Merge remote-tracking branch 'kwolf/for-anthony' into staging

* kwolf/for-anthony: (30 commits)
  qemu-iotests: add tests for streaming error handling
  qemu-iotests: map underscore to dash in QMP argument names
  blkdebug: process all set_state rules in the old state
  stream: add on-error argument
  block: introduce block job error
  iostatus: reorganize io error code
  iostatus: change is_read to a bool
  iostatus: move BlockdevOnError declaration to QAPI
  iostatus: rename BlockErrorAction, BlockQMPEventAction
  qemu-iotests: add test for pausing a streaming operation
  qmp: add block-job-pause and block-job-resume
  block: add support for job pause/resume
  qmp: add 'busy' member to BlockJobInfo
  block: add block_job_query
  block: move job APIs to separate files
  block: fix documentation of block_job_cancel_sync
  qerror/block: introduce QERR_BLOCK_JOB_NOT_ACTIVE
  qemu-iotests: add initial tests for live block commit
  QAPI: add command for live block commit, 'block-commit'
  block: helper function, to find the base image of a chain
  ...

40 files changed, 4923 insertions, 426 deletions

diff --git a/Makefile.objs b/Makefile.objs
index 4412757309..b1f3e22547 100644
--- a/Makefile.objs
+++ b/Makefile.objs
@@ -42,7 +42,8 @@ coroutine-obj-$(CONFIG_WIN32) += coroutine-win32.o
 # block-obj-y is code used by both qemu system emulation and qemu-img
 
 block-obj-y = cutils.o iov.o cache-utils.o qemu-option.o module.o async.o
-block-obj-y += nbd.o block.o aio.o aes.o qemu-config.o qemu-progress.o qemu-sockets.o
+block-obj-y += nbd.o block.o blockjob.o aio.o aes.o qemu-config.o
+block-obj-y += qemu-progress.o qemu-sockets.o uri.o
 block-obj-y += $(coroutine-obj-y) $(qobject-obj-y) $(version-obj-y)
 block-obj-$(CONFIG_POSIX) += posix-aio-compat.o
 block-obj-$(CONFIG_LINUX_AIO) += linux-aio.o
@@ -59,7 +60,7 @@ endif
 # suppress *all* target specific code in case of system emulation, i.e. a
 # single QEMU executable should support all CPUs and machines.
 
-common-obj-y = $(block-obj-y) blockdev.o
+common-obj-y = $(block-obj-y) blockdev.o block/
 common-obj-y += net.o net/
 common-obj-y += qom/
 common-obj-y += readline.o console.o cursor.o
diff --git a/QMP/qmp-events.txt b/QMP/qmp-events.txt
index 287805825f..987c5756b3 100644
--- a/QMP/qmp-events.txt
+++ b/QMP/qmp-events.txt
@@ -50,7 +50,8 @@ Emitted when a block job has been cancelled.
 
 Data:
 
-- "type":     Job type ("stream" for image streaming, json-string)
+- "type":     Job type (json-string; "stream" for image streaming
+                                     "commit" for block commit)
 - "device":   Device name (json-string)
 - "len":      Maximum progress value (json-int)
 - "offset":   Current progress value (json-int)
@@ -73,7 +74,8 @@ Emitted when a block job has completed.
 
 Data:
 
-- "type":     Job type ("stream" for image streaming, json-string)
+- "type":     Job type (json-string; "stream" for image streaming
+                                     "commit" for block commit)
 - "device":   Device name (json-string)
 - "len":      Maximum progress value (json-int)
 - "offset":   Current progress value (json-int)
@@ -94,6 +96,28 @@ Example:
                "speed": 0 },
      "timestamp": { "seconds": 1267061043, "microseconds": 959568 } }
 
+BLOCK_JOB_ERROR
+---------------
+
+Emitted when a block job encounters an error.
+
+Data:
+
+- "device": device name (json-string)
+- "operation": I/O operation (json-string, "read" or "write")
+- "action": action that has been taken, it's one of the following (json-string):
+    "ignore": error has been ignored, the job may fail later
+    "report": error will be reported and the job canceled
+    "stop": error caused job to be paused
+
+Example:
+
+{ "event": "BLOCK_JOB_ERROR",
+    "data": { "device": "ide0-hd1",
+              "operation": "write",
+              "action": "stop" },
+    "timestamp": { "seconds": 1265044230, "microseconds": 450486 } }
+
 DEVICE_TRAY_MOVED
 -----------------
 
diff --git a/aio.c b/aio.c
index 0a9eb10c76..c738a4e15d 100644
--- a/aio.c
+++ b/aio.c
@@ -119,7 +119,7 @@ bool qemu_aio_wait(void)
         return true;
     }
 
-    walking_handlers = 1;
+    walking_handlers++;
 
     FD_ZERO(&rdfds);
     FD_ZERO(&wrfds);
@@ -147,7 +147,7 @@ bool qemu_aio_wait(void)
         }
     }
 
-    walking_handlers = 0;
+    walking_handlers--;
 
     /* No AIO operations?  Get us out of here */
     if (!busy) {
@@ -159,14 +159,14 @@ bool qemu_aio_wait(void)
 
     /* if we have any readable fds, dispatch event */
     if (ret > 0) {
-        walking_handlers = 1;
-
         /* we have to walk very carefully in case
          * qemu_aio_set_fd_handler is called while we're walking */
         node = QLIST_FIRST(&aio_handlers);
         while (node) {
             AioHandler *tmp;
 
+            walking_handlers++;
+
             if (!node->deleted &&
                 FD_ISSET(node->fd, &rdfds) &&
                 node->io_read) {
@@ -181,13 +181,13 @@ bool qemu_aio_wait(void)
             tmp = node;
             node = QLIST_NEXT(node, node);
 
-            if (tmp->deleted) {
+            walking_handlers--;
+
+            if (!walking_handlers && tmp->deleted) {
                 QLIST_REMOVE(tmp, node);
                 g_free(tmp);
             }
         }
-
-        walking_handlers = 0;
     }
 
     return true;
diff --git a/block-migration.c b/block-migration.c
index 7def8ab197..ed933017f9 100644
--- a/block-migration.c
+++ b/block-migration.c
@@ -519,6 +519,8 @@ static void blk_mig_cleanup(void)
     BlkMigDevState *bmds;
     BlkMigBlock *blk;
 
+    bdrv_drain_all();
+
     set_dirty_tracking(0);
 
     while ((bmds = QSIMPLEQ_FIRST(&block_mig_state.bmds_list)) != NULL) {
diff --git a/block.c b/block.c
index 751ebdc06b..c108a76952 100644
--- a/block.c
+++ b/block.c
@@ -26,8 +26,10 @@
 #include "trace.h"
 #include "monitor.h"
 #include "block_int.h"
+#include "blockjob.h"
 #include "module.h"
 #include "qjson.h"
+#include "sysemu.h"
 #include "qemu-coroutine.h"
 #include "qmp-commands.h"
 #include "qemu-timer.h"
@@ -1386,7 +1388,8 @@ void bdrv_set_dev_ops(BlockDriverState *bs, const BlockDevOps *ops,
 }
 
 void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
-                               BlockQMPEventAction action, int is_read)
+                               enum MonitorEvent ev,
+                               BlockErrorAction action, bool is_read)
 {
     QObject *data;
     const char *action_str;
@@ -1409,7 +1412,7 @@ void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
                               bdrv->device_name,
                               action_str,
                               is_read ? "read" : "write");
-    monitor_protocol_event(QEVENT_BLOCK_IO_ERROR, data);
+    monitor_protocol_event(ev, data);
 
     qobject_decref(data);
 }
@@ -1724,6 +1727,149 @@ int bdrv_change_backing_file(BlockDriverState *bs,
     return ret;
 }
 
+/*
+ * Finds the image layer in the chain that has 'bs' as its backing file.
+ *
+ * active is the current topmost image.
+ *
+ * Returns NULL if bs is not found in active's image chain,
+ * or if active == bs.
+ */
+BlockDriverState *bdrv_find_overlay(BlockDriverState *active,
+                                    BlockDriverState *bs)
+{
+    BlockDriverState *overlay = NULL;
+    BlockDriverState *intermediate;
+
+    assert(active != NULL);
+    assert(bs != NULL);
+
+    /* if bs is the same as active, then by definition it has no overlay
+     */
+    if (active == bs) {
+        return NULL;
+    }
+
+    intermediate = active;
+    while (intermediate->backing_hd) {
+        if (intermediate->backing_hd == bs) {
+            overlay = intermediate;
+            break;
+        }
+        intermediate = intermediate->backing_hd;
+    }
+
+    return overlay;
+}
+
+typedef struct BlkIntermediateStates {
+    BlockDriverState *bs;
+    QSIMPLEQ_ENTRY(BlkIntermediateStates) entry;
+} BlkIntermediateStates;
+
+
+/*
+ * Drops images above 'base' up to and including 'top', and sets the image
+ * above 'top' to have base as its backing file.
+ *
+ * Requires that the overlay to 'top' is opened r/w, so that the backing file
+ * information in 'bs' can be properly updated.
+ *
+ * E.g., this will convert the following chain:
+ * bottom <- base <- intermediate <- top <- active
+ *
+ * to
+ *
+ * bottom <- base <- active
+ *
+ * It is allowed for bottom==base, in which case it converts:
+ *
+ * base <- intermediate <- top <- active
+ *
+ * to
+ *
+ * base <- active
+ *
+ * Error conditions:
+ *  if active == top, that is considered an error
+ *
+ */
+int bdrv_drop_intermediate(BlockDriverState *active, BlockDriverState *top,
+                           BlockDriverState *base)
+{
+    BlockDriverState *intermediate;
+    BlockDriverState *base_bs = NULL;
+    BlockDriverState *new_top_bs = NULL;
+    BlkIntermediateStates *intermediate_state, *next;
+    int ret = -EIO;
+
+    QSIMPLEQ_HEAD(states_to_delete, BlkIntermediateStates) states_to_delete;
+    QSIMPLEQ_INIT(&states_to_delete);
+
+    if (!top->drv || !base->drv) {
+        goto exit;
+    }
+
+    new_top_bs = bdrv_find_overlay(active, top);
+
+    if (new_top_bs == NULL) {
+        /* we could not find the image above 'top', this is an error */
+        goto exit;
+    }
+
+    /* special case of new_top_bs->backing_hd already pointing to base - nothing
+     * to do, no intermediate images */
+    if (new_top_bs->backing_hd == base) {
+        ret = 0;
+        goto exit;
+    }
+
+    intermediate = top;
+
+    /* now we will go down through the list, and add each BDS we find
+     * into our deletion queue, until we hit the 'base'
+     */
+    while (intermediate) {
+        intermediate_state = g_malloc0(sizeof(BlkIntermediateStates));
+        intermediate_state->bs = intermediate;
+        QSIMPLEQ_INSERT_TAIL(&states_to_delete, intermediate_state, entry);
+
+        if (intermediate->backing_hd == base) {
+            base_bs = intermediate->backing_hd;
+            break;
+        }
+        intermediate = intermediate->backing_hd;
+    }
+    if (base_bs == NULL) {
+        /* something went wrong, we did not end at the base. safely
+         * unravel everything, and exit with error */
+        goto exit;
+    }
+
+    /* success - we can delete the intermediate states, and link top->base */
+    ret = bdrv_change_backing_file(new_top_bs, base_bs->filename,
+                                   base_bs->drv ? base_bs->drv->format_name : "");
+    if (ret) {
+        goto exit;
+    }
+    new_top_bs->backing_hd = base_bs;
+
+
+    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
+        /* so that bdrv_close() does not recursively close the chain */
+        intermediate_state->bs->backing_hd = NULL;
+        bdrv_delete(intermediate_state->bs);
+    }
+    ret = 0;
+
+exit:
+    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
+        g_free(intermediate_state);
+    }
+    return ret;
+}
+
+
 static int bdrv_check_byte_request(BlockDriverState *bs, int64_t offset,
                                    size_t size)
 {
@@ -2330,18 +2476,51 @@ void bdrv_set_io_limits(BlockDriverState *bs,
     bs->io_limits_enabled = bdrv_io_limits_enabled(bs);
 }
 
-void bdrv_set_on_error(BlockDriverState *bs, BlockErrorAction on_read_error,
-                       BlockErrorAction on_write_error)
+void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
+                       BlockdevOnError on_write_error)
 {
     bs->on_read_error = on_read_error;
     bs->on_write_error = on_write_error;
 }
 
-BlockErrorAction bdrv_get_on_error(BlockDriverState *bs, int is_read)
+BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read)
 {
     return is_read ? bs->on_read_error : bs->on_write_error;
 }
 
+BlockErrorAction bdrv_get_error_action(BlockDriverState *bs, bool is_read, int error)
+{
+    BlockdevOnError on_err = is_read ? bs->on_read_error : bs->on_write_error;
+
+    switch (on_err) {
+    case BLOCKDEV_ON_ERROR_ENOSPC:
+        return (error == ENOSPC) ? BDRV_ACTION_STOP : BDRV_ACTION_REPORT;
+    case BLOCKDEV_ON_ERROR_STOP:
+        return BDRV_ACTION_STOP;
+    case BLOCKDEV_ON_ERROR_REPORT:
+        return BDRV_ACTION_REPORT;
+    case BLOCKDEV_ON_ERROR_IGNORE:
+        return BDRV_ACTION_IGNORE;
+    default:
+        abort();
+    }
+}
+
+/* This is done by device models because, while the block layer knows
+ * about the error, it does not know whether an operation comes from
+ * the device or the block layer (from a job, for example).
+ */
+void bdrv_error_action(BlockDriverState *bs, BlockErrorAction action,
+                       bool is_read, int error)
+{
+    assert(error >= 0);
+    bdrv_emit_qmp_error_event(bs, QEVENT_BLOCK_IO_ERROR, action, is_read);
+    if (action == BDRV_ACTION_STOP) {
+        vm_stop(RUN_STATE_IO_ERROR);
+        bdrv_iostatus_set_err(bs, error);
+    }
+}
+
 int bdrv_is_read_only(BlockDriverState *bs)
 {
     return bs->read_only;
@@ -2974,6 +3153,22 @@ int bdrv_get_backing_file_depth(BlockDriverState *bs)
     return 1 + bdrv_get_backing_file_depth(bs->backing_hd);
 }
 
+BlockDriverState *bdrv_find_base(BlockDriverState *bs)
+{
+    BlockDriverState *curr_bs = NULL;
+
+    if (!bs) {
+        return NULL;
+    }
+
+    curr_bs = bs;
+
+    while (curr_bs->backing_hd) {
+        curr_bs = curr_bs->backing_hd;
+    }
+    return curr_bs;
+}
+
 #define NB_SUFFIXES 4
 
 char *get_human_readable_size(char *buf, int buf_size, int64_t size)
@@ -4049,9 +4244,9 @@ void bdrv_iostatus_enable(BlockDriverState *bs)
 bool bdrv_iostatus_is_enabled(const BlockDriverState *bs)
 {
     return (bs->iostatus_enabled &&
-           (bs->on_write_error == BLOCK_ERR_STOP_ENOSPC ||
-            bs->on_write_error == BLOCK_ERR_STOP_ANY    ||
-            bs->on_read_error == BLOCK_ERR_STOP_ANY));
+           (bs->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
+            bs->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
+            bs->on_read_error == BLOCKDEV_ON_ERROR_STOP));
 }
 
 void bdrv_iostatus_disable(BlockDriverState *bs)
@@ -4066,14 +4261,10 @@ void bdrv_iostatus_reset(BlockDriverState *bs)
     }
 }
 
-/* XXX: Today this is set by device models because it makes the implementation
-   quite simple. However, the block layer knows about the error, so it's
-   possible to implement this without device models being involved */
 void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
 {
-    if (bdrv_iostatus_is_enabled(bs) &&
-        bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
-        assert(error >= 0);
+    assert(bdrv_iostatus_is_enabled(bs));
+    if (bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
         bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                          BLOCK_DEVICE_IO_STATUS_FAILED;
     }
@@ -4247,130 +4438,3 @@ out:
 
     return ret;
 }
-
-void *block_job_create(const BlockJobType *job_type, BlockDriverState *bs,
-                       int64_t speed, BlockDriverCompletionFunc *cb,
-                       void *opaque, Error **errp)
-{
-    BlockJob *job;
-
-    if (bs->job || bdrv_in_use(bs)) {
-        error_set(errp, QERR_DEVICE_IN_USE, bdrv_get_device_name(bs));
-        return NULL;
-    }
-    bdrv_set_in_use(bs, 1);
-
-    job = g_malloc0(job_type->instance_size);
-    job->job_type      = job_type;
-    job->bs            = bs;
-    job->cb            = cb;
-    job->opaque        = opaque;
-    job->busy          = true;
-    bs->job = job;
-
-    /* Only set speed when necessary to avoid NotSupported error */
-    if (speed != 0) {
-        Error *local_err = NULL;
-
-        block_job_set_speed(job, speed, &local_err);
-        if (error_is_set(&local_err)) {
-            bs->job = NULL;
-            g_free(job);
-            bdrv_set_in_use(bs, 0);
-            error_propagate(errp, local_err);
-            return NULL;
-        }
-    }
-    return job;
-}
-
-void block_job_complete(BlockJob *job, int ret)
-{
-    BlockDriverState *bs = job->bs;
-
-    assert(bs->job == job);
-    job->cb(job->opaque, ret);
-    bs->job = NULL;
-    g_free(job);
-    bdrv_set_in_use(bs, 0);
-}
-
-void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp)
-{
-    Error *local_err = NULL;
-
-    if (!job->job_type->set_speed) {
-        error_set(errp, QERR_NOT_SUPPORTED);
-        return;
-    }
-    job->job_type->set_speed(job, speed, &local_err);
-    if (error_is_set(&local_err)) {
-        error_propagate(errp, local_err);
-        return;
-    }
-
-    job->speed = speed;
-}
-
-void block_job_cancel(BlockJob *job)
-{
-    job->cancelled = true;
-    if (job->co && !job->busy) {
-        qemu_coroutine_enter(job->co, NULL);
-    }
-}
-
-bool block_job_is_cancelled(BlockJob *job)
-{
-    return job->cancelled;
-}
-
-struct BlockCancelData {
-    BlockJob *job;
-    BlockDriverCompletionFunc *cb;
-    void *opaque;
-    bool cancelled;
-    int ret;
-};
-
-static void block_job_cancel_cb(void *opaque, int ret)
-{
-    struct BlockCancelData *data = opaque;
-
-    data->cancelled = block_job_is_cancelled(data->job);
-    data->ret = ret;
-    data->cb(data->opaque, ret);
-}
-
-int block_job_cancel_sync(BlockJob *job)
-{
-    struct BlockCancelData data;
-    BlockDriverState *bs = job->bs;
-
-    assert(bs->job == job);
-
-    /* Set up our own callback to store the result and chain to
-     * the original callback.
-     */
-    data.job = job;
-    data.cb = job->cb;
-    data.opaque = job->opaque;
-    data.ret = -EINPROGRESS;
-    job->cb = block_job_cancel_cb;
-    job->opaque = &data;
-    block_job_cancel(job);
-    while (data.ret == -EINPROGRESS) {
-        qemu_aio_wait();
-    }
-    return (data.cancelled && data.ret == 0) ? -ECANCELED : data.ret;
-}
-
-void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns)
-{
-    /* Check cancellation *before* setting busy = false, too!  */
-    if (!block_job_is_cancelled(job)) {
-        job->busy = false;
-        co_sleep_ns(clock, ns);
-        job->busy = true;
-    }
-}
diff --git a/block.h b/block.h
index b1095d8599..e2d89d7bc1 100644
--- a/block.h
+++ b/block.h
@@ -6,9 +6,11 @@
 #include "qemu-option.h"
 #include "qemu-coroutine.h"
 #include "qobject.h"
+#include "qapi-types.h"
 
 /* block.c */
 typedef struct BlockDriver BlockDriver;
+typedef struct BlockJob BlockJob;
 
 typedef struct BlockDriverInfo {
     /* in bytes, 0 if irrelevant */
@@ -89,13 +91,8 @@ typedef struct BlockDevOps {
 #define BDRV_SECTOR_MASK   ~(BDRV_SECTOR_SIZE - 1)
 
 typedef enum {
-    BLOCK_ERR_REPORT, BLOCK_ERR_IGNORE, BLOCK_ERR_STOP_ENOSPC,
-    BLOCK_ERR_STOP_ANY
-} BlockErrorAction;
-
-typedef enum {
     BDRV_ACTION_REPORT, BDRV_ACTION_IGNORE, BDRV_ACTION_STOP
-} BlockQMPEventAction;
+} BlockErrorAction;
 
 typedef QSIMPLEQ_HEAD(BlockReopenQueue, BlockReopenQueueEntry) BlockReopenQueue;
 
@@ -111,8 +108,6 @@ void bdrv_iostatus_reset(BlockDriverState *bs);
 void bdrv_iostatus_disable(BlockDriverState *bs);
 bool bdrv_iostatus_is_enabled(const BlockDriverState *bs);
 void bdrv_iostatus_set_err(BlockDriverState *bs, int error);
-void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
-                               BlockQMPEventAction action, int is_read);
 void bdrv_info_print(Monitor *mon, const QObject *data);
 void bdrv_info(Monitor *mon, QObject **ret_data);
 void bdrv_stats_print(Monitor *mon, const QObject *data);
@@ -203,6 +198,11 @@ int bdrv_commit_all(void);
 int bdrv_change_backing_file(BlockDriverState *bs,
     const char *backing_file, const char *backing_fmt);
 void bdrv_register(BlockDriver *bdrv);
+int bdrv_drop_intermediate(BlockDriverState *active, BlockDriverState *top,
+                           BlockDriverState *base);
+BlockDriverState *bdrv_find_overlay(BlockDriverState *active,
+                                    BlockDriverState *bs);
+BlockDriverState *bdrv_find_base(BlockDriverState *bs);
 
 
 typedef struct BdrvCheckResult {
@@ -277,9 +277,12 @@ int bdrv_has_zero_init(BlockDriverState *bs);
 int bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num, int nb_sectors,
                       int *pnum);
 
-void bdrv_set_on_error(BlockDriverState *bs, BlockErrorAction on_read_error,
-                       BlockErrorAction on_write_error);
-BlockErrorAction bdrv_get_on_error(BlockDriverState *bs, int is_read);
+void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
+                       BlockdevOnError on_write_error);
+BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read);
+BlockErrorAction bdrv_get_error_action(BlockDriverState *bs, bool is_read, int error);
+void bdrv_error_action(BlockDriverState *bs, BlockErrorAction action,
+                       bool is_read, int error);
 int bdrv_is_read_only(BlockDriverState *bs);
 int bdrv_is_sg(BlockDriverState *bs);
 int bdrv_enable_write_cache(BlockDriverState *bs);
diff --git a/block/Makefile.objs b/block/Makefile.objs
index b5754d39bf..554f429d05 100644
--- a/block/Makefile.objs
+++ b/block/Makefile.objs
@@ -3,9 +3,12 @@ block-obj-y += qcow2.o qcow2-refcount.o qcow2-cluster.o qcow2-snapshot.o qcow2-c
 block-obj-y += qed.o qed-gencb.o qed-l2-cache.o qed-table.o qed-cluster.o
 block-obj-y += qed-check.o
 block-obj-y += parallels.o nbd.o blkdebug.o sheepdog.o blkverify.o
-block-obj-y += stream.o
 block-obj-$(CONFIG_WIN32) += raw-win32.o
 block-obj-$(CONFIG_POSIX) += raw-posix.o
 block-obj-$(CONFIG_LIBISCSI) += iscsi.o
 block-obj-$(CONFIG_CURL) += curl.o
 block-obj-$(CONFIG_RBD) += rbd.o
+block-obj-$(CONFIG_GLUSTERFS) += gluster.o
+
+common-obj-y += stream.o
+common-obj-y += commit.o
diff --git a/block/blkdebug.c b/block/blkdebug.c
index 59dcea0650..1206d5256b 100644
--- a/block/blkdebug.c
+++ b/block/blkdebug.c
@@ -28,6 +28,7 @@
 
 typedef struct BDRVBlkdebugState {
     int state;
+    int new_state;
     QLIST_HEAD(, BlkdebugRule) rules[BLKDBG_EVENT_MAX];
     QSIMPLEQ_HEAD(, BlkdebugRule) active_rules;
 } BDRVBlkdebugState;
@@ -403,12 +404,12 @@ static void blkdebug_close(BlockDriverState *bs)
 }
 
 static bool process_rule(BlockDriverState *bs, struct BlkdebugRule *rule,
-    int old_state, bool injected)
+    bool injected)
 {
     BDRVBlkdebugState *s = bs->opaque;
 
     /* Only process rules for the current state */
-    if (rule->state && rule->state != old_state) {
+    if (rule->state && rule->state != s->state) {
         return injected;
     }
 
@@ -423,7 +424,7 @@ static bool process_rule(BlockDriverState *bs, struct BlkdebugRule *rule,
         break;
 
     case ACTION_SET_STATE:
-        s->state = rule->options.set_state.new_state;
+        s->new_state = rule->options.set_state.new_state;
         break;
     }
     return injected;
@@ -433,15 +434,16 @@ static void blkdebug_debug_event(BlockDriverState *bs, BlkDebugEvent event)
 {
     BDRVBlkdebugState *s = bs->opaque;
     struct BlkdebugRule *rule;
-    int old_state = s->state;
     bool injected;
 
     assert((int)event >= 0 && event < BLKDBG_EVENT_MAX);
 
     injected = false;
+    s->new_state = s->state;
     QLIST_FOREACH(rule, &s->rules[event], next) {
-        injected = process_rule(bs, rule, old_state, injected);
+        injected = process_rule(bs, rule, injected);
     }
+    s->state = s->new_state;
 }
 
 static int64_t blkdebug_getlength(BlockDriverState *bs)
diff --git a/block/commit.c b/block/commit.c
new file mode 100644
index 0000000000..733c91403c
--- /dev/null
+++ b/block/commit.c
@@ -0,0 +1,268 @@
+/*
+ * Live block commit
+ *
+ * Copyright Red Hat, Inc. 2012
+ *
+ * Authors:
+ *  Jeff Cody   <jcody@redhat.com>
+ *  Based on stream.c by Stefan Hajnoczi
+ *
+ * This work is licensed under the terms of the GNU LGPL, version 2 or later.
+ * See the COPYING.LIB file in the top-level directory.
+ *
+ */
+
+#include "trace.h"
+#include "block_int.h"
+#include "blockjob.h"
+#include "qemu/ratelimit.h"
+
+enum {
+    /*
+     * Size of data buffer for populating the image file.  This should be large
+     * enough to process multiple clusters in a single call, so that populating
+     * contiguous regions of the image is efficient.
+     */
+    COMMIT_BUFFER_SIZE = 512 * 1024, /* in bytes */
+};
+
+#define SLICE_TIME 100000000ULL /* ns */
+
+typedef struct CommitBlockJob {
+    BlockJob common;
+    RateLimit limit;
+    BlockDriverState *active;
+    BlockDriverState *top;
+    BlockDriverState *base;
+    BlockdevOnError on_error;
+    int base_flags;
+    int orig_overlay_flags;
+} CommitBlockJob;
+
+static int coroutine_fn commit_populate(BlockDriverState *bs,
+                                        BlockDriverState *base,
+                                        int64_t sector_num, int nb_sectors,
+                                        void *buf)
+{
+    int ret = 0;
+
+    ret = bdrv_read(bs, sector_num, buf, nb_sectors);
+    if (ret) {
+        return ret;
+    }
+
+    ret = bdrv_write(base, sector_num, buf, nb_sectors);
+    if (ret) {
+        return ret;
+    }
+
+    return 0;
+}
+
+static void coroutine_fn commit_run(void *opaque)
+{
+    CommitBlockJob *s = opaque;
+    BlockDriverState *active = s->active;
+    BlockDriverState *top = s->top;
+    BlockDriverState *base = s->base;
+    BlockDriverState *overlay_bs = NULL;
+    int64_t sector_num, end;
+    int ret = 0;
+    int n = 0;
+    void *buf;
+    int bytes_written = 0;
+    int64_t base_len;
+
+    ret = s->common.len = bdrv_getlength(top);
+
+
+    if (s->common.len < 0) {
+        goto exit_restore_reopen;
+    }
+
+    ret = base_len = bdrv_getlength(base);
+    if (base_len < 0) {
+        goto exit_restore_reopen;
+    }
+
+    if (base_len < s->common.len) {
+        ret = bdrv_truncate(base, s->common.len);
+        if (ret) {
+            goto exit_restore_reopen;
+        }
+    }
+
+    overlay_bs = bdrv_find_overlay(active, top);
+
+    end = s->common.len >> BDRV_SECTOR_BITS;
+    buf = qemu_blockalign(top, COMMIT_BUFFER_SIZE);
+
+    for (sector_num = 0; sector_num < end; sector_num += n) {
+        uint64_t delay_ns = 0;
+        bool copy;
+
+wait:
+        /* Note that even when no rate limit is applied we need to yield
+         * with no pending I/O here so that qemu_aio_flush() returns.
+         */
+        block_job_sleep_ns(&s->common, rt_clock, delay_ns);
+        if (block_job_is_cancelled(&s->common)) {
+            break;
+        }
+        /* Copy if allocated above the base */
+        ret = bdrv_co_is_allocated_above(top, base, sector_num,
+                                         COMMIT_BUFFER_SIZE / BDRV_SECTOR_SIZE,
+                                         &n);
+        copy = (ret == 1);
+        trace_commit_one_iteration(s, sector_num, n, ret);
+        if (copy) {
+            if (s->common.speed) {
+                delay_ns = ratelimit_calculate_delay(&s->limit, n);
+                if (delay_ns > 0) {
+                    goto wait;
+                }
+            }
+            ret = commit_populate(top, base, sector_num, n, buf);
+            bytes_written += n * BDRV_SECTOR_SIZE;
+        }
+        if (ret < 0) {
+            if (s->on_error == BLOCKDEV_ON_ERROR_STOP ||
+                s->on_error == BLOCKDEV_ON_ERROR_REPORT||
+                (s->on_error == BLOCKDEV_ON_ERROR_ENOSPC && ret == -ENOSPC)) {
+                goto exit_free_buf;
+            } else {
+                n = 0;
+                continue;
+            }
+        }
+        /* Publish progress */
+        s->common.offset += n * BDRV_SECTOR_SIZE;
+    }
+
+    ret = 0;
+
+    if (!block_job_is_cancelled(&s->common) && sector_num == end) {
+        /* success */
+        ret = bdrv_drop_intermediate(active, top, base);
+    }
+
+exit_free_buf:
+    qemu_vfree(buf);
+
+exit_restore_reopen:
+    /* restore base open flags here if appropriate (e.g., change the base back
+     * to r/o). These reopens do not need to be atomic, since we won't abort
+     * even on failure here */
+    if (s->base_flags != bdrv_get_flags(base)) {
+        bdrv_reopen(base, s->base_flags, NULL);
+    }
+    if (s->orig_overlay_flags != bdrv_get_flags(overlay_bs)) {
+        bdrv_reopen(overlay_bs, s->orig_overlay_flags, NULL);
+    }
+
+    block_job_complete(&s->common, ret);
+}
+
+static void commit_set_speed(BlockJob *job, int64_t speed, Error **errp)
+{
+    CommitBlockJob *s = container_of(job, CommitBlockJob, common);
+
+    if (speed < 0) {
+        error_set(errp, QERR_INVALID_PARAMETER, "speed");
+        return;
+    }
+    ratelimit_set_speed(&s->limit, speed / BDRV_SECTOR_SIZE, SLICE_TIME);
+}
+
+static BlockJobType commit_job_type = {
+    .instance_size = sizeof(CommitBlockJob),
+    .job_type      = "commit",
+    .set_speed     = commit_set_speed,
+};
+
+void commit_start(BlockDriverState *bs, BlockDriverState *base,
+                  BlockDriverState *top, int64_t speed,
+                  BlockdevOnError on_error, BlockDriverCompletionFunc *cb,
+                  void *opaque, Error **errp)
+{
+    CommitBlockJob *s;
+    BlockReopenQueue *reopen_queue = NULL;
+    int orig_overlay_flags;
+    int orig_base_flags;
+    BlockDriverState *overlay_bs;
+    Error *local_err = NULL;
+
+    if ((on_error == BLOCKDEV_ON_ERROR_STOP ||
+         on_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
+        !bdrv_iostatus_is_enabled(bs)) {
+        error_set(errp, QERR_INVALID_PARAMETER_COMBINATION);
+        return;
+    }
+
+    /* Once we support top == active layer, remove this check */
+    if (top == bs) {
+        error_setg(errp,
+                   "Top image as the active layer is currently unsupported");
+        return;
+    }
+
+    if (top == base) {
+        error_setg(errp, "Invalid files for merge: top and base are the same");
+        return;
+    }
+
+    /* top and base may be valid, but let's make sure that base is reachable
+     * from top */
+    if (bdrv_find_backing_image(top, base->filename) != base) {
+        error_setg(errp,
+                   "Base (%s) is not reachable from top (%s)",
+                   base->filename, top->filename);
+        return;
+    }
+
+    overlay_bs = bdrv_find_overlay(bs, top);
+
+    if (overlay_bs == NULL) {
+        error_setg(errp, "Could not find overlay image for %s:", top->filename);
+        return;
+    }
+
+    orig_base_flags    = bdrv_get_flags(base);
+    orig_overlay_flags = bdrv_get_flags(overlay_bs);
+
+    /* convert base & overlay_bs to r/w, if necessary */
+    if (!(orig_base_flags & BDRV_O_RDWR)) {
+        reopen_queue = bdrv_reopen_queue(reopen_queue, base,
+                                         orig_base_flags | BDRV_O_RDWR);
+    }
+    if (!(orig_overlay_flags & BDRV_O_RDWR)) {
+        reopen_queue = bdrv_reopen_queue(reopen_queue, overlay_bs,
+                                         orig_overlay_flags | BDRV_O_RDWR);
+    }
+    if (reopen_queue) {
+        bdrv_reopen_multiple(reopen_queue, &local_err);
+        if (local_err != NULL) {
+            error_propagate(errp, local_err);
+            return;
+        }
+    }
+
+
+    s = block_job_create(&commit_job_type, bs, speed, cb, opaque, errp);
+    if (!s) {
+        return;
+    }
+
+    s->base   = base;
+    s->top    = top;
+    s->active = bs;
+
+    s->base_flags          = orig_base_flags;
+    s->orig_overlay_flags  = orig_overlay_flags;
+
+    s->on_error = on_error;
+    s->common.co = qemu_coroutine_create(commit_run);
+
+    trace_commit_start(bs, base, top, s, s->common.co, opaque);
+    qemu_coroutine_enter(s->common.co, s);
+}
diff --git a/block/gluster.c b/block/gluster.c
new file mode 100644
index 0000000000..3588d7377f
--- /dev/null
+++ b/block/gluster.c
@@ -0,0 +1,624 @@
+/*
+ * GlusterFS backend for QEMU
+ *
+ * Copyright (C) 2012 Bharata B Rao <bharata@linux.vnet.ibm.com>
+ *
+ * Pipe handling mechanism in AIO implementation is derived from
+ * block/rbd.c. Hence,
+ *
+ * Copyright (C) 2010-2011 Christian Brunner <chb@muc.de>,
+ *                         Josh Durgin <josh.durgin@dreamhost.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ * Contributions after 2012-01-13 are licensed under the terms of the
+ * GNU GPL, version 2 or (at your option) any later version.
+ */
+#include <glusterfs/api/glfs.h>
+#include "block_int.h"
+#include "qemu_socket.h"
+#include "uri.h"
+
+typedef struct GlusterAIOCB {
+    BlockDriverAIOCB common;
+    int64_t size;
+    int ret;
+    bool *finished;
+    QEMUBH *bh;
+} GlusterAIOCB;
+
+typedef struct BDRVGlusterState {
+    struct glfs *glfs;
+    int fds[2];
+    struct glfs_fd *fd;
+    int qemu_aio_count;
+    int event_reader_pos;
+    GlusterAIOCB *event_acb;
+} BDRVGlusterState;
+
+#define GLUSTER_FD_READ  0
+#define GLUSTER_FD_WRITE 1
+
+typedef struct GlusterConf {
+    char *server;
+    int port;
+    char *volname;
+    char *image;
+    char *transport;
+} GlusterConf;
+
+static void qemu_gluster_gconf_free(GlusterConf *gconf)
+{
+    g_free(gconf->server);
+    g_free(gconf->volname);
+    g_free(gconf->image);
+    g_free(gconf->transport);
+    g_free(gconf);
+}
+
+static int parse_volume_options(GlusterConf *gconf, char *path)
+{
+    char *p, *q;
+
+    if (!path) {
+        return -EINVAL;
+    }
+
+    /* volume */
+    p = q = path + strspn(path, "/");
+    p += strcspn(p, "/");
+    if (*p == '\0') {
+        return -EINVAL;
+    }
+    gconf->volname = g_strndup(q, p - q);
+
+    /* image */
+    p += strspn(p, "/");
+    if (*p == '\0') {
+        return -EINVAL;
+    }
+    gconf->image = g_strdup(p);
+    return 0;
+}
+
+/*
+ * file=gluster[+transport]://[server[:port]]/volname/image[?socket=...]
+ *
+ * 'gluster' is the protocol.
+ *
+ * 'transport' specifies the transport type used to connect to gluster
+ * management daemon (glusterd). Valid transport types are
+ * tcp, unix and rdma. If a transport type isn't specified, then tcp
+ * type is assumed.
+ *
+ * 'server' specifies the server where the volume file specification for
+ * the given volume resides. This can be either hostname, ipv4 address
+ * or ipv6 address. ipv6 address needs to be within square brackets [ ].
+ * If transport type is 'unix', then 'server' field should not be specifed.
+ * The 'socket' field needs to be populated with the path to unix domain
+ * socket.
+ *
+ * 'port' is the port number on which glusterd is listening. This is optional
+ * and if not specified, QEMU will send 0 which will make gluster to use the
+ * default port. If the transport type is unix, then 'port' should not be
+ * specified.
+ *
+ * 'volname' is the name of the gluster volume which contains the VM image.
+ *
+ * 'image' is the path to the actual VM image that resides on gluster volume.
+ *
+ * Examples:
+ *
+ * file=gluster://1.2.3.4/testvol/a.img
+ * file=gluster+tcp://1.2.3.4/testvol/a.img
+ * file=gluster+tcp://1.2.3.4:24007/testvol/dir/a.img
+ * file=gluster+tcp://[1:2:3:4:5:6:7:8]/testvol/dir/a.img
+ * file=gluster+tcp://[1:2:3:4:5:6:7:8]:24007/testvol/dir/a.img
+ * file=gluster+tcp://server.domain.com:24007/testvol/dir/a.img
+ * file=gluster+unix:///testvol/dir/a.img?socket=/tmp/glusterd.socket
+ * file=gluster+rdma://1.2.3.4:24007/testvol/a.img
+ */
+static int qemu_gluster_parseuri(GlusterConf *gconf, const char *filename)
+{
+    URI *uri;
+    QueryParams *qp = NULL;
+    bool is_unix = false;
+    int ret = 0;
+
+    uri = uri_parse(filename);
+    if (!uri) {
+        return -EINVAL;
+    }
+
+    /* transport */
+    if (!strcmp(uri->scheme, "gluster")) {
+        gconf->transport = g_strdup("tcp");
+    } else if (!strcmp(uri->scheme, "gluster+tcp")) {
+        gconf->transport = g_strdup("tcp");
+    } else if (!strcmp(uri->scheme, "gluster+unix")) {
+        gconf->transport = g_strdup("unix");
+        is_unix = true;
+    } else if (!strcmp(uri->scheme, "gluster+rdma")) {
+        gconf->transport = g_strdup("rdma");
+    } else {
+        ret = -EINVAL;
+        goto out;
+    }
+
+    ret = parse_volume_options(gconf, uri->path);
+    if (ret < 0) {
+        goto out;
+    }
+
+    qp = query_params_parse(uri->query);
+    if (qp->n > 1 || (is_unix && !qp->n) || (!is_unix && qp->n)) {
+        ret = -EINVAL;
+        goto out;
+    }
+
+    if (is_unix) {
+        if (uri->server || uri->port) {
+            ret = -EINVAL;
+            goto out;
+        }
+        if (strcmp(qp->p[0].name, "socket")) {
+            ret = -EINVAL;
+            goto out;
+        }
+        gconf->server = g_strdup(qp->p[0].value);
+    } else {
+        gconf->server = g_strdup(uri->server);
+        gconf->port = uri->port;
+    }
+
+out:
+    if (qp) {
+        query_params_free(qp);
+    }
+    uri_free(uri);
+    return ret;
+}
+
+static struct glfs *qemu_gluster_init(GlusterConf *gconf, const char *filename)
+{
+    struct glfs *glfs = NULL;
+    int ret;
+    int old_errno;
+
+    ret = qemu_gluster_parseuri(gconf, filename);
+    if (ret < 0) {
+        error_report("Usage: file=gluster[+transport]://[server[:port]]/"
+            "volname/image[?socket=...]");
+        errno = -ret;
+        goto out;
+    }
+
+    glfs = glfs_new(gconf->volname);
+    if (!glfs) {
+        goto out;
+    }
+
+    ret = glfs_set_volfile_server(glfs, gconf->transport, gconf->server,
+            gconf->port);
+    if (ret < 0) {
+        goto out;
+    }
+
+    /*
+     * TODO: Use GF_LOG_ERROR instead of hard code value of 4 here when
+     * GlusterFS makes GF_LOG_* macros available to libgfapi users.
+     */
+    ret = glfs_set_logging(glfs, "-", 4);
+    if (ret < 0) {
+        goto out;
+    }
+
+    ret = glfs_init(glfs);
+    if (ret) {
+        error_report("Gluster connection failed for server=%s port=%d "
+             "volume=%s image=%s transport=%s\n", gconf->server, gconf->port,
+             gconf->volname, gconf->image, gconf->transport);
+        goto out;
+    }
+    return glfs;
+
+out:
+    if (glfs) {
+        old_errno = errno;
+        glfs_fini(glfs);
+        errno = old_errno;
+    }
+    return NULL;
+}
+
+static void qemu_gluster_complete_aio(GlusterAIOCB *acb, BDRVGlusterState *s)
+{
+    int ret;
+    bool *finished = acb->finished;
+    BlockDriverCompletionFunc *cb = acb->common.cb;
+    void *opaque = acb->common.opaque;
+
+    if (!acb->ret || acb->ret == acb->size) {
+        ret = 0; /* Success */
+    } else if (acb->ret < 0) {
+        ret = acb->ret; /* Read/Write failed */
+    } else {
+        ret = -EIO; /* Partial read/write - fail it */
+    }
+
+    s->qemu_aio_count--;
+    qemu_aio_release(acb);
+    cb(opaque, ret);
+    if (finished) {
+        *finished = true;
+    }
+}
+
+static void qemu_gluster_aio_event_reader(void *opaque)
+{
+    BDRVGlusterState *s = opaque;
+    ssize_t ret;
+
+    do {
+        char *p = (char *)&s->event_acb;
+
+        ret = read(s->fds[GLUSTER_FD_READ], p + s->event_reader_pos,
+                   sizeof(s->event_acb) - s->event_reader_pos);
+        if (ret > 0) {
+            s->event_reader_pos += ret;
+            if (s->event_reader_pos == sizeof(s->event_acb)) {
+                s->event_reader_pos = 0;
+                qemu_gluster_complete_aio(s->event_acb, s);
+            }
+        }
+    } while (ret < 0 && errno == EINTR);
+}
+
+static int qemu_gluster_aio_flush_cb(void *opaque)
+{
+    BDRVGlusterState *s = opaque;
+
+    return (s->qemu_aio_count > 0);
+}
+
+static int qemu_gluster_open(BlockDriverState *bs, const char *filename,
+    int bdrv_flags)
+{
+    BDRVGlusterState *s = bs->opaque;
+    int open_flags = O_BINARY;
+    int ret = 0;
+    GlusterConf *gconf = g_malloc0(sizeof(GlusterConf));
+
+    s->glfs = qemu_gluster_init(gconf, filename);
+    if (!s->glfs) {
+        ret = -errno;
+        goto out;
+    }
+
+    if (bdrv_flags & BDRV_O_RDWR) {
+        open_flags |= O_RDWR;
+    } else {
+        open_flags |= O_RDONLY;
+    }
+
+    if ((bdrv_flags & BDRV_O_NOCACHE)) {
+        open_flags |= O_DIRECT;
+    }
+
+    s->fd = glfs_open(s->glfs, gconf->image, open_flags);
+    if (!s->fd) {
+        ret = -errno;
+        goto out;
+    }
+
+    ret = qemu_pipe(s->fds);
+    if (ret < 0) {
+        ret = -errno;
+        goto out;
+    }
+    fcntl(s->fds[GLUSTER_FD_READ], F_SETFL, O_NONBLOCK);
+    qemu_aio_set_fd_handler(s->fds[GLUSTER_FD_READ],
+        qemu_gluster_aio_event_reader, NULL, qemu_gluster_aio_flush_cb, s);
+
+out:
+    qemu_gluster_gconf_free(gconf);
+    if (!ret) {
+        return ret;
+    }
+    if (s->fd) {
+        glfs_close(s->fd);
+    }
+    if (s->glfs) {
+        glfs_fini(s->glfs);
+    }
+    return ret;
+}
+
+static int qemu_gluster_create(const char *filename,
+        QEMUOptionParameter *options)
+{
+    struct glfs *glfs;
+    struct glfs_fd *fd;
+    int ret = 0;
+    int64_t total_size = 0;
+    GlusterConf *gconf = g_malloc0(sizeof(GlusterConf));
+
+    glfs = qemu_gluster_init(gconf, filename);
+    if (!glfs) {
+        ret = -errno;
+        goto out;
+    }
+
+    while (options && options->name) {
+        if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
+            total_size = options->value.n / BDRV_SECTOR_SIZE;
+        }
+        options++;
+    }
+
+    fd = glfs_creat(glfs, gconf->image,
+        O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, S_IRUSR | S_IWUSR);
+    if (!fd) {
+        ret = -errno;
+    } else {
+        if (glfs_ftruncate(fd, total_size * BDRV_SECTOR_SIZE) != 0) {
+            ret = -errno;
+        }
+        if (glfs_close(fd) != 0) {
+            ret = -errno;
+        }
+    }
+out:
+    qemu_gluster_gconf_free(gconf);
+    if (glfs) {
+        glfs_fini(glfs);
+    }
+    return ret;
+}
+
+static void qemu_gluster_aio_cancel(BlockDriverAIOCB *blockacb)
+{
+    GlusterAIOCB *acb = (GlusterAIOCB *)blockacb;
+    bool finished = false;
+
+    acb->finished = &finished;
+    while (!finished) {
+        qemu_aio_wait();
+    }
+}
+
+static AIOPool gluster_aio_pool = {
+    .aiocb_size = sizeof(GlusterAIOCB),
+    .cancel = qemu_gluster_aio_cancel,
+};
+
+static void gluster_finish_aiocb(struct glfs_fd *fd, ssize_t ret, void *arg)
+{
+    GlusterAIOCB *acb = (GlusterAIOCB *)arg;
+    BlockDriverState *bs = acb->common.bs;
+    BDRVGlusterState *s = bs->opaque;
+    int retval;
+
+    acb->ret = ret;
+    retval = qemu_write_full(s->fds[GLUSTER_FD_WRITE], &acb, sizeof(acb));
+    if (retval != sizeof(acb)) {
+        /*
+         * Gluster AIO callback thread failed to notify the waiting
+         * QEMU thread about IO completion.
+         *
+         * Complete this IO request and make the disk inaccessible for
+         * subsequent reads and writes.
+         */
+        error_report("Gluster failed to notify QEMU about IO completion");
+
+        qemu_mutex_lock_iothread(); /* We are in gluster thread context */
+        acb->common.cb(acb->common.opaque, -EIO);
+        qemu_aio_release(acb);
+        s->qemu_aio_count--;
+        close(s->fds[GLUSTER_FD_READ]);
+        close(s->fds[GLUSTER_FD_WRITE]);
+        qemu_aio_set_fd_handler(s->fds[GLUSTER_FD_READ], NULL, NULL, NULL,
+            NULL);
+        bs->drv = NULL; /* Make the disk inaccessible */
+        qemu_mutex_unlock_iothread();
+    }
+}
+
+static BlockDriverAIOCB *qemu_gluster_aio_rw(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque, int write)
+{
+    int ret;
+    GlusterAIOCB *acb;
+    BDRVGlusterState *s = bs->opaque;
+    size_t size;
+    off_t offset;
+
+    offset = sector_num * BDRV_SECTOR_SIZE;
+    size = nb_sectors * BDRV_SECTOR_SIZE;
+    s->qemu_aio_count++;
+
+    acb = qemu_aio_get(&gluster_aio_pool, bs, cb, opaque);
+    acb->size = size;
+    acb->ret = 0;
+    acb->finished = NULL;
+
+    if (write) {
+        ret = glfs_pwritev_async(s->fd, qiov->iov, qiov->niov, offset, 0,
+            &gluster_finish_aiocb, acb);
+    } else {
+        ret = glfs_preadv_async(s->fd, qiov->iov, qiov->niov, offset, 0,
+            &gluster_finish_aiocb, acb);
+    }
+
+    if (ret < 0) {
+        goto out;
+    }
+    return &acb->common;
+
+out:
+    s->qemu_aio_count--;
+    qemu_aio_release(acb);
+    return NULL;
+}
+
+static BlockDriverAIOCB *qemu_gluster_aio_readv(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    return qemu_gluster_aio_rw(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
+}
+
+static BlockDriverAIOCB *qemu_gluster_aio_writev(BlockDriverState *bs,
+        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    return qemu_gluster_aio_rw(bs, sector_num, qiov, nb_sectors, cb, opaque, 1);
+}
+
+static BlockDriverAIOCB *qemu_gluster_aio_flush(BlockDriverState *bs,
+        BlockDriverCompletionFunc *cb, void *opaque)
+{
+    int ret;
+    GlusterAIOCB *acb;
+    BDRVGlusterState *s = bs->opaque;
+
+    acb = qemu_aio_get(&gluster_aio_pool, bs, cb, opaque);
+    acb->size = 0;
+    acb->ret = 0;
+    acb->finished = NULL;
+    s->qemu_aio_count++;
+
+    ret = glfs_fsync_async(s->fd, &gluster_finish_aiocb, acb);
+    if (ret < 0) {
+        goto out;
+    }
+    return &acb->common;
+
+out:
+    s->qemu_aio_count--;
+    qemu_aio_release(acb);
+    return NULL;
+}
+
+static int64_t qemu_gluster_getlength(BlockDriverState *bs)
+{
+    BDRVGlusterState *s = bs->opaque;
+    int64_t ret;
+
+    ret = glfs_lseek(s->fd, 0, SEEK_END);
+    if (ret < 0) {
+        return -errno;
+    } else {
+        return ret;
+    }
+}
+
+static int64_t qemu_gluster_allocated_file_size(BlockDriverState *bs)
+{
+    BDRVGlusterState *s = bs->opaque;
+    struct stat st;
+    int ret;
+
+    ret = glfs_fstat(s->fd, &st);
+    if (ret < 0) {
+        return -errno;
+    } else {
+        return st.st_blocks * 512;
+    }
+}
+
+static void qemu_gluster_close(BlockDriverState *bs)
+{
+    BDRVGlusterState *s = bs->opaque;
+
+    close(s->fds[GLUSTER_FD_READ]);
+    close(s->fds[GLUSTER_FD_WRITE]);
+    qemu_aio_set_fd_handler(s->fds[GLUSTER_FD_READ], NULL, NULL, NULL, NULL);
+
+    if (s->fd) {
+        glfs_close(s->fd);
+        s->fd = NULL;
+    }
+    glfs_fini(s->glfs);
+}
+
+static QEMUOptionParameter qemu_gluster_create_options[] = {
+    {
+        .name = BLOCK_OPT_SIZE,
+        .type = OPT_SIZE,
+        .help = "Virtual disk size"
+    },
+    { NULL }
+};
+
+static BlockDriver bdrv_gluster = {
+    .format_name                  = "gluster",
+    .protocol_name                = "gluster",
+    .instance_size                = sizeof(BDRVGlusterState),
+    .bdrv_file_open               = qemu_gluster_open,
+    .bdrv_close                   = qemu_gluster_close,
+    .bdrv_create                  = qemu_gluster_create,
+    .bdrv_getlength               = qemu_gluster_getlength,
+    .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size,
+    .bdrv_aio_readv               = qemu_gluster_aio_readv,
+    .bdrv_aio_writev              = qemu_gluster_aio_writev,
+    .bdrv_aio_flush               = qemu_gluster_aio_flush,
+    .create_options               = qemu_gluster_create_options,
+};
+
+static BlockDriver bdrv_gluster_tcp = {
+    .format_name                  = "gluster",
+    .protocol_name                = "gluster+tcp",
+    .instance_size                = sizeof(BDRVGlusterState),
+    .bdrv_file_open               = qemu_gluster_open,
+    .bdrv_close                   = qemu_gluster_close,
+    .bdrv_create                  = qemu_gluster_create,
+    .bdrv_getlength               = qemu_gluster_getlength,
+    .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size,
+    .bdrv_aio_readv               = qemu_gluster_aio_readv,
+    .bdrv_aio_writev              = qemu_gluster_aio_writev,
+    .bdrv_aio_flush               = qemu_gluster_aio_flush,
+    .create_options               = qemu_gluster_create_options,
+};
+
+static BlockDriver bdrv_gluster_unix = {
+    .format_name                  = "gluster",
+    .protocol_name                = "gluster+unix",
+    .instance_size                = sizeof(BDRVGlusterState),
+    .bdrv_file_open               = qemu_gluster_open,
+    .bdrv_close                   = qemu_gluster_close,
+    .bdrv_create                  = qemu_gluster_create,
+    .bdrv_getlength               = qemu_gluster_getlength,
+    .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size,
+    .bdrv_aio_readv               = qemu_gluster_aio_readv,
+    .bdrv_aio_writev              = qemu_gluster_aio_writev,
+    .bdrv_aio_flush               = qemu_gluster_aio_flush,
+    .create_options               = qemu_gluster_create_options,
+};
+
+static BlockDriver bdrv_gluster_rdma = {
+    .format_name                  = "gluster",
+    .protocol_name                = "gluster+rdma",
+    .instance_size                = sizeof(BDRVGlusterState),
+    .bdrv_file_open               = qemu_gluster_open,
+    .bdrv_close                   = qemu_gluster_close,
+    .bdrv_create                  = qemu_gluster_create,
+    .bdrv_getlength               = qemu_gluster_getlength,
+    .bdrv_get_allocated_file_size = qemu_gluster_allocated_file_size,
+    .bdrv_aio_readv               = qemu_gluster_aio_readv,
+    .bdrv_aio_writev              = qemu_gluster_aio_writev,
+    .bdrv_aio_flush               = qemu_gluster_aio_flush,
+    .create_options               = qemu_gluster_create_options,
+};
+
+static void bdrv_gluster_init(void)
+{
+    bdrv_register(&bdrv_gluster_rdma);
+    bdrv_register(&bdrv_gluster_unix);
+    bdrv_register(&bdrv_gluster_tcp);
+    bdrv_register(&bdrv_gluster);
+}
+
+block_init(bdrv_gluster_init);
diff --git a/block/stream.c b/block/stream.c
index c4f87dd5b6..792665276e 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -13,6 +13,7 @@
 
 #include "trace.h"
 #include "block_int.h"
+#include "blockjob.h"
 #include "qemu/ratelimit.h"
 
 enum {
@@ -30,6 +31,7 @@ typedef struct StreamBlockJob {
     BlockJob common;
     RateLimit limit;
     BlockDriverState *base;
+    BlockdevOnError on_error;
     char backing_file_id[1024];
 } StreamBlockJob;
 
@@ -77,6 +79,7 @@ static void coroutine_fn stream_run(void *opaque)
     BlockDriverState *bs = s->common.bs;
     BlockDriverState *base = s->base;
     int64_t sector_num, end;
+    int error = 0;
     int ret = 0;
     int n = 0;
     void *buf;
@@ -141,7 +144,19 @@ wait:
             ret = stream_populate(bs, sector_num, n, buf);
         }
         if (ret < 0) {
-            break;
+            BlockErrorAction action =
+                block_job_error_action(&s->common, s->common.bs, s->on_error,
+                                       true, -ret);
+            if (action == BDRV_ACTION_STOP) {
+                n = 0;
+                continue;
+            }
+            if (error == 0) {
+                error = ret;
+            }
+            if (action == BDRV_ACTION_REPORT) {
+                break;
+            }
         }
         ret = 0;
 
@@ -153,6 +168,9 @@ wait:
         bdrv_disable_copy_on_read(bs);
     }
 
+    /* Do not remove the backing file if an error was there but ignored.  */
+    ret = error;
+
     if (!block_job_is_cancelled(&s->common) && sector_num == end && ret == 0) {
         const char *base_id = NULL, *base_fmt = NULL;
         if (base) {
@@ -188,11 +206,19 @@ static BlockJobType stream_job_type = {
 
 void stream_start(BlockDriverState *bs, BlockDriverState *base,
                   const char *base_id, int64_t speed,
+                  BlockdevOnError on_error,
                   BlockDriverCompletionFunc *cb,
                   void *opaque, Error **errp)
 {
     StreamBlockJob *s;
 
+    if ((on_error == BLOCKDEV_ON_ERROR_STOP ||
+         on_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
+        !bdrv_iostatus_is_enabled(bs)) {
+        error_set(errp, QERR_INVALID_PARAMETER, "on-error");
+        return;
+    }
+
     s = block_job_create(&stream_job_type, bs, speed, cb, opaque, errp);
     if (!s) {
         return;
@@ -203,6 +229,7 @@ void stream_start(BlockDriverState *bs, BlockDriverState *base,
         pstrcpy(s->backing_file_id, sizeof(s->backing_file_id), base_id);
     }
 
+    s->on_error = on_error;
     s->common.co = qemu_coroutine_create(stream_run);
     trace_stream_start(bs, base, s, s->common.co, opaque);
     qemu_coroutine_enter(s->common.co, s);
diff --git a/block_int.h b/block_int.h
index ac4245cb18..f4bae04401 100644
--- a/block_int.h
+++ b/block_int.h
@@ -31,6 +31,7 @@
 #include "qemu-timer.h"
 #include "qapi-types.h"
 #include "qerror.h"
+#include "monitor.h"
 
 #define BLOCK_FLAG_ENCRYPT          1
 #define BLOCK_FLAG_COMPAT6          4
@@ -67,73 +68,6 @@ typedef struct BlockIOBaseValue {
     uint64_t ios[2];
 } BlockIOBaseValue;
 
-typedef struct BlockJob BlockJob;
-
-/**
- * BlockJobType:
- *
- * A class type for block job objects.
- */
-typedef struct BlockJobType {
-    /** Derived BlockJob struct size */
-    size_t instance_size;
-
-    /** String describing the operation, part of query-block-jobs QMP API */
-    const char *job_type;
-
-    /** Optional callback for job types that support setting a speed limit */
-    void (*set_speed)(BlockJob *job, int64_t speed, Error **errp);
-} BlockJobType;
-
-/**
- * BlockJob:
- *
- * Long-running operation on a BlockDriverState.
- */
-struct BlockJob {
-    /** The job type, including the job vtable.  */
-    const BlockJobType *job_type;
-
-    /** The block device on which the job is operating.  */
-    BlockDriverState *bs;
-
-    /**
-     * The coroutine that executes the job.  If not NULL, it is
-     * reentered when busy is false and the job is cancelled.
-     */
-    Coroutine *co;
-
-    /**
-     * Set to true if the job should cancel itself.  The flag must
-     * always be tested just before toggling the busy flag from false
-     * to true.  After a job has been cancelled, it should only yield
-     * if #qemu_aio_wait will ("sooner or later") reenter the coroutine.
-     */
-    bool cancelled;
-
-    /**
-     * Set to false by the job while it is in a quiescent state, where
-     * no I/O is pending and the job has yielded on any condition
-     * that is not detected by #qemu_aio_wait, such as a timer.
-     */
-    bool busy;
-
-    /** Offset that is published by the query-block-jobs QMP API */
-    int64_t offset;
-
-    /** Length that is published by the query-block-jobs QMP API */
-    int64_t len;
-
-    /** Speed that was set with @block_job_set_speed.  */
-    int64_t speed;
-
-    /** The completion function that will be called when the job completes.  */
-    BlockDriverCompletionFunc *cb;
-
-    /** The opaque value that is passed to the completion function.  */
-    void *opaque;
-};
-
 struct BlockDriver {
     const char *format_name;
     int instance_size;
@@ -329,7 +263,7 @@ struct BlockDriverState {
 
     /* NOTE: the following infos are only hints for real hardware
        drivers. They are not used by the block driver */
-    BlockErrorAction on_read_error, on_write_error;
+    BlockdevOnError on_read_error, on_write_error;
     bool iostatus_enabled;
     BlockDeviceIoStatus iostatus;
     char device_name[32];
@@ -353,92 +287,9 @@ void bdrv_set_io_limits(BlockDriverState *bs,
 #ifdef _WIN32
 int is_windows_drive(const char *filename);
 #endif
-
-/**
- * block_job_create:
- * @job_type: The class object for the newly-created job.
- * @bs: The block
- * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
- * @cb: Completion function for the job.
- * @opaque: Opaque pointer value passed to @cb.
- * @errp: Error object.
- *
- * Create a new long-running block device job and return it.  The job
- * will call @cb asynchronously when the job completes.  Note that
- * @bs may have been closed at the time the @cb it is called.  If
- * this is the case, the job may be reported as either cancelled or
- * completed.
- *
- * This function is not part of the public job interface; it should be
- * called from a wrapper that is specific to the job type.
- */
-void *block_job_create(const BlockJobType *job_type, BlockDriverState *bs,
-                       int64_t speed, BlockDriverCompletionFunc *cb,
-                       void *opaque, Error **errp);
-
-/**
- * block_job_sleep_ns:
- * @job: The job that calls the function.
- * @clock: The clock to sleep on.
- * @ns: How many nanoseconds to stop for.
- *
- * Put the job to sleep (assuming that it wasn't canceled) for @ns
- * nanoseconds.  Canceling the job will interrupt the wait immediately.
- */
-void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns);
-
-/**
- * block_job_complete:
- * @job: The job being completed.
- * @ret: The status code.
- *
- * Call the completion function that was registered at creation time, and
- * free @job.
- */
-void block_job_complete(BlockJob *job, int ret);
-
-/**
- * block_job_set_speed:
- * @job: The job to set the speed for.
- * @speed: The new value
- * @errp: Error object.
- *
- * Set a rate-limiting parameter for the job; the actual meaning may
- * vary depending on the job type.
- */
-void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp);
-
-/**
- * block_job_cancel:
- * @job: The job to be canceled.
- *
- * Asynchronously cancel the specified job.
- */
-void block_job_cancel(BlockJob *job);
-
-/**
- * block_job_is_cancelled:
- * @job: The job being queried.
- *
- * Returns whether the job is scheduled for cancellation.
- */
-bool block_job_is_cancelled(BlockJob *job);
-
-/**
- * block_job_cancel:
- * @job: The job to be canceled.
- *
- * Asynchronously cancel the job and wait for it to reach a quiescent
- * state.  Note that the completion callback will still be called
- * asynchronously, hence it is *not* valid to call #bdrv_delete
- * immediately after #block_job_cancel_sync.  Users of block jobs
- * will usually protect the BlockDriverState objects with a reference
- * count, should this be a concern.
- *
- * Returns the return value from the job if the job actually completed
- * during the call, or -ECANCELED if it was canceled.
- */
-int block_job_cancel_sync(BlockJob *job);
+void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
+                               enum MonitorEvent ev,
+                               BlockErrorAction action, bool is_read);
 
 /**
  * stream_start:
@@ -448,6 +299,7 @@ int block_job_cancel_sync(BlockJob *job);
  * @base_id: The file name that will be written to @bs as the new
  * backing file if the job completes.  Ignored if @base is %NULL.
  * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @on_error: The action to take upon error.
  * @cb: Completion function for the job.
  * @opaque: Opaque pointer value passed to @cb.
  * @errp: Error object.
@@ -459,8 +311,24 @@ int block_job_cancel_sync(BlockJob *job);
  * @base_id in the written image and to @base in the live BlockDriverState.
  */
 void stream_start(BlockDriverState *bs, BlockDriverState *base,
-                  const char *base_id, int64_t speed,
+                  const char *base_id, int64_t speed, BlockdevOnError on_error,
                   BlockDriverCompletionFunc *cb,
                   void *opaque, Error **errp);
 
+/**
+ * commit_start:
+ * @bs: Top Block device
+ * @base: Block device that will be written into, and become the new top
+ * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @on_error: The action to take upon error.
+ * @cb: Completion function for the job.
+ * @opaque: Opaque pointer value passed to @cb.
+ * @errp: Error object.
+ *
+ */
+void commit_start(BlockDriverState *bs, BlockDriverState *base,
+                 BlockDriverState *top, int64_t speed,
+                 BlockdevOnError on_error, BlockDriverCompletionFunc *cb,
+                 void *opaque, Error **errp);
+
 #endif /* BLOCK_INT_H */
diff --git a/blockdev.c b/blockdev.c
index e5d450f0bb..5f18dfa97b 100644
--- a/blockdev.c
+++ b/blockdev.c
@@ -9,6 +9,7 @@
 
 #include "blockdev.h"
 #include "hw/block-common.h"
+#include "blockjob.h"
 #include "monitor.h"
 #include "qerror.h"
 #include "qemu-option.h"
@@ -237,16 +238,16 @@ static void drive_put_ref_bh_schedule(DriveInfo *dinfo)
     qemu_bh_schedule(s->bh);
 }
 
-static int parse_block_error_action(const char *buf, int is_read)
+static int parse_block_error_action(const char *buf, bool is_read)
 {
     if (!strcmp(buf, "ignore")) {
-        return BLOCK_ERR_IGNORE;
+        return BLOCKDEV_ON_ERROR_IGNORE;
     } else if (!is_read && !strcmp(buf, "enospc")) {
-        return BLOCK_ERR_STOP_ENOSPC;
+        return BLOCKDEV_ON_ERROR_ENOSPC;
     } else if (!strcmp(buf, "stop")) {
-        return BLOCK_ERR_STOP_ANY;
+        return BLOCKDEV_ON_ERROR_STOP;
     } else if (!strcmp(buf, "report")) {
-        return BLOCK_ERR_REPORT;
+        return BLOCKDEV_ON_ERROR_REPORT;
     } else {
         error_report("'%s' invalid %s error action",
                      buf, is_read ? "read" : "write");
@@ -432,7 +433,7 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
         return NULL;
     }
 
-    on_write_error = BLOCK_ERR_STOP_ENOSPC;
+    on_write_error = BLOCKDEV_ON_ERROR_ENOSPC;
     if ((buf = qemu_opt_get(opts, "werror")) != NULL) {
         if (type != IF_IDE && type != IF_SCSI && type != IF_VIRTIO && type != IF_NONE) {
             error_report("werror is not supported by this bus type");
@@ -445,7 +446,7 @@ DriveInfo *drive_init(QemuOpts *opts, int default_to_scsi)
         }
     }
 
-    on_read_error = BLOCK_ERR_REPORT;
+    on_read_error = BLOCKDEV_ON_ERROR_REPORT;
     if ((buf = qemu_opt_get(opts, "rerror")) != NULL) {
         if (type != IF_IDE && type != IF_VIRTIO && type != IF_SCSI && type != IF_NONE) {
             error_report("rerror is not supported by this bus type");
@@ -805,6 +806,11 @@ void qmp_transaction(BlockdevActionList *dev_list, Error **errp)
     QSIMPLEQ_FOREACH(states, &snap_bdrv_states, entry) {
         /* This removes our old bs from the bdrv_states, and adds the new bs */
         bdrv_append(states->new_bs, states->old_bs);
+        /* We don't need (or want) to use the transactional
+         * bdrv_reopen_multiple() across all the entries at once, because we
+         * don't want to abort all of them if one of them fails the reopen */
+        bdrv_reopen(states->new_bs, states->new_bs->open_flags & ~BDRV_O_RDWR,
+                    NULL);
     }
 
     /* success */
@@ -1065,12 +1071,12 @@ static QObject *qobject_from_block_job(BlockJob *job)
                               job->speed);
 }
 
-static void block_stream_cb(void *opaque, int ret)
+static void block_job_cb(void *opaque, int ret)
 {
     BlockDriverState *bs = opaque;
     QObject *obj;
 
-    trace_block_stream_cb(bs, bs->job, ret);
+    trace_block_job_cb(bs, bs->job, ret);
 
     assert(bs->job);
     obj = qobject_from_block_job(bs->job);
@@ -1090,13 +1096,18 @@ static void block_stream_cb(void *opaque, int ret)
 }
 
 void qmp_block_stream(const char *device, bool has_base,
-                      const char *base, bool has_speed,
-                      int64_t speed, Error **errp)
+                      const char *base, bool has_speed, int64_t speed,
+                      bool has_on_error, BlockdevOnError on_error,
+                      Error **errp)
 {
     BlockDriverState *bs;
     BlockDriverState *base_bs = NULL;
     Error *local_err = NULL;
 
+    if (!has_on_error) {
+        on_error = BLOCKDEV_ON_ERROR_REPORT;
+    }
+
     bs = bdrv_find(device);
     if (!bs) {
         error_set(errp, QERR_DEVICE_NOT_FOUND, device);
@@ -1112,7 +1123,7 @@ void qmp_block_stream(const char *device, bool has_base,
     }
 
     stream_start(bs, base_bs, base, has_speed ? speed : 0,
-                 block_stream_cb, bs, &local_err);
+                 on_error, block_job_cb, bs, &local_err);
     if (error_is_set(&local_err)) {
         error_propagate(errp, local_err);
         return;
@@ -1126,6 +1137,64 @@ void qmp_block_stream(const char *device, bool has_base,
     trace_qmp_block_stream(bs, bs->job);
 }
 
+void qmp_block_commit(const char *device,
+                      bool has_base, const char *base, const char *top,
+                      bool has_speed, int64_t speed,
+                      Error **errp)
+{
+    BlockDriverState *bs;
+    BlockDriverState *base_bs, *top_bs;
+    Error *local_err = NULL;
+    /* This will be part of the QMP command, if/when the
+     * BlockdevOnError change for blkmirror makes it in
+     */
+    BlockdevOnError on_error = BLOCKDEV_ON_ERROR_REPORT;
+
+    /* drain all i/o before commits */
+    bdrv_drain_all();
+
+    bs = bdrv_find(device);
+    if (!bs) {
+        error_set(errp, QERR_DEVICE_NOT_FOUND, device);
+        return;
+    }
+    if (base && has_base) {
+        base_bs = bdrv_find_backing_image(bs, base);
+    } else {
+        base_bs = bdrv_find_base(bs);
+    }
+
+    if (base_bs == NULL) {
+        error_set(errp, QERR_BASE_NOT_FOUND, base ? base : "NULL");
+        return;
+    }
+
+    /* default top_bs is the active layer */
+    top_bs = bs;
+
+    if (top) {
+        if (strcmp(bs->filename, top) != 0) {
+            top_bs = bdrv_find_backing_image(bs, top);
+        }
+    }
+
+    if (top_bs == NULL) {
+        error_setg(errp, "Top image file %s not found", top ? top : "NULL");
+        return;
+    }
+
+    commit_start(bs, base_bs, top_bs, speed, on_error, block_job_cb, bs,
+                &local_err);
+    if (local_err != NULL) {
+        error_propagate(errp, local_err);
+        return;
+    }
+    /* Grab a reference so hotplug does not delete the BlockDriverState from
+     * underneath us.
+     */
+    drive_get_ref(drive_get_by_blockdev(bs));
+}
+
 static BlockJob *find_block_job(const char *device)
 {
     BlockDriverState *bs;
@@ -1142,19 +1211,28 @@ void qmp_block_job_set_speed(const char *device, int64_t speed, Error **errp)
     BlockJob *job = find_block_job(device);
 
     if (!job) {
-        error_set(errp, QERR_DEVICE_NOT_ACTIVE, device);
+        error_set(errp, QERR_BLOCK_JOB_NOT_ACTIVE, device);
         return;
     }
 
     block_job_set_speed(job, speed, errp);
 }
 
-void qmp_block_job_cancel(const char *device, Error **errp)
+void qmp_block_job_cancel(const char *device,
+                          bool has_force, bool force, Error **errp)
 {
     BlockJob *job = find_block_job(device);
 
+    if (!has_force) {
+        force = false;
+    }
+
     if (!job) {
-        error_set(errp, QERR_DEVICE_NOT_ACTIVE, device);
+        error_set(errp, QERR_BLOCK_JOB_NOT_ACTIVE, device);
+        return;
+    }
+    if (job->paused && !force) {
+        error_set(errp, QERR_BLOCK_JOB_PAUSED, device);
         return;
     }
 
@@ -1162,25 +1240,40 @@ void qmp_block_job_cancel(const char *device, Error **errp)
     block_job_cancel(job);
 }
 
+void qmp_block_job_pause(const char *device, Error **errp)
+{
+    BlockJob *job = find_block_job(device);
+
+    if (!job) {
+        error_set(errp, QERR_BLOCK_JOB_NOT_ACTIVE, device);
+        return;
+    }
+
+    trace_qmp_block_job_pause(job);
+    block_job_pause(job);
+}
+
+void qmp_block_job_resume(const char *device, Error **errp)
+{
+    BlockJob *job = find_block_job(device);
+
+    if (!job) {
+        error_set(errp, QERR_BLOCK_JOB_NOT_ACTIVE, device);
+        return;
+    }
+
+    trace_qmp_block_job_resume(job);
+    block_job_resume(job);
+}
+
 static void do_qmp_query_block_jobs_one(void *opaque, BlockDriverState *bs)
 {
     BlockJobInfoList **prev = opaque;
     BlockJob *job = bs->job;
 
     if (job) {
-        BlockJobInfoList *elem;
-        BlockJobInfo *info = g_new(BlockJobInfo, 1);
-        *info = (BlockJobInfo){
-            .type   = g_strdup(job->job_type->job_type),
-            .device = g_strdup(bdrv_get_device_name(bs)),
-            .len    = job->len,
-            .offset = job->offset,
-            .speed  = job->speed,
-        };
-
-        elem = g_new0(BlockJobInfoList, 1);
-        elem->value = info;
-
+        BlockJobInfoList *elem = g_new0(BlockJobInfoList, 1);
+        elem->value = block_job_query(bs->job);
         (*prev)->next = elem;
         *prev = elem;
     }
diff --git a/blockjob.c b/blockjob.c
new file mode 100644
index 0000000000..f55f55a193
--- /dev/null
+++ b/blockjob.c
@@ -0,0 +1,249 @@
+/*
+ * QEMU System Emulator block driver
+ *
+ * Copyright (c) 2011 IBM Corp.
+ * Copyright (c) 2012 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 "config-host.h"
+#include "qemu-common.h"
+#include "trace.h"
+#include "monitor.h"
+#include "block.h"
+#include "blockjob.h"
+#include "block_int.h"
+#include "qjson.h"
+#include "qemu-coroutine.h"
+#include "qmp-commands.h"
+#include "qemu-timer.h"
+
+void *block_job_create(const BlockJobType *job_type, BlockDriverState *bs,
+                       int64_t speed, BlockDriverCompletionFunc *cb,
+                       void *opaque, Error **errp)
+{
+    BlockJob *job;
+
+    if (bs->job || bdrv_in_use(bs)) {
+        error_set(errp, QERR_DEVICE_IN_USE, bdrv_get_device_name(bs));
+        return NULL;
+    }
+    bdrv_set_in_use(bs, 1);
+
+    job = g_malloc0(job_type->instance_size);
+    job->job_type      = job_type;
+    job->bs            = bs;
+    job->cb            = cb;
+    job->opaque        = opaque;
+    job->busy          = true;
+    bs->job = job;
+
+    /* Only set speed when necessary to avoid NotSupported error */
+    if (speed != 0) {
+        Error *local_err = NULL;
+
+        block_job_set_speed(job, speed, &local_err);
+        if (error_is_set(&local_err)) {
+            bs->job = NULL;
+            g_free(job);
+            bdrv_set_in_use(bs, 0);
+            error_propagate(errp, local_err);
+            return NULL;
+        }
+    }
+    return job;
+}
+
+void block_job_complete(BlockJob *job, int ret)
+{
+    BlockDriverState *bs = job->bs;
+
+    assert(bs->job == job);
+    job->cb(job->opaque, ret);
+    bs->job = NULL;
+    g_free(job);
+    bdrv_set_in_use(bs, 0);
+}
+
+void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp)
+{
+    Error *local_err = NULL;
+
+    if (!job->job_type->set_speed) {
+        error_set(errp, QERR_NOT_SUPPORTED);
+        return;
+    }
+    job->job_type->set_speed(job, speed, &local_err);
+    if (error_is_set(&local_err)) {
+        error_propagate(errp, local_err);
+        return;
+    }
+
+    job->speed = speed;
+}
+
+void block_job_pause(BlockJob *job)
+{
+    job->paused = true;
+}
+
+bool block_job_is_paused(BlockJob *job)
+{
+    return job->paused;
+}
+
+void block_job_resume(BlockJob *job)
+{
+    job->paused = false;
+    block_job_iostatus_reset(job);
+    if (job->co && !job->busy) {
+        qemu_coroutine_enter(job->co, NULL);
+    }
+}
+
+void block_job_cancel(BlockJob *job)
+{
+    job->cancelled = true;
+    block_job_resume(job);
+}
+
+bool block_job_is_cancelled(BlockJob *job)
+{
+    return job->cancelled;
+}
+
+void block_job_iostatus_reset(BlockJob *job)
+{
+    job->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
+}
+
+struct BlockCancelData {
+    BlockJob *job;
+    BlockDriverCompletionFunc *cb;
+    void *opaque;
+    bool cancelled;
+    int ret;
+};
+
+static void block_job_cancel_cb(void *opaque, int ret)
+{
+    struct BlockCancelData *data = opaque;
+
+    data->cancelled = block_job_is_cancelled(data->job);
+    data->ret = ret;
+    data->cb(data->opaque, ret);
+}
+
+int block_job_cancel_sync(BlockJob *job)
+{
+    struct BlockCancelData data;
+    BlockDriverState *bs = job->bs;
+
+    assert(bs->job == job);
+
+    /* Set up our own callback to store the result and chain to
+     * the original callback.
+     */
+    data.job = job;
+    data.cb = job->cb;
+    data.opaque = job->opaque;
+    data.ret = -EINPROGRESS;
+    job->cb = block_job_cancel_cb;
+    job->opaque = &data;
+    block_job_cancel(job);
+    while (data.ret == -EINPROGRESS) {
+        qemu_aio_wait();
+    }
+    return (data.cancelled && data.ret == 0) ? -ECANCELED : data.ret;
+}
+
+void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns)
+{
+    assert(job->busy);
+
+    /* Check cancellation *before* setting busy = false, too!  */
+    if (block_job_is_cancelled(job)) {
+        return;
+    }
+
+    job->busy = false;
+    if (block_job_is_paused(job)) {
+        qemu_coroutine_yield();
+    } else {
+        co_sleep_ns(clock, ns);
+    }
+    job->busy = true;
+}
+
+BlockJobInfo *block_job_query(BlockJob *job)
+{
+    BlockJobInfo *info = g_new0(BlockJobInfo, 1);
+    info->type      = g_strdup(job->job_type->job_type);
+    info->device    = g_strdup(bdrv_get_device_name(job->bs));
+    info->len       = job->len;
+    info->busy      = job->busy;
+    info->paused    = job->paused;
+    info->offset    = job->offset;
+    info->speed     = job->speed;
+    info->io_status = job->iostatus;
+    return info;
+}
+
+static void block_job_iostatus_set_err(BlockJob *job, int error)
+{
+    if (job->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
+        job->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
+                                          BLOCK_DEVICE_IO_STATUS_FAILED;
+    }
+}
+
+
+BlockErrorAction block_job_error_action(BlockJob *job, BlockDriverState *bs,
+                                        BlockdevOnError on_err,
+                                        int is_read, int error)
+{
+    BlockErrorAction action;
+
+    switch (on_err) {
+    case BLOCKDEV_ON_ERROR_ENOSPC:
+        action = (error == ENOSPC) ? BDRV_ACTION_STOP : BDRV_ACTION_REPORT;
+        break;
+    case BLOCKDEV_ON_ERROR_STOP:
+        action = BDRV_ACTION_STOP;
+        break;
+    case BLOCKDEV_ON_ERROR_REPORT:
+        action = BDRV_ACTION_REPORT;
+        break;
+    case BLOCKDEV_ON_ERROR_IGNORE:
+        action = BDRV_ACTION_IGNORE;
+        break;
+    default:
+        abort();
+    }
+    bdrv_emit_qmp_error_event(job->bs, QEVENT_BLOCK_JOB_ERROR, action, is_read);
+    if (action == BDRV_ACTION_STOP) {
+        block_job_pause(job);
+        block_job_iostatus_set_err(job, error);
+        if (bs != job->bs) {
+            bdrv_iostatus_set_err(bs, error);
+        }
+    }
+    return action;
+}
diff --git a/blockjob.h b/blockjob.h
new file mode 100644
index 0000000000..930cc3c46a
--- /dev/null
+++ b/blockjob.h
@@ -0,0 +1,243 @@
+/*
+ * Declarations for long-running block device operations
+ *
+ * Copyright (c) 2011 IBM Corp.
+ * Copyright (c) 2012 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.
+ */
+#ifndef BLOCKJOB_H
+#define BLOCKJOB_H 1
+
+#include "block.h"
+
+/**
+ * BlockJobType:
+ *
+ * A class type for block job objects.
+ */
+typedef struct BlockJobType {
+    /** Derived BlockJob struct size */
+    size_t instance_size;
+
+    /** String describing the operation, part of query-block-jobs QMP API */
+    const char *job_type;
+
+    /** Optional callback for job types that support setting a speed limit */
+    void (*set_speed)(BlockJob *job, int64_t speed, Error **errp);
+} BlockJobType;
+
+/**
+ * BlockJob:
+ *
+ * Long-running operation on a BlockDriverState.
+ */
+struct BlockJob {
+    /** The job type, including the job vtable.  */
+    const BlockJobType *job_type;
+
+    /** The block device on which the job is operating.  */
+    BlockDriverState *bs;
+
+    /**
+     * The coroutine that executes the job.  If not NULL, it is
+     * reentered when busy is false and the job is cancelled.
+     */
+    Coroutine *co;
+
+    /**
+     * Set to true if the job should cancel itself.  The flag must
+     * always be tested just before toggling the busy flag from false
+     * to true.  After a job has been cancelled, it should only yield
+     * if #qemu_aio_wait will ("sooner or later") reenter the coroutine.
+     */
+    bool cancelled;
+
+    /**
+     * Set to true if the job is either paused, or will pause itself
+     * as soon as possible (if busy == true).
+     */
+    bool paused;
+
+    /**
+     * Set to false by the job while it is in a quiescent state, where
+     * no I/O is pending and the job has yielded on any condition
+     * that is not detected by #qemu_aio_wait, such as a timer.
+     */
+    bool busy;
+
+    /** Status that is published by the query-block-jobs QMP API */
+    BlockDeviceIoStatus iostatus;
+
+    /** Offset that is published by the query-block-jobs QMP API */
+    int64_t offset;
+
+    /** Length that is published by the query-block-jobs QMP API */
+    int64_t len;
+
+    /** Speed that was set with @block_job_set_speed.  */
+    int64_t speed;
+
+    /** The completion function that will be called when the job completes.  */
+    BlockDriverCompletionFunc *cb;
+
+    /** The opaque value that is passed to the completion function.  */
+    void *opaque;
+};
+
+/**
+ * block_job_create:
+ * @job_type: The class object for the newly-created job.
+ * @bs: The block
+ * @speed: The maximum speed, in bytes per second, or 0 for unlimited.
+ * @cb: Completion function for the job.
+ * @opaque: Opaque pointer value passed to @cb.
+ * @errp: Error object.
+ *
+ * Create a new long-running block device job and return it.  The job
+ * will call @cb asynchronously when the job completes.  Note that
+ * @bs may have been closed at the time the @cb it is called.  If
+ * this is the case, the job may be reported as either cancelled or
+ * completed.
+ *
+ * This function is not part of the public job interface; it should be
+ * called from a wrapper that is specific to the job type.
+ */
+void *block_job_create(const BlockJobType *job_type, BlockDriverState *bs,
+                       int64_t speed, BlockDriverCompletionFunc *cb,
+                       void *opaque, Error **errp);
+
+/**
+ * block_job_sleep_ns:
+ * @job: The job that calls the function.
+ * @clock: The clock to sleep on.
+ * @ns: How many nanoseconds to stop for.
+ *
+ * Put the job to sleep (assuming that it wasn't canceled) for @ns
+ * nanoseconds.  Canceling the job will interrupt the wait immediately.
+ */
+void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns);
+
+/**
+ * block_job_complete:
+ * @job: The job being completed.
+ * @ret: The status code.
+ *
+ * Call the completion function that was registered at creation time, and
+ * free @job.
+ */
+void block_job_complete(BlockJob *job, int ret);
+
+/**
+ * block_job_set_speed:
+ * @job: The job to set the speed for.
+ * @speed: The new value
+ * @errp: Error object.
+ *
+ * Set a rate-limiting parameter for the job; the actual meaning may
+ * vary depending on the job type.
+ */
+void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp);
+
+/**
+ * block_job_cancel:
+ * @job: The job to be canceled.
+ *
+ * Asynchronously cancel the specified job.
+ */
+void block_job_cancel(BlockJob *job);
+
+/**
+ * block_job_is_cancelled:
+ * @job: The job being queried.
+ *
+ * Returns whether the job is scheduled for cancellation.
+ */
+bool block_job_is_cancelled(BlockJob *job);
+
+/**
+ * block_job_query:
+ * @job: The job to get information about.
+ *
+ * Return information about a job.
+ */
+BlockJobInfo *block_job_query(BlockJob *job);
+
+/**
+ * block_job_pause:
+ * @job: The job to be paused.
+ *
+ * Asynchronously pause the specified job.
+ */
+void block_job_pause(BlockJob *job);
+
+/**
+ * block_job_resume:
+ * @job: The job to be resumed.
+ *
+ * Resume the specified job.
+ */
+void block_job_resume(BlockJob *job);
+
+/**
+ * block_job_is_paused:
+ * @job: The job being queried.
+ *
+ * Returns whether the job is currently paused, or will pause
+ * as soon as it reaches a sleeping point.
+ */
+bool block_job_is_paused(BlockJob *job);
+
+/**
+ * block_job_cancel_sync:
+ * @job: The job to be canceled.
+ *
+ * Synchronously cancel the job.  The completion callback is called
+ * before the function returns.  The job may actually complete
+ * instead of canceling itself; the circumstances under which this
+ * happens depend on the kind of job that is active.
+ *
+ * Returns the return value from the job if the job actually completed
+ * during the call, or -ECANCELED if it was canceled.
+ */
+int block_job_cancel_sync(BlockJob *job);
+
+/**
+ * block_job_iostatus_reset:
+ * @job: The job whose I/O status should be reset.
+ *
+ * Reset I/O status on @job.
+ */
+void block_job_iostatus_reset(BlockJob *job);
+
+/**
+ * block_job_error_action:
+ * @job: The job to signal an error for.
+ * @bs: The block device on which to set an I/O error.
+ * @on_err: The error action setting.
+ * @is_read: Whether the operation was a read.
+ * @error: The error that was reported.
+ *
+ * Report an I/O error for a block job and possibly stop the VM.  Return the
+ * action that was selected based on @on_err and @error.
+ */
+BlockErrorAction block_job_error_action(BlockJob *job, BlockDriverState *bs,
+                                        BlockdevOnError on_err,
+                                        int is_read, int error);
+#endif
diff --git a/configure b/configure
index c5e5bb2e81..e58846d5e2 100755
--- a/configure
+++ b/configure
@@ -219,6 +219,7 @@ want_tools="yes"
 libiscsi=""
 coroutine=""
 seccomp=""
+glusterfs=""
 
 # parse CC options first
 for opt do
@@ -856,6 +857,10 @@ for opt do
   ;;
   --disable-seccomp) seccomp="no"
   ;;
+  --disable-glusterfs) glusterfs="no"
+  ;;
+  --enable-glusterfs) glusterfs="yes"
+  ;;
   *) echo "ERROR: unknown option $opt"; show_help="yes"
   ;;
   esac
@@ -1128,6 +1133,8 @@ echo "  --disable-seccomp        disable seccomp support"
 echo "  --enable-seccomp         enables seccomp support"
 echo "  --with-coroutine=BACKEND coroutine backend. Supported options:"
 echo "                           gthread, ucontext, sigaltstack, windows"
+echo "  --enable-glusterfs       enable GlusterFS backend"
+echo "  --disable-glusterfs      disable GlusterFS backend"
 echo ""
 echo "NOTE: The object files are built at the place where configure is launched"
 exit 1
@@ -2303,6 +2310,29 @@ EOF
   fi
 fi
 
+##########################################
+# glusterfs probe
+if test "$glusterfs" != "no" ; then
+  cat > $TMPC <<EOF
+#include <glusterfs/api/glfs.h>
+int main(void) {
+    (void) glfs_new("volume");
+    return 0;
+}
+EOF
+  glusterfs_libs="-lgfapi -lgfrpc -lgfxdr"
+  if compile_prog "" "$glusterfs_libs" ; then
+    glusterfs=yes
+    libs_tools="$glusterfs_libs $libs_tools"
+    libs_softmmu="$glusterfs_libs $libs_softmmu"
+  else
+    if test "$glusterfs" = "yes" ; then
+      feature_not_found "GlusterFS backend support"
+    fi
+    glusterfs=no
+  fi
+fi
+
 #
 # Check for xxxat() functions when we are building linux-user
 # emulator.  This is done because older glibc versions don't
@@ -3170,6 +3200,7 @@ echo "libiscsi support  $libiscsi"
 echo "build guest agent $guest_agent"
 echo "seccomp support   $seccomp"
 echo "coroutine backend $coroutine_backend"
+echo "GlusterFS support $glusterfs"
 
 if test "$sdl_too_old" = "yes"; then
 echo "-> Your SDL version is too old - please upgrade to have SDL support"
@@ -3516,6 +3547,10 @@ if test "$has_environ" = "yes" ; then
   echo "CONFIG_HAS_ENVIRON=y" >> $config_host_mak
 fi
 
+if test "$glusterfs" = "yes" ; then
+  echo "CONFIG_GLUSTERFS=y" >> $config_host_mak
+fi
+
 # USB host support
 case "$usb" in
 linux)
diff --git a/hmp-commands.hx b/hmp-commands.hx
index 0302458df2..e0b537d0cc 100644
--- a/hmp-commands.hx
+++ b/hmp-commands.hx
@@ -99,9 +99,10 @@ ETEXI
 
     {
         .name       = "block_job_cancel",
-        .args_type  = "device:B",
-        .params     = "device",
-        .help       = "stop an active background block operation",
+        .args_type  = "force:-f,device:B",
+        .params     = "[-f] device",
+        .help       = "stop an active background block operation (use -f"
+                      "\n\t\t\t if the operation is currently paused)",
         .mhandler.cmd = hmp_block_job_cancel,
     },
 
@@ -112,6 +113,34 @@ Stop an active block streaming operation.
 ETEXI
 
     {
+        .name       = "block_job_pause",
+        .args_type  = "device:B",
+        .params     = "device",
+        .help       = "pause an active background block operation",
+        .mhandler.cmd = hmp_block_job_pause,
+    },
+
+STEXI
+@item block_job_pause
+@findex block_job_pause
+Pause an active block streaming operation.
+ETEXI
+
+    {
+        .name       = "block_job_resume",
+        .args_type  = "device:B",
+        .params     = "device",
+        .help       = "resume a paused background block operation",
+        .mhandler.cmd = hmp_block_job_resume,
+    },
+
+STEXI
+@item block_job_resume
+@findex block_job_resume
+Resume a paused block streaming operation.
+ETEXI
+
+    {
         .name       = "eject",
         .args_type  = "force:-f,device:B",
         .params     = "[-f] device",
diff --git a/hmp.c b/hmp.c
index 3306bcdbb4..70bdec2433 100644
--- a/hmp.c
+++ b/hmp.c
@@ -930,7 +930,8 @@ void hmp_block_stream(Monitor *mon, const QDict *qdict)
     int64_t speed = qdict_get_try_int(qdict, "speed", 0);
 
     qmp_block_stream(device, base != NULL, base,
-                     qdict_haskey(qdict, "speed"), speed, &error);
+                     qdict_haskey(qdict, "speed"), speed,
+                     BLOCKDEV_ON_ERROR_REPORT, true, &error);
 
     hmp_handle_error(mon, &error);
 }
@@ -950,8 +951,29 @@ void hmp_block_job_cancel(Monitor *mon, const QDict *qdict)
 {
     Error *error = NULL;
     const char *device = qdict_get_str(qdict, "device");
+    bool force = qdict_get_try_bool(qdict, "force", 0);
 
-    qmp_block_job_cancel(device, &error);
+    qmp_block_job_cancel(device, true, force, &error);
+
+    hmp_handle_error(mon, &error);
+}
+
+void hmp_block_job_pause(Monitor *mon, const QDict *qdict)
+{
+    Error *error = NULL;
+    const char *device = qdict_get_str(qdict, "device");
+
+    qmp_block_job_pause(device, &error);
+
+    hmp_handle_error(mon, &error);
+}
+
+void hmp_block_job_resume(Monitor *mon, const QDict *qdict)
+{
+    Error *error = NULL;
+    const char *device = qdict_get_str(qdict, "device");
+
+    qmp_block_job_resume(device, &error);
 
     hmp_handle_error(mon, &error);
 }
diff --git a/hmp.h b/hmp.h
index 48b9c59f8a..71ea384523 100644
--- a/hmp.h
+++ b/hmp.h
@@ -64,6 +64,8 @@ void hmp_block_set_io_throttle(Monitor *mon, const QDict *qdict);
 void hmp_block_stream(Monitor *mon, const QDict *qdict);
 void hmp_block_job_set_speed(Monitor *mon, const QDict *qdict);
 void hmp_block_job_cancel(Monitor *mon, const QDict *qdict);
+void hmp_block_job_pause(Monitor *mon, const QDict *qdict);
+void hmp_block_job_resume(Monitor *mon, const QDict *qdict);
 void hmp_migrate(Monitor *mon, const QDict *qdict);
 void hmp_device_del(Monitor *mon, const QDict *qdict);
 void hmp_dump_guest_memory(Monitor *mon, const QDict *qdict);
diff --git a/hw/fdc.c b/hw/fdc.c
index 08830c1ba2..43b0f20503 100644
--- a/hw/fdc.c
+++ b/hw/fdc.c
@@ -1994,11 +1994,11 @@ static int fdctrl_connect_drives(FDCtrl *fdctrl)
         drive->fdctrl = fdctrl;
 
         if (drive->bs) {
-            if (bdrv_get_on_error(drive->bs, 0) != BLOCK_ERR_STOP_ENOSPC) {
+            if (bdrv_get_on_error(drive->bs, 0) != BLOCKDEV_ON_ERROR_ENOSPC) {
                 error_report("fdc doesn't support drive option werror");
                 return -1;
             }
-            if (bdrv_get_on_error(drive->bs, 1) != BLOCK_ERR_REPORT) {
+            if (bdrv_get_on_error(drive->bs, 1) != BLOCKDEV_ON_ERROR_REPORT) {
                 error_report("fdc doesn't support drive option rerror");
                 return -1;
             }
diff --git a/hw/ide/core.c b/hw/ide/core.c
index d6fb69c634..d683a8cc84 100644
--- a/hw/ide/core.c
+++ b/hw/ide/core.c
@@ -556,32 +556,22 @@ void ide_dma_error(IDEState *s)
 
 static int ide_handle_rw_error(IDEState *s, int error, int op)
 {
-    int is_read = (op & BM_STATUS_RETRY_READ);
-    BlockErrorAction action = bdrv_get_on_error(s->bs, is_read);
+    bool is_read = (op & BM_STATUS_RETRY_READ) != 0;
+    BlockErrorAction action = bdrv_get_error_action(s->bs, is_read, error);
 
-    if (action == BLOCK_ERR_IGNORE) {
-        bdrv_emit_qmp_error_event(s->bs, BDRV_ACTION_IGNORE, is_read);
-        return 0;
-    }
-
-    if ((error == ENOSPC && action == BLOCK_ERR_STOP_ENOSPC)
-            || action == BLOCK_ERR_STOP_ANY) {
+    if (action == BDRV_ACTION_STOP) {
         s->bus->dma->ops->set_unit(s->bus->dma, s->unit);
         s->bus->error_status = op;
-        bdrv_emit_qmp_error_event(s->bs, BDRV_ACTION_STOP, is_read);
-        vm_stop(RUN_STATE_IO_ERROR);
-        bdrv_iostatus_set_err(s->bs, error);
-    } else {
+    } else if (action == BDRV_ACTION_REPORT) {
         if (op & BM_STATUS_DMA_RETRY) {
             dma_buf_commit(s);
             ide_dma_error(s);
         } else {
             ide_rw_error(s);
         }
-        bdrv_emit_qmp_error_event(s->bs, BDRV_ACTION_REPORT, is_read);
     }
-
-    return 1;
+    bdrv_error_action(s->bs, action, is_read, error);
+    return action != BDRV_ACTION_IGNORE;
 }
 
 void ide_dma_cb(void *opaque, int ret)
diff --git a/hw/ide/pci.c b/hw/ide/pci.c
index 88c0942e34..644533f777 100644
--- a/hw/ide/pci.c
+++ b/hw/ide/pci.c
@@ -188,7 +188,7 @@ static void bmdma_restart_bh(void *opaque)
 {
     BMDMAState *bm = opaque;
     IDEBus *bus = bm->bus;
-    int is_read;
+    bool is_read;
     int error_status;
 
     qemu_bh_delete(bm->bh);
@@ -198,7 +198,7 @@ static void bmdma_restart_bh(void *opaque)
         return;
     }
 
-    is_read = !!(bus->error_status & BM_STATUS_RETRY_READ);
+    is_read = (bus->error_status & BM_STATUS_RETRY_READ) != 0;
 
     /* The error status must be cleared before resubmitting the request: The
      * request may fail again, and this case can only be distinguished if the
diff --git a/hw/scsi-disk.c b/hw/scsi-disk.c
index 95e91585e6..99bb02ebf8 100644
--- a/hw/scsi-disk.c
+++ b/hw/scsi-disk.c
@@ -386,23 +386,11 @@ static void scsi_read_data(SCSIRequest *req)
  */
 static int scsi_handle_rw_error(SCSIDiskReq *r, int error)
 {
-    int is_read = (r->req.cmd.xfer == SCSI_XFER_FROM_DEV);
+    bool is_read = (r->req.cmd.xfer == SCSI_XFER_FROM_DEV);
     SCSIDiskState *s = DO_UPCAST(SCSIDiskState, qdev, r->req.dev);
-    BlockErrorAction action = bdrv_get_on_error(s->qdev.conf.bs, is_read);
+    BlockErrorAction action = bdrv_get_error_action(s->qdev.conf.bs, is_read, error);
 
-    if (action == BLOCK_ERR_IGNORE) {
-        bdrv_emit_qmp_error_event(s->qdev.conf.bs, BDRV_ACTION_IGNORE, is_read);
-        return 0;
-    }
-
-    if ((error == ENOSPC && action == BLOCK_ERR_STOP_ENOSPC)
-            || action == BLOCK_ERR_STOP_ANY) {
-
-        bdrv_emit_qmp_error_event(s->qdev.conf.bs, BDRV_ACTION_STOP, is_read);
-        vm_stop(RUN_STATE_IO_ERROR);
-        bdrv_iostatus_set_err(s->qdev.conf.bs, error);
-        scsi_req_retry(&r->req);
-    } else {
+    if (action == BDRV_ACTION_REPORT) {
         switch (error) {
         case ENOMEDIUM:
             scsi_check_condition(r, SENSE_CODE(NO_MEDIUM));
@@ -417,9 +405,12 @@ static int scsi_handle_rw_error(SCSIDiskReq *r, int error)
             scsi_check_condition(r, SENSE_CODE(IO_ERROR));
             break;
         }
-        bdrv_emit_qmp_error_event(s->qdev.conf.bs, BDRV_ACTION_REPORT, is_read);
     }
-    return 1;
+    bdrv_error_action(s->qdev.conf.bs, action, is_read, error);
+    if (action == BDRV_ACTION_STOP) {
+        scsi_req_retry(&r->req);
+    }
+    return action != BDRV_ACTION_IGNORE;
 }
 
 static void scsi_write_complete(void * opaque, int ret)
diff --git a/hw/scsi-generic.c b/hw/scsi-generic.c
index a5eb663ecf..d9045341ba 100644
--- a/hw/scsi-generic.c
+++ b/hw/scsi-generic.c
@@ -400,11 +400,11 @@ static int scsi_generic_initfn(SCSIDevice *s)
         return -1;
     }
 
-    if (bdrv_get_on_error(s->conf.bs, 0) != BLOCK_ERR_STOP_ENOSPC) {
+    if (bdrv_get_on_error(s->conf.bs, 0) != BLOCKDEV_ON_ERROR_ENOSPC) {
         error_report("Device doesn't support drive option werror");
         return -1;
     }
-    if (bdrv_get_on_error(s->conf.bs, 1) != BLOCK_ERR_REPORT) {
+    if (bdrv_get_on_error(s->conf.bs, 1) != BLOCKDEV_ON_ERROR_REPORT) {
         error_report("Device doesn't support drive option rerror");
         return -1;
     }
diff --git a/hw/virtio-blk.c b/hw/virtio-blk.c
index 6f6d172fd0..e25cc96477 100644
--- a/hw/virtio-blk.c
+++ b/hw/virtio-blk.c
@@ -64,31 +64,22 @@ static void virtio_blk_req_complete(VirtIOBlockReq *req, int status)
 }
 
 static int virtio_blk_handle_rw_error(VirtIOBlockReq *req, int error,
-    int is_read)
+    bool is_read)
 {
-    BlockErrorAction action = bdrv_get_on_error(req->dev->bs, is_read);
+    BlockErrorAction action = bdrv_get_error_action(req->dev->bs, is_read, error);
     VirtIOBlock *s = req->dev;
 
-    if (action == BLOCK_ERR_IGNORE) {
-        bdrv_emit_qmp_error_event(s->bs, BDRV_ACTION_IGNORE, is_read);
-        return 0;
-    }
-
-    if ((error == ENOSPC && action == BLOCK_ERR_STOP_ENOSPC)
-            || action == BLOCK_ERR_STOP_ANY) {
+    if (action == BDRV_ACTION_STOP) {
         req->next = s->rq;
         s->rq = req;
-        bdrv_emit_qmp_error_event(s->bs, BDRV_ACTION_STOP, is_read);
-        vm_stop(RUN_STATE_IO_ERROR);
-        bdrv_iostatus_set_err(s->bs, error);
-    } else {
+    } else if (action == BDRV_ACTION_REPORT) {
         virtio_blk_req_complete(req, VIRTIO_BLK_S_IOERR);
         bdrv_acct_done(s->bs, &req->acct);
         g_free(req);
-        bdrv_emit_qmp_error_event(s->bs, BDRV_ACTION_REPORT, is_read);
     }
 
-    return 1;
+    bdrv_error_action(s->bs, action, is_read, error);
+    return action != BDRV_ACTION_IGNORE;
 }
 
 static void virtio_blk_rw_complete(void *opaque, int ret)
@@ -98,7 +89,7 @@ static void virtio_blk_rw_complete(void *opaque, int ret)
     trace_virtio_blk_rw_complete(req, ret);
 
     if (ret) {
-        int is_read = !(ldl_p(&req->out->type) & VIRTIO_BLK_T_OUT);
+        bool is_read = !(ldl_p(&req->out->type) & VIRTIO_BLK_T_OUT);
         if (virtio_blk_handle_rw_error(req, -ret, is_read))
             return;
     }
diff --git a/monitor.c b/monitor.c
index cd735bbbeb..a0e3ffb924 100644
--- a/monitor.c
+++ b/monitor.c
@@ -450,6 +450,7 @@ static const char *monitor_event_names[] = {
     [QEVENT_SPICE_DISCONNECTED] = "SPICE_DISCONNECTED",
     [QEVENT_BLOCK_JOB_COMPLETED] = "BLOCK_JOB_COMPLETED",
     [QEVENT_BLOCK_JOB_CANCELLED] = "BLOCK_JOB_CANCELLED",
+    [QEVENT_BLOCK_JOB_ERROR] = "BLOCK_JOB_ERROR",
     [QEVENT_DEVICE_TRAY_MOVED] = "DEVICE_TRAY_MOVED",
     [QEVENT_SUSPEND] = "SUSPEND",
     [QEVENT_SUSPEND_DISK] = "SUSPEND_DISK",
diff --git a/monitor.h b/monitor.h
index e240c3f42d..b6e7d95a30 100644
--- a/monitor.h
+++ b/monitor.h
@@ -38,6 +38,7 @@ typedef enum MonitorEvent {
     QEVENT_SPICE_DISCONNECTED,
     QEVENT_BLOCK_JOB_COMPLETED,
     QEVENT_BLOCK_JOB_CANCELLED,
+    QEVENT_BLOCK_JOB_ERROR,
     QEVENT_DEVICE_TRAY_MOVED,
     QEVENT_SUSPEND,
     QEVENT_SUSPEND_DISK,
diff --git a/qapi-schema.json b/qapi-schema.json
index f4c21855cf..23abb94a1a 100644
--- a/qapi-schema.json
+++ b/qapi-schema.json
@@ -1113,6 +1113,29 @@
 { 'command': 'query-pci', 'returns': ['PciInfo'] }
 
 ##
+# @BlockdevOnError:
+#
+# An enumeration of possible behaviors for errors on I/O operations.
+# The exact meaning depends on whether the I/O was initiated by a guest
+# or by a block job
+#
+# @report: for guest operations, report the error to the guest;
+#          for jobs, cancel the job
+#
+# @ignore: ignore the error, only report a QMP event (BLOCK_IO_ERROR
+#          or BLOCK_JOB_ERROR)
+#
+# @enospc: same as @stop on ENOSPC, same as @report otherwise.
+#
+# @stop: for guest operations, stop the virtual machine;
+#        for jobs, pause the job
+#
+# Since: 1.3
+##
+{ 'enum': 'BlockdevOnError',
+  'data': ['report', 'ignore', 'enospc', 'stop'] }
+
+##
 # @BlockJobInfo:
 #
 # Information about a long-running block device operation.
@@ -1123,15 +1146,24 @@
 #
 # @len: the maximum progress value
 #
+# @busy: false if the job is known to be in a quiescent state, with
+#        no pending I/O.  Since 1.3.
+#
+# @paused: whether the job is paused or, if @busy is true, will
+#          pause itself as soon as possible.  Since 1.3.
+#
 # @offset: the current progress value
 #
 # @speed: the rate limit, bytes per second
 #
+# @io-status: the status of the job (since 1.3)
+#
 # Since: 1.1
 ##
 { 'type': 'BlockJobInfo',
   'data': {'type': 'str', 'device': 'str', 'len': 'int',
-           'offset': 'int', 'speed': 'int'} }
+           'offset': 'int', 'busy': 'bool', 'paused': 'bool', 'speed': 'int',
+           'io-status': 'BlockDeviceIoStatus'} }
 
 ##
 # @query-block-jobs:
@@ -1493,6 +1525,40 @@
   'returns': 'str' }
 
 ##
+# @block-commit
+#
+# Live commit of data from overlay image nodes into backing nodes - i.e.,
+# writes data between 'top' and 'base' into 'base'.
+#
+# @device:  the name of the device
+#
+# @base:   #optional The file name of the backing image to write data into.
+#                    If not specified, this is the deepest backing image
+#
+# @top:              The file name of the backing image within the image chain,
+#                    which contains the topmost data to be committed down.
+#                    Note, the active layer as 'top' is currently unsupported.
+#
+#                    If top == base, that is an error.
+#
+#
+# @speed:  #optional the maximum speed, in bytes per second
+#
+# Returns: Nothing on success
+#          If commit or stream is already active on this device, DeviceInUse
+#          If @device does not exist, DeviceNotFound
+#          If image commit is not supported by this device, NotSupported
+#          If @base or @top is invalid, a generic error is returned
+#          If @top is the active layer, or omitted, a generic error is returned
+#          If @speed is invalid, InvalidParameter
+#
+# Since: 1.3
+#
+##
+{ 'command': 'block-commit',
+  'data': { 'device': 'str', '*base': 'str', 'top': 'str',
+            '*speed': 'int' } }
+
 # @migrate_cancel
 #
 # Cancel the current executing migration process.
@@ -1828,13 +1894,18 @@
 #
 # @speed:  #optional the maximum speed, in bytes per second
 #
+# @on-error: #optional the action to take on an error (default report).
+#            'stop' and 'enospc' can only be used if the block device
+#            supports io-status (see BlockInfo).  Since 1.3.
+#
 # Returns: Nothing on success
 #          If @device does not exist, DeviceNotFound
 #
 # Since: 1.1
 ##
-{ 'command': 'block-stream', 'data': { 'device': 'str', '*base': 'str',
-                                       '*speed': 'int' } }
+{ 'command': 'block-stream',
+  'data': { 'device': 'str', '*base': 'str', '*speed': 'int',
+            '*on-error': 'BlockdevOnError' } }
 
 ##
 # @block-job-set-speed:
@@ -1878,12 +1949,58 @@
 #
 # @device: the device name
 #
+# @force: #optional whether to allow cancellation of a paused job (default
+#         false).  Since 1.3.
+#
 # Returns: Nothing on success
 #          If no background operation is active on this device, DeviceNotActive
 #
 # Since: 1.1
 ##
-{ 'command': 'block-job-cancel', 'data': { 'device': 'str' } }
+{ 'command': 'block-job-cancel', 'data': { 'device': 'str', '*force': 'bool' } }
+
+##
+# @block-job-pause:
+#
+# Pause an active background block operation.
+#
+# This command returns immediately after marking the active background block
+# operation for pausing.  It is an error to call this command if no
+# operation is in progress.  Pausing an already paused job has no cumulative
+# effect; a single block-job-resume command will resume the job.
+#
+# The operation will pause as soon as possible.  No event is emitted when
+# the operation is actually paused.  Cancelling a paused job automatically
+# resumes it.
+#
+# @device: the device name
+#
+# Returns: Nothing on success
+#          If no background operation is active on this device, DeviceNotActive
+#
+# Since: 1.3
+##
+{ 'command': 'block-job-pause', 'data': { 'device': 'str' } }
+
+##
+# @block-job-resume:
+#
+# Resume an active background block operation.
+#
+# This command returns immediately after resuming a paused background block
+# operation.  It is an error to call this command if no operation is in
+# progress.  Resuming an already running job is not an error.
+#
+# This command also clears the error status of the job.
+#
+# @device: the device name
+#
+# Returns: Nothing on success
+#          If no background operation is active on this device, DeviceNotActive
+#
+# Since: 1.3
+##
+{ 'command': 'block-job-resume', 'data': { 'device': 'str' } }
 
 ##
 # @ObjectTypeInfo:
diff --git a/qemu-tool.c b/qemu-tool.c
index 18205babab..f2f98138ce 100644
--- a/qemu-tool.c
+++ b/qemu-tool.c
@@ -19,6 +19,7 @@
 #include "qemu-log.h"
 #include "migration.h"
 #include "main-loop.h"
+#include "sysemu.h"
 #include "qemu_socket.h"
 #include "slirp/libslirp.h"
 
@@ -37,6 +38,11 @@ const char *qemu_get_vm_name(void)
 
 Monitor *cur_mon;
 
+void vm_stop(RunState state)
+{
+    abort();
+}
+
 int monitor_cur_is_qmp(void)
 {
     return 0;
diff --git a/qerror.h b/qerror.h
index d0a76a4f71..c91708cc3c 100644
--- a/qerror.h
+++ b/qerror.h
@@ -48,6 +48,12 @@ void assert_no_error(Error *err);
 #define QERR_BASE_NOT_FOUND \
     ERROR_CLASS_GENERIC_ERROR, "Base '%s' not found"
 
+#define QERR_BLOCK_JOB_NOT_ACTIVE \
+    ERROR_CLASS_DEVICE_NOT_ACTIVE, "No active block job on device '%s'"
+
+#define QERR_BLOCK_JOB_PAUSED \
+    ERROR_CLASS_GENERIC_ERROR, "The block job for device '%s' is currently paused"
+
 #define QERR_BLOCK_FORMAT_FEATURE_NOT_SUPPORTED \
     ERROR_CLASS_GENERIC_ERROR, "Block format '%s' used by device '%s' does not support feature '%s'"
 
diff --git a/qmp-commands.hx b/qmp-commands.hx
index 36e08d9ffc..1aef27ca14 100644
--- a/qmp-commands.hx
+++ b/qmp-commands.hx
@@ -787,11 +787,17 @@ EQMP
 
     {
         .name       = "block-stream",
-        .args_type  = "device:B,base:s?,speed:o?",
+        .args_type  = "device:B,base:s?,speed:o?,on-error:s?",
         .mhandler.cmd_new = qmp_marshal_input_block_stream,
     },
 
     {
+        .name       = "block-commit",
+        .args_type  = "device:B,base:s?,top:s,speed:o?",
+        .mhandler.cmd_new = qmp_marshal_input_block_commit,
+    },
+
+    {
         .name       = "block-job-set-speed",
         .args_type  = "device:B,speed:o",
         .mhandler.cmd_new = qmp_marshal_input_block_job_set_speed,
@@ -799,10 +805,20 @@ EQMP
 
     {
         .name       = "block-job-cancel",
-        .args_type  = "device:B",
+        .args_type  = "device:B,force:b?",
         .mhandler.cmd_new = qmp_marshal_input_block_job_cancel,
     },
     {
+        .name       = "block-job-pause",
+        .args_type  = "device:B",
+        .mhandler.cmd_new = qmp_marshal_input_block_job_pause,
+    },
+    {
+        .name       = "block-job-resume",
+        .args_type  = "device:B",
+        .mhandler.cmd_new = qmp_marshal_input_block_job_resume,
+    },
+    {
         .name       = "transaction",
         .args_type  = "actions:q",
         .mhandler.cmd_new = qmp_marshal_input_transaction,
diff --git a/tests/qemu-iotests/030 b/tests/qemu-iotests/030
index 55b16f81dd..dd4ef11996 100755
--- a/tests/qemu-iotests/030
+++ b/tests/qemu-iotests/030
@@ -18,6 +18,7 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #
 
+import time
 import os
 import iotests
 from iotests import qemu_img, qemu_io
@@ -98,6 +99,43 @@ class TestSingleDrive(ImageStreamingTestCase):
                          qemu_io('-c', 'map', test_img),
                          'image file map does not match backing file after streaming')
 
+    def test_stream_pause(self):
+        self.assert_no_active_streams()
+
+        result = self.vm.qmp('block-stream', device='drive0')
+        self.assert_qmp(result, 'return', {})
+
+        result = self.vm.qmp('block-job-pause', device='drive0')
+        self.assert_qmp(result, 'return', {})
+
+        time.sleep(1)
+        result = self.vm.qmp('query-block-jobs')
+        offset = self.dictpath(result, 'return[0]/offset')
+
+        time.sleep(1)
+        result = self.vm.qmp('query-block-jobs')
+        self.assert_qmp(result, 'return[0]/offset', offset)
+
+        result = self.vm.qmp('block-job-resume', device='drive0')
+        self.assert_qmp(result, 'return', {})
+
+        completed = False
+        while not completed:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_COMPLETED':
+                    self.assert_qmp(event, 'data/type', 'stream')
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/offset', self.image_len)
+                    self.assert_qmp(event, 'data/len', self.image_len)
+                    completed = True
+
+        self.assert_no_active_streams()
+        self.vm.shutdown()
+
+        self.assertEqual(qemu_io('-c', 'map', backing_img),
+                         qemu_io('-c', 'map', test_img),
+                         'image file map does not match backing file after streaming')
+
     def test_stream_partial(self):
         self.assert_no_active_streams()
 
@@ -157,6 +195,226 @@ class TestSmallerBackingFile(ImageStreamingTestCase):
         self.assert_no_active_streams()
         self.vm.shutdown()
 
+class TestErrors(ImageStreamingTestCase):
+    image_len = 2 * 1024 * 1024 # MB
+
+    # this should match STREAM_BUFFER_SIZE/512 in block/stream.c
+    STREAM_BUFFER_SIZE = 512 * 1024
+
+    def create_blkdebug_file(self, name, event, errno):
+        file = open(name, 'w')
+        file.write('''
+[inject-error]
+state = "1"
+event = "%s"
+errno = "%d"
+immediately = "off"
+once = "on"
+sector = "%d"
+
+[set-state]
+state = "1"
+event = "%s"
+new_state = "2"
+
+[set-state]
+state = "2"
+event = "%s"
+new_state = "1"
+''' % (event, errno, self.STREAM_BUFFER_SIZE / 512, event, event))
+        file.close()
+
+class TestEIO(TestErrors):
+    def setUp(self):
+        self.blkdebug_file = backing_img + ".blkdebug"
+        self.create_image(backing_img, TestErrors.image_len)
+        self.create_blkdebug_file(self.blkdebug_file, "read_aio", 5)
+        qemu_img('create', '-f', iotests.imgfmt,
+                 '-o', 'backing_file=blkdebug:%s:%s,backing_fmt=raw'
+                       % (self.blkdebug_file, backing_img),
+                 test_img)
+        self.vm = iotests.VM().add_drive(test_img)
+        self.vm.launch()
+
+    def tearDown(self):
+        self.vm.shutdown()
+        os.remove(test_img)
+        os.remove(backing_img)
+        os.remove(self.blkdebug_file)
+
+    def test_report(self):
+        self.assert_no_active_streams()
+
+        result = self.vm.qmp('block-stream', device='drive0')
+        self.assert_qmp(result, 'return', {})
+
+        completed = False
+        error = False
+        while not completed:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_ERROR':
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/operation', 'read')
+                    error = True
+                elif event['event'] == 'BLOCK_JOB_COMPLETED':
+                    self.assertTrue(error, 'job completed unexpectedly')
+                    self.assert_qmp(event, 'data/type', 'stream')
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/error', 'Input/output error')
+                    self.assert_qmp(event, 'data/offset', self.STREAM_BUFFER_SIZE)
+                    self.assert_qmp(event, 'data/len', self.image_len)
+                    completed = True
+
+        self.assert_no_active_streams()
+        self.vm.shutdown()
+
+    def test_ignore(self):
+        self.assert_no_active_streams()
+
+        result = self.vm.qmp('block-stream', device='drive0', on_error='ignore')
+        self.assert_qmp(result, 'return', {})
+
+        error = False
+        completed = False
+        while not completed:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_ERROR':
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/operation', 'read')
+                    result = self.vm.qmp('query-block-jobs')
+                    self.assert_qmp(result, 'return[0]/paused', False)
+                    error = True
+                elif event['event'] == 'BLOCK_JOB_COMPLETED':
+                    self.assertTrue(error, 'job completed unexpectedly')
+                    self.assert_qmp(event, 'data/type', 'stream')
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/error', 'Input/output error')
+                    self.assert_qmp(event, 'data/offset', self.image_len)
+                    self.assert_qmp(event, 'data/len', self.image_len)
+                    completed = True
+
+        self.assert_no_active_streams()
+        self.vm.shutdown()
+
+    def test_stop(self):
+        self.assert_no_active_streams()
+
+        result = self.vm.qmp('block-stream', device='drive0', on_error='stop')
+        self.assert_qmp(result, 'return', {})
+
+        error = False
+        completed = False
+        while not completed:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_ERROR':
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/operation', 'read')
+
+                    result = self.vm.qmp('query-block-jobs')
+                    self.assert_qmp(result, 'return[0]/paused', True)
+                    self.assert_qmp(result, 'return[0]/offset', self.STREAM_BUFFER_SIZE)
+                    self.assert_qmp(result, 'return[0]/io-status', 'failed')
+
+                    result = self.vm.qmp('block-job-resume', device='drive0')
+                    self.assert_qmp(result, 'return', {})
+
+                    result = self.vm.qmp('query-block-jobs')
+                    self.assert_qmp(result, 'return[0]/paused', False)
+                    self.assert_qmp(result, 'return[0]/io-status', 'ok')
+                    error = True
+                elif event['event'] == 'BLOCK_JOB_COMPLETED':
+                    self.assertTrue(error, 'job completed unexpectedly')
+                    self.assert_qmp(event, 'data/type', 'stream')
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp_absent(event, 'data/error')
+                    self.assert_qmp(event, 'data/offset', self.image_len)
+                    self.assert_qmp(event, 'data/len', self.image_len)
+                    completed = True
+
+        self.assert_no_active_streams()
+        self.vm.shutdown()
+
+    def test_enospc(self):
+        self.assert_no_active_streams()
+
+        result = self.vm.qmp('block-stream', device='drive0', on_error='enospc')
+        self.assert_qmp(result, 'return', {})
+
+        completed = False
+        error = False
+        while not completed:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_ERROR':
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/operation', 'read')
+                    error = True
+                elif event['event'] == 'BLOCK_JOB_COMPLETED':
+                    self.assertTrue(error, 'job completed unexpectedly')
+                    self.assert_qmp(event, 'data/type', 'stream')
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/error', 'Input/output error')
+                    self.assert_qmp(event, 'data/offset', self.STREAM_BUFFER_SIZE)
+                    self.assert_qmp(event, 'data/len', self.image_len)
+                    completed = True
+
+        self.assert_no_active_streams()
+        self.vm.shutdown()
+
+class TestENOSPC(TestErrors):
+    def setUp(self):
+        self.blkdebug_file = backing_img + ".blkdebug"
+        self.create_image(backing_img, TestErrors.image_len)
+        self.create_blkdebug_file(self.blkdebug_file, "read_aio", 28)
+        qemu_img('create', '-f', iotests.imgfmt,
+                 '-o', 'backing_file=blkdebug:%s:%s,backing_fmt=raw'
+                       % (self.blkdebug_file, backing_img),
+                 test_img)
+        self.vm = iotests.VM().add_drive(test_img)
+        self.vm.launch()
+
+    def tearDown(self):
+        self.vm.shutdown()
+        os.remove(test_img)
+        os.remove(backing_img)
+        os.remove(self.blkdebug_file)
+
+    def test_enospc(self):
+        self.assert_no_active_streams()
+
+        result = self.vm.qmp('block-stream', device='drive0', on_error='enospc')
+        self.assert_qmp(result, 'return', {})
+
+        error = False
+        completed = False
+        while not completed:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_ERROR':
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/operation', 'read')
+
+                    result = self.vm.qmp('query-block-jobs')
+                    self.assert_qmp(result, 'return[0]/paused', True)
+                    self.assert_qmp(result, 'return[0]/offset', self.STREAM_BUFFER_SIZE)
+                    self.assert_qmp(result, 'return[0]/io-status', 'nospace')
+
+                    result = self.vm.qmp('block-job-resume', device='drive0')
+                    self.assert_qmp(result, 'return', {})
+
+                    result = self.vm.qmp('query-block-jobs')
+                    self.assert_qmp(result, 'return[0]/paused', False)
+                    self.assert_qmp(result, 'return[0]/io-status', 'ok')
+                    error = True
+                elif event['event'] == 'BLOCK_JOB_COMPLETED':
+                    self.assertTrue(error, 'job completed unexpectedly')
+                    self.assert_qmp(event, 'data/type', 'stream')
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp_absent(event, 'data/error')
+                    self.assert_qmp(event, 'data/offset', self.image_len)
+                    self.assert_qmp(event, 'data/len', self.image_len)
+                    completed = True
+
+        self.assert_no_active_streams()
+        self.vm.shutdown()
 
 class TestStreamStop(ImageStreamingTestCase):
     image_len = 8 * 1024 * 1024 * 1024 # GB
@@ -173,8 +431,6 @@ class TestStreamStop(ImageStreamingTestCase):
         os.remove(backing_img)
 
     def test_stream_stop(self):
-        import time
-
         self.assert_no_active_streams()
 
         result = self.vm.qmp('block-stream', device='drive0')
diff --git a/tests/qemu-iotests/030.out b/tests/qemu-iotests/030.out
index 2f7d3902f2..fa16b5ccef 100644
--- a/tests/qemu-iotests/030.out
+++ b/tests/qemu-iotests/030.out
@@ -1,5 +1,5 @@
-.......
+.............
 ----------------------------------------------------------------------
-Ran 7 tests
+Ran 13 tests
 
 OK
diff --git a/tests/qemu-iotests/040 b/tests/qemu-iotests/040
new file mode 100755
index 0000000000..258e7eae38
--- /dev/null
+++ b/tests/qemu-iotests/040
@@ -0,0 +1,178 @@
+#!/usr/bin/env python
+#
+# Tests for image block commit.
+#
+# Copyright (C) 2012 IBM, Corp.
+# Copyright (C) 2012 Red Hat, Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program 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 General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+# Test for live block commit
+# Derived from Image Streaming Test 030
+
+import time
+import os
+import iotests
+from iotests import qemu_img, qemu_io
+import struct
+
+backing_img = os.path.join(iotests.test_dir, 'backing.img')
+mid_img = os.path.join(iotests.test_dir, 'mid.img')
+test_img = os.path.join(iotests.test_dir, 'test.img')
+
+class ImageCommitTestCase(iotests.QMPTestCase):
+    '''Abstract base class for image commit test cases'''
+
+    def assert_no_active_commit(self):
+        result = self.vm.qmp('query-block-jobs')
+        self.assert_qmp(result, 'return', [])
+
+    def cancel_and_wait(self, drive='drive0'):
+        '''Cancel a block job and wait for it to finish'''
+        result = self.vm.qmp('block-job-cancel', device=drive)
+        self.assert_qmp(result, 'return', {})
+
+        cancelled = False
+        while not cancelled:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_CANCELLED':
+                    self.assert_qmp(event, 'data/type', 'commit')
+                    self.assert_qmp(event, 'data/device', drive)
+                    cancelled = True
+
+        self.assert_no_active_commit()
+
+    def create_image(self, name, size):
+        file = open(name, 'w')
+        i = 0
+        while i < size:
+            sector = struct.pack('>l504xl', i / 512, i / 512)
+            file.write(sector)
+            i = i + 512
+        file.close()
+
+
+class TestSingleDrive(ImageCommitTestCase):
+    image_len = 1 * 1024 * 1024
+    test_len = 1 * 1024 * 256
+
+    def setUp(self):
+        self.create_image(backing_img, TestSingleDrive.image_len)
+        qemu_img('create', '-f', iotests.imgfmt, '-o', 'backing_file=%s' % backing_img, mid_img)
+        qemu_img('create', '-f', iotests.imgfmt, '-o', 'backing_file=%s' % mid_img, test_img)
+        qemu_io('-c', 'write -P 0xab 0 524288', backing_img)
+        qemu_io('-c', 'write -P 0xef 524288 524288', mid_img)
+        self.vm = iotests.VM().add_drive(test_img)
+        self.vm.launch()
+
+    def tearDown(self):
+        self.vm.shutdown()
+        os.remove(test_img)
+        os.remove(mid_img)
+        os.remove(backing_img)
+
+    def test_commit(self):
+        self.assert_no_active_commit()
+        result = self.vm.qmp('block-commit', device='drive0', top='%s' % mid_img)
+        self.assert_qmp(result, 'return', {})
+
+        completed = False
+        while not completed:
+            for event in self.vm.get_qmp_events(wait=True):
+                if event['event'] == 'BLOCK_JOB_COMPLETED':
+                    self.assert_qmp(event, 'data/type', 'commit')
+                    self.assert_qmp(event, 'data/device', 'drive0')
+                    self.assert_qmp(event, 'data/offset', self.image_len)
+                    self.assert_qmp(event, 'data/len', self.image_len)
+                    completed = True
+
+        self.assert_no_active_commit()
+        self.vm.shutdown()
+
+        self.assertEqual(-1, qemu_io('-c', 'read -P 0xab 0 524288', backing_img).find("verification failed"))
+        self.assertEqual(-1, qemu_io('-c', 'read -P 0xef 524288 524288', backing_img).find("verification failed"))
+
+    def test_device_not_found(self):
+        result = self.vm.qmp('block-commit', device='nonexistent', top='%s' % mid_img)
+        self.assert_qmp(result, 'error/class', 'DeviceNotFound')
+
+    def test_top_same_base(self):
+        self.assert_no_active_commit()
+        result = self.vm.qmp('block-commit', device='drive0', top='%s' % backing_img, base='%s' % backing_img)
+        self.assert_qmp(result, 'error/class', 'GenericError')
+        self.assert_qmp(result, 'error/desc', 'Invalid files for merge: top and base are the same')
+
+    def test_top_invalid(self):
+        self.assert_no_active_commit()
+        result = self.vm.qmp('block-commit', device='drive0', top='badfile', base='%s' % backing_img)
+        self.assert_qmp(result, 'error/class', 'GenericError')
+        self.assert_qmp(result, 'error/desc', 'Top image file badfile not found')
+
+    def test_base_invalid(self):
+        self.assert_no_active_commit()
+        result = self.vm.qmp('block-commit', device='drive0', top='%s' % mid_img, base='badfile')
+        self.assert_qmp(result, 'error/class', 'GenericError')
+        self.assert_qmp(result, 'error/desc', 'Base \'badfile\' not found')
+
+    def test_top_is_active(self):
+        self.assert_no_active_commit()
+        result = self.vm.qmp('block-commit', device='drive0', top='%s' % test_img, base='%s' % backing_img)
+        self.assert_qmp(result, 'error/class', 'GenericError')
+        self.assert_qmp(result, 'error/desc', 'Top image as the active layer is currently unsupported')
+
+    def test_top_and_base_reversed(self):
+        self.assert_no_active_commit()
+        result = self.vm.qmp('block-commit', device='drive0', top='%s' % backing_img, base='%s' % mid_img)
+        self.assert_qmp(result, 'error/class', 'GenericError')
+        self.assert_qmp(result, 'error/desc', 'Base (%(1)s) is not reachable from top (%(2)s)' % {"1" : mid_img, "2" : backing_img})
+
+    def test_top_omitted(self):
+        self.assert_no_active_commit()
+        result = self.vm.qmp('block-commit', device='drive0')
+        self.assert_qmp(result, 'error/class', 'GenericError')
+        self.assert_qmp(result, 'error/desc', "Parameter 'top' is missing")
+
+
+class TestSetSpeed(ImageCommitTestCase):
+    image_len = 80 * 1024 * 1024 # MB
+
+    def setUp(self):
+        qemu_img('create', backing_img, str(TestSetSpeed.image_len))
+        qemu_img('create', '-f', iotests.imgfmt, '-o', 'backing_file=%s' % backing_img, mid_img)
+        qemu_img('create', '-f', iotests.imgfmt, '-o', 'backing_file=%s' % mid_img, test_img)
+        self.vm = iotests.VM().add_drive(test_img)
+        self.vm.launch()
+
+    def tearDown(self):
+        self.vm.shutdown()
+        os.remove(test_img)
+        os.remove(mid_img)
+        os.remove(backing_img)
+
+    def test_set_speed(self):
+        self.assert_no_active_commit()
+
+        result = self.vm.qmp('block-commit', device='drive0', top=mid_img, speed=1024 * 1024)
+        self.assert_qmp(result, 'return', {})
+
+        # Ensure the speed we set was accepted
+        result = self.vm.qmp('query-block-jobs')
+        self.assert_qmp(result, 'return[0]/device', 'drive0')
+        self.assert_qmp(result, 'return[0]/speed', 1024 * 1024)
+
+        self.cancel_and_wait()
+
+
+if __name__ == '__main__':
+    iotests.main(supported_fmts=['qcow2', 'qed'])
diff --git a/tests/qemu-iotests/040.out b/tests/qemu-iotests/040.out
new file mode 100644
index 0000000000..dae404e278
--- /dev/null
+++ b/tests/qemu-iotests/040.out
@@ -0,0 +1,5 @@
+.........
+----------------------------------------------------------------------
+Ran 9 tests
+
+OK
diff --git a/tests/qemu-iotests/group b/tests/qemu-iotests/group
index ebb5ca4b41..66d2ba9689 100644
--- a/tests/qemu-iotests/group
+++ b/tests/qemu-iotests/group
@@ -36,7 +36,7 @@
 027 rw auto quick
 028 rw backing auto
 029 rw auto quick
-030 rw auto
+030 rw auto backing
 031 rw auto quick
 032 rw auto
 033 rw auto
@@ -46,3 +46,4 @@
 037 rw auto backing
 038 rw auto backing
 039 rw auto
+040 rw auto
diff --git a/tests/qemu-iotests/iotests.py b/tests/qemu-iotests/iotests.py
index e05b1d640b..3c60b2d163 100644
--- a/tests/qemu-iotests/iotests.py
+++ b/tests/qemu-iotests/iotests.py
@@ -19,6 +19,7 @@
 import os
 import re
 import subprocess
+import string
 import unittest
 import sys; sys.path.append(os.path.join(os.path.dirname(__file__), '..', '..', 'QMP'))
 import qmp
@@ -96,9 +97,14 @@ class VM(object):
             os.remove(self._qemu_log_path)
             self._popen = None
 
+    underscore_to_dash = string.maketrans('_', '-')
     def qmp(self, cmd, **args):
         '''Invoke a QMP command and return the result dict'''
-        return self._qmp.cmd(cmd, args=args)
+        qmp_args = dict()
+        for k in args.keys():
+            qmp_args[k.translate(self.underscore_to_dash)] = args[k]
+
+        return self._qmp.cmd(cmd, args=qmp_args)
 
     def get_qmp_events(self, wait=False):
         '''Poll for queued QMP events and return a list of dicts'''
@@ -132,6 +138,13 @@ class QMPTestCase(unittest.TestCase):
                     self.fail('invalid index "%s" in path "%s" in "%s"' % (idx, path, str(d)))
         return d
 
+    def assert_qmp_absent(self, d, path):
+        try:
+            result = self.dictpath(d, path)
+        except AssertionError:
+            return
+        self.fail('path "%s" has value "%s"' % (path, str(result)))
+
     def assert_qmp(self, d, path, value):
         '''Assert that the value for a specific path in a QMP dict matches'''
         result = self.dictpath(d, path)
diff --git a/trace-events b/trace-events
index f5b5097552..42b66f19f4 100644
--- a/trace-events
+++ b/trace-events
@@ -74,10 +74,14 @@ bdrv_co_do_copy_on_readv(void *bs, int64_t sector_num, int nb_sectors, int64_t c
 # block/stream.c
 stream_one_iteration(void *s, int64_t sector_num, int nb_sectors, int is_allocated) "s %p sector_num %"PRId64" nb_sectors %d is_allocated %d"
 stream_start(void *bs, void *base, void *s, void *co, void *opaque) "bs %p base %p s %p co %p opaque %p"
+commit_one_iteration(void *s, int64_t sector_num, int nb_sectors, int is_allocated) "s %p sector_num %"PRId64" nb_sectors %d is_allocated %d"
+commit_start(void *bs, void *base, void *top, void *s, void *co, void *opaque) "bs %p base %p top %p s %p co %p opaque %p"
 
 # blockdev.c
 qmp_block_job_cancel(void *job) "job %p"
-block_stream_cb(void *bs, void *job, int ret) "bs %p job %p ret %d"
+qmp_block_job_pause(void *job) "job %p"
+qmp_block_job_resume(void *job) "job %p"
+block_job_cb(void *bs, void *job, int ret) "bs %p job %p ret %d"
 qmp_block_stream(void *bs, void *job) "bs %p job %p"
 
 # hw/virtio-blk.c
diff --git a/uri.c b/uri.c
new file mode 100644
index 0000000000..dd922de339
--- /dev/null
+++ b/uri.c
@@ -0,0 +1,2249 @@
+/**
+ * uri.c: set of generic URI related routines
+ *
+ * Reference: RFCs 3986, 2732 and 2373
+ *
+ * Copyright (C) 1998-2003 Daniel Veillard.  All Rights Reserved.
+ *
+ * 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
+ * DANIEL VEILLARD 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.
+ *
+ * Except as contained in this notice, the name of Daniel Veillard shall not
+ * be used in advertising or otherwise to promote the sale, use or other
+ * dealings in this Software without prior written authorization from him.
+ *
+ * daniel@veillard.com
+ *
+ **
+ *
+ * Copyright (C) 2007, 2009-2010 Red Hat, Inc.
+ *
+ * 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 2.1 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
+ *
+ * Authors:
+ *    Richard W.M. Jones <rjones@redhat.com>
+ *
+ */
+
+#include <glib.h>
+#include <string.h>
+#include <stdio.h>
+
+#include "uri.h"
+
+static void uri_clean(URI *uri);
+
+/*
+ * Old rule from 2396 used in legacy handling code
+ * alpha    = lowalpha | upalpha
+ */
+#define IS_ALPHA(x) (IS_LOWALPHA(x) || IS_UPALPHA(x))
+
+
+/*
+ * lowalpha = "a" | "b" | "c" | "d" | "e" | "f" | "g" | "h" | "i" | "j" |
+ *            "k" | "l" | "m" | "n" | "o" | "p" | "q" | "r" | "s" | "t" |
+ *            "u" | "v" | "w" | "x" | "y" | "z"
+ */
+
+#define IS_LOWALPHA(x) (((x) >= 'a') && ((x) <= 'z'))
+
+/*
+ * upalpha = "A" | "B" | "C" | "D" | "E" | "F" | "G" | "H" | "I" | "J" |
+ *           "K" | "L" | "M" | "N" | "O" | "P" | "Q" | "R" | "S" | "T" |
+ *           "U" | "V" | "W" | "X" | "Y" | "Z"
+ */
+#define IS_UPALPHA(x) (((x) >= 'A') && ((x) <= 'Z'))
+
+#ifdef IS_DIGIT
+#undef IS_DIGIT
+#endif
+/*
+ * digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9"
+ */
+#define IS_DIGIT(x) (((x) >= '0') && ((x) <= '9'))
+
+/*
+ * alphanum = alpha | digit
+ */
+
+#define IS_ALPHANUM(x) (IS_ALPHA(x) || IS_DIGIT(x))
+
+/*
+ * mark = "-" | "_" | "." | "!" | "~" | "*" | "'" | "(" | ")"
+ */
+
+#define IS_MARK(x) (((x) == '-') || ((x) == '_') || ((x) == '.') ||     \
+    ((x) == '!') || ((x) == '~') || ((x) == '*') || ((x) == '\'') ||    \
+    ((x) == '(') || ((x) == ')'))
+
+/*
+ * unwise = "{" | "}" | "|" | "\" | "^" | "`"
+ */
+
+#define IS_UNWISE(p)                                                    \
+      (((*(p) == '{')) || ((*(p) == '}')) || ((*(p) == '|')) ||         \
+       ((*(p) == '\\')) || ((*(p) == '^')) || ((*(p) == '[')) ||        \
+       ((*(p) == ']')) || ((*(p) == '`')))
+/*
+ * reserved = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" | "$" | "," |
+ *            "[" | "]"
+ */
+
+#define IS_RESERVED(x) (((x) == ';') || ((x) == '/') || ((x) == '?') || \
+        ((x) == ':') || ((x) == '@') || ((x) == '&') || ((x) == '=') || \
+        ((x) == '+') || ((x) == '$') || ((x) == ',') || ((x) == '[') || \
+        ((x) == ']'))
+
+/*
+ * unreserved = alphanum | mark
+ */
+
+#define IS_UNRESERVED(x) (IS_ALPHANUM(x) || IS_MARK(x))
+
+/*
+ * Skip to next pointer char, handle escaped sequences
+ */
+
+#define NEXT(p) ((*p == '%')? p += 3 : p++)
+
+/*
+ * Productions from the spec.
+ *
+ *    authority     = server | reg_name
+ *    reg_name      = 1*( unreserved | escaped | "$" | "," |
+ *                        ";" | ":" | "@" | "&" | "=" | "+" )
+ *
+ * path          = [ abs_path | opaque_part ]
+ */
+
+
+/************************************************************************
+ *									*
+ *                         RFC 3986 parser				*
+ *									*
+ ************************************************************************/
+
+#define ISA_DIGIT(p) ((*(p) >= '0') && (*(p) <= '9'))
+#define ISA_ALPHA(p) (((*(p) >= 'a') && (*(p) <= 'z')) ||		\
+                      ((*(p) >= 'A') && (*(p) <= 'Z')))
+#define ISA_HEXDIG(p)							\
+       (ISA_DIGIT(p) || ((*(p) >= 'a') && (*(p) <= 'f')) ||		\
+        ((*(p) >= 'A') && (*(p) <= 'F')))
+
+/*
+ *    sub-delims    = "!" / "$" / "&" / "'" / "(" / ")"
+ *                     / "*" / "+" / "," / ";" / "="
+ */
+#define ISA_SUB_DELIM(p)						\
+      (((*(p) == '!')) || ((*(p) == '$')) || ((*(p) == '&')) ||		\
+       ((*(p) == '(')) || ((*(p) == ')')) || ((*(p) == '*')) ||		\
+       ((*(p) == '+')) || ((*(p) == ',')) || ((*(p) == ';')) ||		\
+       ((*(p) == '=')) || ((*(p) == '\'')))
+
+/*
+ *    gen-delims    = ":" / "/" / "?" / "#" / "[" / "]" / "@"
+ */
+#define ISA_GEN_DELIM(p)						\
+      (((*(p) == ':')) || ((*(p) == '/')) || ((*(p) == '?')) ||         \
+       ((*(p) == '#')) || ((*(p) == '[')) || ((*(p) == ']')) ||         \
+       ((*(p) == '@')))
+
+/*
+ *    reserved      = gen-delims / sub-delims
+ */
+#define ISA_RESERVED(p) (ISA_GEN_DELIM(p) || (ISA_SUB_DELIM(p)))
+
+/*
+ *    unreserved    = ALPHA / DIGIT / "-" / "." / "_" / "~"
+ */
+#define ISA_UNRESERVED(p)						\
+      ((ISA_ALPHA(p)) || (ISA_DIGIT(p)) || ((*(p) == '-')) ||		\
+       ((*(p) == '.')) || ((*(p) == '_')) || ((*(p) == '~')))
+
+/*
+ *    pct-encoded   = "%" HEXDIG HEXDIG
+ */
+#define ISA_PCT_ENCODED(p)						\
+     ((*(p) == '%') && (ISA_HEXDIG(p + 1)) && (ISA_HEXDIG(p + 2)))
+
+/*
+ *    pchar         = unreserved / pct-encoded / sub-delims / ":" / "@"
+ */
+#define ISA_PCHAR(p)							\
+     (ISA_UNRESERVED(p) || ISA_PCT_ENCODED(p) || ISA_SUB_DELIM(p) ||	\
+      ((*(p) == ':')) || ((*(p) == '@')))
+
+/**
+ * rfc3986_parse_scheme:
+ * @uri:  pointer to an URI structure
+ * @str:  pointer to the string to analyze
+ *
+ * Parse an URI scheme
+ *
+ * ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_scheme(URI *uri, const char **str) {
+    const char *cur;
+
+    if (str == NULL)
+	return(-1);
+
+    cur = *str;
+    if (!ISA_ALPHA(cur))
+	return(2);
+    cur++;
+    while (ISA_ALPHA(cur) || ISA_DIGIT(cur) ||
+           (*cur == '+') || (*cur == '-') || (*cur == '.')) cur++;
+    if (uri != NULL) {
+	if (uri->scheme != NULL) g_free(uri->scheme);
+	uri->scheme = g_strndup(*str, cur - *str);
+    }
+    *str = cur;
+    return(0);
+}
+
+/**
+ * rfc3986_parse_fragment:
+ * @uri:  pointer to an URI structure
+ * @str:  pointer to the string to analyze
+ *
+ * Parse the query part of an URI
+ *
+ * fragment      = *( pchar / "/" / "?" )
+ * NOTE: the strict syntax as defined by 3986 does not allow '[' and ']'
+ *       in the fragment identifier but this is used very broadly for
+ *       xpointer scheme selection, so we are allowing it here to not break
+ *       for example all the DocBook processing chains.
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_fragment(URI *uri, const char **str)
+{
+    const char *cur;
+
+    if (str == NULL)
+        return (-1);
+
+    cur = *str;
+
+    while ((ISA_PCHAR(cur)) || (*cur == '/') || (*cur == '?') ||
+           (*cur == '[') || (*cur == ']') ||
+           ((uri != NULL) && (uri->cleanup & 1) && (IS_UNWISE(cur))))
+        NEXT(cur);
+    if (uri != NULL) {
+        if (uri->fragment != NULL)
+            g_free(uri->fragment);
+	if (uri->cleanup & 2)
+	    uri->fragment = g_strndup(*str, cur - *str);
+	else
+	    uri->fragment = uri_string_unescape(*str, cur - *str, NULL);
+    }
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_query:
+ * @uri:  pointer to an URI structure
+ * @str:  pointer to the string to analyze
+ *
+ * Parse the query part of an URI
+ *
+ * query = *uric
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_query(URI *uri, const char **str)
+{
+    const char *cur;
+
+    if (str == NULL)
+        return (-1);
+
+    cur = *str;
+
+    while ((ISA_PCHAR(cur)) || (*cur == '/') || (*cur == '?') ||
+           ((uri != NULL) && (uri->cleanup & 1) && (IS_UNWISE(cur))))
+        NEXT(cur);
+    if (uri != NULL) {
+	if (uri->query != NULL)
+	    g_free (uri->query);
+	uri->query = g_strndup (*str, cur - *str);
+    }
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_port:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse a port  part and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * port          = *DIGIT
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_port(URI *uri, const char **str)
+{
+    const char *cur = *str;
+
+    if (ISA_DIGIT(cur)) {
+	if (uri != NULL)
+	    uri->port = 0;
+	while (ISA_DIGIT(cur)) {
+	    if (uri != NULL)
+		uri->port = uri->port * 10 + (*cur - '0');
+	    cur++;
+	}
+	*str = cur;
+	return(0);
+    }
+    return(1);
+}
+
+/**
+ * rfc3986_parse_user_info:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an user informations part and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * userinfo      = *( unreserved / pct-encoded / sub-delims / ":" )
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_user_info(URI *uri, const char **str)
+{
+    const char *cur;
+
+    cur = *str;
+    while (ISA_UNRESERVED(cur) || ISA_PCT_ENCODED(cur) ||
+           ISA_SUB_DELIM(cur) || (*cur == ':'))
+	NEXT(cur);
+    if (*cur == '@') {
+	if (uri != NULL) {
+	    if (uri->user != NULL) g_free(uri->user);
+	    if (uri->cleanup & 2)
+		uri->user = g_strndup(*str, cur - *str);
+	    else
+		uri->user = uri_string_unescape(*str, cur - *str, NULL);
+	}
+	*str = cur;
+	return(0);
+    }
+    return(1);
+}
+
+/**
+ * rfc3986_parse_dec_octet:
+ * @str:  the string to analyze
+ *
+ *    dec-octet     = DIGIT                 ; 0-9
+ *                  / %x31-39 DIGIT         ; 10-99
+ *                  / "1" 2DIGIT            ; 100-199
+ *                  / "2" %x30-34 DIGIT     ; 200-249
+ *                  / "25" %x30-35          ; 250-255
+ *
+ * Skip a dec-octet.
+ *
+ * Returns 0 if found and skipped, 1 otherwise
+ */
+static int
+rfc3986_parse_dec_octet(const char **str) {
+    const char *cur = *str;
+
+    if (!(ISA_DIGIT(cur)))
+        return(1);
+    if (!ISA_DIGIT(cur+1))
+	cur++;
+    else if ((*cur != '0') && (ISA_DIGIT(cur + 1)) && (!ISA_DIGIT(cur+2)))
+	cur += 2;
+    else if ((*cur == '1') && (ISA_DIGIT(cur + 1)) && (ISA_DIGIT(cur + 2)))
+	cur += 3;
+    else if ((*cur == '2') && (*(cur + 1) >= '0') &&
+	     (*(cur + 1) <= '4') && (ISA_DIGIT(cur + 2)))
+	cur += 3;
+    else if ((*cur == '2') && (*(cur + 1) == '5') &&
+	     (*(cur + 2) >= '0') && (*(cur + 1) <= '5'))
+	cur += 3;
+    else
+        return(1);
+    *str = cur;
+    return(0);
+}
+/**
+ * rfc3986_parse_host:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an host part and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * host          = IP-literal / IPv4address / reg-name
+ * IP-literal    = "[" ( IPv6address / IPvFuture  ) "]"
+ * IPv4address   = dec-octet "." dec-octet "." dec-octet "." dec-octet
+ * reg-name      = *( unreserved / pct-encoded / sub-delims )
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_host(URI *uri, const char **str)
+{
+    const char *cur = *str;
+    const char *host;
+
+    host = cur;
+    /*
+     * IPv6 and future adressing scheme are enclosed between brackets
+     */
+    if (*cur == '[') {
+        cur++;
+	while ((*cur != ']') && (*cur != 0))
+	    cur++;
+	if (*cur != ']')
+	    return(1);
+	cur++;
+	goto found;
+    }
+    /*
+     * try to parse an IPv4
+     */
+    if (ISA_DIGIT(cur)) {
+        if (rfc3986_parse_dec_octet(&cur) != 0)
+	    goto not_ipv4;
+	if (*cur != '.')
+	    goto not_ipv4;
+	cur++;
+        if (rfc3986_parse_dec_octet(&cur) != 0)
+	    goto not_ipv4;
+	if (*cur != '.')
+	    goto not_ipv4;
+        if (rfc3986_parse_dec_octet(&cur) != 0)
+	    goto not_ipv4;
+	if (*cur != '.')
+	    goto not_ipv4;
+        if (rfc3986_parse_dec_octet(&cur) != 0)
+	    goto not_ipv4;
+	goto found;
+not_ipv4:
+        cur = *str;
+    }
+    /*
+     * then this should be a hostname which can be empty
+     */
+    while (ISA_UNRESERVED(cur) || ISA_PCT_ENCODED(cur) || ISA_SUB_DELIM(cur))
+        NEXT(cur);
+found:
+    if (uri != NULL) {
+	if (uri->authority != NULL) g_free(uri->authority);
+	uri->authority = NULL;
+	if (uri->server != NULL) g_free(uri->server);
+	if (cur != host) {
+	    if (uri->cleanup & 2)
+		uri->server = g_strndup(host, cur - host);
+	    else
+		uri->server = uri_string_unescape(host, cur - host, NULL);
+	} else
+	    uri->server = NULL;
+    }
+    *str = cur;
+    return(0);
+}
+
+/**
+ * rfc3986_parse_authority:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an authority part and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * authority     = [ userinfo "@" ] host [ ":" port ]
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_authority(URI *uri, const char **str)
+{
+    const char *cur;
+    int ret;
+
+    cur = *str;
+    /*
+     * try to parse an userinfo and check for the trailing @
+     */
+    ret = rfc3986_parse_user_info(uri, &cur);
+    if ((ret != 0) || (*cur != '@'))
+        cur = *str;
+    else
+        cur++;
+    ret = rfc3986_parse_host(uri, &cur);
+    if (ret != 0) return(ret);
+    if (*cur == ':') {
+        cur++;
+        ret = rfc3986_parse_port(uri, &cur);
+	if (ret != 0) return(ret);
+    }
+    *str = cur;
+    return(0);
+}
+
+/**
+ * rfc3986_parse_segment:
+ * @str:  the string to analyze
+ * @forbid: an optional forbidden character
+ * @empty: allow an empty segment
+ *
+ * Parse a segment and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * segment       = *pchar
+ * segment-nz    = 1*pchar
+ * segment-nz-nc = 1*( unreserved / pct-encoded / sub-delims / "@" )
+ *               ; non-zero-length segment without any colon ":"
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_segment(const char **str, char forbid, int empty)
+{
+    const char *cur;
+
+    cur = *str;
+    if (!ISA_PCHAR(cur)) {
+        if (empty)
+	    return(0);
+	return(1);
+    }
+    while (ISA_PCHAR(cur) && (*cur != forbid))
+        NEXT(cur);
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_path_ab_empty:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an path absolute or empty and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * path-abempty  = *( "/" segment )
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_path_ab_empty(URI *uri, const char **str)
+{
+    const char *cur;
+    int ret;
+
+    cur = *str;
+
+    while (*cur == '/') {
+        cur++;
+	ret = rfc3986_parse_segment(&cur, 0, 1);
+	if (ret != 0) return(ret);
+    }
+    if (uri != NULL) {
+	if (uri->path != NULL) g_free(uri->path);
+        if (*str != cur) {
+            if (uri->cleanup & 2)
+                uri->path = g_strndup(*str, cur - *str);
+            else
+                uri->path = uri_string_unescape(*str, cur - *str, NULL);
+        } else {
+            uri->path = NULL;
+        }
+    }
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_path_absolute:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an path absolute and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * path-absolute = "/" [ segment-nz *( "/" segment ) ]
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_path_absolute(URI *uri, const char **str)
+{
+    const char *cur;
+    int ret;
+
+    cur = *str;
+
+    if (*cur != '/')
+        return(1);
+    cur++;
+    ret = rfc3986_parse_segment(&cur, 0, 0);
+    if (ret == 0) {
+	while (*cur == '/') {
+	    cur++;
+	    ret = rfc3986_parse_segment(&cur, 0, 1);
+	    if (ret != 0) return(ret);
+	}
+    }
+    if (uri != NULL) {
+	if (uri->path != NULL) g_free(uri->path);
+        if (cur != *str) {
+            if (uri->cleanup & 2)
+                uri->path = g_strndup(*str, cur - *str);
+            else
+                uri->path = uri_string_unescape(*str, cur - *str, NULL);
+        } else {
+            uri->path = NULL;
+        }
+    }
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_path_rootless:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an path without root and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * path-rootless = segment-nz *( "/" segment )
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_path_rootless(URI *uri, const char **str)
+{
+    const char *cur;
+    int ret;
+
+    cur = *str;
+
+    ret = rfc3986_parse_segment(&cur, 0, 0);
+    if (ret != 0) return(ret);
+    while (*cur == '/') {
+        cur++;
+	ret = rfc3986_parse_segment(&cur, 0, 1);
+	if (ret != 0) return(ret);
+    }
+    if (uri != NULL) {
+	if (uri->path != NULL) g_free(uri->path);
+        if (cur != *str) {
+            if (uri->cleanup & 2)
+                uri->path = g_strndup(*str, cur - *str);
+            else
+                uri->path = uri_string_unescape(*str, cur - *str, NULL);
+        } else {
+            uri->path = NULL;
+        }
+    }
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_path_no_scheme:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an path which is not a scheme and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * path-noscheme = segment-nz-nc *( "/" segment )
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_path_no_scheme(URI *uri, const char **str)
+{
+    const char *cur;
+    int ret;
+
+    cur = *str;
+
+    ret = rfc3986_parse_segment(&cur, ':', 0);
+    if (ret != 0) return(ret);
+    while (*cur == '/') {
+        cur++;
+	ret = rfc3986_parse_segment(&cur, 0, 1);
+	if (ret != 0) return(ret);
+    }
+    if (uri != NULL) {
+	if (uri->path != NULL) g_free(uri->path);
+        if (cur != *str) {
+            if (uri->cleanup & 2)
+                uri->path = g_strndup(*str, cur - *str);
+            else
+                uri->path = uri_string_unescape(*str, cur - *str, NULL);
+        } else {
+            uri->path = NULL;
+        }
+    }
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_hier_part:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an hierarchical part and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * hier-part     = "//" authority path-abempty
+ *                / path-absolute
+ *                / path-rootless
+ *                / path-empty
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_hier_part(URI *uri, const char **str)
+{
+    const char *cur;
+    int ret;
+
+    cur = *str;
+
+    if ((*cur == '/') && (*(cur + 1) == '/')) {
+        cur += 2;
+	ret = rfc3986_parse_authority(uri, &cur);
+	if (ret != 0) return(ret);
+	ret = rfc3986_parse_path_ab_empty(uri, &cur);
+	if (ret != 0) return(ret);
+	*str = cur;
+	return(0);
+    } else if (*cur == '/') {
+        ret = rfc3986_parse_path_absolute(uri, &cur);
+	if (ret != 0) return(ret);
+    } else if (ISA_PCHAR(cur)) {
+        ret = rfc3986_parse_path_rootless(uri, &cur);
+	if (ret != 0) return(ret);
+    } else {
+	/* path-empty is effectively empty */
+	if (uri != NULL) {
+	    if (uri->path != NULL) g_free(uri->path);
+	    uri->path = NULL;
+	}
+    }
+    *str = cur;
+    return (0);
+}
+
+/**
+ * rfc3986_parse_relative_ref:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an URI string and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * relative-ref  = relative-part [ "?" query ] [ "#" fragment ]
+ * relative-part = "//" authority path-abempty
+ *               / path-absolute
+ *               / path-noscheme
+ *               / path-empty
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_relative_ref(URI *uri, const char *str) {
+    int ret;
+
+    if ((*str == '/') && (*(str + 1) == '/')) {
+        str += 2;
+	ret = rfc3986_parse_authority(uri, &str);
+	if (ret != 0) return(ret);
+	ret = rfc3986_parse_path_ab_empty(uri, &str);
+	if (ret != 0) return(ret);
+    } else if (*str == '/') {
+	ret = rfc3986_parse_path_absolute(uri, &str);
+	if (ret != 0) return(ret);
+    } else if (ISA_PCHAR(str)) {
+        ret = rfc3986_parse_path_no_scheme(uri, &str);
+	if (ret != 0) return(ret);
+    } else {
+	/* path-empty is effectively empty */
+	if (uri != NULL) {
+	    if (uri->path != NULL) g_free(uri->path);
+	    uri->path = NULL;
+	}
+    }
+
+    if (*str == '?') {
+	str++;
+	ret = rfc3986_parse_query(uri, &str);
+	if (ret != 0) return(ret);
+    }
+    if (*str == '#') {
+	str++;
+	ret = rfc3986_parse_fragment(uri, &str);
+	if (ret != 0) return(ret);
+    }
+    if (*str != 0) {
+	uri_clean(uri);
+	return(1);
+    }
+    return(0);
+}
+
+
+/**
+ * rfc3986_parse:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an URI string and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * scheme ":" hier-part [ "?" query ] [ "#" fragment ]
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse(URI *uri, const char *str) {
+    int ret;
+
+    ret = rfc3986_parse_scheme(uri, &str);
+    if (ret != 0) return(ret);
+    if (*str != ':') {
+	return(1);
+    }
+    str++;
+    ret = rfc3986_parse_hier_part(uri, &str);
+    if (ret != 0) return(ret);
+    if (*str == '?') {
+	str++;
+	ret = rfc3986_parse_query(uri, &str);
+	if (ret != 0) return(ret);
+    }
+    if (*str == '#') {
+	str++;
+	ret = rfc3986_parse_fragment(uri, &str);
+	if (ret != 0) return(ret);
+    }
+    if (*str != 0) {
+	uri_clean(uri);
+	return(1);
+    }
+    return(0);
+}
+
+/**
+ * rfc3986_parse_uri_reference:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an URI reference string and fills in the appropriate fields
+ * of the @uri structure
+ *
+ * URI-reference = URI / relative-ref
+ *
+ * Returns 0 or the error code
+ */
+static int
+rfc3986_parse_uri_reference(URI *uri, const char *str) {
+    int ret;
+
+    if (str == NULL)
+	return(-1);
+    uri_clean(uri);
+
+    /*
+     * Try first to parse absolute refs, then fallback to relative if
+     * it fails.
+     */
+    ret = rfc3986_parse(uri, str);
+    if (ret != 0) {
+	uri_clean(uri);
+        ret = rfc3986_parse_relative_ref(uri, str);
+	if (ret != 0) {
+	    uri_clean(uri);
+	    return(ret);
+	}
+    }
+    return(0);
+}
+
+/**
+ * uri_parse:
+ * @str:  the URI string to analyze
+ *
+ * Parse an URI based on RFC 3986
+ *
+ * URI-reference = [ absoluteURI | relativeURI ] [ "#" fragment ]
+ *
+ * Returns a newly built URI or NULL in case of error
+ */
+URI *
+uri_parse(const char *str) {
+    URI *uri;
+    int ret;
+
+    if (str == NULL)
+	return(NULL);
+    uri = uri_new();
+    if (uri != NULL) {
+	ret = rfc3986_parse_uri_reference(uri, str);
+        if (ret) {
+	    uri_free(uri);
+	    return(NULL);
+	}
+    }
+    return(uri);
+}
+
+/**
+ * uri_parse_into:
+ * @uri:  pointer to an URI structure
+ * @str:  the string to analyze
+ *
+ * Parse an URI reference string based on RFC 3986 and fills in the
+ * appropriate fields of the @uri structure
+ *
+ * URI-reference = URI / relative-ref
+ *
+ * Returns 0 or the error code
+ */
+int
+uri_parse_into(URI *uri, const char *str) {
+    return(rfc3986_parse_uri_reference(uri, str));
+}
+
+/**
+ * uri_parse_raw:
+ * @str:  the URI string to analyze
+ * @raw:  if 1 unescaping of URI pieces are disabled
+ *
+ * Parse an URI but allows to keep intact the original fragments.
+ *
+ * URI-reference = URI / relative-ref
+ *
+ * Returns a newly built URI or NULL in case of error
+ */
+URI *
+uri_parse_raw(const char *str, int raw) {
+    URI *uri;
+    int ret;
+
+    if (str == NULL)
+	return(NULL);
+    uri = uri_new();
+    if (uri != NULL) {
+        if (raw) {
+	    uri->cleanup |= 2;
+	}
+	ret = uri_parse_into(uri, str);
+        if (ret) {
+	    uri_free(uri);
+	    return(NULL);
+	}
+    }
+    return(uri);
+}
+
+/************************************************************************
+ *									*
+ *			Generic URI structure functions			*
+ *									*
+ ************************************************************************/
+
+/**
+ * uri_new:
+ *
+ * Simply creates an empty URI
+ *
+ * Returns the new structure or NULL in case of error
+ */
+URI *
+uri_new(void) {
+    URI *ret;
+
+    ret = (URI *) g_malloc(sizeof(URI));
+    memset(ret, 0, sizeof(URI));
+    return(ret);
+}
+
+/**
+ * realloc2n:
+ *
+ * Function to handle properly a reallocation when saving an URI
+ * Also imposes some limit on the length of an URI string output
+ */
+static char *
+realloc2n(char *ret, int *max) {
+    char *temp;
+    int tmp;
+
+    tmp = *max * 2;
+    temp = g_realloc(ret, (tmp + 1));
+    *max = tmp;
+    return(temp);
+}
+
+/**
+ * uri_to_string:
+ * @uri:  pointer to an URI
+ *
+ * Save the URI as an escaped string
+ *
+ * Returns a new string (to be deallocated by caller)
+ */
+char *
+uri_to_string(URI *uri) {
+    char *ret = NULL;
+    char *temp;
+    const char *p;
+    int len;
+    int max;
+
+    if (uri == NULL) return(NULL);
+
+
+    max = 80;
+    ret = g_malloc(max + 1);
+    len = 0;
+
+    if (uri->scheme != NULL) {
+	p = uri->scheme;
+	while (*p != 0) {
+	    if (len >= max) {
+                temp = realloc2n(ret, &max);
+                if (temp == NULL) goto mem_error;
+		ret = temp;
+	    }
+	    ret[len++] = *p++;
+	}
+	if (len >= max) {
+            temp = realloc2n(ret, &max);
+            if (temp == NULL) goto mem_error;
+            ret = temp;
+	}
+	ret[len++] = ':';
+    }
+    if (uri->opaque != NULL) {
+	p = uri->opaque;
+	while (*p != 0) {
+	    if (len + 3 >= max) {
+                temp = realloc2n(ret, &max);
+                if (temp == NULL) goto mem_error;
+                ret = temp;
+	    }
+	    if (IS_RESERVED(*(p)) || IS_UNRESERVED(*(p)))
+		ret[len++] = *p++;
+	    else {
+		int val = *(unsigned char *)p++;
+		int hi = val / 0x10, lo = val % 0x10;
+		ret[len++] = '%';
+		ret[len++] = hi + (hi > 9? 'A'-10 : '0');
+		ret[len++] = lo + (lo > 9? 'A'-10 : '0');
+	    }
+	}
+    } else {
+	if (uri->server != NULL) {
+	    if (len + 3 >= max) {
+                temp = realloc2n(ret, &max);
+                if (temp == NULL) goto mem_error;
+                ret = temp;
+	    }
+	    ret[len++] = '/';
+	    ret[len++] = '/';
+	    if (uri->user != NULL) {
+		p = uri->user;
+		while (*p != 0) {
+		    if (len + 3 >= max) {
+                        temp = realloc2n(ret, &max);
+                        if (temp == NULL) goto mem_error;
+                        ret = temp;
+		    }
+		    if ((IS_UNRESERVED(*(p))) ||
+			((*(p) == ';')) || ((*(p) == ':')) ||
+			((*(p) == '&')) || ((*(p) == '=')) ||
+			((*(p) == '+')) || ((*(p) == '$')) ||
+			((*(p) == ',')))
+			ret[len++] = *p++;
+		    else {
+			int val = *(unsigned char *)p++;
+			int hi = val / 0x10, lo = val % 0x10;
+			ret[len++] = '%';
+			ret[len++] = hi + (hi > 9? 'A'-10 : '0');
+			ret[len++] = lo + (lo > 9? 'A'-10 : '0');
+		    }
+		}
+		if (len + 3 >= max) {
+                    temp = realloc2n(ret, &max);
+                    if (temp == NULL) goto mem_error;
+                    ret = temp;
+		}
+		ret[len++] = '@';
+	    }
+	    p = uri->server;
+	    while (*p != 0) {
+		if (len >= max) {
+                    temp = realloc2n(ret, &max);
+                    if (temp == NULL) goto mem_error;
+                    ret = temp;
+		}
+		ret[len++] = *p++;
+	    }
+	    if (uri->port > 0) {
+		if (len + 10 >= max) {
+                    temp = realloc2n(ret, &max);
+                    if (temp == NULL) goto mem_error;
+                    ret = temp;
+		}
+		len += snprintf(&ret[len], max - len, ":%d", uri->port);
+	    }
+	} else if (uri->authority != NULL) {
+	    if (len + 3 >= max) {
+                temp = realloc2n(ret, &max);
+                if (temp == NULL) goto mem_error;
+                ret = temp;
+	    }
+	    ret[len++] = '/';
+	    ret[len++] = '/';
+	    p = uri->authority;
+	    while (*p != 0) {
+		if (len + 3 >= max) {
+                    temp = realloc2n(ret, &max);
+                    if (temp == NULL) goto mem_error;
+                    ret = temp;
+		}
+		if ((IS_UNRESERVED(*(p))) ||
+                    ((*(p) == '$')) || ((*(p) == ',')) || ((*(p) == ';')) ||
+                    ((*(p) == ':')) || ((*(p) == '@')) || ((*(p) == '&')) ||
+                    ((*(p) == '=')) || ((*(p) == '+')))
+		    ret[len++] = *p++;
+		else {
+		    int val = *(unsigned char *)p++;
+		    int hi = val / 0x10, lo = val % 0x10;
+		    ret[len++] = '%';
+		    ret[len++] = hi + (hi > 9? 'A'-10 : '0');
+		    ret[len++] = lo + (lo > 9? 'A'-10 : '0');
+		}
+	    }
+	} else if (uri->scheme != NULL) {
+	    if (len + 3 >= max) {
+                temp = realloc2n(ret, &max);
+                if (temp == NULL) goto mem_error;
+                ret = temp;
+	    }
+	    ret[len++] = '/';
+	    ret[len++] = '/';
+	}
+	if (uri->path != NULL) {
+	    p = uri->path;
+	    /*
+	     * the colon in file:///d: should not be escaped or
+	     * Windows accesses fail later.
+	     */
+	    if ((uri->scheme != NULL) &&
+		(p[0] == '/') &&
+		(((p[1] >= 'a') && (p[1] <= 'z')) ||
+		 ((p[1] >= 'A') && (p[1] <= 'Z'))) &&
+		(p[2] == ':') &&
+	        (!strcmp(uri->scheme, "file"))) {
+		if (len + 3 >= max) {
+                    temp = realloc2n(ret, &max);
+                    if (temp == NULL) goto mem_error;
+                    ret = temp;
+		}
+		ret[len++] = *p++;
+		ret[len++] = *p++;
+		ret[len++] = *p++;
+	    }
+	    while (*p != 0) {
+		if (len + 3 >= max) {
+                    temp = realloc2n(ret, &max);
+                    if (temp == NULL) goto mem_error;
+                    ret = temp;
+		}
+		if ((IS_UNRESERVED(*(p))) || ((*(p) == '/')) ||
+                    ((*(p) == ';')) || ((*(p) == '@')) || ((*(p) == '&')) ||
+	            ((*(p) == '=')) || ((*(p) == '+')) || ((*(p) == '$')) ||
+	            ((*(p) == ',')))
+		    ret[len++] = *p++;
+		else {
+		    int val = *(unsigned char *)p++;
+		    int hi = val / 0x10, lo = val % 0x10;
+		    ret[len++] = '%';
+		    ret[len++] = hi + (hi > 9? 'A'-10 : '0');
+		    ret[len++] = lo + (lo > 9? 'A'-10 : '0');
+		}
+	    }
+	}
+	if (uri->query != NULL) {
+	    if (len + 1 >= max) {
+                temp = realloc2n(ret, &max);
+                if (temp == NULL) goto mem_error;
+                ret = temp;
+	    }
+	    ret[len++] = '?';
+	    p = uri->query;
+	    while (*p != 0) {
+		if (len + 1 >= max) {
+                    temp = realloc2n(ret, &max);
+                    if (temp == NULL) goto mem_error;
+                    ret = temp;
+		}
+		ret[len++] = *p++;
+	    }
+	}
+    }
+    if (uri->fragment != NULL) {
+	if (len + 3 >= max) {
+            temp = realloc2n(ret, &max);
+            if (temp == NULL) goto mem_error;
+            ret = temp;
+	}
+	ret[len++] = '#';
+	p = uri->fragment;
+	while (*p != 0) {
+	    if (len + 3 >= max) {
+                temp = realloc2n(ret, &max);
+                if (temp == NULL) goto mem_error;
+                ret = temp;
+	    }
+	    if ((IS_UNRESERVED(*(p))) || (IS_RESERVED(*(p))))
+		ret[len++] = *p++;
+	    else {
+		int val = *(unsigned char *)p++;
+		int hi = val / 0x10, lo = val % 0x10;
+		ret[len++] = '%';
+		ret[len++] = hi + (hi > 9? 'A'-10 : '0');
+		ret[len++] = lo + (lo > 9? 'A'-10 : '0');
+	    }
+	}
+    }
+    if (len >= max) {
+        temp = realloc2n(ret, &max);
+        if (temp == NULL) goto mem_error;
+        ret = temp;
+    }
+    ret[len] = 0;
+    return(ret);
+
+mem_error:
+    g_free(ret);
+    return(NULL);
+}
+
+/**
+ * uri_clean:
+ * @uri:  pointer to an URI
+ *
+ * Make sure the URI struct is free of content
+ */
+static void
+uri_clean(URI *uri) {
+    if (uri == NULL) return;
+
+    if (uri->scheme != NULL) g_free(uri->scheme);
+    uri->scheme = NULL;
+    if (uri->server != NULL) g_free(uri->server);
+    uri->server = NULL;
+    if (uri->user != NULL) g_free(uri->user);
+    uri->user = NULL;
+    if (uri->path != NULL) g_free(uri->path);
+    uri->path = NULL;
+    if (uri->fragment != NULL) g_free(uri->fragment);
+    uri->fragment = NULL;
+    if (uri->opaque != NULL) g_free(uri->opaque);
+    uri->opaque = NULL;
+    if (uri->authority != NULL) g_free(uri->authority);
+    uri->authority = NULL;
+    if (uri->query != NULL) g_free(uri->query);
+    uri->query = NULL;
+}
+
+/**
+ * uri_free:
+ * @uri:  pointer to an URI
+ *
+ * Free up the URI struct
+ */
+void
+uri_free(URI *uri) {
+    uri_clean(uri);
+    g_free(uri);
+}
+
+/************************************************************************
+ *									*
+ *			Helper functions				*
+ *									*
+ ************************************************************************/
+
+/**
+ * normalize_uri_path:
+ * @path:  pointer to the path string
+ *
+ * Applies the 5 normalization steps to a path string--that is, RFC 2396
+ * Section 5.2, steps 6.c through 6.g.
+ *
+ * Normalization occurs directly on the string, no new allocation is done
+ *
+ * Returns 0 or an error code
+ */
+static int
+normalize_uri_path(char *path) {
+    char *cur, *out;
+
+    if (path == NULL)
+	return(-1);
+
+    /* Skip all initial "/" chars.  We want to get to the beginning of the
+     * first non-empty segment.
+     */
+    cur = path;
+    while (cur[0] == '/')
+      ++cur;
+    if (cur[0] == '\0')
+      return(0);
+
+    /* Keep everything we've seen so far.  */
+    out = cur;
+
+    /*
+     * Analyze each segment in sequence for cases (c) and (d).
+     */
+    while (cur[0] != '\0') {
+	/*
+	 * c) All occurrences of "./", where "." is a complete path segment,
+	 *    are removed from the buffer string.
+	 */
+	if ((cur[0] == '.') && (cur[1] == '/')) {
+	    cur += 2;
+	    /* '//' normalization should be done at this point too */
+	    while (cur[0] == '/')
+		cur++;
+	    continue;
+	}
+
+	/*
+	 * d) If the buffer string ends with "." as a complete path segment,
+	 *    that "." is removed.
+	 */
+	if ((cur[0] == '.') && (cur[1] == '\0'))
+	    break;
+
+	/* Otherwise keep the segment.  */
+	while (cur[0] != '/') {
+            if (cur[0] == '\0')
+              goto done_cd;
+	    (out++)[0] = (cur++)[0];
+	}
+	/* nomalize // */
+	while ((cur[0] == '/') && (cur[1] == '/'))
+	    cur++;
+
+        (out++)[0] = (cur++)[0];
+    }
+ done_cd:
+    out[0] = '\0';
+
+    /* Reset to the beginning of the first segment for the next sequence.  */
+    cur = path;
+    while (cur[0] == '/')
+      ++cur;
+    if (cur[0] == '\0')
+	return(0);
+
+    /*
+     * Analyze each segment in sequence for cases (e) and (f).
+     *
+     * e) All occurrences of "<segment>/../", where <segment> is a
+     *    complete path segment not equal to "..", are removed from the
+     *    buffer string.  Removal of these path segments is performed
+     *    iteratively, removing the leftmost matching pattern on each
+     *    iteration, until no matching pattern remains.
+     *
+     * f) If the buffer string ends with "<segment>/..", where <segment>
+     *    is a complete path segment not equal to "..", that
+     *    "<segment>/.." is removed.
+     *
+     * To satisfy the "iterative" clause in (e), we need to collapse the
+     * string every time we find something that needs to be removed.  Thus,
+     * we don't need to keep two pointers into the string: we only need a
+     * "current position" pointer.
+     */
+    while (1) {
+        char *segp, *tmp;
+
+        /* At the beginning of each iteration of this loop, "cur" points to
+         * the first character of the segment we want to examine.
+         */
+
+        /* Find the end of the current segment.  */
+        segp = cur;
+        while ((segp[0] != '/') && (segp[0] != '\0'))
+          ++segp;
+
+        /* If this is the last segment, we're done (we need at least two
+         * segments to meet the criteria for the (e) and (f) cases).
+         */
+        if (segp[0] == '\0')
+          break;
+
+        /* If the first segment is "..", or if the next segment _isn't_ "..",
+         * keep this segment and try the next one.
+         */
+        ++segp;
+        if (((cur[0] == '.') && (cur[1] == '.') && (segp == cur+3))
+            || ((segp[0] != '.') || (segp[1] != '.')
+                || ((segp[2] != '/') && (segp[2] != '\0')))) {
+          cur = segp;
+          continue;
+        }
+
+        /* If we get here, remove this segment and the next one and back up
+         * to the previous segment (if there is one), to implement the
+         * "iteratively" clause.  It's pretty much impossible to back up
+         * while maintaining two pointers into the buffer, so just compact
+         * the whole buffer now.
+         */
+
+        /* If this is the end of the buffer, we're done.  */
+        if (segp[2] == '\0') {
+          cur[0] = '\0';
+          break;
+        }
+        /* Valgrind complained, strcpy(cur, segp + 3); */
+        /* string will overlap, do not use strcpy */
+        tmp = cur;
+        segp += 3;
+        while ((*tmp++ = *segp++) != 0)
+          ;
+
+        /* If there are no previous segments, then keep going from here.  */
+        segp = cur;
+        while ((segp > path) && ((--segp)[0] == '/'))
+          ;
+        if (segp == path)
+          continue;
+
+        /* "segp" is pointing to the end of a previous segment; find it's
+         * start.  We need to back up to the previous segment and start
+         * over with that to handle things like "foo/bar/../..".  If we
+         * don't do this, then on the first pass we'll remove the "bar/..",
+         * but be pointing at the second ".." so we won't realize we can also
+         * remove the "foo/..".
+         */
+        cur = segp;
+        while ((cur > path) && (cur[-1] != '/'))
+          --cur;
+    }
+    out[0] = '\0';
+
+    /*
+     * g) If the resulting buffer string still begins with one or more
+     *    complete path segments of "..", then the reference is
+     *    considered to be in error. Implementations may handle this
+     *    error by retaining these components in the resolved path (i.e.,
+     *    treating them as part of the final URI), by removing them from
+     *    the resolved path (i.e., discarding relative levels above the
+     *    root), or by avoiding traversal of the reference.
+     *
+     * We discard them from the final path.
+     */
+    if (path[0] == '/') {
+      cur = path;
+      while ((cur[0] == '/') && (cur[1] == '.') && (cur[2] == '.')
+             && ((cur[3] == '/') || (cur[3] == '\0')))
+	cur += 3;
+
+      if (cur != path) {
+	out = path;
+	while (cur[0] != '\0')
+          (out++)[0] = (cur++)[0];
+	out[0] = 0;
+      }
+    }
+
+    return(0);
+}
+
+static int is_hex(char c) {
+    if (((c >= '0') && (c <= '9')) ||
+        ((c >= 'a') && (c <= 'f')) ||
+        ((c >= 'A') && (c <= 'F')))
+	return(1);
+    return(0);
+}
+
+
+/**
+ * uri_string_unescape:
+ * @str:  the string to unescape
+ * @len:   the length in bytes to unescape (or <= 0 to indicate full string)
+ * @target:  optional destination buffer
+ *
+ * Unescaping routine, but does not check that the string is an URI. The
+ * output is a direct unsigned char translation of %XX values (no encoding)
+ * Note that the length of the result can only be smaller or same size as
+ * the input string.
+ *
+ * Returns a copy of the string, but unescaped, will return NULL only in case
+ * of error
+ */
+char *
+uri_string_unescape(const char *str, int len, char *target) {
+    char *ret, *out;
+    const char *in;
+
+    if (str == NULL)
+	return(NULL);
+    if (len <= 0) len = strlen(str);
+    if (len < 0) return(NULL);
+
+    if (target == NULL) {
+	ret = g_malloc(len + 1);
+    } else
+	ret = target;
+    in = str;
+    out = ret;
+    while(len > 0) {
+	if ((len > 2) && (*in == '%') && (is_hex(in[1])) && (is_hex(in[2]))) {
+	    in++;
+	    if ((*in >= '0') && (*in <= '9'))
+	        *out = (*in - '0');
+	    else if ((*in >= 'a') && (*in <= 'f'))
+	        *out = (*in - 'a') + 10;
+	    else if ((*in >= 'A') && (*in <= 'F'))
+	        *out = (*in - 'A') + 10;
+	    in++;
+	    if ((*in >= '0') && (*in <= '9'))
+	        *out = *out * 16 + (*in - '0');
+	    else if ((*in >= 'a') && (*in <= 'f'))
+	        *out = *out * 16 + (*in - 'a') + 10;
+	    else if ((*in >= 'A') && (*in <= 'F'))
+	        *out = *out * 16 + (*in - 'A') + 10;
+	    in++;
+	    len -= 3;
+	    out++;
+	} else {
+	    *out++ = *in++;
+	    len--;
+	}
+    }
+    *out = 0;
+    return(ret);
+}
+
+/**
+ * uri_string_escape:
+ * @str:  string to escape
+ * @list: exception list string of chars not to escape
+ *
+ * This routine escapes a string to hex, ignoring reserved characters (a-z)
+ * and the characters in the exception list.
+ *
+ * Returns a new escaped string or NULL in case of error.
+ */
+char *
+uri_string_escape(const char *str, const char *list) {
+    char *ret, ch;
+    char *temp;
+    const char *in;
+    int len, out;
+
+    if (str == NULL)
+	return(NULL);
+    if (str[0] == 0)
+	return(g_strdup(str));
+    len = strlen(str);
+    if (!(len > 0)) return(NULL);
+
+    len += 20;
+    ret = g_malloc(len);
+    in = str;
+    out = 0;
+    while(*in != 0) {
+	if (len - out <= 3) {
+            temp = realloc2n(ret, &len);
+	    ret = temp;
+	}
+
+	ch = *in;
+
+	if ((ch != '@') && (!IS_UNRESERVED(ch)) && (!strchr(list, ch))) {
+	    unsigned char val;
+	    ret[out++] = '%';
+	    val = ch >> 4;
+	    if (val <= 9)
+		ret[out++] = '0' + val;
+	    else
+		ret[out++] = 'A' + val - 0xA;
+	    val = ch & 0xF;
+	    if (val <= 9)
+		ret[out++] = '0' + val;
+	    else
+		ret[out++] = 'A' + val - 0xA;
+	    in++;
+	} else {
+	    ret[out++] = *in++;
+	}
+
+    }
+    ret[out] = 0;
+    return(ret);
+}
+
+/************************************************************************
+ *									*
+ *			Public functions				*
+ *									*
+ ************************************************************************/
+
+/**
+ * uri_resolve:
+ * @URI:  the URI instance found in the document
+ * @base:  the base value
+ *
+ * Computes he final URI of the reference done by checking that
+ * the given URI is valid, and building the final URI using the
+ * base URI. This is processed according to section 5.2 of the
+ * RFC 2396
+ *
+ * 5.2. Resolving Relative References to Absolute Form
+ *
+ * Returns a new URI string (to be freed by the caller) or NULL in case
+ *         of error.
+ */
+char *
+uri_resolve(const char *uri, const char *base) {
+    char *val = NULL;
+    int ret, len, indx, cur, out;
+    URI *ref = NULL;
+    URI *bas = NULL;
+    URI *res = NULL;
+
+    /*
+     * 1) The URI reference is parsed into the potential four components and
+     *    fragment identifier, as described in Section 4.3.
+     *
+     *    NOTE that a completely empty URI is treated by modern browsers
+     *    as a reference to "." rather than as a synonym for the current
+     *    URI.  Should we do that here?
+     */
+    if (uri == NULL)
+	ret = -1;
+    else {
+	if (*uri) {
+	    ref = uri_new();
+	    if (ref == NULL)
+		goto done;
+	    ret = uri_parse_into(ref, uri);
+	}
+	else
+	    ret = 0;
+    }
+    if (ret != 0)
+	goto done;
+    if ((ref != NULL) && (ref->scheme != NULL)) {
+	/*
+	 * The URI is absolute don't modify.
+	 */
+	val = g_strdup(uri);
+	goto done;
+    }
+    if (base == NULL)
+	ret = -1;
+    else {
+	bas = uri_new();
+	if (bas == NULL)
+	    goto done;
+	ret = uri_parse_into(bas, base);
+    }
+    if (ret != 0) {
+	if (ref)
+	    val = uri_to_string(ref);
+	goto done;
+    }
+    if (ref == NULL) {
+	/*
+	 * the base fragment must be ignored
+	 */
+	if (bas->fragment != NULL) {
+	    g_free(bas->fragment);
+	    bas->fragment = NULL;
+	}
+	val = uri_to_string(bas);
+	goto done;
+    }
+
+    /*
+     * 2) If the path component is empty and the scheme, authority, and
+     *    query components are undefined, then it is a reference to the
+     *    current document and we are done.  Otherwise, the reference URI's
+     *    query and fragment components are defined as found (or not found)
+     *    within the URI reference and not inherited from the base URI.
+     *
+     *    NOTE that in modern browsers, the parsing differs from the above
+     *    in the following aspect:  the query component is allowed to be
+     *    defined while still treating this as a reference to the current
+     *    document.
+     */
+    res = uri_new();
+    if (res == NULL)
+	goto done;
+    if ((ref->scheme == NULL) && (ref->path == NULL) &&
+	((ref->authority == NULL) && (ref->server == NULL))) {
+	if (bas->scheme != NULL)
+	    res->scheme = g_strdup(bas->scheme);
+	if (bas->authority != NULL)
+	    res->authority = g_strdup(bas->authority);
+	else if (bas->server != NULL) {
+	    res->server = g_strdup(bas->server);
+	    if (bas->user != NULL)
+		res->user = g_strdup(bas->user);
+	    res->port = bas->port;
+	}
+	if (bas->path != NULL)
+	    res->path = g_strdup(bas->path);
+	if (ref->query != NULL)
+	    res->query = g_strdup (ref->query);
+	else if (bas->query != NULL)
+	    res->query = g_strdup(bas->query);
+	if (ref->fragment != NULL)
+	    res->fragment = g_strdup(ref->fragment);
+	goto step_7;
+    }
+
+    /*
+     * 3) If the scheme component is defined, indicating that the reference
+     *    starts with a scheme name, then the reference is interpreted as an
+     *    absolute URI and we are done.  Otherwise, the reference URI's
+     *    scheme is inherited from the base URI's scheme component.
+     */
+    if (ref->scheme != NULL) {
+	val = uri_to_string(ref);
+	goto done;
+    }
+    if (bas->scheme != NULL)
+	res->scheme = g_strdup(bas->scheme);
+
+    if (ref->query != NULL)
+	res->query = g_strdup(ref->query);
+    if (ref->fragment != NULL)
+	res->fragment = g_strdup(ref->fragment);
+
+    /*
+     * 4) If the authority component is defined, then the reference is a
+     *    network-path and we skip to step 7.  Otherwise, the reference
+     *    URI's authority is inherited from the base URI's authority
+     *    component, which will also be undefined if the URI scheme does not
+     *    use an authority component.
+     */
+    if ((ref->authority != NULL) || (ref->server != NULL)) {
+	if (ref->authority != NULL)
+	    res->authority = g_strdup(ref->authority);
+	else {
+	    res->server = g_strdup(ref->server);
+	    if (ref->user != NULL)
+		res->user = g_strdup(ref->user);
+            res->port = ref->port;
+	}
+	if (ref->path != NULL)
+	    res->path = g_strdup(ref->path);
+	goto step_7;
+    }
+    if (bas->authority != NULL)
+	res->authority = g_strdup(bas->authority);
+    else if (bas->server != NULL) {
+	res->server = g_strdup(bas->server);
+	if (bas->user != NULL)
+	    res->user = g_strdup(bas->user);
+	res->port = bas->port;
+    }
+
+    /*
+     * 5) If the path component begins with a slash character ("/"), then
+     *    the reference is an absolute-path and we skip to step 7.
+     */
+    if ((ref->path != NULL) && (ref->path[0] == '/')) {
+	res->path = g_strdup(ref->path);
+	goto step_7;
+    }
+
+
+    /*
+     * 6) If this step is reached, then we are resolving a relative-path
+     *    reference.  The relative path needs to be merged with the base
+     *    URI's path.  Although there are many ways to do this, we will
+     *    describe a simple method using a separate string buffer.
+     *
+     * Allocate a buffer large enough for the result string.
+     */
+    len = 2; /* extra / and 0 */
+    if (ref->path != NULL)
+	len += strlen(ref->path);
+    if (bas->path != NULL)
+	len += strlen(bas->path);
+    res->path = g_malloc(len);
+    res->path[0] = 0;
+
+    /*
+     * a) All but the last segment of the base URI's path component is
+     *    copied to the buffer.  In other words, any characters after the
+     *    last (right-most) slash character, if any, are excluded.
+     */
+    cur = 0;
+    out = 0;
+    if (bas->path != NULL) {
+	while (bas->path[cur] != 0) {
+	    while ((bas->path[cur] != 0) && (bas->path[cur] != '/'))
+		cur++;
+	    if (bas->path[cur] == 0)
+		break;
+
+	    cur++;
+	    while (out < cur) {
+		res->path[out] = bas->path[out];
+		out++;
+	    }
+	}
+    }
+    res->path[out] = 0;
+
+    /*
+     * b) The reference's path component is appended to the buffer
+     *    string.
+     */
+    if (ref->path != NULL && ref->path[0] != 0) {
+	indx = 0;
+	/*
+	 * Ensure the path includes a '/'
+	 */
+	if ((out == 0) && (bas->server != NULL))
+	    res->path[out++] = '/';
+	while (ref->path[indx] != 0) {
+	    res->path[out++] = ref->path[indx++];
+	}
+    }
+    res->path[out] = 0;
+
+    /*
+     * Steps c) to h) are really path normalization steps
+     */
+    normalize_uri_path(res->path);
+
+step_7:
+
+    /*
+     * 7) The resulting URI components, including any inherited from the
+     *    base URI, are recombined to give the absolute form of the URI
+     *    reference.
+     */
+    val = uri_to_string(res);
+
+done:
+    if (ref != NULL)
+	uri_free(ref);
+    if (bas != NULL)
+	uri_free(bas);
+    if (res != NULL)
+	uri_free(res);
+    return(val);
+}
+
+/**
+ * uri_resolve_relative:
+ * @URI:  the URI reference under consideration
+ * @base:  the base value
+ *
+ * Expresses the URI of the reference in terms relative to the
+ * base.  Some examples of this operation include:
+ *     base = "http://site1.com/docs/book1.html"
+ *        URI input                        URI returned
+ *     docs/pic1.gif                    pic1.gif
+ *     docs/img/pic1.gif                img/pic1.gif
+ *     img/pic1.gif                     ../img/pic1.gif
+ *     http://site1.com/docs/pic1.gif   pic1.gif
+ *     http://site2.com/docs/pic1.gif   http://site2.com/docs/pic1.gif
+ *
+ *     base = "docs/book1.html"
+ *        URI input                        URI returned
+ *     docs/pic1.gif                    pic1.gif
+ *     docs/img/pic1.gif                img/pic1.gif
+ *     img/pic1.gif                     ../img/pic1.gif
+ *     http://site1.com/docs/pic1.gif   http://site1.com/docs/pic1.gif
+ *
+ *
+ * Note: if the URI reference is really wierd or complicated, it may be
+ *       worthwhile to first convert it into a "nice" one by calling
+ *       uri_resolve (using 'base') before calling this routine,
+ *       since this routine (for reasonable efficiency) assumes URI has
+ *       already been through some validation.
+ *
+ * Returns a new URI string (to be freed by the caller) or NULL in case
+ * error.
+ */
+char *
+uri_resolve_relative (const char *uri, const char * base)
+{
+    char *val = NULL;
+    int ret;
+    int ix;
+    int pos = 0;
+    int nbslash = 0;
+    int len;
+    URI *ref = NULL;
+    URI *bas = NULL;
+    char *bptr, *uptr, *vptr;
+    int remove_path = 0;
+
+    if ((uri == NULL) || (*uri == 0))
+	return NULL;
+
+    /*
+     * First parse URI into a standard form
+     */
+    ref = uri_new ();
+    if (ref == NULL)
+	return NULL;
+    /* If URI not already in "relative" form */
+    if (uri[0] != '.') {
+	ret = uri_parse_into (ref, uri);
+	if (ret != 0)
+	    goto done;		/* Error in URI, return NULL */
+    } else
+	ref->path = g_strdup(uri);
+
+    /*
+     * Next parse base into the same standard form
+     */
+    if ((base == NULL) || (*base == 0)) {
+	val = g_strdup (uri);
+	goto done;
+    }
+    bas = uri_new ();
+    if (bas == NULL)
+	goto done;
+    if (base[0] != '.') {
+	ret = uri_parse_into (bas, base);
+	if (ret != 0)
+	    goto done;		/* Error in base, return NULL */
+    } else
+	bas->path = g_strdup(base);
+
+    /*
+     * If the scheme / server on the URI differs from the base,
+     * just return the URI
+     */
+    if ((ref->scheme != NULL) &&
+	((bas->scheme == NULL) ||
+	 (strcmp (bas->scheme, ref->scheme)) ||
+	 (strcmp (bas->server, ref->server)))) {
+	val = g_strdup (uri);
+	goto done;
+    }
+    if (!strcmp(bas->path, ref->path)) {
+	val = g_strdup("");
+	goto done;
+    }
+    if (bas->path == NULL) {
+	val = g_strdup(ref->path);
+	goto done;
+    }
+    if (ref->path == NULL) {
+        ref->path = (char *) "/";
+	remove_path = 1;
+    }
+
+    /*
+     * At this point (at last!) we can compare the two paths
+     *
+     * First we take care of the special case where either of the
+     * two path components may be missing (bug 316224)
+     */
+    if (bas->path == NULL) {
+	if (ref->path != NULL) {
+	    uptr = ref->path;
+	    if (*uptr == '/')
+		uptr++;
+	    /* exception characters from uri_to_string */
+	    val = uri_string_escape(uptr, "/;&=+$,");
+	}
+	goto done;
+    }
+    bptr = bas->path;
+    if (ref->path == NULL) {
+	for (ix = 0; bptr[ix] != 0; ix++) {
+	    if (bptr[ix] == '/')
+		nbslash++;
+	}
+	uptr = NULL;
+	len = 1;	/* this is for a string terminator only */
+    } else {
+    /*
+     * Next we compare the two strings and find where they first differ
+     */
+	if ((ref->path[pos] == '.') && (ref->path[pos+1] == '/'))
+            pos += 2;
+	if ((*bptr == '.') && (bptr[1] == '/'))
+            bptr += 2;
+	else if ((*bptr == '/') && (ref->path[pos] != '/'))
+	    bptr++;
+	while ((bptr[pos] == ref->path[pos]) && (bptr[pos] != 0))
+	    pos++;
+
+	if (bptr[pos] == ref->path[pos]) {
+	    val = g_strdup("");
+	    goto done;		/* (I can't imagine why anyone would do this) */
+	}
+
+	/*
+	 * In URI, "back up" to the last '/' encountered.  This will be the
+	 * beginning of the "unique" suffix of URI
+	 */
+	ix = pos;
+	if ((ref->path[ix] == '/') && (ix > 0))
+	    ix--;
+	else if ((ref->path[ix] == 0) && (ix > 1) && (ref->path[ix - 1] == '/'))
+	    ix -= 2;
+	for (; ix > 0; ix--) {
+	    if (ref->path[ix] == '/')
+		break;
+	}
+	if (ix == 0) {
+	    uptr = ref->path;
+	} else {
+	    ix++;
+	    uptr = &ref->path[ix];
+	}
+
+	/*
+	 * In base, count the number of '/' from the differing point
+	 */
+	if (bptr[pos] != ref->path[pos]) {/* check for trivial URI == base */
+	    for (; bptr[ix] != 0; ix++) {
+		if (bptr[ix] == '/')
+		    nbslash++;
+	    }
+	}
+	len = strlen (uptr) + 1;
+    }
+
+    if (nbslash == 0) {
+	if (uptr != NULL)
+	    /* exception characters from uri_to_string */
+	    val = uri_string_escape(uptr, "/;&=+$,");
+	goto done;
+    }
+
+    /*
+     * Allocate just enough space for the returned string -
+     * length of the remainder of the URI, plus enough space
+     * for the "../" groups, plus one for the terminator
+     */
+    val = g_malloc (len + 3 * nbslash);
+    vptr = val;
+    /*
+     * Put in as many "../" as needed
+     */
+    for (; nbslash>0; nbslash--) {
+	*vptr++ = '.';
+	*vptr++ = '.';
+	*vptr++ = '/';
+    }
+    /*
+     * Finish up with the end of the URI
+     */
+    if (uptr != NULL) {
+        if ((vptr > val) && (len > 0) &&
+	    (uptr[0] == '/') && (vptr[-1] == '/')) {
+	    memcpy (vptr, uptr + 1, len - 1);
+	    vptr[len - 2] = 0;
+	} else {
+	    memcpy (vptr, uptr, len);
+	    vptr[len - 1] = 0;
+	}
+    } else {
+	vptr[len - 1] = 0;
+    }
+
+    /* escape the freshly-built path */
+    vptr = val;
+	/* exception characters from uri_to_string */
+    val = uri_string_escape(vptr, "/;&=+$,");
+    g_free(vptr);
+
+done:
+    /*
+     * Free the working variables
+     */
+    if (remove_path != 0)
+        ref->path = NULL;
+    if (ref != NULL)
+	uri_free (ref);
+    if (bas != NULL)
+	uri_free (bas);
+
+    return val;
+}
+
+/*
+ * Utility functions to help parse and assemble query strings.
+ */
+
+struct QueryParams *
+query_params_new (int init_alloc)
+{
+    struct QueryParams *ps;
+
+    if (init_alloc <= 0) init_alloc = 1;
+
+    ps = g_new(QueryParams, 1);
+    ps->n = 0;
+    ps->alloc = init_alloc;
+    ps->p = g_new(QueryParam, ps->alloc);
+
+    return ps;
+}
+
+/* Ensure there is space to store at least one more parameter
+ * at the end of the set.
+ */
+static int
+query_params_append (struct QueryParams *ps,
+               const char *name, const char *value)
+{
+    if (ps->n >= ps->alloc) {
+        ps->p = g_renew(QueryParam, ps->p, ps->alloc * 2);
+        ps->alloc *= 2;
+    }
+
+    ps->p[ps->n].name = g_strdup(name);
+    ps->p[ps->n].value = value ? g_strdup(value) : NULL;
+    ps->p[ps->n].ignore = 0;
+    ps->n++;
+
+    return 0;
+}
+
+void
+query_params_free (struct QueryParams *ps)
+{
+    int i;
+
+    for (i = 0; i < ps->n; ++i) {
+        g_free (ps->p[i].name);
+        g_free (ps->p[i].value);
+    }
+    g_free (ps->p);
+    g_free (ps);
+}
+
+struct QueryParams *
+query_params_parse (const char *query)
+{
+    struct QueryParams *ps;
+    const char *end, *eq;
+
+    ps = query_params_new (0);
+    if (!query || query[0] == '\0') return ps;
+
+    while (*query) {
+        char *name = NULL, *value = NULL;
+
+        /* Find the next separator, or end of the string. */
+        end = strchr (query, '&');
+        if (!end)
+            end = strchr (query, ';');
+        if (!end)
+            end = query + strlen (query);
+
+        /* Find the first '=' character between here and end. */
+        eq = strchr (query, '=');
+        if (eq && eq >= end) eq = NULL;
+
+        /* Empty section (eg. "&&"). */
+        if (end == query)
+            goto next;
+
+        /* If there is no '=' character, then we have just "name"
+         * and consistent with CGI.pm we assume value is "".
+         */
+        else if (!eq) {
+            name = uri_string_unescape (query, end - query, NULL);
+            value = NULL;
+        }
+        /* Or if we have "name=" here (works around annoying
+         * problem when calling uri_string_unescape with len = 0).
+         */
+        else if (eq+1 == end) {
+            name = uri_string_unescape (query, eq - query, NULL);
+            value = g_new0(char, 1);
+        }
+        /* If the '=' character is at the beginning then we have
+         * "=value" and consistent with CGI.pm we _ignore_ this.
+         */
+        else if (query == eq)
+            goto next;
+
+        /* Otherwise it's "name=value". */
+        else {
+            name = uri_string_unescape (query, eq - query, NULL);
+            value = uri_string_unescape (eq+1, end - (eq+1), NULL);
+        }
+
+        /* Append to the parameter set. */
+        query_params_append (ps, name, value);
+        g_free(name);
+        g_free(value);
+
+    next:
+        query = end;
+        if (*query) query ++; /* skip '&' separator */
+    }
+
+    return ps;
+}
diff --git a/uri.h b/uri.h
new file mode 100644
index 0000000000..de99b3bd4b
--- /dev/null
+++ b/uri.h
@@ -0,0 +1,113 @@
+/**
+ * Summary: library of generic URI related routines
+ * Description: library of generic URI related routines
+ *              Implements RFC 2396
+ *
+ * Copyright (C) 1998-2003 Daniel Veillard.  All Rights Reserved.
+ *
+ * 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
+ * DANIEL VEILLARD 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.
+ *
+ * Except as contained in this notice, the name of Daniel Veillard shall not
+ * be used in advertising or otherwise to promote the sale, use or other
+ * dealings in this Software without prior written authorization from him.
+ *
+ * Author: Daniel Veillard
+ **
+ * Copyright (C) 2007 Red Hat, Inc.
+ *
+ * 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 2.1 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA
+ *
+ * Authors:
+ *    Richard W.M. Jones <rjones@redhat.com>
+ *
+ * Utility functions to help parse and assemble query strings.
+ */
+
+#ifndef QEMU_URI_H
+#define QEMU_URI_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * URI:
+ *
+ * A parsed URI reference. This is a struct containing the various fields
+ * as described in RFC 2396 but separated for further processing.
+ */
+typedef struct URI {
+    char *scheme;	/* the URI scheme */
+    char *opaque;	/* opaque part */
+    char *authority;	/* the authority part */
+    char *server;	/* the server part */
+    char *user;		/* the user part */
+    int port;		/* the port number */
+    char *path;		/* the path string */
+    char *fragment;	/* the fragment identifier */
+    int  cleanup;	/* parsing potentially unclean URI */
+    char *query;	/* the query string (as it appears in the URI) */
+} URI;
+
+URI *uri_new(void);
+char *uri_resolve(const char *URI, const char *base);
+char *uri_resolve_relative(const char *URI, const char *base);
+URI *uri_parse(const char *str);
+URI *uri_parse_raw(const char *str, int raw);
+int uri_parse_into(URI *uri, const char *str);
+char *uri_to_string(URI *uri);
+char *uri_string_escape(const char *str, const char *list);
+char *uri_string_unescape(const char *str, int len, char *target);
+void uri_free(URI *uri);
+
+/* Single web service query parameter 'name=value'. */
+typedef struct QueryParam {
+  char *name;			/* Name (unescaped). */
+  char *value;			/* Value (unescaped). */
+  int ignore;			/* Ignore this field in qparam_get_query */
+} QueryParam;
+
+/* Set of parameters. */
+typedef struct QueryParams {
+  int n;			/* number of parameters used */
+  int alloc;			/* allocated space */
+  QueryParam *p;		/* array of parameters */
+} QueryParams;
+
+struct QueryParams *query_params_new (int init_alloc);
+int query_param_append (QueryParams *ps, const char *name, const char *value);
+extern char *query_param_to_string (const QueryParams *ps);
+extern QueryParams *query_params_parse (const char *query);
+extern void query_params_free (QueryParams *ps);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* QEMU_URI_H */