qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCHv5 for 2.9 3/6] pseries: Implement HPT resizing


From: Suraj Jitindar Singh
Subject: Re: [Qemu-devel] [PATCHv5 for 2.9 3/6] pseries: Implement HPT resizing
Date: Wed, 21 Dec 2016 13:55:41 +1100

On Tue, 2016-12-20 at 16:58 +1100, David Gibson wrote:
> This patch implements hypercalls allowing a PAPR guest to resize its
> own
> hash page table.  This will eventually allow for more flexible memory
> hotplug.
> 
> The implementation is partially asynchronous, handled in a special
> thread
> running the hpt_prepare_thread() function.  The state of a pending
> resize
> is stored in SPAPR_MACHINE->pending_hpt.
> 
> The H_RESIZE_HPT_PREPARE hypercall will kick off creation of a new
> HPT, or,
> if one is already in progress, monitor it for completion.  If there
> is an
> existing HPT resize in progress that doesn't match the size specified
> in
> the call, it will cancel it, replacing it with a new one matching the
> given size.
> 
> The H_RESIZE_HPT_COMMIT completes transition to a resized HPT, and
> can only
> be called successfully once H_RESIZE_HPT_PREPARE has successfully
> completed initialization of a new HPT.  The guest must ensure that
> there
> are no concurrent accesses to the existing HPT while this is called
> (this
> effectively means stop_machine() for Linux guests).
> 
> For now H_RESIZE_HPT_COMMIT goes through the whole old HPT, rehashing
> each
> HPTE into the new HPT.  This can have quite high latency, but it
> seems to
> be of the order of typical migration downtime latencies for HPTs of
> size
> up to ~2GiB (which would be used in a 256GiB guest).
> 
> In future we probably want to move more of the rehashing to the
> "prepare"
> phase, by having H_ENTER and other hcalls update both current and
> pending HPTs.  That's a project for another day, but should be
> possible
> without any changes to the guest interface.
> 
> Signed-off-by: David Gibson <address@hidden>
> ---
>  hw/ppc/spapr.c          |   4 +-
>  hw/ppc/spapr_hcall.c    | 353
> +++++++++++++++++++++++++++++++++++++++++++++++-
>  include/hw/ppc/spapr.h  |   6 +
>  target-ppc/mmu-hash64.h |   4 +
>  4 files changed, 361 insertions(+), 6 deletions(-)
> 
> diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
> index 42277a6..129b7d2 100644
> --- a/hw/ppc/spapr.c
> +++ b/hw/ppc/spapr.c
> @@ -93,8 +93,6 @@
>  
>  #define PHANDLE_XICP            0x00001111
>  
> -#define HTAB_SIZE(spapr)        (1ULL << ((spapr)->htab_shift))
> -
>  static XICSState *try_create_xics(const char *type, int nr_servers,
>                                    int nr_irqs, Error **errp)
>  {
> @@ -1055,7 +1053,7 @@ static void close_htab_fd(sPAPRMachineState
> *spapr)
>      spapr->htab_fd = -1;
>  }
>  
> -static int spapr_hpt_shift_for_ramsize(uint64_t ramsize)
> +int spapr_hpt_shift_for_ramsize(uint64_t ramsize)
>  {
>      int shift;
>  
> diff --git a/hw/ppc/spapr_hcall.c b/hw/ppc/spapr_hcall.c
> index 72a9c4d..c0898a4 100644
> --- a/hw/ppc/spapr_hcall.c
> +++ b/hw/ppc/spapr_hcall.c
> @@ -2,6 +2,7 @@
>  #include "qapi/error.h"
>  #include "sysemu/sysemu.h"
>  #include "qemu/log.h"
> +#include "qemu/error-report.h"
>  #include "cpu.h"
>  #include "exec/exec-all.h"
>  #include "helper_regs.h"
> @@ -355,20 +356,327 @@ static target_ulong h_read(PowerPCCPU *cpu,
> sPAPRMachineState *spapr,
>      return H_SUCCESS;
>  }
>  
> +struct sPAPRPendingHPT {
> +    /* These fields are read-only after initialization */
> +    int shift;
> +    QemuThread thread;
> +
> +    /* These fields are protected by the BQL */
> +    bool complete;
> +
> +    /* These fields are private to the preparation thread if
> +     * !complete, otherwise protected by the BQL */
> +    int ret;
> +    void *hpt;
> +};
> +
> +static void free_pending_hpt(sPAPRPendingHPT *pending)
> +{
> +    if (pending->hpt) {
> +        qemu_vfree(pending->hpt);
> +    }
> +
> +    g_free(pending);
> +}
> +
> +static void *hpt_prepare_thread(void *opaque)
> +{
> +    sPAPRPendingHPT *pending = opaque;
> +    size_t size = 1ULL << pending->shift;
> +
> +    pending->hpt = qemu_memalign(size, size);
> +    if (pending->hpt) {
> +        memset(pending->hpt, 0, size);
> +        pending->ret = H_SUCCESS;
> +    } else {
> +        pending->ret = H_NO_MEM;
> +    }
> +
> +    qemu_mutex_lock_iothread();
> +
> +    if (SPAPR_MACHINE(qdev_get_machine())->pending_hpt == pending) {
> +        /* Ready to go */
> +        pending->complete = true;
> +    } else {
> +        /* We've been cancelled, clean ourselves up */
> +        free_pending_hpt(pending);
> +    }
> +
> +    qemu_mutex_unlock_iothread();
> +    return NULL;
> +}
> +
> +/* Must be called with BQL held */
> +static void cancel_hpt_prepare(sPAPRMachineState *spapr)
> +{
> +    sPAPRPendingHPT *pending = spapr->pending_hpt;
> +
> +    /* Let the thread know it's cancelled */
> +    spapr->pending_hpt = NULL;
> +
> +    if (!pending) {
> +        /* Nothing to do */
> +        return;
> +    }
> +
> +    if (!pending->complete) {
> +        /* thread will clean itself up */
> +        return;
> +    }
> +
> +    free_pending_hpt(pending);
> +}
> +
> +static int build_dimm_list(Object *obj, void *opaque)
> +{
> +    GSList **list = opaque;
> +
> +    if (object_dynamic_cast(obj, TYPE_PC_DIMM)) {
> +        DeviceState *dev = DEVICE(obj);
> +        if (dev->realized) { /* only realized DIMMs matter */
> +            *list = g_slist_prepend(*list, dev);
> +        }
> +    }
> +
> +    object_child_foreach(obj, build_dimm_list, opaque);
> +    return 0;
> +}
> +
> +static ram_addr_t get_current_ram_size(void)
> +{
> +    GSList *list = NULL, *item;
> +    ram_addr_t size = ram_size;
> +
> +    build_dimm_list(qdev_get_machine(), &list);
> +    for (item = list; item; item = g_slist_next(item)) {
> +        Object *obj = OBJECT(item->data);
> +        if (!strcmp(object_get_typename(obj), TYPE_PC_DIMM)) {
> +            size += object_property_get_int(obj, PC_DIMM_SIZE_PROP,
> +                                            &error_abort);
> +        }
> +    }
> +    g_slist_free(list);
> +
> +    return size;
> +}
> +
>  static target_ulong h_resize_hpt_prepare(PowerPCCPU *cpu,
>                                           sPAPRMachineState *spapr,
>                                           target_ulong opcode,
>                                           target_ulong *args)
>  {
>      target_ulong flags = args[0];
> -    target_ulong shift = args[1];
> +    int shift = args[1];
> +    sPAPRPendingHPT *pending = spapr->pending_hpt;
>  
>      if (spapr->resize_hpt == SPAPR_RESIZE_HPT_DISABLED) {
>          return H_AUTHORITY;
>      }
>  
>      trace_spapr_h_resize_hpt_prepare(flags, shift);
> -    return H_HARDWARE;
> +
> +    if (flags != 0) {
> +        return H_PARAMETER;
> +    }
> +
> +    if (shift && ((shift < 18) || (shift > 46))) {
> +        return H_PARAMETER;
> +    }
> +
> +    /* We only allow the guest to allocate an HPT one order above
> what
> +     * we'd normally give them (to stop a small guest claiming a
> huge
> +     * chunk of resources in the HPT */
> +    if (shift > (spapr_hpt_shift_for_ramsize(get_current_ram_size())
> + 1)) {
> +        assert(!pending || pending->shift != shift);
Where did we land on keeping this assert or not?
> +        return H_RESOURCE;
> +    }
> +
> +    if (pending) {
> +        /* something already in progress */
> +        if (pending->shift == shift) {
> +            /* and it's suitable */
> +            if (pending->complete) {
> +                return pending->ret;
> +            } else {
> +                return H_LONG_BUSY_ORDER_100_MSEC;
> +            }
> +        }
> +
> +        /* not suitable, cancel and replace */
> +        cancel_hpt_prepare(spapr);
> +    }
> +
> +    if (!shift) {
> +        /* nothing to do */
> +        return H_SUCCESS;
> +    }
> +
> +    /* start new prepare */
> +
> +    pending = g_new0(sPAPRPendingHPT, 1);
> +    pending->shift = shift;
> +    pending->ret = H_HARDWARE;
> +
> +    qemu_thread_create(&pending->thread, "sPAPR HPT prepare",
> +                       hpt_prepare_thread, pending,
> QEMU_THREAD_DETACHED);
> +
> +    spapr->pending_hpt = pending;
> +
> +    /* In theory we could estimate the time more accurately based on
> +     * the new size, but there's not much point */
> +    return H_LONG_BUSY_ORDER_100_MSEC;
> +}
> +
> +static uint64_t new_hpte_load0(void *htab, uint64_t pteg, int slot)
> +{
> +    uint8_t *addr = htab;
> +
> +    addr += pteg * HASH_PTEG_SIZE_64;
> +    addr += slot * HASH_PTE_SIZE_64;
> +    return  ldq_p(addr);
> +}
> +
> +static void new_hpte_store(void *htab, uint64_t pteg, int slot,
> +                           uint64_t pte0, uint64_t pte1)
> +{
> +    uint8_t *addr = htab;
> +
> +    addr += pteg * HASH_PTEG_SIZE_64;
> +    addr += slot * HASH_PTE_SIZE_64;
> +
> +    stq_p(addr, pte0);
> +    stq_p(addr + HASH_PTE_SIZE_64/2, pte1);
> +}
> +
> +static int rehash_hpte(PowerPCCPU *cpu, uint64_t token,
> +                       void *old_hpt, uint64_t oldsize,
> +                       void *new_hpt, uint64_t newsize,
> +                       uint64_t pteg, int slot)
> +{
> +    uint64_t old_hash_mask = (oldsize >> 7) - 1;
> +    uint64_t new_hash_mask = (newsize >> 7) - 1;
> +    target_ulong pte0 = ppc_hash64_load_hpte0(cpu, token, slot);
> +    target_ulong pte1;
> +    uint64_t avpn;
> +    unsigned base_pg_shift;
> +    uint64_t hash, new_pteg, replace_pte0;
> +
> +    if (!(pte0 & HPTE64_V_VALID) || !(pte0 & HPTE64_V_BOLTED)) {
> +        return H_SUCCESS;
> +    }
> +
> +    pte1 = ppc_hash64_load_hpte1(cpu, token, slot);
> +
> +    base_pg_shift = ppc_hash64_hpte_page_shift_noslb(cpu, pte0,
> pte1);
> +    assert(base_pg_shift); /* H_ENTER shouldn't allow a bad encoding
> */
> +    avpn = HPTE64_V_AVPN_VAL(pte0) & ~(((1ULL << base_pg_shift) - 1)
> >> 23);
> +
> +    if (pte0 & HPTE64_V_SECONDARY) {
> +        pteg = ~pteg;
> +    }
> +
> +    if ((pte0 & HPTE64_V_SSIZE) == HPTE64_V_SSIZE_256M) {
> +        uint64_t offset, vsid;
> +
> +        /* We only have 28 - 23 bits of offset in avpn */
> +        offset = (avpn & 0x1f) << 23;
> +        vsid = avpn >> 5;
> +        /* We can find more bits from the pteg value */
> +        if (base_pg_shift < 23) {
> +            offset |= ((vsid ^ pteg) & old_hash_mask) <<
> base_pg_shift;
> +        }
> +
> +        hash = vsid ^ (offset >> base_pg_shift);
> +    } else if ((pte0 & HPTE64_V_SSIZE) == HPTE64_V_SSIZE_1T) {
> +        uint64_t offset, vsid;
> +
> +        /* We only have 40 - 23 bits of seg_off in avpn */
> +        offset = (avpn & 0x1ffff) << 23;
> +        vsid = avpn >> 17;
> +        if (base_pg_shift < 23) {
> +            offset |= ((vsid ^ (vsid << 25) ^ pteg) & old_hash_mask)
> +                << base_pg_shift;
> +        }
> +
> +        hash = vsid ^ (vsid << 25) ^ (offset >> base_pg_shift);
> +    } else {
> +        error_report("rehash_pte: Bad segment size in HPTE");
> +        return H_HARDWARE;
> +    }
> +
> +    new_pteg = hash & new_hash_mask;
> +    if (pte0 & HPTE64_V_SECONDARY) {
> +        assert(~pteg == (hash & old_hash_mask));
> +        new_pteg = ~new_pteg;
> +    } else {
> +        assert(pteg == (hash & old_hash_mask));
> +    }
> +    assert((oldsize != newsize) || (pteg == new_pteg));
> +    replace_pte0 = new_hpte_load0(new_hpt, new_pteg, slot);
> +    /*
> +     * Strictly speaking, we don't need all these tests, since we
> only
> +     * ever rehash bolted HPTEs.  We might in future handle non-
> bolted
> +     * HPTEs, though so make the logic correct for those cases as
> +     * well.
> +     */
> +    if (replace_pte0 & HPTE64_V_VALID) {
> +        assert(newsize < oldsize);
> +        if (replace_pte0 & HPTE64_V_BOLTED) {
> +            if (pte0 & HPTE64_V_BOLTED) {
> +                /* Bolted collision, nothing we can do */
> +                return H_PTEG_FULL;
> +            } else {
> +                /* Discard this hpte */
> +                return H_SUCCESS;
> +            }
> +        }
> +    }
> +
> +    new_hpte_store(new_hpt, new_pteg, slot, pte0, pte1);
> +    return H_SUCCESS;
> +}
> +
> +static int rehash_hpt(PowerPCCPU *cpu,
> +                      void *old_hpt, uint64_t oldsize,
> +                      void *new_hpt, uint64_t newsize)
> +{
> +    CPUPPCState *env = &cpu->env;
> +    uint64_t n_ptegs = oldsize >> 7;
> +    uint64_t pteg;
> +    int slot;
> +    int rc;
> +
> +    assert(env->external_htab == old_hpt);
> +
> +    for (pteg = 0; pteg < n_ptegs; pteg++) {
> +        uint64_t token = ppc_hash64_start_access(cpu, pteg *
> HPTES_PER_GROUP);
> +
> +        if (!token) {
> +            return H_HARDWARE;
> +        }
> +
> +        for (slot = 0; slot < HPTES_PER_GROUP; slot++) {
> +            rc = rehash_hpte(cpu, token, old_hpt, oldsize, new_hpt,
> newsize,
> +                             pteg, slot);
> +            if (rc != H_SUCCESS) {
> +                ppc_hash64_stop_access(cpu, token);
> +                return rc;
> +            }
> +        }
> +        ppc_hash64_stop_access(cpu, token);
> +    }
> +
> +    return H_SUCCESS;
> +}
> +
> +static void pivot_hpt(CPUState *cs, run_on_cpu_data arg)
> +{
> +    sPAPRMachineState *spapr = SPAPR_MACHINE(arg.host_ptr);
> +    PowerPCCPU *cpu = POWERPC_CPU(cs);
> +
> +    cpu_synchronize_state(cs);
> +    ppc_hash64_set_external_hpt(cpu, spapr->htab, spapr->htab_shift,
> +                                &error_fatal);
>  }
>  
>  static target_ulong h_resize_hpt_commit(PowerPCCPU *cpu,
> @@ -378,13 +686,52 @@ static target_ulong
> h_resize_hpt_commit(PowerPCCPU *cpu,
>  {
>      target_ulong flags = args[0];
>      target_ulong shift = args[1];
> +    sPAPRPendingHPT *pending = spapr->pending_hpt;
> +    int rc;
> +    size_t newsize;
>  
>      if (spapr->resize_hpt == SPAPR_RESIZE_HPT_DISABLED) {
>          return H_AUTHORITY;
>      }
>  
>      trace_spapr_h_resize_hpt_commit(flags, shift);
> -    return H_HARDWARE;
> +
> +    if (flags != 0) {
> +        return H_PARAMETER;
> +    }
> +
> +    if (!pending || (pending->shift != shift)) {
> +        /* no matching prepare */
> +        return H_CLOSED;
> +    }
> +
> +    if (!pending->complete) {
> +        /* prepare has not completed */
> +        return H_BUSY;
> +    }
> +
> +    newsize = 1ULL << pending->shift;
> +    rc = rehash_hpt(cpu, spapr->htab, HTAB_SIZE(spapr),
> +                    pending->hpt, newsize);
> +    if (rc == H_SUCCESS) {
> +        CPUState *cs;
> +
> +        qemu_vfree(spapr->htab);
> +        spapr->htab = pending->hpt;
> +        spapr->htab_shift = pending->shift;
> +
> +        CPU_FOREACH(cs) {
> +            run_on_cpu(cs, pivot_hpt, RUN_ON_CPU_HOST_PTR(spapr));
> +        }
> +
> +        pending->hpt = NULL; /* so it's not free()d */
> +    }
> +
> +    /* Clean up */
> +    spapr->pending_hpt = NULL;
> +    free_pending_hpt(pending);
> +
> +    return rc;
>  }
>  
>  static target_ulong h_set_sprg0(PowerPCCPU *cpu, sPAPRMachineState
> *spapr,
> diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
> index d2c758b..954bada 100644
> --- a/include/hw/ppc/spapr.h
> +++ b/include/hw/ppc/spapr.h
> @@ -14,6 +14,7 @@ struct sPAPRNVRAM;
>  typedef struct sPAPRConfigureConnectorState
> sPAPRConfigureConnectorState;
>  typedef struct sPAPREventLogEntry sPAPREventLogEntry;
>  typedef struct sPAPREventSource sPAPREventSource;
> +typedef struct sPAPRPendingHPT sPAPRPendingHPT;
>  
>  #define HPTE64_V_HPTE_DIRTY     0x0000000000000040ULL
>  #define SPAPR_ENTRY_POINT       0x100
> @@ -72,6 +73,8 @@ struct sPAPRMachineState {
>      sPAPRResizeHPT resize_hpt;
>      void *htab;
>      uint32_t htab_shift;
> +    sPAPRPendingHPT *pending_hpt; /* in-progress resize */
> +
>      hwaddr rma_size;
>      int vrma_adjust;
>      ssize_t rtas_size;
> @@ -627,6 +630,7 @@ void
> spapr_hotplug_req_remove_by_count_indexed(sPAPRDRConnectorType
> drc_type,
>                                                 uint32_t count,
> uint32_t index);
>  void *spapr_populate_hotplug_cpu_dt(CPUState *cs, int *fdt_offset,
>                                      sPAPRMachineState *spapr);
> +int spapr_hpt_shift_for_ramsize(uint64_t ramsize);
>  
>  /* rtas-configure-connector state */
>  struct sPAPRConfigureConnectorState {
> @@ -674,4 +678,6 @@ int spapr_rng_populate_dt(void *fdt);
>  
>  void spapr_do_system_reset_on_cpu(CPUState *cs, run_on_cpu_data
> arg);
>  
> +#define HTAB_SIZE(spapr)        (1ULL << ((spapr)->htab_shift))
> +
>  #endif /* HW_SPAPR_H */
> diff --git a/target-ppc/mmu-hash64.h b/target-ppc/mmu-hash64.h
> index ab5d347..4a1b781 100644
> --- a/target-ppc/mmu-hash64.h
> +++ b/target-ppc/mmu-hash64.h
> @@ -60,11 +60,15 @@ void ppc_hash64_update_rmls(CPUPPCState *env);
>  #define HASH_PTE_SIZE_64        16
>  #define HASH_PTEG_SIZE_64       (HASH_PTE_SIZE_64 * HPTES_PER_GROUP)
>  
> +#define HPTE64_V_SSIZE          SLB_VSID_B
> +#define HPTE64_V_SSIZE_256M     SLB_VSID_B_256M
> +#define HPTE64_V_SSIZE_1T       SLB_VSID_B_1T
>  #define HPTE64_V_SSIZE_SHIFT    62
>  #define HPTE64_V_AVPN_SHIFT     7
>  #define HPTE64_V_AVPN           0x3fffffffffffff80ULL
>  #define HPTE64_V_AVPN_VAL(x)    (((x) & HPTE64_V_AVPN) >>
> HPTE64_V_AVPN_SHIFT)
>  #define HPTE64_V_COMPARE(x, y)  (!(((x) ^ (y)) &
> 0xffffffffffffff83ULL))
> +#define HPTE64_V_BOLTED         0x0000000000000010ULL
>  #define HPTE64_V_LARGE          0x0000000000000004ULL
>  #define HPTE64_V_SECONDARY      0x0000000000000002ULL
>  #define HPTE64_V_VALID          0x0000000000000001ULL



reply via email to

[Prev in Thread] Current Thread [Next in Thread]