Series comparison

-[PULL 00/12] tcg patch queue
+[PULL 0/3] tcg patch queue
-The following changes since commit bad76ac319556dab2497429d473b49a237672e1c:
+The following changes since commit 2ecfc0657afa5d29a373271b342f704a1a3c6737:
-  Merge remote-tracking branch 'remotes/vivier2/tags/trivial-branch-pull-request' into staging (2019-10-25 14:17:08 +0100)
+  Merge remote-tracking branch 'remotes/armbru/tags/pull-misc-2020-12-10' into staging (2020-12-10 17:01:05 +0000)
 are available in the Git repository at:
-  https://github.com/rth7680/qemu.git tags/pull-tcg-20191025
+  https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20201210
-for you to fetch changes up to 0ed1bfb046b740b70eed2cf3581e01768703b185:
+for you to fetch changes up to 9e2658d62ebc23efe7df43fc0e306f129510d874:
-  translate-all: Remove tb_alloc (2019-10-25 10:15:25 -0400)
+  accel/tcg: rename tcg-cpus functions to match module name (2020-12-10 17:44:10 -0600)
 ----------------------------------------------------------------
-Improvements for TARGET_PAGE_BITS_VARY
+Split CpusAccel for tcg variants
 Fix for TCI ld16u_i64.
 Fix for segv on icount execute from i/o memory.
 Two misc cleanups.
 ----------------------------------------------------------------
-Alex Bennée (1):
+Claudio Fontana (3):
-      cputlb: ensure _cmmu helper functions follow the naming standard
+      accel/tcg: split CpusAccel into three TCG variants
       accel/tcg: split tcg_start_vcpu_thread
       accel/tcg: rename tcg-cpus functions to match module name
-Clement Deschamps (1):
+ accel/tcg/tcg-cpus-icount.h |  17 ++
-      translate-all: fix uninitialized tb->orig_tb
+ accel/tcg/tcg-cpus-rr.h     |  21 ++
  accel/tcg/tcg-cpus.h        |  12 +-
  accel/tcg/tcg-all.c         |  13 +-
  accel/tcg/tcg-cpus-icount.c | 147 +++++++++++++
  accel/tcg/tcg-cpus-mttcg.c  | 140 ++++++++++++
  accel/tcg/tcg-cpus-rr.c     | 305 ++++++++++++++++++++++++++
  accel/tcg/tcg-cpus.c        | 506 +-------------------------------------------
  softmmu/icount.c            |   2 +-
  accel/tcg/meson.build       |   9 +-
 files changed, 670 insertions(+), 502 deletions(-)
  create mode 100644 accel/tcg/tcg-cpus-icount.h
  create mode 100644 accel/tcg/tcg-cpus-rr.h
  create mode 100644 accel/tcg/tcg-cpus-icount.c
  create mode 100644 accel/tcg/tcg-cpus-mttcg.c
  create mode 100644 accel/tcg/tcg-cpus-rr.c
-Richard Henderson (8):
-      exec: Split out variable page size support to exec-vary.c
-      configure: Detect compiler support for __attribute__((alias))
-      exec: Use const alias for TARGET_PAGE_BITS_VARY
-      exec: Restrict TARGET_PAGE_BITS_VARY assert to CONFIG_DEBUG_TCG
-      exec: Promote TARGET_PAGE_MASK to target_long
-      exec: Cache TARGET_PAGE_MASK for TARGET_PAGE_BITS_VARY
-      cputlb: Fix tlb_vaddr_to_host
-      translate-all: Remove tb_alloc
-Stefan Weil (1):
-      tci: Add implementation for INDEX_op_ld16u_i64
-Wei Yang (1):
-      cpu: use ROUND_UP() to define xxx_PAGE_ALIGN
- Makefile.target                  |   2 +-
- include/exec/cpu-all.h           |  33 +++++++++----
- include/exec/cpu_ldst_template.h |   4 +-
- include/qemu-common.h            |   6 +++
- tcg/tcg.h                        |  20 +++++---
- accel/tcg/cputlb.c               |  26 ++++++++--
- accel/tcg/translate-all.c        |  21 ++------
- exec-vary.c                      | 102 +++++++++++++++++++++++++++++++++++++++
- exec.c                           |  34 -------------
- target/cris/translate_v10.inc.c  |   3 +-
- tcg/tci.c                        |  15 ++++++
- configure                        |  19 ++++++++
-files changed, 208 insertions(+), 77 deletions(-)
- create mode 100644 exec-vary.c

-[PULL 01/12] tci: Add implementation for INDEX_op_ld16u_i64
+Deleted patch
-From: Stefan Weil <sw@weilnetz.de>
-This fixes "make check-tcg" on a Debian x86_64 host.
-Signed-off-by: Stefan Weil <sw@weilnetz.de>
-Tested-by: Thomas Huth <thuth@redhat.com>
-Message-Id: <20190410194838.10123-1-sw@weilnetz.de>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- tcg/tci.c | 15 +++++++++++++++
-file changed, 15 insertions(+)
-diff --git a/tcg/tci.c b/tcg/tci.c
-index XXXXXXX..XXXXXXX 100644
---- a/tcg/tci.c
-+++ b/tcg/tci.c
-@@ -XXX,XX +XXX,XX @@ static void tci_write_reg8(tcg_target_ulong *regs, TCGReg index, uint8_t value)
-     tci_write_reg(regs, index, value);
- }
-+static void
-+tci_write_reg16(tcg_target_ulong *regs, TCGReg index, uint16_t value)
-+{
-+    tci_write_reg(regs, index, value);
-+}
-+
- static void
- tci_write_reg32(tcg_target_ulong *regs, TCGReg index, uint32_t value)
- {
-@@ -XXX,XX +XXX,XX @@ uintptr_t tcg_qemu_tb_exec(CPUArchState *env, uint8_t *tb_ptr)
-             tci_write_reg8(regs, t0, *(uint8_t *)(t1 + t2));
-             break;
-         case INDEX_op_ld8s_i32:
-+            TODO();
-+            break;
-         case INDEX_op_ld16u_i32:
-             TODO();
-             break;
-@@ -XXX,XX +XXX,XX @@ uintptr_t tcg_qemu_tb_exec(CPUArchState *env, uint8_t *tb_ptr)
-             tci_write_reg8(regs, t0, *(uint8_t *)(t1 + t2));
-             break;
-         case INDEX_op_ld8s_i64:
-+            TODO();
-+            break;
-         case INDEX_op_ld16u_i64:
-+            t0 = *tb_ptr++;
-+            t1 = tci_read_r(regs, &tb_ptr);
-+            t2 = tci_read_s32(&tb_ptr);
-+            tci_write_reg16(regs, t0, *(uint16_t *)(t1 + t2));
-+            break;
-         case INDEX_op_ld16s_i64:
-             TODO();
-             break;
---
-.17.1

-[PULL 04/12] exec: Split out variable page size support to exec-vary.c
+[PULL 1/3] accel/tcg: split CpusAccel into three TCG variants
-The next patch will play a trick with "const" that will
+From: Claudio Fontana <cfontana@suse.de>
 confuse the compiler about the uses of target_page_bits
 within exec.c.  Moving everything to a new file prevents
 this confusion.
-No functional change so far.
+split up the CpusAccel tcg_cpus into three TCG variants:
-Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
+tcg_cpus_rr (single threaded, round robin cpus)
-Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
+tcg_cpus_icount (same as rr, but with instruction counting enabled)
-Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
+tcg_cpus_mttcg (multi-threaded cpus)
 Suggested-by: Richard Henderson <richard.henderson@linaro.org>
 Signed-off-by: Claudio Fontana <cfontana@suse.de>
 Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
 Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
 Message-Id: <20201015143217.29337-2-cfontana@suse.de>
 Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
 ---
- Makefile.target       |  2 +-
+ accel/tcg/tcg-cpus-icount.h |  17 ++
- include/qemu-common.h |  6 +++++
+ accel/tcg/tcg-cpus-mttcg.h  |  21 ++
- exec-vary.c           | 57 +++++++++++++++++++++++++++++++++++++++++++
+ accel/tcg/tcg-cpus-rr.h     |  20 ++
- exec.c                | 34 --------------------------
+ accel/tcg/tcg-cpus.h        |  13 +-
-files changed, 64 insertions(+), 35 deletions(-)
+ accel/tcg/tcg-all.c         |   8 +-
- create mode 100644 exec-vary.c
+ accel/tcg/tcg-cpus-icount.c | 147 +++++++++++
  accel/tcg/tcg-cpus-mttcg.c  | 117 +++++++++
  accel/tcg/tcg-cpus-rr.c     | 270 ++++++++++++++++++++
  accel/tcg/tcg-cpus.c        | 484 ++----------------------------------
  softmmu/icount.c            |   2 +-
  accel/tcg/meson.build       |   9 +-
 files changed, 646 insertions(+), 462 deletions(-)
  create mode 100644 accel/tcg/tcg-cpus-icount.h
  create mode 100644 accel/tcg/tcg-cpus-mttcg.h
  create mode 100644 accel/tcg/tcg-cpus-rr.h
  create mode 100644 accel/tcg/tcg-cpus-icount.c
  create mode 100644 accel/tcg/tcg-cpus-mttcg.c
  create mode 100644 accel/tcg/tcg-cpus-rr.c
-diff --git a/Makefile.target b/Makefile.target
+diff --git a/accel/tcg/tcg-cpus-icount.h b/accel/tcg/tcg-cpus-icount.h
 index XXXXXXX..XXXXXXX 100644
 --- a/Makefile.target
 +++ b/Makefile.target
@@ -XXX,XX +XXX,XX @@ obj-y += trace/
  #########################################################
  # cpu emulator library
 -obj-y += exec.o
 +obj-y += exec.o exec-vary.o
  obj-y += accel/
  obj-$(CONFIG_TCG) += tcg/tcg.o tcg/tcg-op.o tcg/tcg-op-vec.o tcg/tcg-op-gvec.o
  obj-$(CONFIG_TCG) += tcg/tcg-common.o tcg/optimize.o
 diff --git a/include/qemu-common.h b/include/qemu-common.h
 index XXXXXXX..XXXXXXX 100644
 --- a/include/qemu-common.h
 +++ b/include/qemu-common.h
@@ -XXX,XX +XXX,XX @@ void cpu_exec_step_atomic(CPUState *cpu);
   */
  bool set_preferred_target_page_bits(int bits);
 +/**
 + * finalize_target_page_bits:
 + * Commit the final value set by set_preferred_target_page_bits.
 + */
 +void finalize_target_page_bits(void);
 +
  /**
   * Sends a (part of) iovec down a socket, yielding when the socket is full, or
   * Receives data into a (part of) iovec from a socket,
 diff --git a/exec-vary.c b/exec-vary.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
-+++ b/exec-vary.c
++++ b/accel/tcg/tcg-cpus-icount.h
 @@ -XXX,XX +XXX,XX @@
 +/*
-+ * Variable page size handling
++ * QEMU TCG Single Threaded vCPUs implementation using instruction counting
 + *
-+ *  Copyright (c) 2003 Fabrice Bellard
++ * Copyright 2020 SUSE LLC
 + *
-+ * This library is free software; you can redistribute it and/or
++ * This work is licensed under the terms of the GNU GPL, version 2 or later.
-+ * modify it under the terms of the GNU Lesser General Public
++ * See the COPYING file in the top-level directory.
-+ * License as published by the Free Software Foundation; either
++ */
-+ * version 2 of the License, or (at your option) any later version.
++
-+ *
++#ifndef TCG_CPUS_ICOUNT_H
-+ * This library is distributed in the hope that it will be useful,
++#define TCG_CPUS_ICOUNT_H
-+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
++
-+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
++void handle_icount_deadline(void);
-+ * Lesser General Public License for more details.
++void prepare_icount_for_run(CPUState *cpu);
-+ *
++void process_icount_data(CPUState *cpu);
-+ * You should have received a copy of the GNU Lesser General Public
++
-+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
++#endif /* TCG_CPUS_ICOUNT_H */
 diff --git a/accel/tcg/tcg-cpus-mttcg.h b/accel/tcg/tcg-cpus-mttcg.h
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/accel/tcg/tcg-cpus-mttcg.h
@@ -XXX,XX +XXX,XX @@
 +/*
 + * QEMU TCG Multi Threaded vCPUs implementation
 + *
 + * Copyright 2020 SUSE LLC
 + *
 + * This work is licensed under the terms of the GNU GPL, version 2 or later.
 + * See the COPYING file in the top-level directory.
 + */
 +
 +#ifndef TCG_CPUS_MTTCG_H
 +#define TCG_CPUS_MTTCG_H
 +
 +/*
 + * In the multi-threaded case each vCPU has its own thread. The TLS
 + * variable current_cpu can be used deep in the code to find the
 + * current CPUState for a given thread.
 + */
 +
 +void *tcg_cpu_thread_fn(void *arg);
 +
 +#endif /* TCG_CPUS_MTTCG_H */
 diff --git a/accel/tcg/tcg-cpus-rr.h b/accel/tcg/tcg-cpus-rr.h
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/accel/tcg/tcg-cpus-rr.h
@@ -XXX,XX +XXX,XX @@
 +/*
 + * QEMU TCG Single Threaded vCPUs implementation
 + *
 + * Copyright 2020 SUSE LLC
 + *
 + * This work is licensed under the terms of the GNU GPL, version 2 or later.
 + * See the COPYING file in the top-level directory.
 + */
 +
 +#ifndef TCG_CPUS_RR_H
 +#define TCG_CPUS_RR_H
 +
 +#define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
 +
 +/* Kick all RR vCPUs. */
 +void qemu_cpu_kick_rr_cpus(CPUState *unused);
 +
 +void *tcg_rr_cpu_thread_fn(void *arg);
 +
 +#endif /* TCG_CPUS_RR_H */
 diff --git a/accel/tcg/tcg-cpus.h b/accel/tcg/tcg-cpus.h
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus.h
 +++ b/accel/tcg/tcg-cpus.h
@@ -XXX,XX +XXX,XX @@
  /*
 - * Accelerator CPUS Interface
 + * QEMU TCG vCPU common functionality
 + *
 + * Functionality common to all TCG vcpu variants: mttcg, rr and icount.
   *
   * Copyright 2020 SUSE LLC
   *
@@ -XXX,XX +XXX,XX @@
  #include "sysemu/cpus.h"
 -extern const CpusAccel tcg_cpus;
 +extern const CpusAccel tcg_cpus_mttcg;
 +extern const CpusAccel tcg_cpus_icount;
 +extern const CpusAccel tcg_cpus_rr;
 +
 +void tcg_start_vcpu_thread(CPUState *cpu);
 +void qemu_tcg_destroy_vcpu(CPUState *cpu);
 +int tcg_cpu_exec(CPUState *cpu);
 +void tcg_handle_interrupt(CPUState *cpu, int mask);
  #endif /* TCG_CPUS_H */
 diff --git a/accel/tcg/tcg-all.c b/accel/tcg/tcg-all.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-all.c
 +++ b/accel/tcg/tcg-all.c
@@ -XXX,XX +XXX,XX @@ static int tcg_init(MachineState *ms)
      tcg_exec_init(s->tb_size * 1024 * 1024);
      mttcg_enabled = s->mttcg_enabled;
 -    cpus_register_accel(&tcg_cpus);
 +    if (mttcg_enabled) {
 +        cpus_register_accel(&tcg_cpus_mttcg);
 +    } else if (icount_enabled()) {
 +        cpus_register_accel(&tcg_cpus_icount);
 +    } else {
 +        cpus_register_accel(&tcg_cpus_rr);
 +    }
      return 0;
  }
 diff --git a/accel/tcg/tcg-cpus-icount.c b/accel/tcg/tcg-cpus-icount.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/accel/tcg/tcg-cpus-icount.c
@@ -XXX,XX +XXX,XX @@
 +/*
 + * QEMU TCG Single Threaded vCPUs implementation using instruction counting
 + *
 + * Copyright (c) 2003-2008 Fabrice Bellard
 + * Copyright (c) 2014 Red Hat Inc.
 + *
 + * Permission is hereby granted, free of charge, to any person obtaining a copy
 + * of this software and associated documentation files (the "Software"), to deal
 + * in the Software without restriction, including without limitation the rights
 + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 + * copies of the Software, and to permit persons to whom the Software is
 + * furnished to do so, subject to the following conditions:
 + *
 + * The above copyright notice and this permission notice shall be included in
 + * all copies or substantial portions of the Software.
 + *
 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 + * THE SOFTWARE.
 + */
 +
 +#include "qemu/osdep.h"
 +#include "qemu-common.h"
++#include "sysemu/tcg.h"
++#include "sysemu/replay.h"
++#include "qemu/main-loop.h"
++#include "qemu/guest-random.h"
 +#include "exec/exec-all.h"
-+
++#include "hw/boards.h"
-+#ifdef TARGET_PAGE_BITS_VARY
++
-+int target_page_bits;
++#include "tcg-cpus.h"
-+bool target_page_bits_decided;
++#include "tcg-cpus-icount.h"
 +#include "tcg-cpus-rr.h"
 +
 +static int64_t tcg_get_icount_limit(void)
 +{
 +    int64_t deadline;
 +
 +    if (replay_mode != REPLAY_MODE_PLAY) {
 +        /*
 +         * Include all the timers, because they may need an attention.
 +         * Too long CPU execution may create unnecessary delay in UI.
 +         */
 +        deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
 +                                              QEMU_TIMER_ATTR_ALL);
 +        /* Check realtime timers, because they help with input processing */
 +        deadline = qemu_soonest_timeout(deadline,
 +                qemu_clock_deadline_ns_all(QEMU_CLOCK_REALTIME,
 +                                           QEMU_TIMER_ATTR_ALL));
 +
 +        /*
 +         * Maintain prior (possibly buggy) behaviour where if no deadline
 +         * was set (as there is no QEMU_CLOCK_VIRTUAL timer) or it is more than
 +         * INT32_MAX nanoseconds ahead, we still use INT32_MAX
 +         * nanoseconds.
 +         */
 +        if ((deadline < 0) || (deadline > INT32_MAX)) {
 +            deadline = INT32_MAX;
 +        }
 +
 +        return icount_round(deadline);
 +    } else {
 +        return replay_get_instructions();
 +    }
 +}
 +
 +static void notify_aio_contexts(void)
 +{
 +    /* Wake up other AioContexts.  */
 +    qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
 +    qemu_clock_run_timers(QEMU_CLOCK_VIRTUAL);
 +}
 +
 +void handle_icount_deadline(void)
 +{
 +    assert(qemu_in_vcpu_thread());
 +    int64_t deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
 +                                                  QEMU_TIMER_ATTR_ALL);
 +
 +    if (deadline == 0) {
 +        notify_aio_contexts();
 +    }
 +}
 +
 +void prepare_icount_for_run(CPUState *cpu)
 +{
 +    int insns_left;
 +
 +    /*
 +     * These should always be cleared by process_icount_data after
 +     * each vCPU execution. However u16.high can be raised
 +     * asynchronously by cpu_exit/cpu_interrupt/tcg_handle_interrupt
 +     */
 +    g_assert(cpu_neg(cpu)->icount_decr.u16.low == 0);
 +    g_assert(cpu->icount_extra == 0);
 +
 +    cpu->icount_budget = tcg_get_icount_limit();
 +    insns_left = MIN(0xffff, cpu->icount_budget);
 +    cpu_neg(cpu)->icount_decr.u16.low = insns_left;
 +    cpu->icount_extra = cpu->icount_budget - insns_left;
 +
 +    replay_mutex_lock();
 +
 +    if (cpu->icount_budget == 0 && replay_has_checkpoint()) {
 +        notify_aio_contexts();
 +    }
 +}
 +
 +void process_icount_data(CPUState *cpu)
 +{
 +    /* Account for executed instructions */
 +    icount_update(cpu);
 +
 +    /* Reset the counters */
 +    cpu_neg(cpu)->icount_decr.u16.low = 0;
 +    cpu->icount_extra = 0;
 +    cpu->icount_budget = 0;
 +
 +    replay_account_executed_instructions();
 +
 +    replay_mutex_unlock();
 +}
 +
 +static void icount_handle_interrupt(CPUState *cpu, int mask)
 +{
 +    int old_mask = cpu->interrupt_request;
 +
 +    tcg_handle_interrupt(cpu, mask);
 +    if (qemu_cpu_is_self(cpu) &&
 +        !cpu->can_do_io
 +        && (mask & ~old_mask) != 0) {
 +        cpu_abort(cpu, "Raised interrupt while not in I/O function");
 +    }
 +}
 +
 +const CpusAccel tcg_cpus_icount = {
 +    .create_vcpu_thread = tcg_start_vcpu_thread,
 +    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
 +
 +    .handle_interrupt = icount_handle_interrupt,
 +    .get_virtual_clock = icount_get,
 +    .get_elapsed_ticks = icount_get,
 +};
 diff --git a/accel/tcg/tcg-cpus-mttcg.c b/accel/tcg/tcg-cpus-mttcg.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/accel/tcg/tcg-cpus-mttcg.c
@@ -XXX,XX +XXX,XX @@
 +/*
 + * QEMU TCG Multi Threaded vCPUs implementation
 + *
 + * Copyright (c) 2003-2008 Fabrice Bellard
 + * Copyright (c) 2014 Red Hat Inc.
 + *
 + * Permission is hereby granted, free of charge, to any person obtaining a copy
 + * of this software and associated documentation files (the "Software"), to deal
 + * in the Software without restriction, including without limitation the rights
 + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 + * copies of the Software, and to permit persons to whom the Software is
 + * furnished to do so, subject to the following conditions:
 + *
 + * The above copyright notice and this permission notice shall be included in
 + * all copies or substantial portions of the Software.
 + *
 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 + * THE SOFTWARE.
 + */
 +
 +#include "qemu/osdep.h"
 +#include "qemu-common.h"
 +#include "sysemu/tcg.h"
 +#include "sysemu/replay.h"
 +#include "qemu/main-loop.h"
 +#include "qemu/guest-random.h"
 +#include "exec/exec-all.h"
 +#include "hw/boards.h"
 +
 +#include "tcg-cpus.h"
 +#include "tcg-cpus-mttcg.h"
 +
 +/*
 + * In the multi-threaded case each vCPU has its own thread. The TLS
 + * variable current_cpu can be used deep in the code to find the
 + * current CPUState for a given thread.
 + */
 +
 +void *tcg_cpu_thread_fn(void *arg)
 +{
 +    CPUState *cpu = arg;
 +
 +    assert(tcg_enabled());
 +    g_assert(!icount_enabled());
 +
 +    rcu_register_thread();
 +    tcg_register_thread();
 +
 +    qemu_mutex_lock_iothread();
 +    qemu_thread_get_self(cpu->thread);
 +
 +    cpu->thread_id = qemu_get_thread_id();
 +    cpu->can_do_io = 1;
 +    current_cpu = cpu;
 +    cpu_thread_signal_created(cpu);
 +    qemu_guest_random_seed_thread_part2(cpu->random_seed);
 +
 +    /* process any pending work */
 +    cpu->exit_request = 1;
 +
 +    do {
 +        if (cpu_can_run(cpu)) {
 +            int r;
 +            qemu_mutex_unlock_iothread();
 +            r = tcg_cpu_exec(cpu);
 +            qemu_mutex_lock_iothread();
 +            switch (r) {
 +            case EXCP_DEBUG:
 +                cpu_handle_guest_debug(cpu);
 +                break;
 +            case EXCP_HALTED:
 +                /*
 +                 * during start-up the vCPU is reset and the thread is
 +                 * kicked several times. If we don't ensure we go back
 +                 * to sleep in the halted state we won't cleanly
 +                 * start-up when the vCPU is enabled.
 +                 *
 +                 * cpu->halted should ensure we sleep in wait_io_event
 +                 */
 +                g_assert(cpu->halted);
 +                break;
 +            case EXCP_ATOMIC:
 +                qemu_mutex_unlock_iothread();
 +                cpu_exec_step_atomic(cpu);
 +                qemu_mutex_lock_iothread();
 +            default:
 +                /* Ignore everything else? */
 +                break;
 +            }
 +        }
 +
 +        qatomic_mb_set(&cpu->exit_request, 0);
 +        qemu_wait_io_event(cpu);
 +    } while (!cpu->unplug || cpu_can_run(cpu));
 +
 +    qemu_tcg_destroy_vcpu(cpu);
 +    qemu_mutex_unlock_iothread();
 +    rcu_unregister_thread();
 +    return NULL;
 +}
 +
 +static void mttcg_kick_vcpu_thread(CPUState *cpu)
 +{
 +    cpu_exit(cpu);
 +}
 +
 +const CpusAccel tcg_cpus_mttcg = {
 +    .create_vcpu_thread = tcg_start_vcpu_thread,
 +    .kick_vcpu_thread = mttcg_kick_vcpu_thread,
 +
 +    .handle_interrupt = tcg_handle_interrupt,
 +};
 diff --git a/accel/tcg/tcg-cpus-rr.c b/accel/tcg/tcg-cpus-rr.c
 new file mode 100644
 index XXXXXXX..XXXXXXX
 --- /dev/null
 +++ b/accel/tcg/tcg-cpus-rr.c
@@ -XXX,XX +XXX,XX @@
 +/*
 + * QEMU TCG Single Threaded vCPUs implementation
 + *
 + * Copyright (c) 2003-2008 Fabrice Bellard
 + * Copyright (c) 2014 Red Hat Inc.
 + *
 + * Permission is hereby granted, free of charge, to any person obtaining a copy
 + * of this software and associated documentation files (the "Software"), to deal
 + * in the Software without restriction, including without limitation the rights
 + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 + * copies of the Software, and to permit persons to whom the Software is
 + * furnished to do so, subject to the following conditions:
 + *
 + * The above copyright notice and this permission notice shall be included in
 + * all copies or substantial portions of the Software.
 + *
 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 + * THE SOFTWARE.
 + */
 +
 +#include "qemu/osdep.h"
 +#include "qemu-common.h"
 +#include "sysemu/tcg.h"
 +#include "sysemu/replay.h"
 +#include "qemu/main-loop.h"
 +#include "qemu/guest-random.h"
 +#include "exec/exec-all.h"
 +#include "hw/boards.h"
 +
 +#include "tcg-cpus.h"
 +#include "tcg-cpus-rr.h"
 +#include "tcg-cpus-icount.h"
 +
 +/* Kick all RR vCPUs */
 +void qemu_cpu_kick_rr_cpus(CPUState *unused)
 +{
 +    CPUState *cpu;
 +
 +    CPU_FOREACH(cpu) {
 +        cpu_exit(cpu);
 +    };
 +}
 +
 +/*
 + * TCG vCPU kick timer
 + *
 + * The kick timer is responsible for moving single threaded vCPU
 + * emulation on to the next vCPU. If more than one vCPU is running a
 + * timer event with force a cpu->exit so the next vCPU can get
 + * scheduled.
 + *
 + * The timer is removed if all vCPUs are idle and restarted again once
 + * idleness is complete.
 + */
 +
 +static QEMUTimer *tcg_kick_vcpu_timer;
 +static CPUState *tcg_current_rr_cpu;
 +
 +#define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
 +
 +static inline int64_t qemu_tcg_next_kick(void)
 +{
 +    return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + TCG_KICK_PERIOD;
 +}
 +
 +/* Kick the currently round-robin scheduled vCPU to next */
 +static void qemu_cpu_kick_rr_next_cpu(void)
 +{
 +    CPUState *cpu;
 +    do {
 +        cpu = qatomic_mb_read(&tcg_current_rr_cpu);
 +        if (cpu) {
 +            cpu_exit(cpu);
 +        }
 +    } while (cpu != qatomic_mb_read(&tcg_current_rr_cpu));
 +}
 +
 +static void kick_tcg_thread(void *opaque)
 +{
 +    timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
 +    qemu_cpu_kick_rr_next_cpu();
 +}
 +
 +static void start_tcg_kick_timer(void)
 +{
 +    if (!tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
 +        tcg_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
 +                                           kick_tcg_thread, NULL);
 +    }
 +    if (tcg_kick_vcpu_timer && !timer_pending(tcg_kick_vcpu_timer)) {
 +        timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
 +    }
 +}
 +
 +static void stop_tcg_kick_timer(void)
 +{
 +    if (tcg_kick_vcpu_timer && timer_pending(tcg_kick_vcpu_timer)) {
 +        timer_del(tcg_kick_vcpu_timer);
 +    }
 +}
 +
 +static void qemu_tcg_rr_wait_io_event(void)
 +{
 +    CPUState *cpu;
 +
 +    while (all_cpu_threads_idle()) {
 +        stop_tcg_kick_timer();
 +        qemu_cond_wait_iothread(first_cpu->halt_cond);
 +    }
 +
 +    start_tcg_kick_timer();
 +
 +    CPU_FOREACH(cpu) {
 +        qemu_wait_io_event_common(cpu);
 +    }
 +}
 +
 +/*
 + * Destroy any remaining vCPUs which have been unplugged and have
 + * finished running
 + */
 +static void deal_with_unplugged_cpus(void)
 +{
 +    CPUState *cpu;
 +
 +    CPU_FOREACH(cpu) {
 +        if (cpu->unplug && !cpu_can_run(cpu)) {
 +            qemu_tcg_destroy_vcpu(cpu);
 +            break;
 +        }
 +    }
 +}
 +
 +/*
 + * In the single-threaded case each vCPU is simulated in turn. If
 + * there is more than a single vCPU we create a simple timer to kick
 + * the vCPU and ensure we don't get stuck in a tight loop in one vCPU.
 + * This is done explicitly rather than relying on side-effects
 + * elsewhere.
 + */
 +
 +void *tcg_rr_cpu_thread_fn(void *arg)
 +{
 +    CPUState *cpu = arg;
 +
 +    assert(tcg_enabled());
 +    rcu_register_thread();
 +    tcg_register_thread();
 +
 +    qemu_mutex_lock_iothread();
 +    qemu_thread_get_self(cpu->thread);
 +
 +    cpu->thread_id = qemu_get_thread_id();
 +    cpu->can_do_io = 1;
 +    cpu_thread_signal_created(cpu);
 +    qemu_guest_random_seed_thread_part2(cpu->random_seed);
 +
 +    /* wait for initial kick-off after machine start */
 +    while (first_cpu->stopped) {
 +        qemu_cond_wait_iothread(first_cpu->halt_cond);
 +
 +        /* process any pending work */
 +        CPU_FOREACH(cpu) {
 +            current_cpu = cpu;
 +            qemu_wait_io_event_common(cpu);
 +        }
 +    }
 +
 +    start_tcg_kick_timer();
 +
 +    cpu = first_cpu;
 +
 +    /* process any pending work */
 +    cpu->exit_request = 1;
 +
 +    while (1) {
 +        qemu_mutex_unlock_iothread();
 +        replay_mutex_lock();
 +        qemu_mutex_lock_iothread();
 +
 +        if (icount_enabled()) {
 +            /* Account partial waits to QEMU_CLOCK_VIRTUAL.  */
 +            icount_account_warp_timer();
 +            /*
 +             * Run the timers here.  This is much more efficient than
 +             * waking up the I/O thread and waiting for completion.
 +             */
 +            handle_icount_deadline();
 +        }
 +
 +        replay_mutex_unlock();
 +
 +        if (!cpu) {
 +            cpu = first_cpu;
 +        }
 +
 +        while (cpu && cpu_work_list_empty(cpu) && !cpu->exit_request) {
 +
 +            qatomic_mb_set(&tcg_current_rr_cpu, cpu);
 +            current_cpu = cpu;
 +
 +            qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
 +                              (cpu->singlestep_enabled & SSTEP_NOTIMER) == 0);
 +
 +            if (cpu_can_run(cpu)) {
 +                int r;
 +
 +                qemu_mutex_unlock_iothread();
 +                if (icount_enabled()) {
 +                    prepare_icount_for_run(cpu);
 +                }
 +                r = tcg_cpu_exec(cpu);
 +                if (icount_enabled()) {
 +                    process_icount_data(cpu);
 +                }
 +                qemu_mutex_lock_iothread();
 +
 +                if (r == EXCP_DEBUG) {
 +                    cpu_handle_guest_debug(cpu);
 +                    break;
 +                } else if (r == EXCP_ATOMIC) {
 +                    qemu_mutex_unlock_iothread();
 +                    cpu_exec_step_atomic(cpu);
 +                    qemu_mutex_lock_iothread();
 +                    break;
 +                }
 +            } else if (cpu->stop) {
 +                if (cpu->unplug) {
 +                    cpu = CPU_NEXT(cpu);
 +                }
 +                break;
 +            }
 +
 +            cpu = CPU_NEXT(cpu);
 +        } /* while (cpu && !cpu->exit_request).. */
 +
 +        /* Does not need qatomic_mb_set because a spurious wakeup is okay.  */
 +        qatomic_set(&tcg_current_rr_cpu, NULL);
 +
 +        if (cpu && cpu->exit_request) {
 +            qatomic_mb_set(&cpu->exit_request, 0);
 +        }
 +
 +        if (icount_enabled() && all_cpu_threads_idle()) {
 +            /*
 +             * When all cpus are sleeping (e.g in WFI), to avoid a deadlock
 +             * in the main_loop, wake it up in order to start the warp timer.
 +             */
 +            qemu_notify_event();
 +        }
 +
 +        qemu_tcg_rr_wait_io_event();
 +        deal_with_unplugged_cpus();
 +    }
 +
 +    rcu_unregister_thread();
 +    return NULL;
 +}
 +
 +const CpusAccel tcg_cpus_rr = {
 +    .create_vcpu_thread = tcg_start_vcpu_thread,
 +    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
 +
 +    .handle_interrupt = tcg_handle_interrupt,
 +};
 diff --git a/accel/tcg/tcg-cpus.c b/accel/tcg/tcg-cpus.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus.c
 +++ b/accel/tcg/tcg-cpus.c
@@ -XXX,XX +XXX,XX @@
  /*
 - * QEMU System Emulator
 + * QEMU TCG vCPU common functionality
 + *
 + * Functionality common to all TCG vCPU variants: mttcg, rr and icount.
   *
   * Copyright (c) 2003-2008 Fabrice Bellard
   * Copyright (c) 2014 Red Hat Inc.
@@ -XXX,XX +XXX,XX @@
  #include "hw/boards.h"
  #include "tcg-cpus.h"
 +#include "tcg-cpus-mttcg.h"
 +#include "tcg-cpus-rr.h"
 -/* Kick all RR vCPUs */
 -static void qemu_cpu_kick_rr_cpus(void)
 -{
 -    CPUState *cpu;
 +/* common functionality among all TCG variants */
 -    CPU_FOREACH(cpu) {
 -        cpu_exit(cpu);
 -    };
 -}
 -
 -static void tcg_kick_vcpu_thread(CPUState *cpu)
 -{
 -    if (qemu_tcg_mttcg_enabled()) {
 -        cpu_exit(cpu);
 -    } else {
 -        qemu_cpu_kick_rr_cpus();
 -    }
 -}
 -
 -/*
 - * TCG vCPU kick timer
 - *
 - * The kick timer is responsible for moving single threaded vCPU
 - * emulation on to the next vCPU. If more than one vCPU is running a
 - * timer event with force a cpu->exit so the next vCPU can get
 - * scheduled.
 - *
 - * The timer is removed if all vCPUs are idle and restarted again once
 - * idleness is complete.
 - */
 -
 -static QEMUTimer *tcg_kick_vcpu_timer;
 -static CPUState *tcg_current_rr_cpu;
 -
 -#define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
 -
 -static inline int64_t qemu_tcg_next_kick(void)
 -{
 -    return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + TCG_KICK_PERIOD;
 -}
 -
 -/* Kick the currently round-robin scheduled vCPU to next */
 -static void qemu_cpu_kick_rr_next_cpu(void)
 -{
 -    CPUState *cpu;
 -    do {
 -        cpu = qatomic_mb_read(&tcg_current_rr_cpu);
 -        if (cpu) {
 -            cpu_exit(cpu);
 -        }
 -    } while (cpu != qatomic_mb_read(&tcg_current_rr_cpu));
 -}
 -
 -static void kick_tcg_thread(void *opaque)
 -{
 -    timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
 -    qemu_cpu_kick_rr_next_cpu();
 -}
 -
 -static void start_tcg_kick_timer(void)
 -{
 -    assert(!mttcg_enabled);
 -    if (!tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
 -        tcg_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
 -                                           kick_tcg_thread, NULL);
 -    }
 -    if (tcg_kick_vcpu_timer && !timer_pending(tcg_kick_vcpu_timer)) {
 -        timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
 -    }
 -}
 -
 -static void stop_tcg_kick_timer(void)
 -{
 -    assert(!mttcg_enabled);
 -    if (tcg_kick_vcpu_timer && timer_pending(tcg_kick_vcpu_timer)) {
 -        timer_del(tcg_kick_vcpu_timer);
 -    }
 -}
 -
 -static void qemu_tcg_destroy_vcpu(CPUState *cpu)
 -{
 -}
 -
 -static void qemu_tcg_rr_wait_io_event(void)
 -{
 -    CPUState *cpu;
 -
 -    while (all_cpu_threads_idle()) {
 -        stop_tcg_kick_timer();
 -        qemu_cond_wait_iothread(first_cpu->halt_cond);
 -    }
 -
 -    start_tcg_kick_timer();
 -
 -    CPU_FOREACH(cpu) {
 -        qemu_wait_io_event_common(cpu);
 -    }
 -}
 -
 -static int64_t tcg_get_icount_limit(void)
 -{
 -    int64_t deadline;
 -
 -    if (replay_mode != REPLAY_MODE_PLAY) {
 -        /*
 -         * Include all the timers, because they may need an attention.
 -         * Too long CPU execution may create unnecessary delay in UI.
 -         */
 -        deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
 -                                              QEMU_TIMER_ATTR_ALL);
 -        /* Check realtime timers, because they help with input processing */
 -        deadline = qemu_soonest_timeout(deadline,
 -                qemu_clock_deadline_ns_all(QEMU_CLOCK_REALTIME,
 -                                           QEMU_TIMER_ATTR_ALL));
 -
 -        /*
 -         * Maintain prior (possibly buggy) behaviour where if no deadline
 -         * was set (as there is no QEMU_CLOCK_VIRTUAL timer) or it is more than
 -         * INT32_MAX nanoseconds ahead, we still use INT32_MAX
 -         * nanoseconds.
 -         */
 -        if ((deadline < 0) || (deadline > INT32_MAX)) {
 -            deadline = INT32_MAX;
 -        }
 -
 -        return icount_round(deadline);
 -    } else {
 -        return replay_get_instructions();
 -    }
 -}
 -
 -static void notify_aio_contexts(void)
 -{
 -    /* Wake up other AioContexts.  */
 -    qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
 -    qemu_clock_run_timers(QEMU_CLOCK_VIRTUAL);
 -}
 -
 -static void handle_icount_deadline(void)
 -{
 -    assert(qemu_in_vcpu_thread());
 -    if (icount_enabled()) {
 -        int64_t deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
 -                                                      QEMU_TIMER_ATTR_ALL);
 -
 -        if (deadline == 0) {
 -            notify_aio_contexts();
 -        }
 -    }
 -}
 -
 -static void prepare_icount_for_run(CPUState *cpu)
 -{
 -    if (icount_enabled()) {
 -        int insns_left;
 -
 -        /*
 -         * These should always be cleared by process_icount_data after
 -         * each vCPU execution. However u16.high can be raised
 -         * asynchronously by cpu_exit/cpu_interrupt/tcg_handle_interrupt
 -         */
 -        g_assert(cpu_neg(cpu)->icount_decr.u16.low == 0);
 -        g_assert(cpu->icount_extra == 0);
 -
 -        cpu->icount_budget = tcg_get_icount_limit();
 -        insns_left = MIN(0xffff, cpu->icount_budget);
 -        cpu_neg(cpu)->icount_decr.u16.low = insns_left;
 -        cpu->icount_extra = cpu->icount_budget - insns_left;
 -
 -        replay_mutex_lock();
 -
 -        if (cpu->icount_budget == 0 && replay_has_checkpoint()) {
 -            notify_aio_contexts();
 -        }
 -    }
 -}
 -
 -static void process_icount_data(CPUState *cpu)
 -{
 -    if (icount_enabled()) {
 -        /* Account for executed instructions */
 -        icount_update(cpu);
 -
 -        /* Reset the counters */
 -        cpu_neg(cpu)->icount_decr.u16.low = 0;
 -        cpu->icount_extra = 0;
 -        cpu->icount_budget = 0;
 -
 -        replay_account_executed_instructions();
 -
 -        replay_mutex_unlock();
 -    }
 -}
 -
 -static int tcg_cpu_exec(CPUState *cpu)
 -{
 -    int ret;
 -#ifdef CONFIG_PROFILER
 -    int64_t ti;
 -#endif
 -
 -    assert(tcg_enabled());
 -#ifdef CONFIG_PROFILER
 -    ti = profile_getclock();
 -#endif
 -    cpu_exec_start(cpu);
 -    ret = cpu_exec(cpu);
 -    cpu_exec_end(cpu);
 -#ifdef CONFIG_PROFILER
 -    qatomic_set(&tcg_ctx->prof.cpu_exec_time,
 -                tcg_ctx->prof.cpu_exec_time + profile_getclock() - ti);
 -#endif
 -    return ret;
 -}
 -
 -/*
 - * Destroy any remaining vCPUs which have been unplugged and have
 - * finished running
 - */
 -static void deal_with_unplugged_cpus(void)
 -{
 -    CPUState *cpu;
 -
 -    CPU_FOREACH(cpu) {
 -        if (cpu->unplug && !cpu_can_run(cpu)) {
 -            qemu_tcg_destroy_vcpu(cpu);
 -            cpu_thread_signal_destroyed(cpu);
 -            break;
 -        }
 -    }
 -}
 -
 -/*
 - * Single-threaded TCG
 - *
 - * In the single-threaded case each vCPU is simulated in turn. If
 - * there is more than a single vCPU we create a simple timer to kick
 - * the vCPU and ensure we don't get stuck in a tight loop in one vCPU.
 - * This is done explicitly rather than relying on side-effects
 - * elsewhere.
 - */
 -
 -static void *tcg_rr_cpu_thread_fn(void *arg)
 -{
 -    CPUState *cpu = arg;
 -
 -    assert(tcg_enabled());
 -    rcu_register_thread();
 -    tcg_register_thread();
 -
 -    qemu_mutex_lock_iothread();
 -    qemu_thread_get_self(cpu->thread);
 -
 -    cpu->thread_id = qemu_get_thread_id();
 -    cpu->can_do_io = 1;
 -    cpu_thread_signal_created(cpu);
 -    qemu_guest_random_seed_thread_part2(cpu->random_seed);
 -
 -    /* wait for initial kick-off after machine start */
 -    while (first_cpu->stopped) {
 -        qemu_cond_wait_iothread(first_cpu->halt_cond);
 -
 -        /* process any pending work */
 -        CPU_FOREACH(cpu) {
 -            current_cpu = cpu;
 -            qemu_wait_io_event_common(cpu);
 -        }
 -    }
 -
 -    start_tcg_kick_timer();
 -
 -    cpu = first_cpu;
 -
 -    /* process any pending work */
 -    cpu->exit_request = 1;
 -
 -    while (1) {
 -        qemu_mutex_unlock_iothread();
 -        replay_mutex_lock();
 -        qemu_mutex_lock_iothread();
 -        /* Account partial waits to QEMU_CLOCK_VIRTUAL.  */
 -        icount_account_warp_timer();
 -
 -        /*
 -         * Run the timers here.  This is much more efficient than
 -         * waking up the I/O thread and waiting for completion.
 -         */
 -        handle_icount_deadline();
 -
 -        replay_mutex_unlock();
 -
 -        if (!cpu) {
 -            cpu = first_cpu;
 -        }
 -
 -        while (cpu && cpu_work_list_empty(cpu) && !cpu->exit_request) {
 -
 -            qatomic_mb_set(&tcg_current_rr_cpu, cpu);
 -            current_cpu = cpu;
 -
 -            qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
 -                              (cpu->singlestep_enabled & SSTEP_NOTIMER) == 0);
 -
 -            if (cpu_can_run(cpu)) {
 -                int r;
 -
 -                qemu_mutex_unlock_iothread();
 -                prepare_icount_for_run(cpu);
 -
 -                r = tcg_cpu_exec(cpu);
 -
 -                process_icount_data(cpu);
 -                qemu_mutex_lock_iothread();
 -
 -                if (r == EXCP_DEBUG) {
 -                    cpu_handle_guest_debug(cpu);
 -                    break;
 -                } else if (r == EXCP_ATOMIC) {
 -                    qemu_mutex_unlock_iothread();
 -                    cpu_exec_step_atomic(cpu);
 -                    qemu_mutex_lock_iothread();
 -                    break;
 -                }
 -            } else if (cpu->stop) {
 -                if (cpu->unplug) {
 -                    cpu = CPU_NEXT(cpu);
 -                }
 -                break;
 -            }
 -
 -            cpu = CPU_NEXT(cpu);
 -        } /* while (cpu && !cpu->exit_request).. */
 -
 -        /* Does not need qatomic_mb_set because a spurious wakeup is okay.  */
 -        qatomic_set(&tcg_current_rr_cpu, NULL);
 -
 -        if (cpu && cpu->exit_request) {
 -            qatomic_mb_set(&cpu->exit_request, 0);
 -        }
 -
 -        if (icount_enabled() && all_cpu_threads_idle()) {
 -            /*
 -             * When all cpus are sleeping (e.g in WFI), to avoid a deadlock
 -             * in the main_loop, wake it up in order to start the warp timer.
 -             */
 -            qemu_notify_event();
 -        }
 -
 -        qemu_tcg_rr_wait_io_event();
 -        deal_with_unplugged_cpus();
 -    }
 -
 -    rcu_unregister_thread();
 -    return NULL;
 -}
 -
 -/*
 - * Multi-threaded TCG
 - *
 - * In the multi-threaded case each vCPU has its own thread. The TLS
 - * variable current_cpu can be used deep in the code to find the
 - * current CPUState for a given thread.
 - */
 -
 -static void *tcg_cpu_thread_fn(void *arg)
 -{
 -    CPUState *cpu = arg;
 -
 -    assert(tcg_enabled());
 -    g_assert(!icount_enabled());
 -
 -    rcu_register_thread();
 -    tcg_register_thread();
 -
 -    qemu_mutex_lock_iothread();
 -    qemu_thread_get_self(cpu->thread);
 -
 -    cpu->thread_id = qemu_get_thread_id();
 -    cpu->can_do_io = 1;
 -    current_cpu = cpu;
 -    cpu_thread_signal_created(cpu);
 -    qemu_guest_random_seed_thread_part2(cpu->random_seed);
 -
 -    /* process any pending work */
 -    cpu->exit_request = 1;
 -
 -    do {
 -        if (cpu_can_run(cpu)) {
 -            int r;
 -            qemu_mutex_unlock_iothread();
 -            r = tcg_cpu_exec(cpu);
 -            qemu_mutex_lock_iothread();
 -            switch (r) {
 -            case EXCP_DEBUG:
 -                cpu_handle_guest_debug(cpu);
 -                break;
 -            case EXCP_HALTED:
 -                /*
 -                 * during start-up the vCPU is reset and the thread is
 -                 * kicked several times. If we don't ensure we go back
 -                 * to sleep in the halted state we won't cleanly
 -                 * start-up when the vCPU is enabled.
 -                 *
 -                 * cpu->halted should ensure we sleep in wait_io_event
 -                 */
 -                g_assert(cpu->halted);
 -                break;
 -            case EXCP_ATOMIC:
 -                qemu_mutex_unlock_iothread();
 -                cpu_exec_step_atomic(cpu);
 -                qemu_mutex_lock_iothread();
 -            default:
 -                /* Ignore everything else? */
 -                break;
 -            }
 -        }
 -
 -        qatomic_mb_set(&cpu->exit_request, 0);
 -        qemu_wait_io_event(cpu);
 -    } while (!cpu->unplug || cpu_can_run(cpu));
 -
 -    qemu_tcg_destroy_vcpu(cpu);
 -    cpu_thread_signal_destroyed(cpu);
 -    qemu_mutex_unlock_iothread();
 -    rcu_unregister_thread();
 -    return NULL;
 -}
 -
 -static void tcg_start_vcpu_thread(CPUState *cpu)
 +void tcg_start_vcpu_thread(CPUState *cpu)
  {
      char thread_name[VCPU_THREAD_NAME_SIZE];
      static QemuCond *single_tcg_halt_cond;
@@ -XXX,XX +XXX,XX @@ static void tcg_start_vcpu_thread(CPUState *cpu)
      }
  }
 -static int64_t tcg_get_virtual_clock(void)
 +void qemu_tcg_destroy_vcpu(CPUState *cpu)
  {
 -    if (icount_enabled()) {
 -        return icount_get();
 -    }
 -    return cpu_get_clock();
 +    cpu_thread_signal_destroyed(cpu);
  }
 -static int64_t tcg_get_elapsed_ticks(void)
 +int tcg_cpu_exec(CPUState *cpu)
  {
 -    if (icount_enabled()) {
 -        return icount_get();
 -    }
 -    return cpu_get_ticks();
 +    int ret;
 +#ifdef CONFIG_PROFILER
 +    int64_t ti;
 +#endif
-+
++    assert(tcg_enabled());
-+bool set_preferred_target_page_bits(int bits)
++#ifdef CONFIG_PROFILER
-+{
++    ti = profile_getclock();
 +    /*
 +     * The target page size is the lowest common denominator for all
 +     * the CPUs in the system, so we can only make it smaller, never
 +     * larger. And we can't make it smaller once we've committed to
 +     * a particular size.
 +     */
 +#ifdef TARGET_PAGE_BITS_VARY
 +    assert(bits >= TARGET_PAGE_BITS_MIN);
 +    if (target_page_bits == 0 || target_page_bits > bits) {
 +        if (target_page_bits_decided) {
 +            return false;
 +        }
 +        target_page_bits = bits;
 +    }
 +#endif
-+    return true;
++    cpu_exec_start(cpu);
-+}
++    ret = cpu_exec(cpu);
-+
++    cpu_exec_end(cpu);
-+void finalize_target_page_bits(void)
++#ifdef CONFIG_PROFILER
-+{
++    qatomic_set(&tcg_ctx->prof.cpu_exec_time,
-+#ifdef TARGET_PAGE_BITS_VARY
++                tcg_ctx->prof.cpu_exec_time + profile_getclock() - ti);
 +    if (target_page_bits == 0) {
 +        target_page_bits = TARGET_PAGE_BITS_MIN;
 +    }
 +    target_page_bits_decided = true;
 +#endif
-+}
++    return ret;
-diff --git a/exec.c b/exec.c
+ }
  /* mask must never be zero, except for A20 change call */
 -static void tcg_handle_interrupt(CPUState *cpu, int mask)
 +void tcg_handle_interrupt(CPUState *cpu, int mask)
  {
 -    int old_mask;
      g_assert(qemu_mutex_iothread_locked());
 -    old_mask = cpu->interrupt_request;
      cpu->interrupt_request |= mask;
      /*
@@ -XXX,XX +XXX,XX @@ static void tcg_handle_interrupt(CPUState *cpu, int mask)
          qemu_cpu_kick(cpu);
      } else {
          qatomic_set(&cpu_neg(cpu)->icount_decr.u16.high, -1);
 -        if (icount_enabled() &&
 -            !cpu->can_do_io
 -            && (mask & ~old_mask) != 0) {
 -            cpu_abort(cpu, "Raised interrupt while not in I/O function");
 -        }
      }
  }
 -
 -const CpusAccel tcg_cpus = {
 -    .create_vcpu_thread = tcg_start_vcpu_thread,
 -    .kick_vcpu_thread = tcg_kick_vcpu_thread,
 -
 -    .handle_interrupt = tcg_handle_interrupt,
 -
 -    .get_virtual_clock = tcg_get_virtual_clock,
 -    .get_elapsed_ticks = tcg_get_elapsed_ticks,
 -};
 diff --git a/softmmu/icount.c b/softmmu/icount.c
 index XXXXXXX..XXXXXXX 100644
---- a/exec.c
+--- a/softmmu/icount.c
-+++ b/exec.c
++++ b/softmmu/icount.c
-@@ -XXX,XX +XXX,XX @@ AddressSpace address_space_memory;
+@@ -XXX,XX +XXX,XX @@ void icount_start_warp_timer(void)
- static MemoryRegion io_mem_unassigned;
- #endif
+ void icount_account_warp_timer(void)
+ {
--#ifdef TARGET_PAGE_BITS_VARY
+-    if (!icount_enabled() || !icount_sleep) {
--int target_page_bits;
++    if (!icount_sleep) {
--bool target_page_bits_decided;
+         return;
--#endif
+     }
--
- CPUTailQ cpus = QTAILQ_HEAD_INITIALIZER(cpus);
+diff --git a/accel/tcg/meson.build b/accel/tcg/meson.build
+index XXXXXXX..XXXXXXX 100644
- /* current CPU in the current thread. It is only valid inside
+--- a/accel/tcg/meson.build
-@@ -XXX,XX +XXX,XX @@ int use_icount;
++++ b/accel/tcg/meson.build
- uintptr_t qemu_host_page_size;
+@@ -XXX,XX +XXX,XX @@ tcg_ss.add(when: 'CONFIG_SOFTMMU', if_false: files('user-exec-stub.c'))
- intptr_t qemu_host_page_mask;
+ tcg_ss.add(when: 'CONFIG_PLUGIN', if_true: [files('plugin-gen.c'), libdl])
+ specific_ss.add_all(when: 'CONFIG_TCG', if_true: tcg_ss)
--bool set_preferred_target_page_bits(int bits)
--{
+-specific_ss.add(when: ['CONFIG_SOFTMMU', 'CONFIG_TCG'], if_true: files('tcg-all.c', 'cputlb.c', 'tcg-cpus.c'))
--    /* The target page size is the lowest common denominator for all
++specific_ss.add(when: ['CONFIG_SOFTMMU', 'CONFIG_TCG'], if_true: files(
--     * the CPUs in the system, so we can only make it smaller, never
++  'tcg-all.c',
--     * larger. And we can't make it smaller once we've committed to
++  'cputlb.c',
--     * a particular size.
++  'tcg-cpus.c',
--     */
++  'tcg-cpus-mttcg.c',
--#ifdef TARGET_PAGE_BITS_VARY
++  'tcg-cpus-icount.c',
--    assert(bits >= TARGET_PAGE_BITS_MIN);
++  'tcg-cpus-rr.c'
--    if (target_page_bits == 0 || target_page_bits > bits) {
++))
 -        if (target_page_bits_decided) {
 -            return false;
 -        }
 -        target_page_bits = bits;
 -    }
 -#endif
 -    return true;
 -}
 -
  #if !defined(CONFIG_USER_ONLY)
 -static void finalize_target_page_bits(void)
 -{
 -#ifdef TARGET_PAGE_BITS_VARY
 -    if (target_page_bits == 0) {
 -        target_page_bits = TARGET_PAGE_BITS_MIN;
 -    }
 -    target_page_bits_decided = true;
 -#endif
 -}
 -
  typedef struct PhysPageEntry PhysPageEntry;
  struct PhysPageEntry {
 --
-.17.1
+.25.1

-[PULL 02/12] cputlb: ensure _cmmu helper functions follow the naming standard
+[PULL 2/3] accel/tcg: split tcg_start_vcpu_thread
-From: Alex Bennée <alex.bennee@linaro.org>
+From: Claudio Fontana <cfontana@suse.de>
-We document this in docs/devel/load-stores.rst so lets follow it. The
+after the initial split into 3 tcg variants, we proceed to also
-bit and 64 bit access functions have historically not included the
+split tcg_start_vcpu_thread.
-sign so we leave those as is. We also introduce some signed helpers
-which are used for loading immediate values in the translator.
+We actually split it in 2 this time, since the icount variant
+just uses the round robin function.
-Fixes: 282dffc8
-Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
+Suggested-by: Richard Henderson <richard.henderson@linaro.org>
-Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
+Signed-off-by: Claudio Fontana <cfontana@suse.de>
-Message-Id: <20191021150910.23216-1-alex.bennee@linaro.org>
+Message-Id: <20201015143217.29337-3-cfontana@suse.de>
 Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
 ---
- include/exec/cpu_ldst_template.h |  4 ++--
+ accel/tcg/tcg-cpus-mttcg.h  | 21 --------------
- tcg/tcg.h                        | 20 ++++++++++++++------
+ accel/tcg/tcg-cpus-rr.h     |  3 +-
- accel/tcg/cputlb.c               | 24 +++++++++++++++++++++---
+ accel/tcg/tcg-cpus.h        |  1 -
- target/cris/translate_v10.inc.c  |  3 +--
+ accel/tcg/tcg-all.c         |  5 ++++
-files changed, 38 insertions(+), 13 deletions(-)
+ accel/tcg/tcg-cpus-icount.c |  2 +-
+ accel/tcg/tcg-cpus-mttcg.c  | 29 +++++++++++++++++--
-diff --git a/include/exec/cpu_ldst_template.h b/include/exec/cpu_ldst_template.h
+ accel/tcg/tcg-cpus-rr.c     | 39 +++++++++++++++++++++++--
-index XXXXXXX..XXXXXXX 100644
+ accel/tcg/tcg-cpus.c        | 58 -------------------------------------
---- a/include/exec/cpu_ldst_template.h
+files changed, 71 insertions(+), 87 deletions(-)
-+++ b/include/exec/cpu_ldst_template.h
+ delete mode 100644 accel/tcg/tcg-cpus-mttcg.h
-@@ -XXX,XX +XXX,XX @@
- #ifdef SOFTMMU_CODE_ACCESS
+diff --git a/accel/tcg/tcg-cpus-mttcg.h b/accel/tcg/tcg-cpus-mttcg.h
- #define ADDR_READ addr_code
+deleted file mode 100644
- #define MMUSUFFIX _cmmu
+index XXXXXXX..XXXXXXX
--#define URETSUFFIX SUFFIX
+--- a/accel/tcg/tcg-cpus-mttcg.h
--#define SRETSUFFIX SUFFIX
++++ /dev/null
-+#define URETSUFFIX USUFFIX
+@@ -XXX,XX +XXX,XX @@
-+#define SRETSUFFIX glue(s, SUFFIX)
+-/*
- #else
+- * QEMU TCG Multi Threaded vCPUs implementation
- #define ADDR_READ addr_read
+- *
- #define MMUSUFFIX _mmu
+- * Copyright 2020 SUSE LLC
-diff --git a/tcg/tcg.h b/tcg/tcg.h
+- *
-index XXXXXXX..XXXXXXX 100644
+- * This work is licensed under the terms of the GNU GPL, version 2 or later.
---- a/tcg/tcg.h
+- * See the COPYING file in the top-level directory.
-+++ b/tcg/tcg.h
+- */
-@@ -XXX,XX +XXX,XX @@ void helper_be_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val,
+-
- void helper_be_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val,
+-#ifndef TCG_CPUS_MTTCG_H
-                        TCGMemOpIdx oi, uintptr_t retaddr);
+-#define TCG_CPUS_MTTCG_H
+-
--uint8_t helper_ret_ldb_cmmu(CPUArchState *env, target_ulong addr,
+-/*
-+uint8_t helper_ret_ldub_cmmu(CPUArchState *env, target_ulong addr,
+- * In the multi-threaded case each vCPU has its own thread. The TLS
-                             TCGMemOpIdx oi, uintptr_t retaddr);
+- * variable current_cpu can be used deep in the code to find the
--uint16_t helper_le_ldw_cmmu(CPUArchState *env, target_ulong addr,
+- * current CPUState for a given thread.
-+int8_t helper_ret_ldsb_cmmu(CPUArchState *env, target_ulong addr,
+- */
-                             TCGMemOpIdx oi, uintptr_t retaddr);
+-
-+uint16_t helper_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
+-void *tcg_cpu_thread_fn(void *arg);
-+                             TCGMemOpIdx oi, uintptr_t retaddr);
+-
-+int16_t helper_le_ldsw_cmmu(CPUArchState *env, target_ulong addr,
+-#endif /* TCG_CPUS_MTTCG_H */
-+                             TCGMemOpIdx oi, uintptr_t retaddr);
+diff --git a/accel/tcg/tcg-cpus-rr.h b/accel/tcg/tcg-cpus-rr.h
- uint32_t helper_le_ldl_cmmu(CPUArchState *env, target_ulong addr,
+index XXXXXXX..XXXXXXX 100644
-                             TCGMemOpIdx oi, uintptr_t retaddr);
+--- a/accel/tcg/tcg-cpus-rr.h
- uint64_t helper_le_ldq_cmmu(CPUArchState *env, target_ulong addr,
++++ b/accel/tcg/tcg-cpus-rr.h
-                             TCGMemOpIdx oi, uintptr_t retaddr);
+@@ -XXX,XX +XXX,XX @@
--uint16_t helper_be_ldw_cmmu(CPUArchState *env, target_ulong addr,
+ /* Kick all RR vCPUs. */
--                            TCGMemOpIdx oi, uintptr_t retaddr);
+ void qemu_cpu_kick_rr_cpus(CPUState *unused);
-+uint16_t helper_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
-+                             TCGMemOpIdx oi, uintptr_t retaddr);
+-void *tcg_rr_cpu_thread_fn(void *arg);
-+int16_t helper_be_ldsw_cmmu(CPUArchState *env, target_ulong addr,
++/* start the round robin vcpu thread */
-+                             TCGMemOpIdx oi, uintptr_t retaddr);
++void rr_start_vcpu_thread(CPUState *cpu);
- uint32_t helper_be_ldl_cmmu(CPUArchState *env, target_ulong addr,
-                             TCGMemOpIdx oi, uintptr_t retaddr);
+ #endif /* TCG_CPUS_RR_H */
- uint64_t helper_be_ldq_cmmu(CPUArchState *env, target_ulong addr,
+diff --git a/accel/tcg/tcg-cpus.h b/accel/tcg/tcg-cpus.h
-@@ -XXX,XX +XXX,XX @@ uint64_t helper_be_ldq_cmmu(CPUArchState *env, target_ulong addr,
+index XXXXXXX..XXXXXXX 100644
- # define helper_ret_stw_mmu   helper_be_stw_mmu
+--- a/accel/tcg/tcg-cpus.h
- # define helper_ret_stl_mmu   helper_be_stl_mmu
++++ b/accel/tcg/tcg-cpus.h
- # define helper_ret_stq_mmu   helper_be_stq_mmu
+@@ -XXX,XX +XXX,XX @@ extern const CpusAccel tcg_cpus_mttcg;
--# define helper_ret_ldw_cmmu  helper_be_ldw_cmmu
+ extern const CpusAccel tcg_cpus_icount;
-+# define helper_ret_lduw_cmmu  helper_be_lduw_cmmu
+ extern const CpusAccel tcg_cpus_rr;
-+# define helper_ret_ldsw_cmmu  helper_be_ldsw_cmmu
- # define helper_ret_ldl_cmmu  helper_be_ldl_cmmu
+-void tcg_start_vcpu_thread(CPUState *cpu);
- # define helper_ret_ldq_cmmu  helper_be_ldq_cmmu
+ void qemu_tcg_destroy_vcpu(CPUState *cpu);
- #else
+ int tcg_cpu_exec(CPUState *cpu);
-@@ -XXX,XX +XXX,XX @@ uint64_t helper_be_ldq_cmmu(CPUArchState *env, target_ulong addr,
+ void tcg_handle_interrupt(CPUState *cpu, int mask);
- # define helper_ret_stw_mmu   helper_le_stw_mmu
+diff --git a/accel/tcg/tcg-all.c b/accel/tcg/tcg-all.c
- # define helper_ret_stl_mmu   helper_le_stl_mmu
+index XXXXXXX..XXXXXXX 100644
- # define helper_ret_stq_mmu   helper_le_stq_mmu
+--- a/accel/tcg/tcg-all.c
--# define helper_ret_ldw_cmmu  helper_le_ldw_cmmu
++++ b/accel/tcg/tcg-all.c
-+# define helper_ret_lduw_cmmu  helper_le_lduw_cmmu
+@@ -XXX,XX +XXX,XX @@ static int tcg_init(MachineState *ms)
-+# define helper_ret_ldsw_cmmu  helper_le_ldsw_cmmu
+     tcg_exec_init(s->tb_size * 1024 * 1024);
- # define helper_ret_ldl_cmmu  helper_le_ldl_cmmu
+     mttcg_enabled = s->mttcg_enabled;
- # define helper_ret_ldq_cmmu  helper_le_ldq_cmmu
- #endif
++    /*
-diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
++     * Initialize TCG regions
-index XXXXXXX..XXXXXXX 100644
++     */
---- a/accel/tcg/cputlb.c
++    tcg_region_init();
-+++ b/accel/tcg/cputlb.c
++
-@@ -XXX,XX +XXX,XX @@ static uint64_t full_ldub_cmmu(CPUArchState *env, target_ulong addr,
+     if (mttcg_enabled) {
-     return load_helper(env, addr, oi, retaddr, MO_8, true, full_ldub_cmmu);
+         cpus_register_accel(&tcg_cpus_mttcg);
      } else if (icount_enabled()) {
 diff --git a/accel/tcg/tcg-cpus-icount.c b/accel/tcg/tcg-cpus-icount.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus-icount.c
 +++ b/accel/tcg/tcg-cpus-icount.c
@@ -XXX,XX +XXX,XX @@ static void icount_handle_interrupt(CPUState *cpu, int mask)
  }
--uint8_t helper_ret_ldb_cmmu(CPUArchState *env, target_ulong addr,
+ const CpusAccel tcg_cpus_icount = {
-+uint8_t helper_ret_ldub_cmmu(CPUArchState *env, target_ulong addr,
+-    .create_vcpu_thread = tcg_start_vcpu_thread,
-                             TCGMemOpIdx oi, uintptr_t retaddr)
++    .create_vcpu_thread = rr_start_vcpu_thread,
      .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
      .handle_interrupt = icount_handle_interrupt,
 diff --git a/accel/tcg/tcg-cpus-mttcg.c b/accel/tcg/tcg-cpus-mttcg.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus-mttcg.c
 +++ b/accel/tcg/tcg-cpus-mttcg.c
@@ -XXX,XX +XXX,XX @@
  #include "hw/boards.h"
  #include "tcg-cpus.h"
 -#include "tcg-cpus-mttcg.h"
  /*
   * In the multi-threaded case each vCPU has its own thread. The TLS
@@ -XXX,XX +XXX,XX @@
   * current CPUState for a given thread.
   */
 -void *tcg_cpu_thread_fn(void *arg)
 +static void *tcg_cpu_thread_fn(void *arg)
  {
-     return full_ldub_cmmu(env, addr, oi, retaddr);
+     CPUState *cpu = arg;
@@ -XXX,XX +XXX,XX @@ static void mttcg_kick_vcpu_thread(CPUState *cpu)
      cpu_exit(cpu);
  }
-+int8_t helper_ret_ldsb_cmmu(CPUArchState *env, target_ulong addr,
++static void mttcg_start_vcpu_thread(CPUState *cpu)
 +                            TCGMemOpIdx oi, uintptr_t retaddr)
 +{
-+    return (int8_t) full_ldub_cmmu(env, addr, oi, retaddr);
++    char thread_name[VCPU_THREAD_NAME_SIZE];
 +
 +    g_assert(tcg_enabled());
 +
 +    parallel_cpus = (current_machine->smp.max_cpus > 1);
 +
 +    cpu->thread = g_malloc0(sizeof(QemuThread));
 +    cpu->halt_cond = g_malloc0(sizeof(QemuCond));
 +    qemu_cond_init(cpu->halt_cond);
 +
 +    /* create a thread per vCPU with TCG (MTTCG) */
 +    snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
 +             cpu->cpu_index);
 +
 +    qemu_thread_create(cpu->thread, thread_name, tcg_cpu_thread_fn,
 +                       cpu, QEMU_THREAD_JOINABLE);
 +
 +#ifdef _WIN32
 +    cpu->hThread = qemu_thread_get_handle(cpu->thread);
 +#endif
 +}
 +
- static uint64_t full_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
+ const CpusAccel tcg_cpus_mttcg = {
-                                   TCGMemOpIdx oi, uintptr_t retaddr)
+-    .create_vcpu_thread = tcg_start_vcpu_thread,
 +    .create_vcpu_thread = mttcg_start_vcpu_thread,
      .kick_vcpu_thread = mttcg_kick_vcpu_thread,
      .handle_interrupt = tcg_handle_interrupt,
 diff --git a/accel/tcg/tcg-cpus-rr.c b/accel/tcg/tcg-cpus-rr.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus-rr.c
 +++ b/accel/tcg/tcg-cpus-rr.c
@@ -XXX,XX +XXX,XX @@ static void deal_with_unplugged_cpus(void)
   * elsewhere.
   */
 -void *tcg_rr_cpu_thread_fn(void *arg)
 +static void *tcg_rr_cpu_thread_fn(void *arg)
  {
-@@ -XXX,XX +XXX,XX @@ static uint64_t full_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
+     CPUState *cpu = arg;
-                        full_le_lduw_cmmu);
@@ -XXX,XX +XXX,XX @@ void *tcg_rr_cpu_thread_fn(void *arg)
      return NULL;
  }
--uint16_t helper_le_ldw_cmmu(CPUArchState *env, target_ulong addr,
++void rr_start_vcpu_thread(CPUState *cpu)
-+uint16_t helper_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
++{
-                             TCGMemOpIdx oi, uintptr_t retaddr)
++    char thread_name[VCPU_THREAD_NAME_SIZE];
 +    static QemuCond *single_tcg_halt_cond;
 +    static QemuThread *single_tcg_cpu_thread;
 +
 +    g_assert(tcg_enabled());
 +    parallel_cpus = false;
 +
 +    if (!single_tcg_cpu_thread) {
 +        cpu->thread = g_malloc0(sizeof(QemuThread));
 +        cpu->halt_cond = g_malloc0(sizeof(QemuCond));
 +        qemu_cond_init(cpu->halt_cond);
 +
 +        /* share a single thread for all cpus with TCG */
 +        snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "ALL CPUs/TCG");
 +        qemu_thread_create(cpu->thread, thread_name,
 +                           tcg_rr_cpu_thread_fn,
 +                           cpu, QEMU_THREAD_JOINABLE);
 +
 +        single_tcg_halt_cond = cpu->halt_cond;
 +        single_tcg_cpu_thread = cpu->thread;
 +#ifdef _WIN32
 +        cpu->hThread = qemu_thread_get_handle(cpu->thread);
 +#endif
 +    } else {
 +        /* we share the thread */
 +        cpu->thread = single_tcg_cpu_thread;
 +        cpu->halt_cond = single_tcg_halt_cond;
 +        cpu->thread_id = first_cpu->thread_id;
 +        cpu->can_do_io = 1;
 +        cpu->created = true;
 +    }
 +}
 +
  const CpusAccel tcg_cpus_rr = {
 -    .create_vcpu_thread = tcg_start_vcpu_thread,
 +    .create_vcpu_thread = rr_start_vcpu_thread,
      .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
      .handle_interrupt = tcg_handle_interrupt,
 diff --git a/accel/tcg/tcg-cpus.c b/accel/tcg/tcg-cpus.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus.c
 +++ b/accel/tcg/tcg-cpus.c
@@ -XXX,XX +XXX,XX @@
  #include "hw/boards.h"
  #include "tcg-cpus.h"
 -#include "tcg-cpus-mttcg.h"
 -#include "tcg-cpus-rr.h"
  /* common functionality among all TCG variants */
 -void tcg_start_vcpu_thread(CPUState *cpu)
 -{
 -    char thread_name[VCPU_THREAD_NAME_SIZE];
 -    static QemuCond *single_tcg_halt_cond;
 -    static QemuThread *single_tcg_cpu_thread;
 -    static int tcg_region_inited;
 -
 -    assert(tcg_enabled());
 -    /*
 -     * Initialize TCG regions--once. Now is a good time, because:
 -     * (1) TCG's init context, prologue and target globals have been set up.
 -     * (2) qemu_tcg_mttcg_enabled() works now (TCG init code runs before the
 -     *     -accel flag is processed, so the check doesn't work then).
 -     */
 -    if (!tcg_region_inited) {
 -        tcg_region_inited = 1;
 -        tcg_region_init();
 -        parallel_cpus = qemu_tcg_mttcg_enabled() && current_machine->smp.max_cpus > 1;
 -    }
 -
 -    if (qemu_tcg_mttcg_enabled() || !single_tcg_cpu_thread) {
 -        cpu->thread = g_malloc0(sizeof(QemuThread));
 -        cpu->halt_cond = g_malloc0(sizeof(QemuCond));
 -        qemu_cond_init(cpu->halt_cond);
 -
 -        if (qemu_tcg_mttcg_enabled()) {
 -            /* create a thread per vCPU with TCG (MTTCG) */
 -            snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
 -                 cpu->cpu_index);
 -
 -            qemu_thread_create(cpu->thread, thread_name, tcg_cpu_thread_fn,
 -                               cpu, QEMU_THREAD_JOINABLE);
 -
 -        } else {
 -            /* share a single thread for all cpus with TCG */
 -            snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "ALL CPUs/TCG");
 -            qemu_thread_create(cpu->thread, thread_name,
 -                               tcg_rr_cpu_thread_fn,
 -                               cpu, QEMU_THREAD_JOINABLE);
 -
 -            single_tcg_halt_cond = cpu->halt_cond;
 -            single_tcg_cpu_thread = cpu->thread;
 -        }
 -#ifdef _WIN32
 -        cpu->hThread = qemu_thread_get_handle(cpu->thread);
 -#endif
 -    } else {
 -        /* For non-MTTCG cases we share the thread */
 -        cpu->thread = single_tcg_cpu_thread;
 -        cpu->halt_cond = single_tcg_halt_cond;
 -        cpu->thread_id = first_cpu->thread_id;
 -        cpu->can_do_io = 1;
 -        cpu->created = true;
 -    }
 -}
 -
  void qemu_tcg_destroy_vcpu(CPUState *cpu)
  {
-     return full_le_lduw_cmmu(env, addr, oi, retaddr);
+     cpu_thread_signal_destroyed(cpu);
  }
 +int16_t helper_le_ldsw_cmmu(CPUArchState *env, target_ulong addr,
 +                            TCGMemOpIdx oi, uintptr_t retaddr)
 +{
 +    return (int16_t) full_le_lduw_cmmu(env, addr, oi, retaddr);
 +}
 +
  static uint64_t full_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
                                    TCGMemOpIdx oi, uintptr_t retaddr)
  {
@@ -XXX,XX +XXX,XX @@ static uint64_t full_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
                         full_be_lduw_cmmu);
  }
 -uint16_t helper_be_ldw_cmmu(CPUArchState *env, target_ulong addr,
 +uint16_t helper_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
                              TCGMemOpIdx oi, uintptr_t retaddr)
  {
      return full_be_lduw_cmmu(env, addr, oi, retaddr);
  }
 +int16_t helper_be_ldsw_cmmu(CPUArchState *env, target_ulong addr,
 +                            TCGMemOpIdx oi, uintptr_t retaddr)
 +{
 +    return (int16_t) full_be_lduw_cmmu(env, addr, oi, retaddr);
 +}
 +
  static uint64_t full_le_ldul_cmmu(CPUArchState *env, target_ulong addr,
                                    TCGMemOpIdx oi, uintptr_t retaddr)
  {
 diff --git a/target/cris/translate_v10.inc.c b/target/cris/translate_v10.inc.c
 index XXXXXXX..XXXXXXX 100644
 --- a/target/cris/translate_v10.inc.c
 +++ b/target/cris/translate_v10.inc.c
@@ -XXX,XX +XXX,XX @@ static unsigned int dec10_ind(CPUCRISState *env, DisasContext *dc)
          case CRISV10_IND_BCC_M:
              cris_cc_mask(dc, 0);
 -            imm = cpu_ldsw_code(env, dc->pc + 2);
 -            simm = (int16_t)imm;
 +            simm = cpu_ldsw_code(env, dc->pc + 2);
              simm += 4;
              LOG_DIS("bcc_m: b%s %x\n", cc_name(dc->cond), dc->pc + simm);
 --
-.17.1
+.25.1

-[PULL 03/12] cpu: use ROUND_UP() to define xxx_PAGE_ALIGN
+Deleted patch
-From: Wei Yang <richardw.yang@linux.intel.com>
-Use ROUND_UP() to define, which is a little bit easy to read.
-Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
-Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
-Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
-Reviewed-by: Juan Quintela <quintela@redhat.com>
-Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
-Message-Id: <20191013021145.16011-2-richardw.yang@linux.intel.com>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- include/exec/cpu-all.h | 7 +++----
-file changed, 3 insertions(+), 4 deletions(-)
-diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
-index XXXXXXX..XXXXXXX 100644
---- a/include/exec/cpu-all.h
-+++ b/include/exec/cpu-all.h
-@@ -XXX,XX +XXX,XX @@ extern int target_page_bits;
- #define TARGET_PAGE_SIZE (1 << TARGET_PAGE_BITS)
- #define TARGET_PAGE_MASK ~(TARGET_PAGE_SIZE - 1)
--#define TARGET_PAGE_ALIGN(addr) (((addr) + TARGET_PAGE_SIZE - 1) & TARGET_PAGE_MASK)
-+#define TARGET_PAGE_ALIGN(addr) ROUND_UP((addr), TARGET_PAGE_SIZE)
- /* Using intptr_t ensures that qemu_*_page_mask is sign-extended even
-  * when intptr_t is 32-bit and we are aligning a long long.
-@@ -XXX,XX +XXX,XX @@ extern int target_page_bits;
- extern uintptr_t qemu_host_page_size;
- extern intptr_t qemu_host_page_mask;
--#define HOST_PAGE_ALIGN(addr) (((addr) + qemu_host_page_size - 1) & qemu_host_page_mask)
--#define REAL_HOST_PAGE_ALIGN(addr) (((addr) + qemu_real_host_page_size - 1) & \
--                                    qemu_real_host_page_mask)
-+#define HOST_PAGE_ALIGN(addr) ROUND_UP((addr), qemu_host_page_size)
-+#define REAL_HOST_PAGE_ALIGN(addr) ROUND_UP((addr), qemu_real_host_page_size)
- /* same as PROT_xxx */
- #define PAGE_READ      0x0001
---
-.17.1

-[PULL 05/12] configure: Detect compiler support for __attribute__((alias))
+Deleted patch
-Such support is present almost everywhere, except for Xcode 9.
-It is added in Xcode 10, but travis uses xcode9 by default,
-so we should support it for a while yet.
-Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- configure | 19 +++++++++++++++++++
-file changed, 19 insertions(+)
-diff --git a/configure b/configure
-index XXXXXXX..XXXXXXX 100755
---- a/configure
-+++ b/configure
-@@ -XXX,XX +XXX,XX @@ if compile_prog "" "" ; then
-   vector16=yes
- fi
-+########################################
-+# See if __attribute__((alias)) is supported.
-+# This false for Xcode 9, but has been remedied for Xcode 10.
-+# Unfortunately, travis uses Xcode 9 by default.
-+
-+attralias=no
-+cat > $TMPC << EOF
-+int x = 1;
-+extern const int y __attribute__((alias("x")));
-+int main(void) { return 0; }
-+EOF
-+if compile_prog "" "" ; then
-+    attralias=yes
-+fi
-+
- ########################################
- # check if getauxval is available.
-@@ -XXX,XX +XXX,XX @@ if test "$vector16" = "yes" ; then
-   echo "CONFIG_VECTOR16=y" >> $config_host_mak
- fi
-+if test "$attralias" = "yes" ; then
-+  echo "CONFIG_ATTRIBUTE_ALIAS=y" >> $config_host_mak
-+fi
-+
- if test "$getauxval" = "yes" ; then
-   echo "CONFIG_GETAUXVAL=y" >> $config_host_mak
- fi
---
-.17.1

-[PULL 06/12] exec: Use const alias for TARGET_PAGE_BITS_VARY
+Deleted patch
-Using a variable that is declared "const" for this tells the
-compiler that it may read the value once and assume that it
-does not change across function calls.
-For target_page_size, this means we have only one assert per
-function, and one read of the variable.
-This reduces the size of qemu-system-aarch64 by 8k.
-Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
-Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- include/exec/cpu-all.h | 14 +++++++---
- exec-vary.c            | 60 ++++++++++++++++++++++++++++++++++++------
-files changed, 62 insertions(+), 12 deletions(-)
-diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
-index XXXXXXX..XXXXXXX 100644
---- a/include/exec/cpu-all.h
-+++ b/include/exec/cpu-all.h
-@@ -XXX,XX +XXX,XX @@ static inline void stl_phys_notdirty(AddressSpace *as, hwaddr addr, uint32_t val
- /* page related stuff */
- #ifdef TARGET_PAGE_BITS_VARY
--extern bool target_page_bits_decided;
--extern int target_page_bits;
--#define TARGET_PAGE_BITS ({ assert(target_page_bits_decided); \
--                            target_page_bits; })
-+typedef struct {
-+    bool decided;
-+    int bits;
-+} TargetPageBits;
-+# if defined(CONFIG_ATTRIBUTE_ALIAS) || !defined(IN_EXEC_VARY)
-+extern const TargetPageBits target_page;
-+#else
-+extern TargetPageBits target_page;
-+# endif
-+#define TARGET_PAGE_BITS (assert(target_page.decided), target_page.bits)
- #else
- #define TARGET_PAGE_BITS_MIN TARGET_PAGE_BITS
- #endif
-diff --git a/exec-vary.c b/exec-vary.c
-index XXXXXXX..XXXXXXX 100644
---- a/exec-vary.c
-+++ b/exec-vary.c
-@@ -XXX,XX +XXX,XX @@
- #include "qemu/osdep.h"
- #include "qemu-common.h"
-+
-+#define IN_EXEC_VARY 1
-+
- #include "exec/exec-all.h"
- #ifdef TARGET_PAGE_BITS_VARY
--int target_page_bits;
--bool target_page_bits_decided;
-+# ifdef CONFIG_ATTRIBUTE_ALIAS
-+/*
-+ * We want to declare the "target_page" variable as const, which tells
-+ * the compiler that it can cache any value that it reads across calls.
-+ * This avoids multiple assertions and multiple reads within any one user.
-+ *
-+ * This works because we initialize the target_page data very early, in a
-+ * location far removed from the functions that require the final results.
-+ *
-+ * This also requires that we have a non-constant symbol by which we can
-+ * perform the actual initialization, and which forces the data to be
-+ * allocated within writable memory.  Thus "init_target_page", and we use
-+ * that symbol exclusively in the two functions that initialize this value.
-+ *
-+ * The "target_page" symbol is created as an alias of "init_target_page".
-+ */
-+static TargetPageBits init_target_page;
-+
-+/*
-+ * Note that this is *not* a redundant decl, this is the definition of
-+ * the "target_page" symbol.  The syntax for this definition requires
-+ * the use of the extern keyword.  This seems to be a GCC bug in
-+ * either the syntax for the alias attribute or in -Wredundant-decls.
-+ *
-+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91765
-+ */
-+#  pragma GCC diagnostic push
-+#  pragma GCC diagnostic ignored "-Wredundant-decls"
-+
-+extern const TargetPageBits target_page
-+    __attribute__((alias("init_target_page")));
-+
-+#  pragma GCC diagnostic pop
-+# else
-+/*
-+ * When aliases are not supported then we force two different declarations,
-+ * by way of suppressing the header declaration with IN_EXEC_VARY.
-+ * We assume that on such an old compiler, LTO cannot be used, and so the
-+ * compiler cannot not detect the mismatched declarations, and all is well.
-+ */
-+TargetPageBits target_page;
-+#  define init_target_page target_page
-+# endif
- #endif
- bool set_preferred_target_page_bits(int bits)
-@@ -XXX,XX +XXX,XX @@ bool set_preferred_target_page_bits(int bits)
-      */
- #ifdef TARGET_PAGE_BITS_VARY
-     assert(bits >= TARGET_PAGE_BITS_MIN);
--    if (target_page_bits == 0 || target_page_bits > bits) {
--        if (target_page_bits_decided) {
-+    if (init_target_page.bits == 0 || init_target_page.bits > bits) {
-+        if (init_target_page.decided) {
-             return false;
-         }
--        target_page_bits = bits;
-+        init_target_page.bits = bits;
-     }
- #endif
-     return true;
-@@ -XXX,XX +XXX,XX @@ bool set_preferred_target_page_bits(int bits)
- void finalize_target_page_bits(void)
- {
- #ifdef TARGET_PAGE_BITS_VARY
--    if (target_page_bits == 0) {
--        target_page_bits = TARGET_PAGE_BITS_MIN;
-+    if (init_target_page.bits == 0) {
-+        init_target_page.bits = TARGET_PAGE_BITS_MIN;
-     }
--    target_page_bits_decided = true;
-+    init_target_page.decided = true;
- #endif
- }
---
-.17.1

-[PULL 07/12] exec: Restrict TARGET_PAGE_BITS_VARY assert to CONFIG_DEBUG_TCG
+Deleted patch
-This reduces the size of a release build by about 10k.
-Noticably, within the tlb miss helpers.
-Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
-Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
-Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- include/exec/cpu-all.h | 6 +++++-
-file changed, 5 insertions(+), 1 deletion(-)
-diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
-index XXXXXXX..XXXXXXX 100644
---- a/include/exec/cpu-all.h
-+++ b/include/exec/cpu-all.h
-@@ -XXX,XX +XXX,XX @@ extern const TargetPageBits target_page;
- #else
- extern TargetPageBits target_page;
- # endif
--#define TARGET_PAGE_BITS (assert(target_page.decided), target_page.bits)
-+# ifdef CONFIG_DEBUG_TCG
-+#  define TARGET_PAGE_BITS (assert(target_page.decided), target_page.bits)
-+# else
-+#  define TARGET_PAGE_BITS target_page.bits
-+# endif
- #else
- #define TARGET_PAGE_BITS_MIN TARGET_PAGE_BITS
- #endif
---
-.17.1

-[PULL 08/12] exec: Promote TARGET_PAGE_MASK to target_long
+Deleted patch
-There are some uint64_t uses that expect TARGET_PAGE_MASK to
-extend for a 32-bit, so this must continue to be a signed type.
-Define based on TARGET_PAGE_BITS not TARGET_PAGE_SIZE; this
-will make a following patch more clear.
-This should not have a functional effect so far.
-Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
-Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- include/exec/cpu-all.h | 2 +-
-file changed, 1 insertion(+), 1 deletion(-)
-diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
-index XXXXXXX..XXXXXXX 100644
---- a/include/exec/cpu-all.h
-+++ b/include/exec/cpu-all.h
-@@ -XXX,XX +XXX,XX @@ extern TargetPageBits target_page;
- #endif
- #define TARGET_PAGE_SIZE (1 << TARGET_PAGE_BITS)
--#define TARGET_PAGE_MASK ~(TARGET_PAGE_SIZE - 1)
-+#define TARGET_PAGE_MASK ((target_long)-1 << TARGET_PAGE_BITS)
- #define TARGET_PAGE_ALIGN(addr) ROUND_UP((addr), TARGET_PAGE_SIZE)
- /* Using intptr_t ensures that qemu_*_page_mask is sign-extended even
---
-.17.1

-[PULL 09/12] exec: Cache TARGET_PAGE_MASK for TARGET_PAGE_BITS_VARY
+Deleted patch
-This eliminates a set of runtime shifts.  It turns out that we
-require TARGET_PAGE_MASK more often than TARGET_PAGE_SIZE, so
-redefine TARGET_PAGE_SIZE based on TARGET_PAGE_MASK instead of
-the other way around.
-Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
-Reviewed-by: David Hildenbrand <david@redhat.com>
-Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- include/exec/cpu-all.h | 8 ++++++--
- exec-vary.c            | 1 +
-files changed, 7 insertions(+), 2 deletions(-)
-diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
-index XXXXXXX..XXXXXXX 100644
---- a/include/exec/cpu-all.h
-+++ b/include/exec/cpu-all.h
-@@ -XXX,XX +XXX,XX @@ static inline void stl_phys_notdirty(AddressSpace *as, hwaddr addr, uint32_t val
- typedef struct {
-     bool decided;
-     int bits;
-+    target_long mask;
- } TargetPageBits;
- # if defined(CONFIG_ATTRIBUTE_ALIAS) || !defined(IN_EXEC_VARY)
- extern const TargetPageBits target_page;
-@@ -XXX,XX +XXX,XX @@ extern TargetPageBits target_page;
- # endif
- # ifdef CONFIG_DEBUG_TCG
- #  define TARGET_PAGE_BITS (assert(target_page.decided), target_page.bits)
-+#  define TARGET_PAGE_MASK (assert(target_page.decided), target_page.mask)
- # else
- #  define TARGET_PAGE_BITS target_page.bits
-+#  define TARGET_PAGE_MASK target_page.mask
- # endif
-+# define TARGET_PAGE_SIZE  (-(int)TARGET_PAGE_MASK)
- #else
- #define TARGET_PAGE_BITS_MIN TARGET_PAGE_BITS
-+#define TARGET_PAGE_SIZE   (1 << TARGET_PAGE_BITS)
-+#define TARGET_PAGE_MASK   ((target_long)-1 << TARGET_PAGE_BITS)
- #endif
--#define TARGET_PAGE_SIZE (1 << TARGET_PAGE_BITS)
--#define TARGET_PAGE_MASK ((target_long)-1 << TARGET_PAGE_BITS)
- #define TARGET_PAGE_ALIGN(addr) ROUND_UP((addr), TARGET_PAGE_SIZE)
- /* Using intptr_t ensures that qemu_*_page_mask is sign-extended even
-diff --git a/exec-vary.c b/exec-vary.c
-index XXXXXXX..XXXXXXX 100644
---- a/exec-vary.c
-+++ b/exec-vary.c
-@@ -XXX,XX +XXX,XX @@ void finalize_target_page_bits(void)
-         init_target_page.bits = TARGET_PAGE_BITS_MIN;
-     }
-     init_target_page.decided = true;
-+    init_target_page.mask = (target_long)-1 << init_target_page.bits;
- #endif
- }
---
-.17.1

-[PULL 10/12] cputlb: Fix tlb_vaddr_to_host
+Deleted patch
-Using uintptr_t instead of target_ulong meant that, for 64-bit guest
-and 32-bit host, we truncated the guest address comparator and so may
-not hit the tlb when we should.
-Fixes: 4811e9095c0
-Reviewed-by: David Hildenbrand <david@redhat.com>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- accel/tcg/cputlb.c | 2 +-
-file changed, 1 insertion(+), 1 deletion(-)
-diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
-index XXXXXXX..XXXXXXX 100644
---- a/accel/tcg/cputlb.c
-+++ b/accel/tcg/cputlb.c
-@@ -XXX,XX +XXX,XX @@ void *tlb_vaddr_to_host(CPUArchState *env, abi_ptr addr,
-                         MMUAccessType access_type, int mmu_idx)
- {
-     CPUTLBEntry *entry = tlb_entry(env, mmu_idx, addr);
--    uintptr_t tlb_addr, page;
-+    target_ulong tlb_addr, page;
-     size_t elt_ofs;
-     switch (access_type) {
---
-.17.1

-[PULL 11/12] translate-all: fix uninitialized tb->orig_tb
+Deleted patch
-From: Clement Deschamps <clement.deschamps@greensocs.com>
-This fixes a segmentation fault in icount mode when executing
-from an IO region.
-TB is marked as CF_NOCACHE but tb->orig_tb is not initialized
-(equals previous value in code_gen_buffer).
-The issue happens in cpu_io_recompile() when it tries to invalidate orig_tb.
-Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
-Signed-off-by: Clement Deschamps <clement.deschamps@greensocs.com>
-Message-Id: <20191022140016.918371-1-clement.deschamps@greensocs.com>
-Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
----
- accel/tcg/translate-all.c | 1 +
-file changed, 1 insertion(+)
-diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
-index XXXXXXX..XXXXXXX 100644
---- a/accel/tcg/translate-all.c
-+++ b/accel/tcg/translate-all.c
-@@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu,
-     tb->cs_base = cs_base;
-     tb->flags = flags;
-     tb->cflags = cflags;
-+    tb->orig_tb = NULL;
-     tb->trace_vcpu_dstate = *cpu->trace_dstate;
-     tcg_ctx->tb_cflags = cflags;
-  tb_overflow:
---
-.17.1

-[PULL 12/12] translate-all: Remove tb_alloc
+[PULL 3/3] accel/tcg: rename tcg-cpus functions to match module name
-Since 2ac01d6dafab, this function does only two things: assert a
+From: Claudio Fontana <cfontana@suse.de>
 lock is held, and call tcg_tb_alloc.  It is used exactly once,
 and its user has already done the assert.
-Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
+Signed-off-by: Claudio Fontana <cfontana@suse.de>
-Reviewed-by: Clement Deschamps <clement.deschamps@greensocs.com>
+Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
 Message-Id: <20201015143217.29337-4-cfontana@suse.de>
 Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
 ---
- accel/tcg/translate-all.c | 20 ++------------------
+ accel/tcg/tcg-cpus-icount.h |  6 +--
-file changed, 2 insertions(+), 18 deletions(-)
+ accel/tcg/tcg-cpus-rr.h     |  2 +-
  accel/tcg/tcg-cpus.h        |  6 +--
  accel/tcg/tcg-cpus-icount.c | 24 ++++++------
  accel/tcg/tcg-cpus-mttcg.c  | 10 ++---
  accel/tcg/tcg-cpus-rr.c     | 74 ++++++++++++++++++-------------------
  accel/tcg/tcg-cpus.c        |  6 +--
 files changed, 64 insertions(+), 64 deletions(-)
-diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
+diff --git a/accel/tcg/tcg-cpus-icount.h b/accel/tcg/tcg-cpus-icount.h
 index XXXXXXX..XXXXXXX 100644
---- a/accel/tcg/translate-all.c
+--- a/accel/tcg/tcg-cpus-icount.h
-+++ b/accel/tcg/translate-all.c
++++ b/accel/tcg/tcg-cpus-icount.h
-@@ -XXX,XX +XXX,XX @@ void tcg_exec_init(unsigned long tb_size)
+@@ -XXX,XX +XXX,XX @@
- #endif
+ #ifndef TCG_CPUS_ICOUNT_H
- }
+ #define TCG_CPUS_ICOUNT_H
--/*
+-void handle_icount_deadline(void);
-- * Allocate a new translation block. Flush the translation buffer if
+-void prepare_icount_for_run(CPUState *cpu);
-- * too many translation blocks or too much generated code.
+-void process_icount_data(CPUState *cpu);
-- */
++void icount_handle_deadline(void);
--static TranslationBlock *tb_alloc(target_ulong pc)
++void icount_prepare_for_run(CPUState *cpu);
--{
++void icount_process_data(CPUState *cpu);
--    TranslationBlock *tb;
--
+ #endif /* TCG_CPUS_ICOUNT_H */
--    assert_memory_lock();
+diff --git a/accel/tcg/tcg-cpus-rr.h b/accel/tcg/tcg-cpus-rr.h
--
+index XXXXXXX..XXXXXXX 100644
--    tb = tcg_tb_alloc(tcg_ctx);
+--- a/accel/tcg/tcg-cpus-rr.h
--    if (unlikely(tb == NULL)) {
++++ b/accel/tcg/tcg-cpus-rr.h
--        return NULL;
+@@ -XXX,XX +XXX,XX @@
--    }
+ #define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
--    return tb;
--}
+ /* Kick all RR vCPUs. */
--
+-void qemu_cpu_kick_rr_cpus(CPUState *unused);
- /* call with @p->lock held */
++void rr_kick_vcpu_thread(CPUState *unused);
- static inline void invalidate_page_bitmap(PageDesc *p)
- {
+ /* start the round robin vcpu thread */
-@@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu,
+ void rr_start_vcpu_thread(CPUState *cpu);
-     TCGProfile *prof = &tcg_ctx->prof;
+diff --git a/accel/tcg/tcg-cpus.h b/accel/tcg/tcg-cpus.h
-     int64_t ti;
+index XXXXXXX..XXXXXXX 100644
- #endif
+--- a/accel/tcg/tcg-cpus.h
-+
++++ b/accel/tcg/tcg-cpus.h
-     assert_memory_lock();
+@@ -XXX,XX +XXX,XX @@ extern const CpusAccel tcg_cpus_mttcg;
+ extern const CpusAccel tcg_cpus_icount;
-     phys_pc = get_page_addr_code(env, pc);
+ extern const CpusAccel tcg_cpus_rr;
-@@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu,
-     }
+-void qemu_tcg_destroy_vcpu(CPUState *cpu);
+-int tcg_cpu_exec(CPUState *cpu);
-  buffer_overflow:
+-void tcg_handle_interrupt(CPUState *cpu, int mask);
--    tb = tb_alloc(pc);
++void tcg_cpus_destroy(CPUState *cpu);
-+    tb = tcg_tb_alloc(tcg_ctx);
++int tcg_cpus_exec(CPUState *cpu);
-     if (unlikely(!tb)) {
++void tcg_cpus_handle_interrupt(CPUState *cpu, int mask);
-         /* flush must be done */
-         tb_flush(cpu);
+ #endif /* TCG_CPUS_H */
 diff --git a/accel/tcg/tcg-cpus-icount.c b/accel/tcg/tcg-cpus-icount.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus-icount.c
 +++ b/accel/tcg/tcg-cpus-icount.c
@@ -XXX,XX +XXX,XX @@
  #include "tcg-cpus-icount.h"
  #include "tcg-cpus-rr.h"
 -static int64_t tcg_get_icount_limit(void)
 +static int64_t icount_get_limit(void)
  {
      int64_t deadline;
@@ -XXX,XX +XXX,XX @@ static int64_t tcg_get_icount_limit(void)
      }
  }
 -static void notify_aio_contexts(void)
 +static void icount_notify_aio_contexts(void)
  {
      /* Wake up other AioContexts.  */
      qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
      qemu_clock_run_timers(QEMU_CLOCK_VIRTUAL);
  }
 -void handle_icount_deadline(void)
 +void icount_handle_deadline(void)
  {
      assert(qemu_in_vcpu_thread());
      int64_t deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
                                                    QEMU_TIMER_ATTR_ALL);
      if (deadline == 0) {
 -        notify_aio_contexts();
 +        icount_notify_aio_contexts();
      }
  }
 -void prepare_icount_for_run(CPUState *cpu)
 +void icount_prepare_for_run(CPUState *cpu)
  {
      int insns_left;
      /*
 -     * These should always be cleared by process_icount_data after
 +     * These should always be cleared by icount_process_data after
       * each vCPU execution. However u16.high can be raised
 -     * asynchronously by cpu_exit/cpu_interrupt/tcg_handle_interrupt
 +     * asynchronously by cpu_exit/cpu_interrupt/tcg_cpus_handle_interrupt
       */
      g_assert(cpu_neg(cpu)->icount_decr.u16.low == 0);
      g_assert(cpu->icount_extra == 0);
 -    cpu->icount_budget = tcg_get_icount_limit();
 +    cpu->icount_budget = icount_get_limit();
      insns_left = MIN(0xffff, cpu->icount_budget);
      cpu_neg(cpu)->icount_decr.u16.low = insns_left;
      cpu->icount_extra = cpu->icount_budget - insns_left;
@@ -XXX,XX +XXX,XX @@ void prepare_icount_for_run(CPUState *cpu)
      replay_mutex_lock();
      if (cpu->icount_budget == 0 && replay_has_checkpoint()) {
 -        notify_aio_contexts();
 +        icount_notify_aio_contexts();
      }
  }
 -void process_icount_data(CPUState *cpu)
 +void icount_process_data(CPUState *cpu)
  {
      /* Account for executed instructions */
      icount_update(cpu);
@@ -XXX,XX +XXX,XX @@ static void icount_handle_interrupt(CPUState *cpu, int mask)
  {
      int old_mask = cpu->interrupt_request;
 -    tcg_handle_interrupt(cpu, mask);
 +    tcg_cpus_handle_interrupt(cpu, mask);
      if (qemu_cpu_is_self(cpu) &&
          !cpu->can_do_io
          && (mask & ~old_mask) != 0) {
@@ -XXX,XX +XXX,XX @@ static void icount_handle_interrupt(CPUState *cpu, int mask)
  const CpusAccel tcg_cpus_icount = {
      .create_vcpu_thread = rr_start_vcpu_thread,
 -    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
 +    .kick_vcpu_thread = rr_kick_vcpu_thread,
      .handle_interrupt = icount_handle_interrupt,
      .get_virtual_clock = icount_get,
 diff --git a/accel/tcg/tcg-cpus-mttcg.c b/accel/tcg/tcg-cpus-mttcg.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus-mttcg.c
 +++ b/accel/tcg/tcg-cpus-mttcg.c
@@ -XXX,XX +XXX,XX @@
   * current CPUState for a given thread.
   */
 -static void *tcg_cpu_thread_fn(void *arg)
 +static void *mttcg_cpu_thread_fn(void *arg)
  {
      CPUState *cpu = arg;
@@ -XXX,XX +XXX,XX @@ static void *tcg_cpu_thread_fn(void *arg)
          if (cpu_can_run(cpu)) {
              int r;
              qemu_mutex_unlock_iothread();
 -            r = tcg_cpu_exec(cpu);
 +            r = tcg_cpus_exec(cpu);
              qemu_mutex_lock_iothread();
              switch (r) {
              case EXCP_DEBUG:
@@ -XXX,XX +XXX,XX @@ static void *tcg_cpu_thread_fn(void *arg)
          qemu_wait_io_event(cpu);
      } while (!cpu->unplug || cpu_can_run(cpu));
 -    qemu_tcg_destroy_vcpu(cpu);
 +    tcg_cpus_destroy(cpu);
      qemu_mutex_unlock_iothread();
      rcu_unregister_thread();
      return NULL;
@@ -XXX,XX +XXX,XX @@ static void mttcg_start_vcpu_thread(CPUState *cpu)
      snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
               cpu->cpu_index);
 -    qemu_thread_create(cpu->thread, thread_name, tcg_cpu_thread_fn,
 +    qemu_thread_create(cpu->thread, thread_name, mttcg_cpu_thread_fn,
                         cpu, QEMU_THREAD_JOINABLE);
  #ifdef _WIN32
@@ -XXX,XX +XXX,XX @@ const CpusAccel tcg_cpus_mttcg = {
      .create_vcpu_thread = mttcg_start_vcpu_thread,
      .kick_vcpu_thread = mttcg_kick_vcpu_thread,
 -    .handle_interrupt = tcg_handle_interrupt,
 +    .handle_interrupt = tcg_cpus_handle_interrupt,
  };
 diff --git a/accel/tcg/tcg-cpus-rr.c b/accel/tcg/tcg-cpus-rr.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus-rr.c
 +++ b/accel/tcg/tcg-cpus-rr.c
@@ -XXX,XX +XXX,XX @@
  #include "tcg-cpus-icount.h"
  /* Kick all RR vCPUs */
 -void qemu_cpu_kick_rr_cpus(CPUState *unused)
 +void rr_kick_vcpu_thread(CPUState *unused)
  {
      CPUState *cpu;
@@ -XXX,XX +XXX,XX @@ void qemu_cpu_kick_rr_cpus(CPUState *unused)
   * idleness is complete.
   */
 -static QEMUTimer *tcg_kick_vcpu_timer;
 -static CPUState *tcg_current_rr_cpu;
 +static QEMUTimer *rr_kick_vcpu_timer;
 +static CPUState *rr_current_cpu;
  #define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
 -static inline int64_t qemu_tcg_next_kick(void)
 +static inline int64_t rr_next_kick_time(void)
  {
      return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + TCG_KICK_PERIOD;
  }
  /* Kick the currently round-robin scheduled vCPU to next */
 -static void qemu_cpu_kick_rr_next_cpu(void)
 +static void rr_kick_next_cpu(void)
  {
      CPUState *cpu;
      do {
 -        cpu = qatomic_mb_read(&tcg_current_rr_cpu);
 +        cpu = qatomic_mb_read(&rr_current_cpu);
          if (cpu) {
              cpu_exit(cpu);
          }
 -    } while (cpu != qatomic_mb_read(&tcg_current_rr_cpu));
 +    } while (cpu != qatomic_mb_read(&rr_current_cpu));
  }
 -static void kick_tcg_thread(void *opaque)
 +static void rr_kick_thread(void *opaque)
  {
 -    timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
 -    qemu_cpu_kick_rr_next_cpu();
 +    timer_mod(rr_kick_vcpu_timer, rr_next_kick_time());
 +    rr_kick_next_cpu();
  }
 -static void start_tcg_kick_timer(void)
 +static void rr_start_kick_timer(void)
  {
 -    if (!tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
 -        tcg_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
 -                                           kick_tcg_thread, NULL);
 +    if (!rr_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
 +        rr_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
 +                                           rr_kick_thread, NULL);
      }
 -    if (tcg_kick_vcpu_timer && !timer_pending(tcg_kick_vcpu_timer)) {
 -        timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
 +    if (rr_kick_vcpu_timer && !timer_pending(rr_kick_vcpu_timer)) {
 +        timer_mod(rr_kick_vcpu_timer, rr_next_kick_time());
      }
  }
 -static void stop_tcg_kick_timer(void)
 +static void rr_stop_kick_timer(void)
  {
 -    if (tcg_kick_vcpu_timer && timer_pending(tcg_kick_vcpu_timer)) {
 -        timer_del(tcg_kick_vcpu_timer);
 +    if (rr_kick_vcpu_timer && timer_pending(rr_kick_vcpu_timer)) {
 +        timer_del(rr_kick_vcpu_timer);
      }
  }
 -static void qemu_tcg_rr_wait_io_event(void)
 +static void rr_wait_io_event(void)
  {
      CPUState *cpu;
      while (all_cpu_threads_idle()) {
 -        stop_tcg_kick_timer();
 +        rr_stop_kick_timer();
          qemu_cond_wait_iothread(first_cpu->halt_cond);
      }
 -    start_tcg_kick_timer();
 +    rr_start_kick_timer();
      CPU_FOREACH(cpu) {
          qemu_wait_io_event_common(cpu);
@@ -XXX,XX +XXX,XX @@ static void qemu_tcg_rr_wait_io_event(void)
   * Destroy any remaining vCPUs which have been unplugged and have
   * finished running
   */
 -static void deal_with_unplugged_cpus(void)
 +static void rr_deal_with_unplugged_cpus(void)
  {
      CPUState *cpu;
      CPU_FOREACH(cpu) {
          if (cpu->unplug && !cpu_can_run(cpu)) {
 -            qemu_tcg_destroy_vcpu(cpu);
 +            tcg_cpus_destroy(cpu);
              break;
          }
      }
@@ -XXX,XX +XXX,XX @@ static void deal_with_unplugged_cpus(void)
   * elsewhere.
   */
 -static void *tcg_rr_cpu_thread_fn(void *arg)
 +static void *rr_cpu_thread_fn(void *arg)
  {
      CPUState *cpu = arg;
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
          }
      }
 -    start_tcg_kick_timer();
 +    rr_start_kick_timer();
      cpu = first_cpu;
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
               * Run the timers here.  This is much more efficient than
               * waking up the I/O thread and waiting for completion.
               */
 -            handle_icount_deadline();
 +            icount_handle_deadline();
          }
          replay_mutex_unlock();
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
          while (cpu && cpu_work_list_empty(cpu) && !cpu->exit_request) {
 -            qatomic_mb_set(&tcg_current_rr_cpu, cpu);
 +            qatomic_mb_set(&rr_current_cpu, cpu);
              current_cpu = cpu;
              qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
                  qemu_mutex_unlock_iothread();
                  if (icount_enabled()) {
 -                    prepare_icount_for_run(cpu);
 +                    icount_prepare_for_run(cpu);
                  }
 -                r = tcg_cpu_exec(cpu);
 +                r = tcg_cpus_exec(cpu);
                  if (icount_enabled()) {
 -                    process_icount_data(cpu);
 +                    icount_process_data(cpu);
                  }
                  qemu_mutex_lock_iothread();
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
          } /* while (cpu && !cpu->exit_request).. */
          /* Does not need qatomic_mb_set because a spurious wakeup is okay.  */
 -        qatomic_set(&tcg_current_rr_cpu, NULL);
 +        qatomic_set(&rr_current_cpu, NULL);
          if (cpu && cpu->exit_request) {
              qatomic_mb_set(&cpu->exit_request, 0);
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
              qemu_notify_event();
          }
 -        qemu_tcg_rr_wait_io_event();
 -        deal_with_unplugged_cpus();
 +        rr_wait_io_event();
 +        rr_deal_with_unplugged_cpus();
      }
      rcu_unregister_thread();
@@ -XXX,XX +XXX,XX @@ void rr_start_vcpu_thread(CPUState *cpu)
          /* share a single thread for all cpus with TCG */
          snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "ALL CPUs/TCG");
          qemu_thread_create(cpu->thread, thread_name,
 -                           tcg_rr_cpu_thread_fn,
 +                           rr_cpu_thread_fn,
                             cpu, QEMU_THREAD_JOINABLE);
          single_tcg_halt_cond = cpu->halt_cond;
@@ -XXX,XX +XXX,XX @@ void rr_start_vcpu_thread(CPUState *cpu)
  const CpusAccel tcg_cpus_rr = {
      .create_vcpu_thread = rr_start_vcpu_thread,
 -    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
 +    .kick_vcpu_thread = rr_kick_vcpu_thread,
 -    .handle_interrupt = tcg_handle_interrupt,
 +    .handle_interrupt = tcg_cpus_handle_interrupt,
  };
 diff --git a/accel/tcg/tcg-cpus.c b/accel/tcg/tcg-cpus.c
 index XXXXXXX..XXXXXXX 100644
 --- a/accel/tcg/tcg-cpus.c
 +++ b/accel/tcg/tcg-cpus.c
@@ -XXX,XX +XXX,XX @@
  /* common functionality among all TCG variants */
 -void qemu_tcg_destroy_vcpu(CPUState *cpu)
 +void tcg_cpus_destroy(CPUState *cpu)
  {
      cpu_thread_signal_destroyed(cpu);
  }
 -int tcg_cpu_exec(CPUState *cpu)
 +int tcg_cpus_exec(CPUState *cpu)
  {
      int ret;
  #ifdef CONFIG_PROFILER
@@ -XXX,XX +XXX,XX @@ int tcg_cpu_exec(CPUState *cpu)
  }
  /* mask must never be zero, except for A20 change call */
 -void tcg_handle_interrupt(CPUState *cpu, int mask)
 +void tcg_cpus_handle_interrupt(CPUState *cpu, int mask)
  {
      g_assert(qemu_mutex_iothread_locked());
 --
-.17.1
+.25.1

The following changes since commit bad76ac319556dab2497429d473b49a237672e1c:

Merge remote-tracking branch 'remotes/vivier2/tags/trivial-branch-pull-request' into staging (2019-10-25 14:17:08 +0100)

are available in the Git repository at:

https://github.com/rth7680/qemu.git tags/pull-tcg-20191025

for you to fetch changes up to 0ed1bfb046b740b70eed2cf3581e01768703b185:

translate-all: Remove tb_alloc (2019-10-25 10:15:25 -0400)

----------------------------------------------------------------
Improvements for TARGET_PAGE_BITS_VARY
Fix for TCI ld16u_i64.
Fix for segv on icount execute from i/o memory.
Two misc cleanups.

----------------------------------------------------------------
Alex Bennée (1):
      cputlb: ensure _cmmu helper functions follow the naming standard

Clement Deschamps (1):
      translate-all: fix uninitialized tb->orig_tb

Richard Henderson (8):
      exec: Split out variable page size support to exec-vary.c
      configure: Detect compiler support for __attribute__((alias))
      exec: Use const alias for TARGET_PAGE_BITS_VARY
      exec: Restrict TARGET_PAGE_BITS_VARY assert to CONFIG_DEBUG_TCG
      exec: Promote TARGET_PAGE_MASK to target_long
      exec: Cache TARGET_PAGE_MASK for TARGET_PAGE_BITS_VARY
      cputlb: Fix tlb_vaddr_to_host
      translate-all: Remove tb_alloc

Stefan Weil (1):
      tci: Add implementation for INDEX_op_ld16u_i64

Wei Yang (1):
      cpu: use ROUND_UP() to define xxx_PAGE_ALIGN

From: Stefan Weil <sw@weilnetz.de>

This fixes "make check-tcg" on a Debian x86_64 host.

Signed-off-by: Stefan Weil <sw@weilnetz.de>
Tested-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20190410194838.10123-1-sw@weilnetz.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/tci.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/tcg/tci.c b/tcg/tci.c
index XXXXXXX..XXXXXXX 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -XXX,XX +XXX,XX @@ static void tci_write_reg8(tcg_target_ulong *regs, TCGReg index, uint8_t value)
     tci_write_reg(regs, index, value);
 }
 
+static void
+tci_write_reg16(tcg_target_ulong *regs, TCGReg index, uint16_t value)
+{
+    tci_write_reg(regs, index, value);
+}
+
 static void
 tci_write_reg32(tcg_target_ulong *regs, TCGReg index, uint32_t value)
 {
@@ -XXX,XX +XXX,XX @@ uintptr_t tcg_qemu_tb_exec(CPUArchState *env, uint8_t *tb_ptr)
             tci_write_reg8(regs, t0, *(uint8_t *)(t1 + t2));
             break;
         case INDEX_op_ld8s_i32:
+            TODO();
+            break;
         case INDEX_op_ld16u_i32:
             TODO();
             break;
@@ -XXX,XX +XXX,XX @@ uintptr_t tcg_qemu_tb_exec(CPUArchState *env, uint8_t *tb_ptr)
             tci_write_reg8(regs, t0, *(uint8_t *)(t1 + t2));
             break;
         case INDEX_op_ld8s_i64:
+            TODO();
+            break;
         case INDEX_op_ld16u_i64:
+            t0 = *tb_ptr++;
+            t1 = tci_read_r(regs, &tb_ptr);
+            t2 = tci_read_s32(&tb_ptr);
+            tci_write_reg16(regs, t0, *(uint16_t *)(t1 + t2));
+            break;
         case INDEX_op_ld16s_i64:
             TODO();
             break;
-- 
2.17.1

From: Alex Bennée <alex.bennee@linaro.org>

We document this in docs/devel/load-stores.rst so lets follow it. The
32 bit and 64 bit access functions have historically not included the
sign so we leave those as is. We also introduce some signed helpers
which are used for loading immediate values in the translator.

Fixes: 282dffc8
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20191021150910.23216-1-alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/exec/cpu_ldst_template.h |  4 ++--
 tcg/tcg.h                        | 20 ++++++++++++++------
 accel/tcg/cputlb.c               | 24 +++++++++++++++++++++---
 target/cris/translate_v10.inc.c  |  3 +--
 4 files changed, 38 insertions(+), 13 deletions(-)

diff --git a/include/exec/cpu_ldst_template.h b/include/exec/cpu_ldst_template.h
index XXXXXXX..XXXXXXX 100644
--- a/include/exec/cpu_ldst_template.h
+++ b/include/exec/cpu_ldst_template.h
@@ -XXX,XX +XXX,XX @@
 #ifdef SOFTMMU_CODE_ACCESS
 #define ADDR_READ addr_code
 #define MMUSUFFIX _cmmu
-#define URETSUFFIX SUFFIX
-#define SRETSUFFIX SUFFIX
+#define URETSUFFIX USUFFIX
+#define SRETSUFFIX glue(s, SUFFIX)
 #else
 #define ADDR_READ addr_read
 #define MMUSUFFIX _mmu
diff --git a/tcg/tcg.h b/tcg/tcg.h
index XXXXXXX..XXXXXXX 100644
--- a/tcg/tcg.h
+++ b/tcg/tcg.h
@@ -XXX,XX +XXX,XX @@ void helper_be_stl_mmu(CPUArchState *env, target_ulong addr, uint32_t val,
 void helper_be_stq_mmu(CPUArchState *env, target_ulong addr, uint64_t val,
                        TCGMemOpIdx oi, uintptr_t retaddr);
 
-uint8_t helper_ret_ldb_cmmu(CPUArchState *env, target_ulong addr,
+uint8_t helper_ret_ldub_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr);
-uint16_t helper_le_ldw_cmmu(CPUArchState *env, target_ulong addr,
+int8_t helper_ret_ldsb_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr);
+uint16_t helper_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
+                             TCGMemOpIdx oi, uintptr_t retaddr);
+int16_t helper_le_ldsw_cmmu(CPUArchState *env, target_ulong addr,
+                             TCGMemOpIdx oi, uintptr_t retaddr);
 uint32_t helper_le_ldl_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr);
 uint64_t helper_le_ldq_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr);
-uint16_t helper_be_ldw_cmmu(CPUArchState *env, target_ulong addr,
-                            TCGMemOpIdx oi, uintptr_t retaddr);
+uint16_t helper_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
+                             TCGMemOpIdx oi, uintptr_t retaddr);
+int16_t helper_be_ldsw_cmmu(CPUArchState *env, target_ulong addr,
+                             TCGMemOpIdx oi, uintptr_t retaddr);
 uint32_t helper_be_ldl_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr);
 uint64_t helper_be_ldq_cmmu(CPUArchState *env, target_ulong addr,
@@ -XXX,XX +XXX,XX @@ uint64_t helper_be_ldq_cmmu(CPUArchState *env, target_ulong addr,
 # define helper_ret_stw_mmu   helper_be_stw_mmu
 # define helper_ret_stl_mmu   helper_be_stl_mmu
 # define helper_ret_stq_mmu   helper_be_stq_mmu
-# define helper_ret_ldw_cmmu  helper_be_ldw_cmmu
+# define helper_ret_lduw_cmmu  helper_be_lduw_cmmu
+# define helper_ret_ldsw_cmmu  helper_be_ldsw_cmmu
 # define helper_ret_ldl_cmmu  helper_be_ldl_cmmu
 # define helper_ret_ldq_cmmu  helper_be_ldq_cmmu
 #else
@@ -XXX,XX +XXX,XX @@ uint64_t helper_be_ldq_cmmu(CPUArchState *env, target_ulong addr,
 # define helper_ret_stw_mmu   helper_le_stw_mmu
 # define helper_ret_stl_mmu   helper_le_stl_mmu
 # define helper_ret_stq_mmu   helper_le_stq_mmu
-# define helper_ret_ldw_cmmu  helper_le_ldw_cmmu
+# define helper_ret_lduw_cmmu  helper_le_lduw_cmmu
+# define helper_ret_ldsw_cmmu  helper_le_ldsw_cmmu
 # define helper_ret_ldl_cmmu  helper_le_ldl_cmmu
 # define helper_ret_ldq_cmmu  helper_le_ldq_cmmu
 #endif
diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/cputlb.c
+++ b/accel/tcg/cputlb.c
@@ -XXX,XX +XXX,XX @@ static uint64_t full_ldub_cmmu(CPUArchState *env, target_ulong addr,
     return load_helper(env, addr, oi, retaddr, MO_8, true, full_ldub_cmmu);
 }
 
-uint8_t helper_ret_ldb_cmmu(CPUArchState *env, target_ulong addr,
+uint8_t helper_ret_ldub_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr)
 {
     return full_ldub_cmmu(env, addr, oi, retaddr);
 }
 
+int8_t helper_ret_ldsb_cmmu(CPUArchState *env, target_ulong addr,
+                            TCGMemOpIdx oi, uintptr_t retaddr)
+{
+    return (int8_t) full_ldub_cmmu(env, addr, oi, retaddr);
+}
+
 static uint64_t full_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
                                   TCGMemOpIdx oi, uintptr_t retaddr)
 {
@@ -XXX,XX +XXX,XX @@ static uint64_t full_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
                        full_le_lduw_cmmu);
 }
 
-uint16_t helper_le_ldw_cmmu(CPUArchState *env, target_ulong addr,
+uint16_t helper_le_lduw_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr)
 {
     return full_le_lduw_cmmu(env, addr, oi, retaddr);
 }
 
+int16_t helper_le_ldsw_cmmu(CPUArchState *env, target_ulong addr,
+                            TCGMemOpIdx oi, uintptr_t retaddr)
+{
+    return (int16_t) full_le_lduw_cmmu(env, addr, oi, retaddr);
+}
+
 static uint64_t full_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
                                   TCGMemOpIdx oi, uintptr_t retaddr)
 {
@@ -XXX,XX +XXX,XX @@ static uint64_t full_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
                        full_be_lduw_cmmu);
 }
 
-uint16_t helper_be_ldw_cmmu(CPUArchState *env, target_ulong addr,
+uint16_t helper_be_lduw_cmmu(CPUArchState *env, target_ulong addr,
                             TCGMemOpIdx oi, uintptr_t retaddr)
 {
     return full_be_lduw_cmmu(env, addr, oi, retaddr);
 }
 
+int16_t helper_be_ldsw_cmmu(CPUArchState *env, target_ulong addr,
+                            TCGMemOpIdx oi, uintptr_t retaddr)
+{
+    return (int16_t) full_be_lduw_cmmu(env, addr, oi, retaddr);
+}
+
 static uint64_t full_le_ldul_cmmu(CPUArchState *env, target_ulong addr,
                                   TCGMemOpIdx oi, uintptr_t retaddr)
 {
diff --git a/target/cris/translate_v10.inc.c b/target/cris/translate_v10.inc.c
index XXXXXXX..XXXXXXX 100644
--- a/target/cris/translate_v10.inc.c
+++ b/target/cris/translate_v10.inc.c
@@ -XXX,XX +XXX,XX @@ static unsigned int dec10_ind(CPUCRISState *env, DisasContext *dc)
         case CRISV10_IND_BCC_M:
 
             cris_cc_mask(dc, 0);
-            imm = cpu_ldsw_code(env, dc->pc + 2);
-            simm = (int16_t)imm;
+            simm = cpu_ldsw_code(env, dc->pc + 2);
             simm += 4;
 
             LOG_DIS("bcc_m: b%s %x\n", cc_name(dc->cond), dc->pc + simm);
-- 
2.17.1

From: Wei Yang <richardw.yang@linux.intel.com>

Use ROUND_UP() to define, which is a little bit easy to read.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Message-Id: <20191013021145.16011-2-richardw.yang@linux.intel.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/exec/cpu-all.h | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
index XXXXXXX..XXXXXXX 100644
--- a/include/exec/cpu-all.h
+++ b/include/exec/cpu-all.h
@@ -XXX,XX +XXX,XX @@ extern int target_page_bits;
 
 #define TARGET_PAGE_SIZE (1 << TARGET_PAGE_BITS)
 #define TARGET_PAGE_MASK ~(TARGET_PAGE_SIZE - 1)
-#define TARGET_PAGE_ALIGN(addr) (((addr) + TARGET_PAGE_SIZE - 1) & TARGET_PAGE_MASK)
+#define TARGET_PAGE_ALIGN(addr) ROUND_UP((addr), TARGET_PAGE_SIZE)
 
 /* Using intptr_t ensures that qemu_*_page_mask is sign-extended even
  * when intptr_t is 32-bit and we are aligning a long long.
@@ -XXX,XX +XXX,XX @@ extern int target_page_bits;
 extern uintptr_t qemu_host_page_size;
 extern intptr_t qemu_host_page_mask;
 
-#define HOST_PAGE_ALIGN(addr) (((addr) + qemu_host_page_size - 1) & qemu_host_page_mask)
-#define REAL_HOST_PAGE_ALIGN(addr) (((addr) + qemu_real_host_page_size - 1) & \
-                                    qemu_real_host_page_mask)
+#define HOST_PAGE_ALIGN(addr) ROUND_UP((addr), qemu_host_page_size)
+#define REAL_HOST_PAGE_ALIGN(addr) ROUND_UP((addr), qemu_real_host_page_size)
 
 /* same as PROT_xxx */
 #define PAGE_READ      0x0001
-- 
2.17.1

The next patch will play a trick with "const" that will
confuse the compiler about the uses of target_page_bits
within exec.c.  Moving everything to a new file prevents
this confusion.

No functional change so far.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 Makefile.target       |  2 +-
 include/qemu-common.h |  6 +++++
 exec-vary.c           | 57 +++++++++++++++++++++++++++++++++++++++++++
 exec.c                | 34 --------------------------
 4 files changed, 64 insertions(+), 35 deletions(-)
 create mode 100644 exec-vary.c

diff --git a/Makefile.target b/Makefile.target
index XXXXXXX..XXXXXXX 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -XXX,XX +XXX,XX @@ obj-y += trace/
 
 #########################################################
 # cpu emulator library
-obj-y += exec.o
+obj-y += exec.o exec-vary.o
 obj-y += accel/
 obj-$(CONFIG_TCG) += tcg/tcg.o tcg/tcg-op.o tcg/tcg-op-vec.o tcg/tcg-op-gvec.o
 obj-$(CONFIG_TCG) += tcg/tcg-common.o tcg/optimize.o
diff --git a/include/qemu-common.h b/include/qemu-common.h
index XXXXXXX..XXXXXXX 100644
--- a/include/qemu-common.h
+++ b/include/qemu-common.h
@@ -XXX,XX +XXX,XX @@ void cpu_exec_step_atomic(CPUState *cpu);
  */
 bool set_preferred_target_page_bits(int bits);
 
+/**
+ * finalize_target_page_bits:
+ * Commit the final value set by set_preferred_target_page_bits.
+ */
+void finalize_target_page_bits(void);
+
 /**
  * Sends a (part of) iovec down a socket, yielding when the socket is full, or
  * Receives data into a (part of) iovec from a socket,
diff --git a/exec-vary.c b/exec-vary.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/exec-vary.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * Variable page size handling
+ *
+ *  Copyright (c) 2003 Fabrice Bellard
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+#include "exec/exec-all.h"
+
+#ifdef TARGET_PAGE_BITS_VARY
+int target_page_bits;
+bool target_page_bits_decided;
+#endif
+
+bool set_preferred_target_page_bits(int bits)
+{
+    /*
+     * The target page size is the lowest common denominator for all
+     * the CPUs in the system, so we can only make it smaller, never
+     * larger. And we can't make it smaller once we've committed to
+     * a particular size.
+     */
+#ifdef TARGET_PAGE_BITS_VARY
+    assert(bits >= TARGET_PAGE_BITS_MIN);
+    if (target_page_bits == 0 || target_page_bits > bits) {
+        if (target_page_bits_decided) {
+            return false;
+        }
+        target_page_bits = bits;
+    }
+#endif
+    return true;
+}
+
+void finalize_target_page_bits(void)
+{
+#ifdef TARGET_PAGE_BITS_VARY
+    if (target_page_bits == 0) {
+        target_page_bits = TARGET_PAGE_BITS_MIN;
+    }
+    target_page_bits_decided = true;
+#endif
+}
diff --git a/exec.c b/exec.c
index XXXXXXX..XXXXXXX 100644
--- a/exec.c
+++ b/exec.c
@@ -XXX,XX +XXX,XX @@ AddressSpace address_space_memory;
 static MemoryRegion io_mem_unassigned;
 #endif
 
-#ifdef TARGET_PAGE_BITS_VARY
-int target_page_bits;
-bool target_page_bits_decided;
-#endif
-
 CPUTailQ cpus = QTAILQ_HEAD_INITIALIZER(cpus);
 
 /* current CPU in the current thread. It is only valid inside
@@ -XXX,XX +XXX,XX @@ int use_icount;
 uintptr_t qemu_host_page_size;
 intptr_t qemu_host_page_mask;
 
-bool set_preferred_target_page_bits(int bits)
-{
-    /* The target page size is the lowest common denominator for all
-     * the CPUs in the system, so we can only make it smaller, never
-     * larger. And we can't make it smaller once we've committed to
-     * a particular size.
-     */
-#ifdef TARGET_PAGE_BITS_VARY
-    assert(bits >= TARGET_PAGE_BITS_MIN);
-    if (target_page_bits == 0 || target_page_bits > bits) {
-        if (target_page_bits_decided) {
-            return false;
-        }
-        target_page_bits = bits;
-    }
-#endif
-    return true;
-}
-
 #if !defined(CONFIG_USER_ONLY)
 
-static void finalize_target_page_bits(void)
-{
-#ifdef TARGET_PAGE_BITS_VARY
-    if (target_page_bits == 0) {
-        target_page_bits = TARGET_PAGE_BITS_MIN;
-    }
-    target_page_bits_decided = true;
-#endif
-}
-
 typedef struct PhysPageEntry PhysPageEntry;
 
 struct PhysPageEntry {
-- 
2.17.1

Such support is present almost everywhere, except for Xcode 9.
It is added in Xcode 10, but travis uses xcode9 by default,
so we should support it for a while yet.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 configure | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/configure b/configure
index XXXXXXX..XXXXXXX 100755
--- a/configure
+++ b/configure
@@ -XXX,XX +XXX,XX @@ if compile_prog "" "" ; then
   vector16=yes
 fi
 
+########################################
+# See if __attribute__((alias)) is supported.
+# This false for Xcode 9, but has been remedied for Xcode 10.
+# Unfortunately, travis uses Xcode 9 by default.
+
+attralias=no
+cat > $TMPC << EOF
+int x = 1;
+extern const int y __attribute__((alias("x")));
+int main(void) { return 0; }
+EOF
+if compile_prog "" "" ; then
+    attralias=yes
+fi
+
 ########################################
 # check if getauxval is available.
 
@@ -XXX,XX +XXX,XX @@ if test "$vector16" = "yes" ; then
   echo "CONFIG_VECTOR16=y" >> $config_host_mak
 fi
 
+if test "$attralias" = "yes" ; then
+  echo "CONFIG_ATTRIBUTE_ALIAS=y" >> $config_host_mak
+fi
+
 if test "$getauxval" = "yes" ; then
   echo "CONFIG_GETAUXVAL=y" >> $config_host_mak
 fi
-- 
2.17.1

Using a variable that is declared "const" for this tells the
compiler that it may read the value once and assume that it
does not change across function calls.

For target_page_size, this means we have only one assert per
function, and one read of the variable.

This reduces the size of qemu-system-aarch64 by 8k.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/exec/cpu-all.h | 14 +++++++---
 exec-vary.c            | 60 ++++++++++++++++++++++++++++++++++++------
 2 files changed, 62 insertions(+), 12 deletions(-)

diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
index XXXXXXX..XXXXXXX 100644
--- a/include/exec/cpu-all.h
+++ b/include/exec/cpu-all.h
@@ -XXX,XX +XXX,XX @@ static inline void stl_phys_notdirty(AddressSpace *as, hwaddr addr, uint32_t val
 /* page related stuff */
 
 #ifdef TARGET_PAGE_BITS_VARY
-extern bool target_page_bits_decided;
-extern int target_page_bits;
-#define TARGET_PAGE_BITS ({ assert(target_page_bits_decided); \
-                            target_page_bits; })
+typedef struct {
+    bool decided;
+    int bits;
+} TargetPageBits;
+# if defined(CONFIG_ATTRIBUTE_ALIAS) || !defined(IN_EXEC_VARY)
+extern const TargetPageBits target_page;
+#else
+extern TargetPageBits target_page;
+# endif
+#define TARGET_PAGE_BITS (assert(target_page.decided), target_page.bits)
 #else
 #define TARGET_PAGE_BITS_MIN TARGET_PAGE_BITS
 #endif
diff --git a/exec-vary.c b/exec-vary.c
index XXXXXXX..XXXXXXX 100644
--- a/exec-vary.c
+++ b/exec-vary.c
@@ -XXX,XX +XXX,XX @@
 
 #include "qemu/osdep.h"
 #include "qemu-common.h"
+
+#define IN_EXEC_VARY 1
+
 #include "exec/exec-all.h"
 
 #ifdef TARGET_PAGE_BITS_VARY
-int target_page_bits;
-bool target_page_bits_decided;
+# ifdef CONFIG_ATTRIBUTE_ALIAS
+/*
+ * We want to declare the "target_page" variable as const, which tells
+ * the compiler that it can cache any value that it reads across calls.
+ * This avoids multiple assertions and multiple reads within any one user.
+ *
+ * This works because we initialize the target_page data very early, in a
+ * location far removed from the functions that require the final results.
+ *
+ * This also requires that we have a non-constant symbol by which we can
+ * perform the actual initialization, and which forces the data to be
+ * allocated within writable memory.  Thus "init_target_page", and we use
+ * that symbol exclusively in the two functions that initialize this value.
+ *
+ * The "target_page" symbol is created as an alias of "init_target_page".
+ */
+static TargetPageBits init_target_page;
+
+/*
+ * Note that this is *not* a redundant decl, this is the definition of
+ * the "target_page" symbol.  The syntax for this definition requires
+ * the use of the extern keyword.  This seems to be a GCC bug in
+ * either the syntax for the alias attribute or in -Wredundant-decls.
+ *
+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91765
+ */
+#  pragma GCC diagnostic push
+#  pragma GCC diagnostic ignored "-Wredundant-decls"
+
+extern const TargetPageBits target_page
+    __attribute__((alias("init_target_page")));
+
+#  pragma GCC diagnostic pop
+# else
+/*
+ * When aliases are not supported then we force two different declarations,
+ * by way of suppressing the header declaration with IN_EXEC_VARY.
+ * We assume that on such an old compiler, LTO cannot be used, and so the
+ * compiler cannot not detect the mismatched declarations, and all is well.
+ */
+TargetPageBits target_page;
+#  define init_target_page target_page
+# endif
 #endif
 
 bool set_preferred_target_page_bits(int bits)
@@ -XXX,XX +XXX,XX @@ bool set_preferred_target_page_bits(int bits)
      */
 #ifdef TARGET_PAGE_BITS_VARY
     assert(bits >= TARGET_PAGE_BITS_MIN);
-    if (target_page_bits == 0 || target_page_bits > bits) {
-        if (target_page_bits_decided) {
+    if (init_target_page.bits == 0 || init_target_page.bits > bits) {
+        if (init_target_page.decided) {
             return false;
         }
-        target_page_bits = bits;
+        init_target_page.bits = bits;
     }
 #endif
     return true;
@@ -XXX,XX +XXX,XX @@ bool set_preferred_target_page_bits(int bits)
 void finalize_target_page_bits(void)
 {
 #ifdef TARGET_PAGE_BITS_VARY
-    if (target_page_bits == 0) {
-        target_page_bits = TARGET_PAGE_BITS_MIN;
+    if (init_target_page.bits == 0) {
+        init_target_page.bits = TARGET_PAGE_BITS_MIN;
     }
-    target_page_bits_decided = true;
+    init_target_page.decided = true;
 #endif
 }
-- 
2.17.1

There are some uint64_t uses that expect TARGET_PAGE_MASK to
extend for a 32-bit, so this must continue to be a signed type.
Define based on TARGET_PAGE_BITS not TARGET_PAGE_SIZE; this
will make a following patch more clear.

This should not have a functional effect so far.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/exec/cpu-all.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
index XXXXXXX..XXXXXXX 100644
--- a/include/exec/cpu-all.h
+++ b/include/exec/cpu-all.h
@@ -XXX,XX +XXX,XX @@ extern TargetPageBits target_page;
 #endif
 
 #define TARGET_PAGE_SIZE (1 << TARGET_PAGE_BITS)
-#define TARGET_PAGE_MASK ~(TARGET_PAGE_SIZE - 1)
+#define TARGET_PAGE_MASK ((target_long)-1 << TARGET_PAGE_BITS)
 #define TARGET_PAGE_ALIGN(addr) ROUND_UP((addr), TARGET_PAGE_SIZE)
 
 /* Using intptr_t ensures that qemu_*_page_mask is sign-extended even
-- 
2.17.1

This eliminates a set of runtime shifts.  It turns out that we
require TARGET_PAGE_MASK more often than TARGET_PAGE_SIZE, so
redefine TARGET_PAGE_SIZE based on TARGET_PAGE_MASK instead of
the other way around.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/exec/cpu-all.h | 8 ++++++--
 exec-vary.c            | 1 +
 2 files changed, 7 insertions(+), 2 deletions(-)

From: Clement Deschamps <clement.deschamps@greensocs.com>

This fixes a segmentation fault in icount mode when executing
from an IO region.

TB is marked as CF_NOCACHE but tb->orig_tb is not initialized
(equals previous value in code_gen_buffer).

The issue happens in cpu_io_recompile() when it tries to invalidate orig_tb.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Clement Deschamps <clement.deschamps@greensocs.com>
Message-Id: <20191022140016.918371-1-clement.deschamps@greensocs.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/translate-all.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu,
     tb->cs_base = cs_base;
     tb->flags = flags;
     tb->cflags = cflags;
+    tb->orig_tb = NULL;
     tb->trace_vcpu_dstate = *cpu->trace_dstate;
     tcg_ctx->tb_cflags = cflags;
  tb_overflow:
-- 
2.17.1

Since 2ac01d6dafab, this function does only two things: assert a
lock is held, and call tcg_tb_alloc.  It is used exactly once,
and its user has already done the assert.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Clement Deschamps <clement.deschamps@greensocs.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/translate-all.c | 20 ++------------------
 1 file changed, 2 insertions(+), 18 deletions(-)

diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -XXX,XX +XXX,XX @@ void tcg_exec_init(unsigned long tb_size)
 #endif
 }
 
-/*
- * Allocate a new translation block. Flush the translation buffer if
- * too many translation blocks or too much generated code.
- */
-static TranslationBlock *tb_alloc(target_ulong pc)
-{
-    TranslationBlock *tb;
-
-    assert_memory_lock();
-
-    tb = tcg_tb_alloc(tcg_ctx);
-    if (unlikely(tb == NULL)) {
-        return NULL;
-    }
-    return tb;
-}
-
 /* call with @p->lock held */
 static inline void invalidate_page_bitmap(PageDesc *p)
 {
@@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu,
     TCGProfile *prof = &tcg_ctx->prof;
     int64_t ti;
 #endif
+
     assert_memory_lock();
 
     phys_pc = get_page_addr_code(env, pc);
@@ -XXX,XX +XXX,XX @@ TranslationBlock *tb_gen_code(CPUState *cpu,
     }
 
  buffer_overflow:
-    tb = tb_alloc(pc);
+    tb = tcg_tb_alloc(tcg_ctx);
     if (unlikely(!tb)) {
         /* flush must be done */
         tb_flush(cpu);
-- 
2.17.1

The following changes since commit 2ecfc0657afa5d29a373271b342f704a1a3c6737:

Merge remote-tracking branch 'remotes/armbru/tags/pull-misc-2020-12-10' into staging (2020-12-10 17:01:05 +0000)

are available in the Git repository at:

https://gitlab.com/rth7680/qemu.git tags/pull-tcg-20201210

for you to fetch changes up to 9e2658d62ebc23efe7df43fc0e306f129510d874:

accel/tcg: rename tcg-cpus functions to match module name (2020-12-10 17:44:10 -0600)

----------------------------------------------------------------
Split CpusAccel for tcg variants

----------------------------------------------------------------
Claudio Fontana (3):
      accel/tcg: split CpusAccel into three TCG variants
      accel/tcg: split tcg_start_vcpu_thread
      accel/tcg: rename tcg-cpus functions to match module name

accel/tcg/tcg-cpus-icount.h |  17 ++
 accel/tcg/tcg-cpus-rr.h     |  21 ++
 accel/tcg/tcg-cpus.h        |  12 +-
 accel/tcg/tcg-all.c         |  13 +-
 accel/tcg/tcg-cpus-icount.c | 147 +++++++++++++
 accel/tcg/tcg-cpus-mttcg.c  | 140 ++++++++++++
 accel/tcg/tcg-cpus-rr.c     | 305 ++++++++++++++++++++++++++
 accel/tcg/tcg-cpus.c        | 506 +-------------------------------------------
 softmmu/icount.c            |   2 +-
 accel/tcg/meson.build       |   9 +-
 10 files changed, 670 insertions(+), 502 deletions(-)
 create mode 100644 accel/tcg/tcg-cpus-icount.h
 create mode 100644 accel/tcg/tcg-cpus-rr.h
 create mode 100644 accel/tcg/tcg-cpus-icount.c
 create mode 100644 accel/tcg/tcg-cpus-mttcg.c
 create mode 100644 accel/tcg/tcg-cpus-rr.c

From: Claudio Fontana <cfontana@suse.de>

split up the CpusAccel tcg_cpus into three TCG variants:

tcg_cpus_rr (single threaded, round robin cpus)
tcg_cpus_icount (same as rr, but with instruction counting enabled)
tcg_cpus_mttcg (multi-threaded cpus)

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201015143217.29337-2-cfontana@suse.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/tcg-cpus-icount.h |  17 ++
 accel/tcg/tcg-cpus-mttcg.h  |  21 ++
 accel/tcg/tcg-cpus-rr.h     |  20 ++
 accel/tcg/tcg-cpus.h        |  13 +-
 accel/tcg/tcg-all.c         |   8 +-
 accel/tcg/tcg-cpus-icount.c | 147 +++++++++++
 accel/tcg/tcg-cpus-mttcg.c  | 117 +++++++++
 accel/tcg/tcg-cpus-rr.c     | 270 ++++++++++++++++++++
 accel/tcg/tcg-cpus.c        | 484 ++----------------------------------
 softmmu/icount.c            |   2 +-
 accel/tcg/meson.build       |   9 +-
 11 files changed, 646 insertions(+), 462 deletions(-)
 create mode 100644 accel/tcg/tcg-cpus-icount.h
 create mode 100644 accel/tcg/tcg-cpus-mttcg.h
 create mode 100644 accel/tcg/tcg-cpus-rr.h
 create mode 100644 accel/tcg/tcg-cpus-icount.c
 create mode 100644 accel/tcg/tcg-cpus-mttcg.c
 create mode 100644 accel/tcg/tcg-cpus-rr.c

diff --git a/accel/tcg/tcg-cpus-icount.h b/accel/tcg/tcg-cpus-icount.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/accel/tcg/tcg-cpus-icount.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * QEMU TCG Single Threaded vCPUs implementation using instruction counting
+ *
+ * Copyright 2020 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef TCG_CPUS_ICOUNT_H
+#define TCG_CPUS_ICOUNT_H
+
+void handle_icount_deadline(void);
+void prepare_icount_for_run(CPUState *cpu);
+void process_icount_data(CPUState *cpu);
+
+#endif /* TCG_CPUS_ICOUNT_H */
diff --git a/accel/tcg/tcg-cpus-mttcg.h b/accel/tcg/tcg-cpus-mttcg.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/accel/tcg/tcg-cpus-mttcg.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * QEMU TCG Multi Threaded vCPUs implementation
+ *
+ * Copyright 2020 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef TCG_CPUS_MTTCG_H
+#define TCG_CPUS_MTTCG_H
+
+/*
+ * In the multi-threaded case each vCPU has its own thread. The TLS
+ * variable current_cpu can be used deep in the code to find the
+ * current CPUState for a given thread.
+ */
+
+void *tcg_cpu_thread_fn(void *arg);
+
+#endif /* TCG_CPUS_MTTCG_H */
diff --git a/accel/tcg/tcg-cpus-rr.h b/accel/tcg/tcg-cpus-rr.h
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/accel/tcg/tcg-cpus-rr.h
@@ -XXX,XX +XXX,XX @@
+/*
+ * QEMU TCG Single Threaded vCPUs implementation
+ *
+ * Copyright 2020 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef TCG_CPUS_RR_H
+#define TCG_CPUS_RR_H
+
+#define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
+
+/* Kick all RR vCPUs. */
+void qemu_cpu_kick_rr_cpus(CPUState *unused);
+
+void *tcg_rr_cpu_thread_fn(void *arg);
+
+#endif /* TCG_CPUS_RR_H */
diff --git a/accel/tcg/tcg-cpus.h b/accel/tcg/tcg-cpus.h
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus.h
+++ b/accel/tcg/tcg-cpus.h
@@ -XXX,XX +XXX,XX @@
 /*
- * Accelerator CPUS Interface
+ * QEMU TCG vCPU common functionality
+ *
+ * Functionality common to all TCG vcpu variants: mttcg, rr and icount.
  *
  * Copyright 2020 SUSE LLC
  *
@@ -XXX,XX +XXX,XX @@
 
 #include "sysemu/cpus.h"
 
-extern const CpusAccel tcg_cpus;
+extern const CpusAccel tcg_cpus_mttcg;
+extern const CpusAccel tcg_cpus_icount;
+extern const CpusAccel tcg_cpus_rr;
+
+void tcg_start_vcpu_thread(CPUState *cpu);
+void qemu_tcg_destroy_vcpu(CPUState *cpu);
+int tcg_cpu_exec(CPUState *cpu);
+void tcg_handle_interrupt(CPUState *cpu, int mask);
 
 #endif /* TCG_CPUS_H */
diff --git a/accel/tcg/tcg-all.c b/accel/tcg/tcg-all.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-all.c
+++ b/accel/tcg/tcg-all.c
@@ -XXX,XX +XXX,XX @@ static int tcg_init(MachineState *ms)
 
     tcg_exec_init(s->tb_size * 1024 * 1024);
     mttcg_enabled = s->mttcg_enabled;
-    cpus_register_accel(&tcg_cpus);
 
+    if (mttcg_enabled) {
+        cpus_register_accel(&tcg_cpus_mttcg);
+    } else if (icount_enabled()) {
+        cpus_register_accel(&tcg_cpus_icount);
+    } else {
+        cpus_register_accel(&tcg_cpus_rr);
+    }
     return 0;
 }
 
diff --git a/accel/tcg/tcg-cpus-icount.c b/accel/tcg/tcg-cpus-icount.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/accel/tcg/tcg-cpus-icount.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * QEMU TCG Single Threaded vCPUs implementation using instruction counting
+ *
+ * Copyright (c) 2003-2008 Fabrice Bellard
+ * Copyright (c) 2014 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+#include "sysemu/tcg.h"
+#include "sysemu/replay.h"
+#include "qemu/main-loop.h"
+#include "qemu/guest-random.h"
+#include "exec/exec-all.h"
+#include "hw/boards.h"
+
+#include "tcg-cpus.h"
+#include "tcg-cpus-icount.h"
+#include "tcg-cpus-rr.h"
+
+static int64_t tcg_get_icount_limit(void)
+{
+    int64_t deadline;
+
+    if (replay_mode != REPLAY_MODE_PLAY) {
+        /*
+         * Include all the timers, because they may need an attention.
+         * Too long CPU execution may create unnecessary delay in UI.
+         */
+        deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
+                                              QEMU_TIMER_ATTR_ALL);
+        /* Check realtime timers, because they help with input processing */
+        deadline = qemu_soonest_timeout(deadline,
+                qemu_clock_deadline_ns_all(QEMU_CLOCK_REALTIME,
+                                           QEMU_TIMER_ATTR_ALL));
+
+        /*
+         * Maintain prior (possibly buggy) behaviour where if no deadline
+         * was set (as there is no QEMU_CLOCK_VIRTUAL timer) or it is more than
+         * INT32_MAX nanoseconds ahead, we still use INT32_MAX
+         * nanoseconds.
+         */
+        if ((deadline < 0) || (deadline > INT32_MAX)) {
+            deadline = INT32_MAX;
+        }
+
+        return icount_round(deadline);
+    } else {
+        return replay_get_instructions();
+    }
+}
+
+static void notify_aio_contexts(void)
+{
+    /* Wake up other AioContexts.  */
+    qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
+    qemu_clock_run_timers(QEMU_CLOCK_VIRTUAL);
+}
+
+void handle_icount_deadline(void)
+{
+    assert(qemu_in_vcpu_thread());
+    int64_t deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
+                                                  QEMU_TIMER_ATTR_ALL);
+
+    if (deadline == 0) {
+        notify_aio_contexts();
+    }
+}
+
+void prepare_icount_for_run(CPUState *cpu)
+{
+    int insns_left;
+
+    /*
+     * These should always be cleared by process_icount_data after
+     * each vCPU execution. However u16.high can be raised
+     * asynchronously by cpu_exit/cpu_interrupt/tcg_handle_interrupt
+     */
+    g_assert(cpu_neg(cpu)->icount_decr.u16.low == 0);
+    g_assert(cpu->icount_extra == 0);
+
+    cpu->icount_budget = tcg_get_icount_limit();
+    insns_left = MIN(0xffff, cpu->icount_budget);
+    cpu_neg(cpu)->icount_decr.u16.low = insns_left;
+    cpu->icount_extra = cpu->icount_budget - insns_left;
+
+    replay_mutex_lock();
+
+    if (cpu->icount_budget == 0 && replay_has_checkpoint()) {
+        notify_aio_contexts();
+    }
+}
+
+void process_icount_data(CPUState *cpu)
+{
+    /* Account for executed instructions */
+    icount_update(cpu);
+
+    /* Reset the counters */
+    cpu_neg(cpu)->icount_decr.u16.low = 0;
+    cpu->icount_extra = 0;
+    cpu->icount_budget = 0;
+
+    replay_account_executed_instructions();
+
+    replay_mutex_unlock();
+}
+
+static void icount_handle_interrupt(CPUState *cpu, int mask)
+{
+    int old_mask = cpu->interrupt_request;
+
+    tcg_handle_interrupt(cpu, mask);
+    if (qemu_cpu_is_self(cpu) &&
+        !cpu->can_do_io
+        && (mask & ~old_mask) != 0) {
+        cpu_abort(cpu, "Raised interrupt while not in I/O function");
+    }
+}
+
+const CpusAccel tcg_cpus_icount = {
+    .create_vcpu_thread = tcg_start_vcpu_thread,
+    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
+
+    .handle_interrupt = icount_handle_interrupt,
+    .get_virtual_clock = icount_get,
+    .get_elapsed_ticks = icount_get,
+};
diff --git a/accel/tcg/tcg-cpus-mttcg.c b/accel/tcg/tcg-cpus-mttcg.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/accel/tcg/tcg-cpus-mttcg.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * QEMU TCG Multi Threaded vCPUs implementation
+ *
+ * Copyright (c) 2003-2008 Fabrice Bellard
+ * Copyright (c) 2014 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+#include "sysemu/tcg.h"
+#include "sysemu/replay.h"
+#include "qemu/main-loop.h"
+#include "qemu/guest-random.h"
+#include "exec/exec-all.h"
+#include "hw/boards.h"
+
+#include "tcg-cpus.h"
+#include "tcg-cpus-mttcg.h"
+
+/*
+ * In the multi-threaded case each vCPU has its own thread. The TLS
+ * variable current_cpu can be used deep in the code to find the
+ * current CPUState for a given thread.
+ */
+
+void *tcg_cpu_thread_fn(void *arg)
+{
+    CPUState *cpu = arg;
+
+    assert(tcg_enabled());
+    g_assert(!icount_enabled());
+
+    rcu_register_thread();
+    tcg_register_thread();
+
+    qemu_mutex_lock_iothread();
+    qemu_thread_get_self(cpu->thread);
+
+    cpu->thread_id = qemu_get_thread_id();
+    cpu->can_do_io = 1;
+    current_cpu = cpu;
+    cpu_thread_signal_created(cpu);
+    qemu_guest_random_seed_thread_part2(cpu->random_seed);
+
+    /* process any pending work */
+    cpu->exit_request = 1;
+
+    do {
+        if (cpu_can_run(cpu)) {
+            int r;
+            qemu_mutex_unlock_iothread();
+            r = tcg_cpu_exec(cpu);
+            qemu_mutex_lock_iothread();
+            switch (r) {
+            case EXCP_DEBUG:
+                cpu_handle_guest_debug(cpu);
+                break;
+            case EXCP_HALTED:
+                /*
+                 * during start-up the vCPU is reset and the thread is
+                 * kicked several times. If we don't ensure we go back
+                 * to sleep in the halted state we won't cleanly
+                 * start-up when the vCPU is enabled.
+                 *
+                 * cpu->halted should ensure we sleep in wait_io_event
+                 */
+                g_assert(cpu->halted);
+                break;
+            case EXCP_ATOMIC:
+                qemu_mutex_unlock_iothread();
+                cpu_exec_step_atomic(cpu);
+                qemu_mutex_lock_iothread();
+            default:
+                /* Ignore everything else? */
+                break;
+            }
+        }
+
+        qatomic_mb_set(&cpu->exit_request, 0);
+        qemu_wait_io_event(cpu);
+    } while (!cpu->unplug || cpu_can_run(cpu));
+
+    qemu_tcg_destroy_vcpu(cpu);
+    qemu_mutex_unlock_iothread();
+    rcu_unregister_thread();
+    return NULL;
+}
+
+static void mttcg_kick_vcpu_thread(CPUState *cpu)
+{
+    cpu_exit(cpu);
+}
+
+const CpusAccel tcg_cpus_mttcg = {
+    .create_vcpu_thread = tcg_start_vcpu_thread,
+    .kick_vcpu_thread = mttcg_kick_vcpu_thread,
+
+    .handle_interrupt = tcg_handle_interrupt,
+};
diff --git a/accel/tcg/tcg-cpus-rr.c b/accel/tcg/tcg-cpus-rr.c
new file mode 100644
index XXXXXXX..XXXXXXX
--- /dev/null
+++ b/accel/tcg/tcg-cpus-rr.c
@@ -XXX,XX +XXX,XX @@
+/*
+ * QEMU TCG Single Threaded vCPUs implementation
+ *
+ * Copyright (c) 2003-2008 Fabrice Bellard
+ * Copyright (c) 2014 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "qemu/osdep.h"
+#include "qemu-common.h"
+#include "sysemu/tcg.h"
+#include "sysemu/replay.h"
+#include "qemu/main-loop.h"
+#include "qemu/guest-random.h"
+#include "exec/exec-all.h"
+#include "hw/boards.h"
+
+#include "tcg-cpus.h"
+#include "tcg-cpus-rr.h"
+#include "tcg-cpus-icount.h"
+
+/* Kick all RR vCPUs */
+void qemu_cpu_kick_rr_cpus(CPUState *unused)
+{
+    CPUState *cpu;
+
+    CPU_FOREACH(cpu) {
+        cpu_exit(cpu);
+    };
+}
+
+/*
+ * TCG vCPU kick timer
+ *
+ * The kick timer is responsible for moving single threaded vCPU
+ * emulation on to the next vCPU. If more than one vCPU is running a
+ * timer event with force a cpu->exit so the next vCPU can get
+ * scheduled.
+ *
+ * The timer is removed if all vCPUs are idle and restarted again once
+ * idleness is complete.
+ */
+
+static QEMUTimer *tcg_kick_vcpu_timer;
+static CPUState *tcg_current_rr_cpu;
+
+#define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
+
+static inline int64_t qemu_tcg_next_kick(void)
+{
+    return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + TCG_KICK_PERIOD;
+}
+
+/* Kick the currently round-robin scheduled vCPU to next */
+static void qemu_cpu_kick_rr_next_cpu(void)
+{
+    CPUState *cpu;
+    do {
+        cpu = qatomic_mb_read(&tcg_current_rr_cpu);
+        if (cpu) {
+            cpu_exit(cpu);
+        }
+    } while (cpu != qatomic_mb_read(&tcg_current_rr_cpu));
+}
+
+static void kick_tcg_thread(void *opaque)
+{
+    timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
+    qemu_cpu_kick_rr_next_cpu();
+}
+
+static void start_tcg_kick_timer(void)
+{
+    if (!tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
+        tcg_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
+                                           kick_tcg_thread, NULL);
+    }
+    if (tcg_kick_vcpu_timer && !timer_pending(tcg_kick_vcpu_timer)) {
+        timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
+    }
+}
+
+static void stop_tcg_kick_timer(void)
+{
+    if (tcg_kick_vcpu_timer && timer_pending(tcg_kick_vcpu_timer)) {
+        timer_del(tcg_kick_vcpu_timer);
+    }
+}
+
+static void qemu_tcg_rr_wait_io_event(void)
+{
+    CPUState *cpu;
+
+    while (all_cpu_threads_idle()) {
+        stop_tcg_kick_timer();
+        qemu_cond_wait_iothread(first_cpu->halt_cond);
+    }
+
+    start_tcg_kick_timer();
+
+    CPU_FOREACH(cpu) {
+        qemu_wait_io_event_common(cpu);
+    }
+}
+
+/*
+ * Destroy any remaining vCPUs which have been unplugged and have
+ * finished running
+ */
+static void deal_with_unplugged_cpus(void)
+{
+    CPUState *cpu;
+
+    CPU_FOREACH(cpu) {
+        if (cpu->unplug && !cpu_can_run(cpu)) {
+            qemu_tcg_destroy_vcpu(cpu);
+            break;
+        }
+    }
+}
+
+/*
+ * In the single-threaded case each vCPU is simulated in turn. If
+ * there is more than a single vCPU we create a simple timer to kick
+ * the vCPU and ensure we don't get stuck in a tight loop in one vCPU.
+ * This is done explicitly rather than relying on side-effects
+ * elsewhere.
+ */
+
+void *tcg_rr_cpu_thread_fn(void *arg)
+{
+    CPUState *cpu = arg;
+
+    assert(tcg_enabled());
+    rcu_register_thread();
+    tcg_register_thread();
+
+    qemu_mutex_lock_iothread();
+    qemu_thread_get_self(cpu->thread);
+
+    cpu->thread_id = qemu_get_thread_id();
+    cpu->can_do_io = 1;
+    cpu_thread_signal_created(cpu);
+    qemu_guest_random_seed_thread_part2(cpu->random_seed);
+
+    /* wait for initial kick-off after machine start */
+    while (first_cpu->stopped) {
+        qemu_cond_wait_iothread(first_cpu->halt_cond);
+
+        /* process any pending work */
+        CPU_FOREACH(cpu) {
+            current_cpu = cpu;
+            qemu_wait_io_event_common(cpu);
+        }
+    }
+
+    start_tcg_kick_timer();
+
+    cpu = first_cpu;
+
+    /* process any pending work */
+    cpu->exit_request = 1;
+
+    while (1) {
+        qemu_mutex_unlock_iothread();
+        replay_mutex_lock();
+        qemu_mutex_lock_iothread();
+
+        if (icount_enabled()) {
+            /* Account partial waits to QEMU_CLOCK_VIRTUAL.  */
+            icount_account_warp_timer();
+            /*
+             * Run the timers here.  This is much more efficient than
+             * waking up the I/O thread and waiting for completion.
+             */
+            handle_icount_deadline();
+        }
+
+        replay_mutex_unlock();
+
+        if (!cpu) {
+            cpu = first_cpu;
+        }
+
+        while (cpu && cpu_work_list_empty(cpu) && !cpu->exit_request) {
+
+            qatomic_mb_set(&tcg_current_rr_cpu, cpu);
+            current_cpu = cpu;
+
+            qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
+                              (cpu->singlestep_enabled & SSTEP_NOTIMER) == 0);
+
+            if (cpu_can_run(cpu)) {
+                int r;
+
+                qemu_mutex_unlock_iothread();
+                if (icount_enabled()) {
+                    prepare_icount_for_run(cpu);
+                }
+                r = tcg_cpu_exec(cpu);
+                if (icount_enabled()) {
+                    process_icount_data(cpu);
+                }
+                qemu_mutex_lock_iothread();
+
+                if (r == EXCP_DEBUG) {
+                    cpu_handle_guest_debug(cpu);
+                    break;
+                } else if (r == EXCP_ATOMIC) {
+                    qemu_mutex_unlock_iothread();
+                    cpu_exec_step_atomic(cpu);
+                    qemu_mutex_lock_iothread();
+                    break;
+                }
+            } else if (cpu->stop) {
+                if (cpu->unplug) {
+                    cpu = CPU_NEXT(cpu);
+                }
+                break;
+            }
+
+            cpu = CPU_NEXT(cpu);
+        } /* while (cpu && !cpu->exit_request).. */
+
+        /* Does not need qatomic_mb_set because a spurious wakeup is okay.  */
+        qatomic_set(&tcg_current_rr_cpu, NULL);
+
+        if (cpu && cpu->exit_request) {
+            qatomic_mb_set(&cpu->exit_request, 0);
+        }
+
+        if (icount_enabled() && all_cpu_threads_idle()) {
+            /*
+             * When all cpus are sleeping (e.g in WFI), to avoid a deadlock
+             * in the main_loop, wake it up in order to start the warp timer.
+             */
+            qemu_notify_event();
+        }
+
+        qemu_tcg_rr_wait_io_event();
+        deal_with_unplugged_cpus();
+    }
+
+    rcu_unregister_thread();
+    return NULL;
+}
+
+const CpusAccel tcg_cpus_rr = {
+    .create_vcpu_thread = tcg_start_vcpu_thread,
+    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
+
+    .handle_interrupt = tcg_handle_interrupt,
+};
diff --git a/accel/tcg/tcg-cpus.c b/accel/tcg/tcg-cpus.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus.c
+++ b/accel/tcg/tcg-cpus.c
@@ -XXX,XX +XXX,XX @@
 /*
- * QEMU System Emulator
+ * QEMU TCG vCPU common functionality
+ *
+ * Functionality common to all TCG vCPU variants: mttcg, rr and icount.
  *
  * Copyright (c) 2003-2008 Fabrice Bellard
  * Copyright (c) 2014 Red Hat Inc.
@@ -XXX,XX +XXX,XX @@
 #include "hw/boards.h"
 
 #include "tcg-cpus.h"
+#include "tcg-cpus-mttcg.h"
+#include "tcg-cpus-rr.h"
 
-/* Kick all RR vCPUs */
-static void qemu_cpu_kick_rr_cpus(void)
-{
-    CPUState *cpu;
+/* common functionality among all TCG variants */
 
-    CPU_FOREACH(cpu) {
-        cpu_exit(cpu);
-    };
-}
-
-static void tcg_kick_vcpu_thread(CPUState *cpu)
-{
-    if (qemu_tcg_mttcg_enabled()) {
-        cpu_exit(cpu);
-    } else {
-        qemu_cpu_kick_rr_cpus();
-    }
-}
-
-/*
- * TCG vCPU kick timer
- *
- * The kick timer is responsible for moving single threaded vCPU
- * emulation on to the next vCPU. If more than one vCPU is running a
- * timer event with force a cpu->exit so the next vCPU can get
- * scheduled.
- *
- * The timer is removed if all vCPUs are idle and restarted again once
- * idleness is complete.
- */
-
-static QEMUTimer *tcg_kick_vcpu_timer;
-static CPUState *tcg_current_rr_cpu;
-
-#define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
-
-static inline int64_t qemu_tcg_next_kick(void)
-{
-    return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + TCG_KICK_PERIOD;
-}
-
-/* Kick the currently round-robin scheduled vCPU to next */
-static void qemu_cpu_kick_rr_next_cpu(void)
-{
-    CPUState *cpu;
-    do {
-        cpu = qatomic_mb_read(&tcg_current_rr_cpu);
-        if (cpu) {
-            cpu_exit(cpu);
-        }
-    } while (cpu != qatomic_mb_read(&tcg_current_rr_cpu));
-}
-
-static void kick_tcg_thread(void *opaque)
-{
-    timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
-    qemu_cpu_kick_rr_next_cpu();
-}
-
-static void start_tcg_kick_timer(void)
-{
-    assert(!mttcg_enabled);
-    if (!tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
-        tcg_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
-                                           kick_tcg_thread, NULL);
-    }
-    if (tcg_kick_vcpu_timer && !timer_pending(tcg_kick_vcpu_timer)) {
-        timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
-    }
-}
-
-static void stop_tcg_kick_timer(void)
-{
-    assert(!mttcg_enabled);
-    if (tcg_kick_vcpu_timer && timer_pending(tcg_kick_vcpu_timer)) {
-        timer_del(tcg_kick_vcpu_timer);
-    }
-}
-
-static void qemu_tcg_destroy_vcpu(CPUState *cpu)
-{
-}
-
-static void qemu_tcg_rr_wait_io_event(void)
-{
-    CPUState *cpu;
-
-    while (all_cpu_threads_idle()) {
-        stop_tcg_kick_timer();
-        qemu_cond_wait_iothread(first_cpu->halt_cond);
-    }
-
-    start_tcg_kick_timer();
-
-    CPU_FOREACH(cpu) {
-        qemu_wait_io_event_common(cpu);
-    }
-}
-
-static int64_t tcg_get_icount_limit(void)
-{
-    int64_t deadline;
-
-    if (replay_mode != REPLAY_MODE_PLAY) {
-        /*
-         * Include all the timers, because they may need an attention.
-         * Too long CPU execution may create unnecessary delay in UI.
-         */
-        deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
-                                              QEMU_TIMER_ATTR_ALL);
-        /* Check realtime timers, because they help with input processing */
-        deadline = qemu_soonest_timeout(deadline,
-                qemu_clock_deadline_ns_all(QEMU_CLOCK_REALTIME,
-                                           QEMU_TIMER_ATTR_ALL));
-
-        /*
-         * Maintain prior (possibly buggy) behaviour where if no deadline
-         * was set (as there is no QEMU_CLOCK_VIRTUAL timer) or it is more than
-         * INT32_MAX nanoseconds ahead, we still use INT32_MAX
-         * nanoseconds.
-         */
-        if ((deadline < 0) || (deadline > INT32_MAX)) {
-            deadline = INT32_MAX;
-        }
-
-        return icount_round(deadline);
-    } else {
-        return replay_get_instructions();
-    }
-}
-
-static void notify_aio_contexts(void)
-{
-    /* Wake up other AioContexts.  */
-    qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
-    qemu_clock_run_timers(QEMU_CLOCK_VIRTUAL);
-}
-
-static void handle_icount_deadline(void)
-{
-    assert(qemu_in_vcpu_thread());
-    if (icount_enabled()) {
-        int64_t deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
-                                                      QEMU_TIMER_ATTR_ALL);
-
-        if (deadline == 0) {
-            notify_aio_contexts();
-        }
-    }
-}
-
-static void prepare_icount_for_run(CPUState *cpu)
-{
-    if (icount_enabled()) {
-        int insns_left;
-
-        /*
-         * These should always be cleared by process_icount_data after
-         * each vCPU execution. However u16.high can be raised
-         * asynchronously by cpu_exit/cpu_interrupt/tcg_handle_interrupt
-         */
-        g_assert(cpu_neg(cpu)->icount_decr.u16.low == 0);
-        g_assert(cpu->icount_extra == 0);
-
-        cpu->icount_budget = tcg_get_icount_limit();
-        insns_left = MIN(0xffff, cpu->icount_budget);
-        cpu_neg(cpu)->icount_decr.u16.low = insns_left;
-        cpu->icount_extra = cpu->icount_budget - insns_left;
-
-        replay_mutex_lock();
-
-        if (cpu->icount_budget == 0 && replay_has_checkpoint()) {
-            notify_aio_contexts();
-        }
-    }
-}
-
-static void process_icount_data(CPUState *cpu)
-{
-    if (icount_enabled()) {
-        /* Account for executed instructions */
-        icount_update(cpu);
-
-        /* Reset the counters */
-        cpu_neg(cpu)->icount_decr.u16.low = 0;
-        cpu->icount_extra = 0;
-        cpu->icount_budget = 0;
-
-        replay_account_executed_instructions();
-
-        replay_mutex_unlock();
-    }
-}
-
-static int tcg_cpu_exec(CPUState *cpu)
-{
-    int ret;
-#ifdef CONFIG_PROFILER
-    int64_t ti;
-#endif
-
-    assert(tcg_enabled());
-#ifdef CONFIG_PROFILER
-    ti = profile_getclock();
-#endif
-    cpu_exec_start(cpu);
-    ret = cpu_exec(cpu);
-    cpu_exec_end(cpu);
-#ifdef CONFIG_PROFILER
-    qatomic_set(&tcg_ctx->prof.cpu_exec_time,
-                tcg_ctx->prof.cpu_exec_time + profile_getclock() - ti);
-#endif
-    return ret;
-}
-
-/*
- * Destroy any remaining vCPUs which have been unplugged and have
- * finished running
- */
-static void deal_with_unplugged_cpus(void)
-{
-    CPUState *cpu;
-
-    CPU_FOREACH(cpu) {
-        if (cpu->unplug && !cpu_can_run(cpu)) {
-            qemu_tcg_destroy_vcpu(cpu);
-            cpu_thread_signal_destroyed(cpu);
-            break;
-        }
-    }
-}
-
-/*
- * Single-threaded TCG
- *
- * In the single-threaded case each vCPU is simulated in turn. If
- * there is more than a single vCPU we create a simple timer to kick
- * the vCPU and ensure we don't get stuck in a tight loop in one vCPU.
- * This is done explicitly rather than relying on side-effects
- * elsewhere.
- */
-
-static void *tcg_rr_cpu_thread_fn(void *arg)
-{
-    CPUState *cpu = arg;
-
-    assert(tcg_enabled());
-    rcu_register_thread();
-    tcg_register_thread();
-
-    qemu_mutex_lock_iothread();
-    qemu_thread_get_self(cpu->thread);
-
-    cpu->thread_id = qemu_get_thread_id();
-    cpu->can_do_io = 1;
-    cpu_thread_signal_created(cpu);
-    qemu_guest_random_seed_thread_part2(cpu->random_seed);
-
-    /* wait for initial kick-off after machine start */
-    while (first_cpu->stopped) {
-        qemu_cond_wait_iothread(first_cpu->halt_cond);
-
-        /* process any pending work */
-        CPU_FOREACH(cpu) {
-            current_cpu = cpu;
-            qemu_wait_io_event_common(cpu);
-        }
-    }
-
-    start_tcg_kick_timer();
-
-    cpu = first_cpu;
-
-    /* process any pending work */
-    cpu->exit_request = 1;
-
-    while (1) {
-        qemu_mutex_unlock_iothread();
-        replay_mutex_lock();
-        qemu_mutex_lock_iothread();
-        /* Account partial waits to QEMU_CLOCK_VIRTUAL.  */
-        icount_account_warp_timer();
-
-        /*
-         * Run the timers here.  This is much more efficient than
-         * waking up the I/O thread and waiting for completion.
-         */
-        handle_icount_deadline();
-
-        replay_mutex_unlock();
-
-        if (!cpu) {
-            cpu = first_cpu;
-        }
-
-        while (cpu && cpu_work_list_empty(cpu) && !cpu->exit_request) {
-
-            qatomic_mb_set(&tcg_current_rr_cpu, cpu);
-            current_cpu = cpu;
-
-            qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
-                              (cpu->singlestep_enabled & SSTEP_NOTIMER) == 0);
-
-            if (cpu_can_run(cpu)) {
-                int r;
-
-                qemu_mutex_unlock_iothread();
-                prepare_icount_for_run(cpu);
-
-                r = tcg_cpu_exec(cpu);
-
-                process_icount_data(cpu);
-                qemu_mutex_lock_iothread();
-
-                if (r == EXCP_DEBUG) {
-                    cpu_handle_guest_debug(cpu);
-                    break;
-                } else if (r == EXCP_ATOMIC) {
-                    qemu_mutex_unlock_iothread();
-                    cpu_exec_step_atomic(cpu);
-                    qemu_mutex_lock_iothread();
-                    break;
-                }
-            } else if (cpu->stop) {
-                if (cpu->unplug) {
-                    cpu = CPU_NEXT(cpu);
-                }
-                break;
-            }
-
-            cpu = CPU_NEXT(cpu);
-        } /* while (cpu && !cpu->exit_request).. */
-
-        /* Does not need qatomic_mb_set because a spurious wakeup is okay.  */
-        qatomic_set(&tcg_current_rr_cpu, NULL);
-
-        if (cpu && cpu->exit_request) {
-            qatomic_mb_set(&cpu->exit_request, 0);
-        }
-
-        if (icount_enabled() && all_cpu_threads_idle()) {
-            /*
-             * When all cpus are sleeping (e.g in WFI), to avoid a deadlock
-             * in the main_loop, wake it up in order to start the warp timer.
-             */
-            qemu_notify_event();
-        }
-
-        qemu_tcg_rr_wait_io_event();
-        deal_with_unplugged_cpus();
-    }
-
-    rcu_unregister_thread();
-    return NULL;
-}
-
-/*
- * Multi-threaded TCG
- *
- * In the multi-threaded case each vCPU has its own thread. The TLS
- * variable current_cpu can be used deep in the code to find the
- * current CPUState for a given thread.
- */
-
-static void *tcg_cpu_thread_fn(void *arg)
-{
-    CPUState *cpu = arg;
-
-    assert(tcg_enabled());
-    g_assert(!icount_enabled());
-
-    rcu_register_thread();
-    tcg_register_thread();
-
-    qemu_mutex_lock_iothread();
-    qemu_thread_get_self(cpu->thread);
-
-    cpu->thread_id = qemu_get_thread_id();
-    cpu->can_do_io = 1;
-    current_cpu = cpu;
-    cpu_thread_signal_created(cpu);
-    qemu_guest_random_seed_thread_part2(cpu->random_seed);
-
-    /* process any pending work */
-    cpu->exit_request = 1;
-
-    do {
-        if (cpu_can_run(cpu)) {
-            int r;
-            qemu_mutex_unlock_iothread();
-            r = tcg_cpu_exec(cpu);
-            qemu_mutex_lock_iothread();
-            switch (r) {
-            case EXCP_DEBUG:
-                cpu_handle_guest_debug(cpu);
-                break;
-            case EXCP_HALTED:
-                /*
-                 * during start-up the vCPU is reset and the thread is
-                 * kicked several times. If we don't ensure we go back
-                 * to sleep in the halted state we won't cleanly
-                 * start-up when the vCPU is enabled.
-                 *
-                 * cpu->halted should ensure we sleep in wait_io_event
-                 */
-                g_assert(cpu->halted);
-                break;
-            case EXCP_ATOMIC:
-                qemu_mutex_unlock_iothread();
-                cpu_exec_step_atomic(cpu);
-                qemu_mutex_lock_iothread();
-            default:
-                /* Ignore everything else? */
-                break;
-            }
-        }
-
-        qatomic_mb_set(&cpu->exit_request, 0);
-        qemu_wait_io_event(cpu);
-    } while (!cpu->unplug || cpu_can_run(cpu));
-
-    qemu_tcg_destroy_vcpu(cpu);
-    cpu_thread_signal_destroyed(cpu);
-    qemu_mutex_unlock_iothread();
-    rcu_unregister_thread();
-    return NULL;
-}
-
-static void tcg_start_vcpu_thread(CPUState *cpu)
+void tcg_start_vcpu_thread(CPUState *cpu)
 {
     char thread_name[VCPU_THREAD_NAME_SIZE];
     static QemuCond *single_tcg_halt_cond;
@@ -XXX,XX +XXX,XX @@ static void tcg_start_vcpu_thread(CPUState *cpu)
     }
 }
 
-static int64_t tcg_get_virtual_clock(void)
+void qemu_tcg_destroy_vcpu(CPUState *cpu)
 {
-    if (icount_enabled()) {
-        return icount_get();
-    }
-    return cpu_get_clock();
+    cpu_thread_signal_destroyed(cpu);
 }
 
-static int64_t tcg_get_elapsed_ticks(void)
+int tcg_cpu_exec(CPUState *cpu)
 {
-    if (icount_enabled()) {
-        return icount_get();
-    }
-    return cpu_get_ticks();
+    int ret;
+#ifdef CONFIG_PROFILER
+    int64_t ti;
+#endif
+    assert(tcg_enabled());
+#ifdef CONFIG_PROFILER
+    ti = profile_getclock();
+#endif
+    cpu_exec_start(cpu);
+    ret = cpu_exec(cpu);
+    cpu_exec_end(cpu);
+#ifdef CONFIG_PROFILER
+    qatomic_set(&tcg_ctx->prof.cpu_exec_time,
+                tcg_ctx->prof.cpu_exec_time + profile_getclock() - ti);
+#endif
+    return ret;
 }
 
 /* mask must never be zero, except for A20 change call */
-static void tcg_handle_interrupt(CPUState *cpu, int mask)
+void tcg_handle_interrupt(CPUState *cpu, int mask)
 {
-    int old_mask;
     g_assert(qemu_mutex_iothread_locked());
 
-    old_mask = cpu->interrupt_request;
     cpu->interrupt_request |= mask;
 
     /*
@@ -XXX,XX +XXX,XX @@ static void tcg_handle_interrupt(CPUState *cpu, int mask)
         qemu_cpu_kick(cpu);
     } else {
         qatomic_set(&cpu_neg(cpu)->icount_decr.u16.high, -1);
-        if (icount_enabled() &&
-            !cpu->can_do_io
-            && (mask & ~old_mask) != 0) {
-            cpu_abort(cpu, "Raised interrupt while not in I/O function");
-        }
     }
 }
-
-const CpusAccel tcg_cpus = {
-    .create_vcpu_thread = tcg_start_vcpu_thread,
-    .kick_vcpu_thread = tcg_kick_vcpu_thread,
-
-    .handle_interrupt = tcg_handle_interrupt,
-
-    .get_virtual_clock = tcg_get_virtual_clock,
-    .get_elapsed_ticks = tcg_get_elapsed_ticks,
-};
diff --git a/softmmu/icount.c b/softmmu/icount.c
index XXXXXXX..XXXXXXX 100644
--- a/softmmu/icount.c
+++ b/softmmu/icount.c
@@ -XXX,XX +XXX,XX @@ void icount_start_warp_timer(void)
 
 void icount_account_warp_timer(void)
 {
-    if (!icount_enabled() || !icount_sleep) {
+    if (!icount_sleep) {
         return;
     }
 
diff --git a/accel/tcg/meson.build b/accel/tcg/meson.build
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/meson.build
+++ b/accel/tcg/meson.build
@@ -XXX,XX +XXX,XX @@ tcg_ss.add(when: 'CONFIG_SOFTMMU', if_false: files('user-exec-stub.c'))
 tcg_ss.add(when: 'CONFIG_PLUGIN', if_true: [files('plugin-gen.c'), libdl])
 specific_ss.add_all(when: 'CONFIG_TCG', if_true: tcg_ss)
 
-specific_ss.add(when: ['CONFIG_SOFTMMU', 'CONFIG_TCG'], if_true: files('tcg-all.c', 'cputlb.c', 'tcg-cpus.c'))
+specific_ss.add(when: ['CONFIG_SOFTMMU', 'CONFIG_TCG'], if_true: files(
+  'tcg-all.c',
+  'cputlb.c',
+  'tcg-cpus.c',
+  'tcg-cpus-mttcg.c',
+  'tcg-cpus-icount.c',
+  'tcg-cpus-rr.c'
+))
-- 
2.25.1

From: Claudio Fontana <cfontana@suse.de>

after the initial split into 3 tcg variants, we proceed to also
split tcg_start_vcpu_thread.

We actually split it in 2 this time, since the icount variant
just uses the round robin function.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Claudio Fontana <cfontana@suse.de>
Message-Id: <20201015143217.29337-3-cfontana@suse.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/tcg-cpus-mttcg.h  | 21 --------------
 accel/tcg/tcg-cpus-rr.h     |  3 +-
 accel/tcg/tcg-cpus.h        |  1 -
 accel/tcg/tcg-all.c         |  5 ++++
 accel/tcg/tcg-cpus-icount.c |  2 +-
 accel/tcg/tcg-cpus-mttcg.c  | 29 +++++++++++++++++--
 accel/tcg/tcg-cpus-rr.c     | 39 +++++++++++++++++++++++--
 accel/tcg/tcg-cpus.c        | 58 -------------------------------------
 8 files changed, 71 insertions(+), 87 deletions(-)
 delete mode 100644 accel/tcg/tcg-cpus-mttcg.h

diff --git a/accel/tcg/tcg-cpus-mttcg.h b/accel/tcg/tcg-cpus-mttcg.h
deleted file mode 100644
index XXXXXXX..XXXXXXX
--- a/accel/tcg/tcg-cpus-mttcg.h
+++ /dev/null
@@ -XXX,XX +XXX,XX @@
-/*
- * QEMU TCG Multi Threaded vCPUs implementation
- *
- * Copyright 2020 SUSE LLC
- *
- * This work is licensed under the terms of the GNU GPL, version 2 or later.
- * See the COPYING file in the top-level directory.
- */
-
-#ifndef TCG_CPUS_MTTCG_H
-#define TCG_CPUS_MTTCG_H
-
-/*
- * In the multi-threaded case each vCPU has its own thread. The TLS
- * variable current_cpu can be used deep in the code to find the
- * current CPUState for a given thread.
- */
-
-void *tcg_cpu_thread_fn(void *arg);
-
-#endif /* TCG_CPUS_MTTCG_H */
diff --git a/accel/tcg/tcg-cpus-rr.h b/accel/tcg/tcg-cpus-rr.h
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-rr.h
+++ b/accel/tcg/tcg-cpus-rr.h
@@ -XXX,XX +XXX,XX @@
 /* Kick all RR vCPUs. */
 void qemu_cpu_kick_rr_cpus(CPUState *unused);
 
-void *tcg_rr_cpu_thread_fn(void *arg);
+/* start the round robin vcpu thread */
+void rr_start_vcpu_thread(CPUState *cpu);
 
 #endif /* TCG_CPUS_RR_H */
diff --git a/accel/tcg/tcg-cpus.h b/accel/tcg/tcg-cpus.h
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus.h
+++ b/accel/tcg/tcg-cpus.h
@@ -XXX,XX +XXX,XX @@ extern const CpusAccel tcg_cpus_mttcg;
 extern const CpusAccel tcg_cpus_icount;
 extern const CpusAccel tcg_cpus_rr;
 
-void tcg_start_vcpu_thread(CPUState *cpu);
 void qemu_tcg_destroy_vcpu(CPUState *cpu);
 int tcg_cpu_exec(CPUState *cpu);
 void tcg_handle_interrupt(CPUState *cpu, int mask);
diff --git a/accel/tcg/tcg-all.c b/accel/tcg/tcg-all.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-all.c
+++ b/accel/tcg/tcg-all.c
@@ -XXX,XX +XXX,XX @@ static int tcg_init(MachineState *ms)
     tcg_exec_init(s->tb_size * 1024 * 1024);
     mttcg_enabled = s->mttcg_enabled;
 
+    /*
+     * Initialize TCG regions
+     */
+    tcg_region_init();
+
     if (mttcg_enabled) {
         cpus_register_accel(&tcg_cpus_mttcg);
     } else if (icount_enabled()) {
diff --git a/accel/tcg/tcg-cpus-icount.c b/accel/tcg/tcg-cpus-icount.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-icount.c
+++ b/accel/tcg/tcg-cpus-icount.c
@@ -XXX,XX +XXX,XX @@ static void icount_handle_interrupt(CPUState *cpu, int mask)
 }
 
 const CpusAccel tcg_cpus_icount = {
-    .create_vcpu_thread = tcg_start_vcpu_thread,
+    .create_vcpu_thread = rr_start_vcpu_thread,
     .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
 
     .handle_interrupt = icount_handle_interrupt,
diff --git a/accel/tcg/tcg-cpus-mttcg.c b/accel/tcg/tcg-cpus-mttcg.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-mttcg.c
+++ b/accel/tcg/tcg-cpus-mttcg.c
@@ -XXX,XX +XXX,XX @@
 #include "hw/boards.h"
 
 #include "tcg-cpus.h"
-#include "tcg-cpus-mttcg.h"
 
 /*
  * In the multi-threaded case each vCPU has its own thread. The TLS
@@ -XXX,XX +XXX,XX @@
  * current CPUState for a given thread.
  */
 
-void *tcg_cpu_thread_fn(void *arg)
+static void *tcg_cpu_thread_fn(void *arg)
 {
     CPUState *cpu = arg;
 
@@ -XXX,XX +XXX,XX @@ static void mttcg_kick_vcpu_thread(CPUState *cpu)
     cpu_exit(cpu);
 }
 
+static void mttcg_start_vcpu_thread(CPUState *cpu)
+{
+    char thread_name[VCPU_THREAD_NAME_SIZE];
+
+    g_assert(tcg_enabled());
+
+    parallel_cpus = (current_machine->smp.max_cpus > 1);
+
+    cpu->thread = g_malloc0(sizeof(QemuThread));
+    cpu->halt_cond = g_malloc0(sizeof(QemuCond));
+    qemu_cond_init(cpu->halt_cond);
+
+    /* create a thread per vCPU with TCG (MTTCG) */
+    snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
+             cpu->cpu_index);
+
+    qemu_thread_create(cpu->thread, thread_name, tcg_cpu_thread_fn,
+                       cpu, QEMU_THREAD_JOINABLE);
+
+#ifdef _WIN32
+    cpu->hThread = qemu_thread_get_handle(cpu->thread);
+#endif
+}
+
 const CpusAccel tcg_cpus_mttcg = {
-    .create_vcpu_thread = tcg_start_vcpu_thread,
+    .create_vcpu_thread = mttcg_start_vcpu_thread,
     .kick_vcpu_thread = mttcg_kick_vcpu_thread,
 
     .handle_interrupt = tcg_handle_interrupt,
diff --git a/accel/tcg/tcg-cpus-rr.c b/accel/tcg/tcg-cpus-rr.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-rr.c
+++ b/accel/tcg/tcg-cpus-rr.c
@@ -XXX,XX +XXX,XX @@ static void deal_with_unplugged_cpus(void)
  * elsewhere.
  */
 
-void *tcg_rr_cpu_thread_fn(void *arg)
+static void *tcg_rr_cpu_thread_fn(void *arg)
 {
     CPUState *cpu = arg;
 
@@ -XXX,XX +XXX,XX @@ void *tcg_rr_cpu_thread_fn(void *arg)
     return NULL;
 }
 
+void rr_start_vcpu_thread(CPUState *cpu)
+{
+    char thread_name[VCPU_THREAD_NAME_SIZE];
+    static QemuCond *single_tcg_halt_cond;
+    static QemuThread *single_tcg_cpu_thread;
+
+    g_assert(tcg_enabled());
+    parallel_cpus = false;
+
+    if (!single_tcg_cpu_thread) {
+        cpu->thread = g_malloc0(sizeof(QemuThread));
+        cpu->halt_cond = g_malloc0(sizeof(QemuCond));
+        qemu_cond_init(cpu->halt_cond);
+
+        /* share a single thread for all cpus with TCG */
+        snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "ALL CPUs/TCG");
+        qemu_thread_create(cpu->thread, thread_name,
+                           tcg_rr_cpu_thread_fn,
+                           cpu, QEMU_THREAD_JOINABLE);
+
+        single_tcg_halt_cond = cpu->halt_cond;
+        single_tcg_cpu_thread = cpu->thread;
+#ifdef _WIN32
+        cpu->hThread = qemu_thread_get_handle(cpu->thread);
+#endif
+    } else {
+        /* we share the thread */
+        cpu->thread = single_tcg_cpu_thread;
+        cpu->halt_cond = single_tcg_halt_cond;
+        cpu->thread_id = first_cpu->thread_id;
+        cpu->can_do_io = 1;
+        cpu->created = true;
+    }
+}
+
 const CpusAccel tcg_cpus_rr = {
-    .create_vcpu_thread = tcg_start_vcpu_thread,
+    .create_vcpu_thread = rr_start_vcpu_thread,
     .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
 
     .handle_interrupt = tcg_handle_interrupt,
diff --git a/accel/tcg/tcg-cpus.c b/accel/tcg/tcg-cpus.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus.c
+++ b/accel/tcg/tcg-cpus.c
@@ -XXX,XX +XXX,XX @@
 #include "hw/boards.h"
 
 #include "tcg-cpus.h"
-#include "tcg-cpus-mttcg.h"
-#include "tcg-cpus-rr.h"
 
 /* common functionality among all TCG variants */
 
-void tcg_start_vcpu_thread(CPUState *cpu)
-{
-    char thread_name[VCPU_THREAD_NAME_SIZE];
-    static QemuCond *single_tcg_halt_cond;
-    static QemuThread *single_tcg_cpu_thread;
-    static int tcg_region_inited;
-
-    assert(tcg_enabled());
-    /*
-     * Initialize TCG regions--once. Now is a good time, because:
-     * (1) TCG's init context, prologue and target globals have been set up.
-     * (2) qemu_tcg_mttcg_enabled() works now (TCG init code runs before the
-     *     -accel flag is processed, so the check doesn't work then).
-     */
-    if (!tcg_region_inited) {
-        tcg_region_inited = 1;
-        tcg_region_init();
-        parallel_cpus = qemu_tcg_mttcg_enabled() && current_machine->smp.max_cpus > 1;
-    }
-
-    if (qemu_tcg_mttcg_enabled() || !single_tcg_cpu_thread) {
-        cpu->thread = g_malloc0(sizeof(QemuThread));
-        cpu->halt_cond = g_malloc0(sizeof(QemuCond));
-        qemu_cond_init(cpu->halt_cond);
-
-        if (qemu_tcg_mttcg_enabled()) {
-            /* create a thread per vCPU with TCG (MTTCG) */
-            snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
-                 cpu->cpu_index);
-
-            qemu_thread_create(cpu->thread, thread_name, tcg_cpu_thread_fn,
-                               cpu, QEMU_THREAD_JOINABLE);
-
-        } else {
-            /* share a single thread for all cpus with TCG */
-            snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "ALL CPUs/TCG");
-            qemu_thread_create(cpu->thread, thread_name,
-                               tcg_rr_cpu_thread_fn,
-                               cpu, QEMU_THREAD_JOINABLE);
-
-            single_tcg_halt_cond = cpu->halt_cond;
-            single_tcg_cpu_thread = cpu->thread;
-        }
-#ifdef _WIN32
-        cpu->hThread = qemu_thread_get_handle(cpu->thread);
-#endif
-    } else {
-        /* For non-MTTCG cases we share the thread */
-        cpu->thread = single_tcg_cpu_thread;
-        cpu->halt_cond = single_tcg_halt_cond;
-        cpu->thread_id = first_cpu->thread_id;
-        cpu->can_do_io = 1;
-        cpu->created = true;
-    }
-}
-
 void qemu_tcg_destroy_vcpu(CPUState *cpu)
 {
     cpu_thread_signal_destroyed(cpu);
-- 
2.25.1

From: Claudio Fontana <cfontana@suse.de>

Signed-off-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201015143217.29337-4-cfontana@suse.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/tcg-cpus-icount.h |  6 +--
 accel/tcg/tcg-cpus-rr.h     |  2 +-
 accel/tcg/tcg-cpus.h        |  6 +--
 accel/tcg/tcg-cpus-icount.c | 24 ++++++------
 accel/tcg/tcg-cpus-mttcg.c  | 10 ++---
 accel/tcg/tcg-cpus-rr.c     | 74 ++++++++++++++++++-------------------
 accel/tcg/tcg-cpus.c        |  6 +--
 7 files changed, 64 insertions(+), 64 deletions(-)

diff --git a/accel/tcg/tcg-cpus-icount.h b/accel/tcg/tcg-cpus-icount.h
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-icount.h
+++ b/accel/tcg/tcg-cpus-icount.h
@@ -XXX,XX +XXX,XX @@
 #ifndef TCG_CPUS_ICOUNT_H
 #define TCG_CPUS_ICOUNT_H
 
-void handle_icount_deadline(void);
-void prepare_icount_for_run(CPUState *cpu);
-void process_icount_data(CPUState *cpu);
+void icount_handle_deadline(void);
+void icount_prepare_for_run(CPUState *cpu);
+void icount_process_data(CPUState *cpu);
 
 #endif /* TCG_CPUS_ICOUNT_H */
diff --git a/accel/tcg/tcg-cpus-rr.h b/accel/tcg/tcg-cpus-rr.h
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-rr.h
+++ b/accel/tcg/tcg-cpus-rr.h
@@ -XXX,XX +XXX,XX @@
 #define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
 
 /* Kick all RR vCPUs. */
-void qemu_cpu_kick_rr_cpus(CPUState *unused);
+void rr_kick_vcpu_thread(CPUState *unused);
 
 /* start the round robin vcpu thread */
 void rr_start_vcpu_thread(CPUState *cpu);
diff --git a/accel/tcg/tcg-cpus.h b/accel/tcg/tcg-cpus.h
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus.h
+++ b/accel/tcg/tcg-cpus.h
@@ -XXX,XX +XXX,XX @@ extern const CpusAccel tcg_cpus_mttcg;
 extern const CpusAccel tcg_cpus_icount;
 extern const CpusAccel tcg_cpus_rr;
 
-void qemu_tcg_destroy_vcpu(CPUState *cpu);
-int tcg_cpu_exec(CPUState *cpu);
-void tcg_handle_interrupt(CPUState *cpu, int mask);
+void tcg_cpus_destroy(CPUState *cpu);
+int tcg_cpus_exec(CPUState *cpu);
+void tcg_cpus_handle_interrupt(CPUState *cpu, int mask);
 
 #endif /* TCG_CPUS_H */
diff --git a/accel/tcg/tcg-cpus-icount.c b/accel/tcg/tcg-cpus-icount.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-icount.c
+++ b/accel/tcg/tcg-cpus-icount.c
@@ -XXX,XX +XXX,XX @@
 #include "tcg-cpus-icount.h"
 #include "tcg-cpus-rr.h"
 
-static int64_t tcg_get_icount_limit(void)
+static int64_t icount_get_limit(void)
 {
     int64_t deadline;
 
@@ -XXX,XX +XXX,XX @@ static int64_t tcg_get_icount_limit(void)
     }
 }
 
-static void notify_aio_contexts(void)
+static void icount_notify_aio_contexts(void)
 {
     /* Wake up other AioContexts.  */
     qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
     qemu_clock_run_timers(QEMU_CLOCK_VIRTUAL);
 }
 
-void handle_icount_deadline(void)
+void icount_handle_deadline(void)
 {
     assert(qemu_in_vcpu_thread());
     int64_t deadline = qemu_clock_deadline_ns_all(QEMU_CLOCK_VIRTUAL,
                                                   QEMU_TIMER_ATTR_ALL);
 
     if (deadline == 0) {
-        notify_aio_contexts();
+        icount_notify_aio_contexts();
     }
 }
 
-void prepare_icount_for_run(CPUState *cpu)
+void icount_prepare_for_run(CPUState *cpu)
 {
     int insns_left;
 
     /*
-     * These should always be cleared by process_icount_data after
+     * These should always be cleared by icount_process_data after
      * each vCPU execution. However u16.high can be raised
-     * asynchronously by cpu_exit/cpu_interrupt/tcg_handle_interrupt
+     * asynchronously by cpu_exit/cpu_interrupt/tcg_cpus_handle_interrupt
      */
     g_assert(cpu_neg(cpu)->icount_decr.u16.low == 0);
     g_assert(cpu->icount_extra == 0);
 
-    cpu->icount_budget = tcg_get_icount_limit();
+    cpu->icount_budget = icount_get_limit();
     insns_left = MIN(0xffff, cpu->icount_budget);
     cpu_neg(cpu)->icount_decr.u16.low = insns_left;
     cpu->icount_extra = cpu->icount_budget - insns_left;
@@ -XXX,XX +XXX,XX @@ void prepare_icount_for_run(CPUState *cpu)
     replay_mutex_lock();
 
     if (cpu->icount_budget == 0 && replay_has_checkpoint()) {
-        notify_aio_contexts();
+        icount_notify_aio_contexts();
     }
 }
 
-void process_icount_data(CPUState *cpu)
+void icount_process_data(CPUState *cpu)
 {
     /* Account for executed instructions */
     icount_update(cpu);
@@ -XXX,XX +XXX,XX @@ static void icount_handle_interrupt(CPUState *cpu, int mask)
 {
     int old_mask = cpu->interrupt_request;
 
-    tcg_handle_interrupt(cpu, mask);
+    tcg_cpus_handle_interrupt(cpu, mask);
     if (qemu_cpu_is_self(cpu) &&
         !cpu->can_do_io
         && (mask & ~old_mask) != 0) {
@@ -XXX,XX +XXX,XX @@ static void icount_handle_interrupt(CPUState *cpu, int mask)
 
 const CpusAccel tcg_cpus_icount = {
     .create_vcpu_thread = rr_start_vcpu_thread,
-    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
+    .kick_vcpu_thread = rr_kick_vcpu_thread,
 
     .handle_interrupt = icount_handle_interrupt,
     .get_virtual_clock = icount_get,
diff --git a/accel/tcg/tcg-cpus-mttcg.c b/accel/tcg/tcg-cpus-mttcg.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-mttcg.c
+++ b/accel/tcg/tcg-cpus-mttcg.c
@@ -XXX,XX +XXX,XX @@
  * current CPUState for a given thread.
  */
 
-static void *tcg_cpu_thread_fn(void *arg)
+static void *mttcg_cpu_thread_fn(void *arg)
 {
     CPUState *cpu = arg;
 
@@ -XXX,XX +XXX,XX @@ static void *tcg_cpu_thread_fn(void *arg)
         if (cpu_can_run(cpu)) {
             int r;
             qemu_mutex_unlock_iothread();
-            r = tcg_cpu_exec(cpu);
+            r = tcg_cpus_exec(cpu);
             qemu_mutex_lock_iothread();
             switch (r) {
             case EXCP_DEBUG:
@@ -XXX,XX +XXX,XX @@ static void *tcg_cpu_thread_fn(void *arg)
         qemu_wait_io_event(cpu);
     } while (!cpu->unplug || cpu_can_run(cpu));
 
-    qemu_tcg_destroy_vcpu(cpu);
+    tcg_cpus_destroy(cpu);
     qemu_mutex_unlock_iothread();
     rcu_unregister_thread();
     return NULL;
@@ -XXX,XX +XXX,XX @@ static void mttcg_start_vcpu_thread(CPUState *cpu)
     snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
              cpu->cpu_index);
 
-    qemu_thread_create(cpu->thread, thread_name, tcg_cpu_thread_fn,
+    qemu_thread_create(cpu->thread, thread_name, mttcg_cpu_thread_fn,
                        cpu, QEMU_THREAD_JOINABLE);
 
 #ifdef _WIN32
@@ -XXX,XX +XXX,XX @@ const CpusAccel tcg_cpus_mttcg = {
     .create_vcpu_thread = mttcg_start_vcpu_thread,
     .kick_vcpu_thread = mttcg_kick_vcpu_thread,
 
-    .handle_interrupt = tcg_handle_interrupt,
+    .handle_interrupt = tcg_cpus_handle_interrupt,
 };
diff --git a/accel/tcg/tcg-cpus-rr.c b/accel/tcg/tcg-cpus-rr.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus-rr.c
+++ b/accel/tcg/tcg-cpus-rr.c
@@ -XXX,XX +XXX,XX @@
 #include "tcg-cpus-icount.h"
 
 /* Kick all RR vCPUs */
-void qemu_cpu_kick_rr_cpus(CPUState *unused)
+void rr_kick_vcpu_thread(CPUState *unused)
 {
     CPUState *cpu;
 
@@ -XXX,XX +XXX,XX @@ void qemu_cpu_kick_rr_cpus(CPUState *unused)
  * idleness is complete.
  */
 
-static QEMUTimer *tcg_kick_vcpu_timer;
-static CPUState *tcg_current_rr_cpu;
+static QEMUTimer *rr_kick_vcpu_timer;
+static CPUState *rr_current_cpu;
 
 #define TCG_KICK_PERIOD (NANOSECONDS_PER_SECOND / 10)
 
-static inline int64_t qemu_tcg_next_kick(void)
+static inline int64_t rr_next_kick_time(void)
 {
     return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + TCG_KICK_PERIOD;
 }
 
 /* Kick the currently round-robin scheduled vCPU to next */
-static void qemu_cpu_kick_rr_next_cpu(void)
+static void rr_kick_next_cpu(void)
 {
     CPUState *cpu;
     do {
-        cpu = qatomic_mb_read(&tcg_current_rr_cpu);
+        cpu = qatomic_mb_read(&rr_current_cpu);
         if (cpu) {
             cpu_exit(cpu);
         }
-    } while (cpu != qatomic_mb_read(&tcg_current_rr_cpu));
+    } while (cpu != qatomic_mb_read(&rr_current_cpu));
 }
 
-static void kick_tcg_thread(void *opaque)
+static void rr_kick_thread(void *opaque)
 {
-    timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
-    qemu_cpu_kick_rr_next_cpu();
+    timer_mod(rr_kick_vcpu_timer, rr_next_kick_time());
+    rr_kick_next_cpu();
 }
 
-static void start_tcg_kick_timer(void)
+static void rr_start_kick_timer(void)
 {
-    if (!tcg_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
-        tcg_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
-                                           kick_tcg_thread, NULL);
+    if (!rr_kick_vcpu_timer && CPU_NEXT(first_cpu)) {
+        rr_kick_vcpu_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
+                                           rr_kick_thread, NULL);
     }
-    if (tcg_kick_vcpu_timer && !timer_pending(tcg_kick_vcpu_timer)) {
-        timer_mod(tcg_kick_vcpu_timer, qemu_tcg_next_kick());
+    if (rr_kick_vcpu_timer && !timer_pending(rr_kick_vcpu_timer)) {
+        timer_mod(rr_kick_vcpu_timer, rr_next_kick_time());
     }
 }
 
-static void stop_tcg_kick_timer(void)
+static void rr_stop_kick_timer(void)
 {
-    if (tcg_kick_vcpu_timer && timer_pending(tcg_kick_vcpu_timer)) {
-        timer_del(tcg_kick_vcpu_timer);
+    if (rr_kick_vcpu_timer && timer_pending(rr_kick_vcpu_timer)) {
+        timer_del(rr_kick_vcpu_timer);
     }
 }
 
-static void qemu_tcg_rr_wait_io_event(void)
+static void rr_wait_io_event(void)
 {
     CPUState *cpu;
 
     while (all_cpu_threads_idle()) {
-        stop_tcg_kick_timer();
+        rr_stop_kick_timer();
         qemu_cond_wait_iothread(first_cpu->halt_cond);
     }
 
-    start_tcg_kick_timer();
+    rr_start_kick_timer();
 
     CPU_FOREACH(cpu) {
         qemu_wait_io_event_common(cpu);
@@ -XXX,XX +XXX,XX @@ static void qemu_tcg_rr_wait_io_event(void)
  * Destroy any remaining vCPUs which have been unplugged and have
  * finished running
  */
-static void deal_with_unplugged_cpus(void)
+static void rr_deal_with_unplugged_cpus(void)
 {
     CPUState *cpu;
 
     CPU_FOREACH(cpu) {
         if (cpu->unplug && !cpu_can_run(cpu)) {
-            qemu_tcg_destroy_vcpu(cpu);
+            tcg_cpus_destroy(cpu);
             break;
         }
     }
@@ -XXX,XX +XXX,XX @@ static void deal_with_unplugged_cpus(void)
  * elsewhere.
  */
 
-static void *tcg_rr_cpu_thread_fn(void *arg)
+static void *rr_cpu_thread_fn(void *arg)
 {
     CPUState *cpu = arg;
 
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
         }
     }
 
-    start_tcg_kick_timer();
+    rr_start_kick_timer();
 
     cpu = first_cpu;
 
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
              * Run the timers here.  This is much more efficient than
              * waking up the I/O thread and waiting for completion.
              */
-            handle_icount_deadline();
+            icount_handle_deadline();
         }
 
         replay_mutex_unlock();
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
 
         while (cpu && cpu_work_list_empty(cpu) && !cpu->exit_request) {
 
-            qatomic_mb_set(&tcg_current_rr_cpu, cpu);
+            qatomic_mb_set(&rr_current_cpu, cpu);
             current_cpu = cpu;
 
             qemu_clock_enable(QEMU_CLOCK_VIRTUAL,
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
 
                 qemu_mutex_unlock_iothread();
                 if (icount_enabled()) {
-                    prepare_icount_for_run(cpu);
+                    icount_prepare_for_run(cpu);
                 }
-                r = tcg_cpu_exec(cpu);
+                r = tcg_cpus_exec(cpu);
                 if (icount_enabled()) {
-                    process_icount_data(cpu);
+                    icount_process_data(cpu);
                 }
                 qemu_mutex_lock_iothread();
 
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
         } /* while (cpu && !cpu->exit_request).. */
 
         /* Does not need qatomic_mb_set because a spurious wakeup is okay.  */
-        qatomic_set(&tcg_current_rr_cpu, NULL);
+        qatomic_set(&rr_current_cpu, NULL);
 
         if (cpu && cpu->exit_request) {
             qatomic_mb_set(&cpu->exit_request, 0);
@@ -XXX,XX +XXX,XX @@ static void *tcg_rr_cpu_thread_fn(void *arg)
             qemu_notify_event();
         }
 
-        qemu_tcg_rr_wait_io_event();
-        deal_with_unplugged_cpus();
+        rr_wait_io_event();
+        rr_deal_with_unplugged_cpus();
     }
 
     rcu_unregister_thread();
@@ -XXX,XX +XXX,XX @@ void rr_start_vcpu_thread(CPUState *cpu)
         /* share a single thread for all cpus with TCG */
         snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "ALL CPUs/TCG");
         qemu_thread_create(cpu->thread, thread_name,
-                           tcg_rr_cpu_thread_fn,
+                           rr_cpu_thread_fn,
                            cpu, QEMU_THREAD_JOINABLE);
 
         single_tcg_halt_cond = cpu->halt_cond;
@@ -XXX,XX +XXX,XX @@ void rr_start_vcpu_thread(CPUState *cpu)
 
 const CpusAccel tcg_cpus_rr = {
     .create_vcpu_thread = rr_start_vcpu_thread,
-    .kick_vcpu_thread = qemu_cpu_kick_rr_cpus,
+    .kick_vcpu_thread = rr_kick_vcpu_thread,
 
-    .handle_interrupt = tcg_handle_interrupt,
+    .handle_interrupt = tcg_cpus_handle_interrupt,
 };
diff --git a/accel/tcg/tcg-cpus.c b/accel/tcg/tcg-cpus.c
index XXXXXXX..XXXXXXX 100644
--- a/accel/tcg/tcg-cpus.c
+++ b/accel/tcg/tcg-cpus.c
@@ -XXX,XX +XXX,XX @@
 
 /* common functionality among all TCG variants */
 
-void qemu_tcg_destroy_vcpu(CPUState *cpu)
+void tcg_cpus_destroy(CPUState *cpu)
 {
     cpu_thread_signal_destroyed(cpu);
 }
 
-int tcg_cpu_exec(CPUState *cpu)
+int tcg_cpus_exec(CPUState *cpu)
 {
     int ret;
 #ifdef CONFIG_PROFILER
@@ -XXX,XX +XXX,XX @@ int tcg_cpu_exec(CPUState *cpu)
 }
 
 /* mask must never be zero, except for A20 change call */
-void tcg_handle_interrupt(CPUState *cpu, int mask)
+void tcg_cpus_handle_interrupt(CPUState *cpu, int mask)
 {
     g_assert(qemu_mutex_iothread_locked());
 
-- 
2.25.1