
> > +
> > +    Note: Administratively disabled CPUs will appear to the guest as
> > +    unavailable, and any attempt to bring them online must go through QMP/HMP
> > +    commands like 'device_set'.
> > 
> >     To control reporting of CPU topology information, values of the topology
> >     parameters can be specified. Machines may only support a subset of the
> > @@ -425,6 +455,24 @@ SRST
> > 
> >         -smp 2
> > 
> > +    Examples using 'disabledcpus':
> > +
> > +    For a board without CPU hotplug, enable 4 CPUs at boot and provision
> > +    2 additional administratively disabled CPUs (maximum is derived
> > +    implicitly as 6 = 4 + 2):
> > +
> > +    ::
> > +
> > +        -smp cpus=4,disabledcpus=2
> > +
> > +    For a board that supports CPU hotplug and 'disabledcpus', enable 4 CPUs
> > +    at boot, provision 2 administratively disabled CPUs, and allow hotplug of
> > +    2 more CPUs (for a maximum of 8):
> > +
> > +    ::
> > +
> > +        -smp cpus=4,disabledcpus=2,maxcpus=8
> > +
> >     Note: The cluster topology will only be generated in ACPI and exposed
> >     to guest if it's explicitly specified in -smp.
> > ERST
> > diff --git a/system/vl.c b/system/vl.c
> > index 3b7057e6c6..2f0fd21a1f 100644
> > --- a/system/vl.c
> > +++ b/system/vl.c
> > @@ -736,6 +736,9 @@ static QemuOptsList qemu_smp_opts = {
> >         {
> >             .name = "cpus",
> >             .type = QEMU_OPT_NUMBER,
> > +        }, {
> > +            .name = "disabledcpus",
> > +            .type = QEMU_OPT_NUMBER,
> >         }, {
> >             .name = "drawers",
> >             .type = QEMU_OPT_NUMBER,
> > -- 
> > 2.34.1
> >   
>