[PATCH V8 20/39] cpr: restart mode

[Steve Sistare]

Provide the cpr-save restart mode, which preserves the guest VM across a
restart of the qemu process.  After cpr-save, the caller passes qemu
command-line arguments to cpr-exec, which directly exec's the new qemu
binary.  The arguments must include -S so new qemu starts in a paused state.
The caller resumes the guest by calling cpr-load.

To use the restart mode, guest RAM must be backed by a memory-backend-file
with share=on.  The '-cpr-enable restart' option causes secondary guest
ram blocks (those not specified on the command line) to be allocated by
mmap'ing a memfd.  The memfd values are saved in special cpr state which
is retrieved after exec, and are kept open across exec, after which they
are retrieved and re-mmap'd.  Hence guest RAM is preserved in place, albeit
with new virtual addresses in the qemu process.

The restart mode supports vfio devices and memory-backend-memfd in
subsequent patches.

cpr-exec syntax:
  { 'command': 'cpr-exec', 'data': { 'argv': [ 'str' ] } }

Add the restart mode:
  { 'enum': 'CprMode', 'data': [ 'reboot', 'restart' ] }

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
---
 migration/cpr.c   | 35 +++++++++++++++++++++++++++++++++++
 qapi/cpr.json     | 26 +++++++++++++++++++++++++-
 qemu-options.hx   |  2 +-
 softmmu/physmem.c | 46 +++++++++++++++++++++++++++++++++++++++++++++-
 trace-events      |  1 +
 5 files changed, 107 insertions(+), 3 deletions(-)

diff --git a/migration/cpr.c b/migration/cpr.c
index 1cc8738..8b3fffd 100644
--- a/migration/cpr.c
+++ b/migration/cpr.c
@@ -22,6 +22,7 @@ static int cpr_enabled_modes;
 void cpr_init(int modes)
 {
     cpr_enabled_modes = modes;
+    cpr_state_load(&error_fatal);
 }
 
 bool cpr_enabled(CprMode mode)
@@ -153,6 +154,37 @@ err:
     cpr_set_mode(CPR_MODE_NONE);
 }
 
+static int preserve_fd(const char *name, int id, int fd, void *opaque)
+{
+    qemu_clear_cloexec(fd);
+    return 0;
+}
+
+static int unpreserve_fd(const char *name, int id, int fd, void *opaque)
+{
+    qemu_set_cloexec(fd);
+    return 0;
+}
+
+void qmp_cpr_exec(strList *args, Error **errp)
+{
+    if (!runstate_check(RUN_STATE_SAVE_VM)) {
+        error_setg(errp, "runstate is not save-vm");
+        return;
+    }
+    if (cpr_get_mode() != CPR_MODE_RESTART) {
+        error_setg(errp, "cpr-exec requires cpr-save with restart mode");
+        return;
+    }
+
+    cpr_walk_fd(preserve_fd, 0);
+    if (cpr_state_save(errp)) {
+        return;
+    }
+
+    assert(qemu_system_exec_request(args, errp) == 0);
+}
+
 void qmp_cpr_load(const char *filename, CprMode mode, Error **errp)
 {
     QEMUFile *f;
@@ -189,6 +221,9 @@ void qmp_cpr_load(const char *filename, CprMode mode, Error 
**errp)
         goto out;
     }
 
+    /* Clear cloexec to prevent fd leaks until the next cpr-save */
+    cpr_walk_fd(unpreserve_fd, 0);
+
     state = global_state_get_runstate();
     if (state == RUN_STATE_RUNNING) {
         vm_start();
diff --git a/qapi/cpr.json b/qapi/cpr.json
index 11c6f88..47ee4ff 100644
--- a/qapi/cpr.json
+++ b/qapi/cpr.json
@@ -15,11 +15,12 @@
 # @CprMode:
 #
 # @reboot: checkpoint can be cpr-load'ed after a host reboot.
+# @restart: checkpoint can be cpr-load'ed after restarting qemu.
 #
 # Since: 7.1
 ##
 { 'enum': 'CprMode',
-  'data': [ 'none', 'reboot' ] }
+  'data': [ 'none', 'reboot', 'restart' ] }
 
 ##
 # @cpr-save:
@@ -38,6 +39,11 @@
 # issue the quit command, reboot the system, start qemu using the same
 # arguments plus -S, and issue the cpr-load command.
 #
+# If @mode is 'restart', the checkpoint remains valid after restarting
+# qemu using a subsequent cpr-exec.  Guest RAM must be backed by a
+# memory-backend-file with share=on.
+# To resume from the checkpoint, issue the cpr-load command.
+#
 # @filename: name of checkpoint file
 # @mode: @CprMode mode
 #
@@ -48,6 +54,24 @@
             'mode': 'CprMode' } }
 
 ##
+# @cpr-exec:
+#
+# Restart qemu by directly exec'ing @argv[0], replacing the qemu process.
+# The PID remains the same.  Must be called after cpr-save restart.
+#
+# @argv[0] should be the path of a new qemu binary, or a prefix command that
+# in turn exec's the new qemu binary.  The arguments must match those used
+# to initially start qemu, plus the -S option so new qemu starts in a paused
+# state.
+#
+# @argv: arguments to be passed to exec().
+#
+# Since: 7.1
+##
+{ 'command': 'cpr-exec',
+  'data': { 'argv': [ 'str' ] } }
+
+##
 # @cpr-load:
 #
 # Load a virtual machine from the checkpoint file @filename that was created
diff --git a/qemu-options.hx b/qemu-options.hx
index 6e51c33..1b49360 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -4484,7 +4484,7 @@ SRST
 ERST
 
 DEF("cpr-enable", HAS_ARG, QEMU_OPTION_cpr_enable, \
-    "-cpr-enable reboot    enable the cpr mode\n",
+    "-cpr-enable reboot|restart    enable the cpr mode\n",
     QEMU_ARCH_ALL)
 SRST
 ``-cpr-enable reboot``
diff --git a/softmmu/physmem.c b/softmmu/physmem.c
index 822c424..412cc80 100644
--- a/softmmu/physmem.c
+++ b/softmmu/physmem.c
@@ -44,6 +44,7 @@
 #include "qemu/qemu-print.h"
 #include "qemu/log.h"
 #include "qemu/memalign.h"
+#include "qemu/memfd.h"
 #include "exec/memory.h"
 #include "exec/ioport.h"
 #include "sysemu/dma.h"
@@ -1962,6 +1963,40 @@ static void dirty_memory_extend(ram_addr_t old_ram_size,
     }
 }
 
+static bool memory_region_is_backend(MemoryRegion *mr)
+{
+    return !!object_dynamic_cast(mr->parent_obj.parent, TYPE_MEMORY_BACKEND);
+}
+
+static void *qemu_anon_memfd_alloc(RAMBlock *rb, size_t maxlen, Error **errp)
+{
+    size_t len, align;
+    void *addr;
+    struct MemoryRegion *mr = rb->mr;
+    const char *name = memory_region_name(mr);
+    int mfd = cpr_find_memfd(name, &len, &maxlen, &align);
+
+    if (mfd >= 0) {
+        rb->used_length = len;
+        rb->max_length = maxlen;
+        mr->align = align;
+    } else {
+        len = rb->used_length;
+        maxlen = rb->max_length;
+        mr->align = QEMU_VMALLOC_ALIGN;
+        mfd = qemu_memfd_create(name, maxlen + mr->align, 0, 0, 0, errp);
+        if (mfd < 0) {
+            return NULL;
+        }
+        cpr_save_memfd(name, mfd, len, maxlen, mr->align);
+    }
+    rb->flags |= RAM_SHARED;
+    qemu_set_cloexec(mfd);
+    addr = file_ram_alloc(rb, maxlen, mfd, false, false, 0, errp);
+    trace_anon_memfd_alloc(name, maxlen, addr, mfd);
+    return addr;
+}
+
 static void ram_block_add(RAMBlock *new_block, Error **errp)
 {
     const bool noreserve = qemu_ram_is_noreserve(new_block);
@@ -1986,6 +2021,14 @@ static void ram_block_add(RAMBlock *new_block, Error 
**errp)
                 qemu_mutex_unlock_ramlist();
                 return;
             }
+        } else if (cpr_enabled(CPR_MODE_RESTART) &&
+                   !memory_region_is_backend(new_block->mr)) {
+            new_block->host = qemu_anon_memfd_alloc(new_block,
+                                                    new_block->max_length,
+                                                    errp);
+            if (!new_block->host) {
+                return;
+            }
         } else {
             new_block->host = qemu_anon_ram_alloc(new_block->max_length,
                                                   &new_block->mr->align,
@@ -1997,8 +2040,8 @@ static void ram_block_add(RAMBlock *new_block, Error 
**errp)
                 qemu_mutex_unlock_ramlist();
                 return;
             }
-            memory_try_enable_merging(new_block->host, new_block->max_length);
         }
+        memory_try_enable_merging(new_block->host, new_block->max_length);
     }
 
     new_ram_size = MAX(old_ram_size,
@@ -2231,6 +2274,7 @@ void qemu_ram_free(RAMBlock *block)
     }
 
     qemu_mutex_lock_ramlist();
+    cpr_delete_memfd(memory_region_name(block->mr));
     QLIST_REMOVE_RCU(block, next);
     ram_list.mru_block = NULL;
     /* Write list before version */
diff --git a/trace-events b/trace-events
index bc71006..07369bb 100644
--- a/trace-events
+++ b/trace-events
@@ -45,6 +45,7 @@ ram_block_discard_range(const char *rbname, void *hva, size_t 
length, bool need_
 # accel/tcg/cputlb.c
 memory_notdirty_write_access(uint64_t vaddr, uint64_t ram_addr, unsigned size) 
"0x%" PRIx64 " ram_addr 0x%" PRIx64 " size %u"
 memory_notdirty_set_dirty(uint64_t vaddr) "0x%" PRIx64
+anon_memfd_alloc(const char *name, size_t size, void *ptr, int fd) "%s size 
%zu ptr %p fd %d"
 
 # gdbstub.c
 gdbstub_op_start(const char *device) "Starting gdbstub using device %s"
-- 
1.8.3.1