On Mon, 2016-12-12 at 15:06 +1100, David Gibson wrote:
> We've now implemented a PAPR extension allowing PAPR guest to resize
> their hash page table (HPT) during runtime.
>
> This patch makes use of that facility to allocate smaller HPTs by
> default.
> Specifically when a guest is aware of the HPT resize facility, qemu
> sizes
> the HPT to the initial memory size, rather than the maximum memory
> size on
> the assumption that the guest will resize its HPT if necessary for
> hot
> plugged memory.
>
> When the initial memory size is much smaller than the maximum memory
> size
> (a common configuration with e.g. oVirt / RHEV) then this can save
> significant memory on the HPT.
>
> If the guest does *not* advertise HPT resize awareness when it makes
> the
> ibm,client-architecture-support call, qemu resizes the HPT for
> maxmimum
> memory size (unless it's been configured not to allow such guests at
> all).
>
> For now we make that reallocation assuming the guest has not yet used
> the
> HPT at all. That's true in practice, but not, strictly, an
> architectural
> or PAPR requirement. If we need to in future we can fix this by
> having
> the client-architecture-support call reboot the guest with the
> revised
> HPT size (the client-architecture-support call is explicitly
> permitted to
> trigger a reboot in this way).
>
> Signed-off-by: David Gibson <da...@gibson.dropbear.id.au>
> ---
> hw/ppc/spapr.c | 21 ++++++++++++++++-----
> hw/ppc/spapr_hcall.c | 32 ++++++++++++++++++++++++++++++++
> include/hw/ppc/spapr.h | 2 ++
> include/hw/ppc/spapr_ovec.h | 1 +
> 4 files changed, 51 insertions(+), 5 deletions(-)
>
> diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
> index f05d0e5..51b189d 100644
> --- a/hw/ppc/spapr.c
> +++ b/hw/ppc/spapr.c
> @@ -1066,8 +1066,8 @@ int spapr_hpt_shift_for_ramsize(uint64_t
> ramsize)
> return shift;
> }
>
> -static void spapr_reallocate_hpt(sPAPRMachineState *spapr, int
> shift,
> - Error **errp)
> +void spapr_reallocate_hpt(sPAPRMachineState *spapr, int shift,
> + Error **errp)
> {
> long rc;
>
> @@ -1140,14 +1140,20 @@ static void ppc_spapr_reset(void)
> hwaddr rtas_addr, fdt_addr;
> void *fdt;
> int rc;
> + int hpt_shift;
>
> /* Check for unknown sysbus devices */
> foreach_dynamic_sysbus_device(find_unknown_sysbus_device, NULL);
>
> /* Allocate and/or reset the hash page table */
> - spapr_reallocate_hpt(spapr,
> - spapr_hpt_shift_for_ramsize(machine-
> >maxram_size),
> - &error_fatal);
> + if ((spapr->resize_hpt == SPAPR_RESIZE_HPT_DISABLED)
> + || (spapr->cas_reboot
> + && !spapr_ovec_test(spapr->ov5_cas, OV5_HPT_RESIZE))) {
> + hpt_shift = spapr_hpt_shift_for_ramsize(machine-
> >maxram_size);
> + } else {
> + hpt_shift = spapr_hpt_shift_for_ramsize(machine->ram_size);
> + }
> + spapr_reallocate_hpt(spapr, hpt_shift, &error_fatal);
>
> /* Update the RMA size if necessary */
> if (spapr->vrma_adjust) {
> @@ -1941,6 +1947,11 @@ static void ppc_spapr_init(MachineState
> *machine)
> spapr_ovec_set(spapr->ov5, OV5_HP_EVT);
> }
>
> + /* advertise support for HPT resizing */
> + if (spapr->resize_hpt != SPAPR_RESIZE_HPT_DISABLED) {
> + spapr_ovec_set(spapr->ov5, OV5_HPT_RESIZE);
> + }
> +
> /* init CPUs */
> if (machine->cpu_model == NULL) {
> machine->cpu_model = kvm_enabled() ? "host" : smc-
> >tcg_default_cpu;
> diff --git a/hw/ppc/spapr_hcall.c b/hw/ppc/spapr_hcall.c
> index 1e89061..5bebfc3 100644
> --- a/hw/ppc/spapr_hcall.c
> +++ b/hw/ppc/spapr_hcall.c
> @@ -1415,6 +1415,38 @@ static target_ulong
> h_client_architecture_support(PowerPCCPU *cpu_,
>
> ov5_guest = spapr_ovec_parse_vector(ov_table, 5);
>
> + /*
> + * HPT resizing is a bit of a special case, because when enabled
> + * we assume the guest will support it until it says it doesn't,
> + * instead of assuming it won't support it until it says it
> does.
> + * Strictly speaking that approach could break for guests which
> + * don't make a CAS call, but those are so old we don't care
> about
> + * them. Without that assumption we'd have to make at least a
> + * temporary allocation of an HPT sized for max memory, which
> + * could be impossibly difficult under KVM HV if maxram is
> large.
> + */
> + if (!spapr_ovec_test(ov5_guest, OV5_HPT_RESIZE)) {
> + int maxshift = spapr_hpt_shift_for_ramsize(MACHINE(spapr)-
> >maxram_size);
> +
> + if (spapr->resize_hpt == SPAPR_RESIZE_HPT_REQUIRED) {
> + error_report(
> + "h_client_architecture_support: Guest doesn't
> support HPT resizing, but resize-hpt=required");
> + exit(1);
> + }
> +
> + if (spapr->htab_shift < maxshift) {
> + CPUState *cs;
> + /* Guest doesn't know about HPT resizing, so we
> + * pre-emptively resize for the maximum permitted
> RAM. At
> + * the point this is called, nothing should have been
> + * entered into the existing HPT */
> + spapr_reallocate_hpt(spapr, maxshift, &error_fatal);
> + CPU_FOREACH(cs) {
> + run_on_cpu(cs, pivot_hpt,
> RUN_ON_CPU_HOST_PTR(spapr));
> + }
> + }
> + }
> +
> /* NOTE: there are actually a number of ov5 bits where input
> from the
> * guest is always zero, and the platform/QEMU enables them
> independently
> * of guest input. To model these properly we'd want some sort
> of mask,
> diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
> index 954bada..ff3bf87 100644
> --- a/include/hw/ppc/spapr.h
> +++ b/include/hw/ppc/spapr.h
> @@ -631,6 +631,8 @@ void
> spapr_hotplug_req_remove_by_count_indexed(sPAPRDRConnectorType
> drc_type,
> void *spapr_populate_hotplug_cpu_dt(CPUState *cs, int *fdt_offset,
> sPAPRMachineState *spapr);
> int spapr_hpt_shift_for_ramsize(uint64_t ramsize);
> +void spapr_reallocate_hpt(sPAPRMachineState *spapr, int shift,
> + Error **errp);
>
> /* rtas-configure-connector state */
> struct sPAPRConfigureConnectorState {
> diff --git a/include/hw/ppc/spapr_ovec.h
> b/include/hw/ppc/spapr_ovec.h
> index 355a344..f5fed87 100644
> --- a/include/hw/ppc/spapr_ovec.h
> +++ b/include/hw/ppc/spapr_ovec.h
> @@ -47,6 +47,7 @@ typedef struct sPAPROptionVector sPAPROptionVector;
> #define OV5_DRCONF_MEMORY OV_BIT(2, 2)
> #define OV5_FORM1_AFFINITY OV_BIT(5, 0)
> #define OV5_HP_EVT OV_BIT(6, 5)
> +#define OV5_HPT_RESIZE OV_BIT(6, 7)
>
> /* interfaces */
> sPAPROptionVector *spapr_ovec_new(void);
Reviewed-by: Suraj Jitindar Singh <sjitindarsi...@gmail.com>
