36 files changed, 605 insertions, 402 deletions

diff --git a/Changelog b/Changelog
index 67d0efd3a..4feb1867e 100644
--- a/Changelog
+++ b/Changelog
@@ -1,3 +1,53 @@
+version 0.12.4
+ - Workaround for broken OSS_GETVERSION on FreeBSD, part two (Juergen Lock)
+ - oss: fix fragment setting (malc)
+ - oss: issue OSS_GETVERSION ioctl only when needed (malc)
+ - oss: refactor code around policy setting (malc)
+ - oss: workaround for cases when OSS_GETVERSION is not defined (malc)
+ - block: Free iovec arrays allocated by multiwrite_merge() (Stefan Hajnoczi)
+ - lsi: fix segfault in lsi_command_complete (Gerd Hoffmann)
+ - lsi: pass lsi_request to lsi_reselect (Gerd Hoffmann)
+ - lsi: move dma_len+dma_buf into lsi_request (Gerd Hoffmann)
+ - lsi: move current_dev into lsi_request (Gerd Hoffmann)
+ - lsi: have lsi_request for the whole life time of the request. (Gerd Hoffmann)
+ - lsi: use QTAILQ for lsi_queue (Gerd Hoffmann)
+ - tcp/mips: Change TCG_AREG0 (fp -> s0) (Stefan Weil)
+ - sh_pci: fix memory and I/O access (Aurelien Jarno)
+ - Fix incoming migration with iothread (Marcelo Tosatti)
+ - Fix SIGFPE for vnc display of width/height = 1 (Chris Webb)
+ - net: remove broken net_set_boot_mask() boot device validation (Eduardo Habkost)
+ - qcow2: Remove request from in-flight list after error (Kevin Wolf)
+ - qcow2: Don't ignore immediate read/write failures (Kevin Wolf)
+ - block: Fix multiwrite memory leak in error case (Kevin Wolf)
+ - block: Fix error code in multiwrite for immediate failures (Kevin Wolf)
+ - block: Fix multiwrite error handling (Kevin Wolf)
+ - scsi-disk: fix buffer overflow (Gerd Hoffmann)
+ - qcow2: Rewrite alloc_refcount_block/grow_refcount_table (Kevin Wolf)
+ - qcow2: Factor next_refcount_table_size out (Kevin Wolf)
+ - block: avoid creating too large iovecs in multiwrite_merge (Christoph Hellwig)
+ - json-parser: Fix segfault on malformed input (Kevin Wolf)
+ - linux-user: switch default ppc64 CPU to 970fx from 970 (Aurelien Jarno)
+ - target-sh4: MMU: fix store queue addresses (Aurelien Jarno)
+ - target-sh4: MMU: fix ITLB priviledge check (Aurelien Jarno)
+ - target-sh4: MMU: fix mem_idx computation (Aurelien Jarno)
+ - sh7750: handle MMUCR TI bit (Aurelien Jarno)
+ - UHCI spurious interrut fix (Paul Brook)
+ - tcg/mips: fix branch offset during retranslation (Aurelien Jarno)
+ - tcg/arm: correctly save/restore registers in prologue/epilogue (Aurelien Jarno)
+ - workaround for cmd646 bmdma register access while no dma is active (Igor V. Kovalenko)
+ - Fix corner case in chardev udp: parameter (Jan Kiszka)
+ - Don't set default monitor when there is a mux'ed one (Jan Kiszka)
+ - spelling typo (compatibilty) in hw/fw_cfg.c (Vagrant Cascadian)
+ - fdc: fix drive property handling. (Gerd Hoffmann)
+ - target-i386: fix commit c22549204a6edc431e8e4358e61bd56386ff6957 (TeLeMan)
+ - target-i386: fix SIB decoding with index = 4 (Aurelien Jarno)
+ - Fix segfault with ram_size > 4095M without kvm (Ryan Harper)
+ - target-i386: Fix long jumps/calls in long mode with REX.W set (malc)
+ - target-i386: fix lddqu SSE instruction (Aurelien Jarno)
+ - qemu-char.c: drop debug printfs from qemu_chr_parse_compat (Jan Kiszka)
+ - fix undefined shifts by >32 (Paolo Bonzini)
+ - Fix qemu -net user,hostfwd= example (Aurelien Jarno)
+
 version 0.12.3
   - kvm: Fix eflags corruption in kvm mode (Jan Kiszka)
   - qcow2: Fix access after end of array (Kevin Wolf)
diff --git a/VERSION b/VERSION
index aa22d3ce3..e01e0ddd8 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-0.12.3
+0.12.4
diff --git a/audio/ossaudio.c b/audio/ossaudio.c
index 4002f1467..42bffaeaf 100644
--- a/audio/ossaudio.c
+++ b/audio/ossaudio.c
@@ -38,6 +38,10 @@
 #define AUDIO_CAP "oss"
 #include "audio_int.h"
 
+#if defined OSS_GETVERSION && defined SNDCTL_DSP_POLICY
+#define USE_DSP_POLICY
+#endif
+
 typedef struct OSSVoiceOut {
     HWVoiceOut hw;
     void *pcm_buf;
@@ -236,14 +240,39 @@ static void oss_dump_info (struct oss_params *req, struct oss_params *obt)
 }
 #endif
 
+#ifdef USE_DSP_POLICY
+static int oss_get_version (int fd, int *version, const char *typ)
+{
+    if (ioctl (fd, OSS_GETVERSION, &version)) {
+#if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
+        /*
+         * Looks like atm (20100109) FreeBSD knows OSS_GETVERSION
+         * since 7.x, but currently only on the mixer device (or in
+         * the Linuxolator), and in the native version that part of
+         * the code is in fact never reached so the ioctl fails anyway.
+         * Until this is fixed, just check the errno and if its what
+         * FreeBSD's sound drivers return atm assume they are new enough.
+         */
+        if (errno == EINVAL) {
+            *version = 0x040000;
+            return 0;
+        }
+#endif
+        oss_logerr2 (errno, typ, "Failed to get OSS version\n");
+        return -1;
+    }
+    return 0;
+}
+#endif
+
 static int oss_open (int in, struct oss_params *req,
                      struct oss_params *obt, int *pfd)
 {
     int fd;
-    int version;
     int oflags = conf.exclusive ? O_EXCL : 0;
     audio_buf_info abinfo;
     int fmt, freq, nchannels;
+    int setfragment = 1;
     const char *dspname = in ? conf.devpath_in : conf.devpath_out;
     const char *typ = in ? "ADC" : "DAC";
 
@@ -281,27 +310,30 @@ static int oss_open (int in, struct oss_params *req,
         goto err;
     }
 
-    if (ioctl (fd, OSS_GETVERSION, &version)) {
-        oss_logerr2 (errno, typ, "Failed to get OSS version\n");
-        version = 0;
-    }
+#ifdef USE_DSP_POLICY
+    if (conf.policy >= 0) {
+        int version;
 
-    if (conf.debug) {
-        dolog ("OSS version = %#x\n", version);
-    }
+        if (!oss_get_version (fd, &version, typ)) {
+            if (conf.debug) {
+                dolog ("OSS version = %#x\n", version);
+            }
 
-#ifdef SNDCTL_DSP_POLICY
-    if (conf.policy >= 0 && version >= 0x040000) {
-        int policy = conf.policy;
-        if (ioctl (fd, SNDCTL_DSP_POLICY, &policy)) {
-            oss_logerr2 (errno, typ, "Failed to set timing policy to %d\n",
-                         conf.policy);
-            goto err;
+            if (version >= 0x040000) {
+                int policy = conf.policy;
+                if (ioctl (fd, SNDCTL_DSP_POLICY, &policy)) {
+                    oss_logerr2 (errno, typ,
+                                 "Failed to set timing policy to %d\n",
+                                 conf.policy);
+                    goto err;
+                }
+                setfragment = 0;
+            }
         }
     }
-    else
 #endif
-    {
+
+    if (setfragment) {
         int mmmmssss = (req->nfrags << 16) | ctz32 (req->fragsize);
         if (ioctl (fd, SNDCTL_DSP_SETFRAGMENT, &mmmmssss)) {
             oss_logerr2 (errno, typ, "Failed to set buffer length (%d, %d)\n",
@@ -857,7 +889,7 @@ static struct audio_option oss_options[] = {
         .valp  = &conf.exclusive,
         .descr = "Open device in exclusive mode (vmix wont work)"
     },
-#ifdef SNDCTL_DSP_POLICY
+#ifdef USE_DSP_POLICY
     {
         .name  = "POLICY",
         .tag   = AUD_OPT_INT,
diff --git a/block.c b/block.c
index 97af3f577..7326bfeee 100644
--- a/block.c
+++ b/block.c
@@ -1608,6 +1608,9 @@ static void multiwrite_user_cb(MultiwriteCB *mcb)
 
     for (i = 0; i < mcb->num_callbacks; i++) {
         mcb->callbacks[i].cb(mcb->callbacks[i].opaque, mcb->error);
+        if (mcb->callbacks[i].free_qiov) {
+            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
+        }
         qemu_free(mcb->callbacks[i].free_qiov);
         qemu_vfree(mcb->callbacks[i].free_buf);
     }
@@ -1617,7 +1620,7 @@ static void multiwrite_cb(void *opaque, int ret)
 {
     MultiwriteCB *mcb = opaque;
 
-    if (ret < 0) {
+    if (ret < 0 && !mcb->error) {
         mcb->error = ret;
         multiwrite_user_cb(mcb);
     }
@@ -1669,6 +1672,10 @@ static int multiwrite_merge(BlockDriverState *bs, BlockRequest *reqs,
             merge = bs->drv->bdrv_merge_requests(bs, &reqs[outidx], &reqs[i]);
         }
 
+        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
+            merge = 0;
+        }
+
         if (merge) {
             size_t size;
             QEMUIOVector *qiov = qemu_mallocz(sizeof(*qiov));
@@ -1754,10 +1761,11 @@ int bdrv_aio_multiwrite(BlockDriverState *bs, BlockRequest *reqs, int num_reqs)
             // submitted yet. Otherwise we'll wait for the submitted AIOs to
             // complete and report the error in the callback.
             if (mcb->num_requests == 0) {
-                reqs[i].error = EIO;
+                reqs[i].error = -EIO;
                 goto fail;
             } else {
-                mcb->error = EIO;
+                mcb->num_requests++;
+                multiwrite_cb(mcb, -EIO);
                 break;
             }
         } else {
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index b13b6935f..c7057b166 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -811,6 +811,7 @@ int qcow2_alloc_cluster_offset(BlockDriverState *bs, uint64_t offset,
 
     cluster_offset = qcow2_alloc_clusters(bs, nb_clusters * s->cluster_size);
     if (cluster_offset < 0) {
+        QLIST_REMOVE(m, next_in_flight);
         return cluster_offset;
     }
 
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index c2a5c0471..5ebbcb63d 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -27,7 +27,7 @@
 #include "block/qcow2.h"
 
 static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size);
-static int update_refcount(BlockDriverState *bs,
+static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
                             int64_t offset, int64_t length,
                             int addend);
 
@@ -123,124 +123,266 @@ static int get_refcount(BlockDriverState *bs, int64_t cluster_index)
     return be16_to_cpu(s->refcount_block_cache[block_index]);
 }
 
-static int grow_refcount_table(BlockDriverState *bs, int min_size)
+/*
+ * Rounds the refcount table size up to avoid growing the table for each single
+ * refcount block that is allocated.
+ */
+static unsigned int next_refcount_table_size(BDRVQcowState *s,
+    unsigned int min_size)
+{
+    unsigned int min_clusters = (min_size >> (s->cluster_bits - 3)) + 1;
+    unsigned int refcount_table_clusters =
+        MAX(1, s->refcount_table_size >> (s->cluster_bits - 3));
+
+    while (min_clusters > refcount_table_clusters) {
+        refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2;
+    }
+
+    return refcount_table_clusters << (s->cluster_bits - 3);
+}
+
+
+/* Checks if two offsets are described by the same refcount block */
+static int in_same_refcount_block(BDRVQcowState *s, uint64_t offset_a,
+    uint64_t offset_b)
+{
+    uint64_t block_a = offset_a >> (2 * s->cluster_bits - REFCOUNT_SHIFT);
+    uint64_t block_b = offset_b >> (2 * s->cluster_bits - REFCOUNT_SHIFT);
+
+    return (block_a == block_b);
+}
+
+/*
+ * Loads a refcount block. If it doesn't exist yet, it is allocated first
+ * (including growing the refcount table if needed).
+ *
+ * Returns the offset of the refcount block on success or -errno in error case
+ */
+static int64_t alloc_refcount_block(BlockDriverState *bs, int64_t cluster_index)
 {
     BDRVQcowState *s = bs->opaque;
-    int new_table_size, new_table_size2, refcount_table_clusters, i, ret;
-    uint64_t *new_table;
-    int64_t table_offset;
-    uint8_t data[12];
-    int old_table_size;
-    int64_t old_table_offset;
+    unsigned int refcount_table_index;
+    int ret;
 
-    if (min_size <= s->refcount_table_size)
-        return 0;
-    /* compute new table size */
-    refcount_table_clusters = s->refcount_table_size >> (s->cluster_bits - 3);
-    for(;;) {
-        if (refcount_table_clusters == 0) {
-            refcount_table_clusters = 1;
-        } else {
-            refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2;
+    /* Find the refcount block for the given cluster */
+    refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
+
+    if (refcount_table_index < s->refcount_table_size) {
+
+        uint64_t refcount_block_offset =
+            s->refcount_table[refcount_table_index];
+
+        /* If it's already there, we're done */
+        if (refcount_block_offset) {
+            if (refcount_block_offset != s->refcount_block_cache_offset) {
+                ret = load_refcount_block(bs, refcount_block_offset);
+                if (ret < 0) {
+                    return ret;
+                }
+            }
+            return refcount_block_offset;
+        }
+    }
+
+    /*
+     * If we came here, we need to allocate something. Something is at least
+     * a cluster for the new refcount block. It may also include a new refcount
+     * table if the old refcount table is too small.
+     *
+     * Note that allocating clusters here needs some special care:
+     *
+     * - We can't use the normal qcow2_alloc_clusters(), it would try to
+     *   increase the refcount and very likely we would end up with an endless
+     *   recursion. Instead we must place the refcount blocks in a way that
+     *   they can describe them themselves.
+     *
+     * - We need to consider that at this point we are inside update_refcounts
+     *   and doing the initial refcount increase. This means that some clusters
+     *   have already been allocated by the caller, but their refcount isn't
+     *   accurate yet. free_cluster_index tells us where this allocation ends
+     *   as long as we don't overwrite it by freeing clusters.
+     *
+     * - alloc_clusters_noref and qcow2_free_clusters may load a different
+     *   refcount block into the cache
+     */
+
+    if (cache_refcount_updates) {
+        ret = write_refcount_block(s);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
+    /* Allocate the refcount block itself and mark it as used */
+    uint64_t new_block = alloc_clusters_noref(bs, s->cluster_size);
+    memset(s->refcount_block_cache, 0, s->cluster_size);
+    s->refcount_block_cache_offset = new_block;
+
+#ifdef DEBUG_ALLOC2
+    fprintf(stderr, "qcow2: Allocate refcount block %d for %" PRIx64
+        " at %" PRIx64 "\n",
+        refcount_table_index, cluster_index << s->cluster_bits, new_block);
+#endif
+
+    if (in_same_refcount_block(s, new_block, cluster_index << s->cluster_bits)) {
+        /* The block describes itself, need to update the cache */
+        int block_index = (new_block >> s->cluster_bits) &
+            ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1);
+        s->refcount_block_cache[block_index] = cpu_to_be16(1);
+    } else {
+        /* Described somewhere else. This can recurse at most twice before we
+         * arrive at a block that describes itself. */
+        ret = update_refcount(bs, new_block, s->cluster_size, 1);
+        if (ret < 0) {
+            goto fail_block;
+        }
+    }
+
+    /* Now the new refcount block needs to be written to disk */
+    ret = bdrv_pwrite(s->hd, new_block, s->refcount_block_cache,
+        s->cluster_size);
+    if (ret < 0) {
+        goto fail_block;
+    }
+
+    /* If the refcount table is big enough, just hook the block up there */
+    if (refcount_table_index < s->refcount_table_size) {
+        uint64_t data64 = cpu_to_be64(new_block);
+        ret = bdrv_pwrite(s->hd,
+            s->refcount_table_offset + refcount_table_index * sizeof(uint64_t),
+            &data64, sizeof(data64));
+        if (ret < 0) {
+            goto fail_block;
         }
-        new_table_size = refcount_table_clusters << (s->cluster_bits - 3);
-        if (min_size <= new_table_size)
-            break;
+
+        s->refcount_table[refcount_table_index] = new_block;
+        return new_block;
     }
+
+    /*
+     * If we come here, we need to grow the refcount table. Again, a new
+     * refcount table needs some space and we can't simply allocate to avoid
+     * endless recursion.
+     *
+     * Therefore let's grab new refcount blocks at the end of the image, which
+     * will describe themselves and the new refcount table. This way we can
+     * reference them only in the new table and do the switch to the new
+     * refcount table at once without producing an inconsistent state in
+     * between.
+     */
+    /* Calculate the number of refcount blocks needed so far */
+    uint64_t refcount_block_clusters = 1 << (s->cluster_bits - REFCOUNT_SHIFT);
+    uint64_t blocks_used = (s->free_cluster_index +
+        refcount_block_clusters - 1) / refcount_block_clusters;
+
+    /* And now we need at least one block more for the new metadata */
+    uint64_t table_size = next_refcount_table_size(s, blocks_used + 1);
+    uint64_t last_table_size;
+    uint64_t blocks_clusters;
+    do {
+        uint64_t table_clusters = size_to_clusters(s, table_size);
+        blocks_clusters = 1 +
+            ((table_clusters + refcount_block_clusters - 1)
+            / refcount_block_clusters);
+        uint64_t meta_clusters = table_clusters + blocks_clusters;
+
+        last_table_size = table_size;
+        table_size = next_refcount_table_size(s, blocks_used +
+            ((meta_clusters + refcount_block_clusters - 1)
+            / refcount_block_clusters));
+
+    } while (last_table_size != table_size);
+
 #ifdef DEBUG_ALLOC2
-    printf("grow_refcount_table from %d to %d\n",
-           s->refcount_table_size,
-           new_table_size);
+    fprintf(stderr, "qcow2: Grow refcount table %" PRId32 " => %" PRId64 "\n",
+        s->refcount_table_size, table_size);
 #endif
-    new_table_size2 = new_table_size * sizeof(uint64_t);
-    new_table = qemu_mallocz(new_table_size2);
+
+    /* Create the new refcount table and blocks */
+    uint64_t meta_offset = (blocks_used * refcount_block_clusters) *
+        s->cluster_size;
+    uint64_t table_offset = meta_offset + blocks_clusters * s->cluster_size;
+    uint16_t *new_blocks = qemu_mallocz(blocks_clusters * s->cluster_size);
+    uint64_t *new_table = qemu_mallocz(table_size * sizeof(uint64_t));
+
+    assert(meta_offset >= (s->free_cluster_index * s->cluster_size));
+
+    /* Fill the new refcount table */
     memcpy(new_table, s->refcount_table,
-           s->refcount_table_size * sizeof(uint64_t));
-    for(i = 0; i < s->refcount_table_size; i++)
+        s->refcount_table_size * sizeof(uint64_t));
+    new_table[refcount_table_index] = new_block;
+
+    int i;
+    for (i = 0; i < blocks_clusters; i++) {
+        new_table[blocks_used + i] = meta_offset + (i * s->cluster_size);
+    }
+
+    /* Fill the refcount blocks */
+    uint64_t table_clusters = size_to_clusters(s, table_size * sizeof(uint64_t));
+    int block = 0;
+    for (i = 0; i < table_clusters + blocks_clusters; i++) {
+        new_blocks[block++] = cpu_to_be16(1);
+    }
+
+    /* Write refcount blocks to disk */
+    ret = bdrv_pwrite(s->hd, meta_offset, new_blocks,
+        blocks_clusters * s->cluster_size);
+    qemu_free(new_blocks);
+    if (ret < 0) {
+        goto fail_table;
+    }
+
+    /* Write refcount table to disk */
+    for(i = 0; i < table_size; i++) {
         cpu_to_be64s(&new_table[i]);
-    /* Note: we cannot update the refcount now to avoid recursion */
-    table_offset = alloc_clusters_noref(bs, new_table_size2);
-    ret = bdrv_pwrite(s->hd, table_offset, new_table, new_table_size2);
-    if (ret != new_table_size2)
-        goto fail;
-    for(i = 0; i < s->refcount_table_size; i++)
-        be64_to_cpus(&new_table[i]);
+    }
 
+    ret = bdrv_pwrite(s->hd, table_offset, new_table,
+        table_size * sizeof(uint64_t));
+    if (ret < 0) {
+        goto fail_table;
+    }
+
+    for(i = 0; i < table_size; i++) {
+        cpu_to_be64s(&new_table[i]);
+    }
+
+    /* Hook up the new refcount table in the qcow2 header */
+    uint8_t data[12];
     cpu_to_be64w((uint64_t*)data, table_offset);
-    cpu_to_be32w((uint32_t*)(data + 8), refcount_table_clusters);
+    cpu_to_be32w((uint32_t*)(data + 8), table_clusters);
     ret = bdrv_pwrite(s->hd, offsetof(QCowHeader, refcount_table_offset),
-                    data, sizeof(data));
-    if (ret != sizeof(data)) {
-        goto fail;
+        data, sizeof(data));
+    if (ret < 0) {
+        goto fail_table;
     }
 
+    /* And switch it in memory */
+    uint64_t old_table_offset = s->refcount_table_offset;
+    uint64_t old_table_size = s->refcount_table_size;
+
     qemu_free(s->refcount_table);
-    old_table_offset = s->refcount_table_offset;
-    old_table_size = s->refcount_table_size;
     s->refcount_table = new_table;
-    s->refcount_table_size = new_table_size;
+    s->refcount_table_size = table_size;
     s->refcount_table_offset = table_offset;
 
-    update_refcount(bs, table_offset, new_table_size2, 1);
+    /* Free old table. Remember, we must not change free_cluster_index */
+    uint64_t old_free_cluster_index = s->free_cluster_index;
     qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t));
-    return 0;
- fail:
-    qemu_free(new_table);
-    return ret < 0 ? ret : -EIO;
-}
+    s->free_cluster_index = old_free_cluster_index;
 
-
-static int64_t alloc_refcount_block(BlockDriverState *bs, int64_t cluster_index)
-{
-    BDRVQcowState *s = bs->opaque;
-    int64_t offset, refcount_block_offset;
-    unsigned int refcount_table_index;
-    int ret;
-    uint64_t data64;
-    int cache = cache_refcount_updates;
-
-    /* Find L1 index and grow refcount table if needed */
-    refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
-    if (refcount_table_index >= s->refcount_table_size) {
-        ret = grow_refcount_table(bs, refcount_table_index + 1);
-        if (ret < 0)
-            return ret;
+    ret = load_refcount_block(bs, new_block);
+    if (ret < 0) {
+        goto fail_block;
     }
 
-    /* Load or allocate the refcount block */
-    refcount_block_offset = s->refcount_table[refcount_table_index];
-    if (!refcount_block_offset) {
-        if (cache_refcount_updates) {
-            write_refcount_block(s);
-            cache_refcount_updates = 0;
-        }
-        /* create a new refcount block */
-        /* Note: we cannot update the refcount now to avoid recursion */
-        offset = alloc_clusters_noref(bs, s->cluster_size);
-        memset(s->refcount_block_cache, 0, s->cluster_size);
-        ret = bdrv_pwrite(s->hd, offset, s->refcount_block_cache, s->cluster_size);
-        if (ret != s->cluster_size)
-            return -EINVAL;
-        s->refcount_table[refcount_table_index] = offset;
-        data64 = cpu_to_be64(offset);
-        ret = bdrv_pwrite(s->hd, s->refcount_table_offset +
-                          refcount_table_index * sizeof(uint64_t),
-                          &data64, sizeof(data64));
-        if (ret != sizeof(data64))
-            return -EINVAL;
-
-        refcount_block_offset = offset;
-        s->refcount_block_cache_offset = offset;
-        update_refcount(bs, offset, s->cluster_size, 1);
-        cache_refcount_updates = cache;
-    } else {
-        if (refcount_block_offset != s->refcount_block_cache_offset) {
-            if (load_refcount_block(bs, refcount_block_offset) < 0)
-                return -EIO;
-        }
-    }
+    return new_block;
 
-    return refcount_block_offset;
+fail_table:
+    qemu_free(new_table);
+fail_block:
+    s->refcount_block_cache_offset = 0;
+    return ret;
 }
 
 #define REFCOUNTS_PER_SECTOR (512 >> REFCOUNT_SHIFT)
diff --git a/block/qcow2.c b/block/qcow2.c
index 4ae8f193d..5d33d6c79 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -467,8 +467,10 @@ static void qcow_aio_read_cb(void *opaque, int ret)
         acb->hd_aiocb = bdrv_aio_readv(s->hd,
                             (acb->cluster_offset >> 9) + index_in_cluster,
                             &acb->hd_qiov, acb->n, qcow_aio_read_cb, acb);
-        if (acb->hd_aiocb == NULL)
+        if (acb->hd_aiocb == NULL) {
+            ret = -EIO;
             goto done;
+        }
     }
 
     return;
@@ -620,11 +622,17 @@ static void qcow_aio_write_cb(void *opaque, int ret)
                                     (acb->cluster_offset >> 9) + index_in_cluster,
                                     &acb->hd_qiov, acb->n,
                                     qcow_aio_write_cb, acb);
-    if (acb->hd_aiocb == NULL)
-        goto done;
+    if (acb->hd_aiocb == NULL) {
+        ret = -EIO;
+        goto fail;
+    }
 
     return;
 
+fail:
+    if (acb->l2meta.nb_clusters != 0) {
+        QLIST_REMOVE(&acb->l2meta, next_in_flight);
+    }
 done:
     if (acb->qiov->niov > 1)
         qemu_vfree(acb->orig_buf);
diff --git a/default-configs/sh4-softmmu.mak b/default-configs/sh4-softmmu.mak
index 4f912eccd..a89f52b16 100644
--- a/default-configs/sh4-softmmu.mak
+++ b/default-configs/sh4-softmmu.mak
@@ -2,3 +2,4 @@
 
 CONFIG_USB_OHCI=y
 CONFIG_PTIMER=y
+CONFIG_ISA_MMIO=y
diff --git a/default-configs/sh4eb-softmmu.mak b/default-configs/sh4eb-softmmu.mak
index 93d0c76f2..762385be3 100644
--- a/default-configs/sh4eb-softmmu.mak
+++ b/default-configs/sh4eb-softmmu.mak
@@ -2,3 +2,4 @@
 
 CONFIG_USB_OHCI=y
 CONFIG_PTIMER=y
+CONFIG_ISA_MMIO=y
diff --git a/dyngen-exec.h b/dyngen-exec.h
index 0353f36cd..3f1c43196 100644
--- a/dyngen-exec.h
+++ b/dyngen-exec.h
@@ -69,9 +69,9 @@ extern int printf(const char *, ...);
 #define AREG1 "r14"
 #define AREG2 "r15"
 #elif defined(__mips__)
-#define AREG0 "fp"
-#define AREG1 "s0"
-#define AREG2 "s1"
+#define AREG0 "s0"
+#define AREG1 "s1"
+#define AREG2 "fp"
 #elif defined(__sparc__)
 #ifdef CONFIG_SOLARIS
 #define AREG0 "g2"
diff --git a/hw/fdc.c b/hw/fdc.c
index 0579b0372..b29136552 100644
--- a/hw/fdc.c
+++ b/hw/fdc.c
@@ -1860,8 +1860,12 @@ fdctrl_t *fdctrl_init_isa(DriveInfo **fds)
     ISADevice *dev;
 
     dev = isa_create("isa-fdc");
-    qdev_prop_set_drive(&dev->qdev, "driveA", fds[0]);
-    qdev_prop_set_drive(&dev->qdev, "driveB", fds[1]);
+    if (fds[0]) {
+        qdev_prop_set_drive(&dev->qdev, "driveA", fds[0]);
+    }
+    if (fds[1]) {
+        qdev_prop_set_drive(&dev->qdev, "driveB", fds[1]);
+    }
     if (qdev_init(&dev->qdev) < 0)
         return NULL;
     return &(DO_UPCAST(fdctrl_isabus_t, busdev, dev)->state);
@@ -1879,8 +1883,12 @@ fdctrl_t *fdctrl_init_sysbus(qemu_irq irq, int dma_chann,
     sys = DO_UPCAST(fdctrl_sysbus_t, busdev.qdev, dev);
     fdctrl = &sys->state;
     fdctrl->dma_chann = dma_chann; /* FIXME */
-    qdev_prop_set_drive(dev, "driveA", fds[0]);
-    qdev_prop_set_drive(dev, "driveB", fds[1]);
+    if (fds[0]) {
+        qdev_prop_set_drive(dev, "driveA", fds[0]);
+    }
+    if (fds[1]) {
+        qdev_prop_set_drive(dev, "driveB", fds[1]);
+    }
     qdev_init_nofail(dev);
     sysbus_connect_irq(&sys->busdev, 0, irq);
     sysbus_mmio_map(&sys->busdev, 0, mmio_base);
@@ -1896,7 +1904,9 @@ fdctrl_t *sun4m_fdctrl_init (qemu_irq irq, target_phys_addr_t io_base,
     fdctrl_t *fdctrl;
 
     dev = qdev_create(NULL, "SUNW,fdtwo");
-    qdev_prop_set_drive(dev, "drive", fds[0]);
+    if (fds[0]) {
+        qdev_prop_set_drive(dev, "drive", fds[0]);
+    }
     qdev_init_nofail(dev);
     sys = DO_UPCAST(fdctrl_sysbus_t, busdev.qdev, dev);
     fdctrl = &sys->state;
diff --git a/hw/fw_cfg.c b/hw/fw_cfg.c
index ea120ba55..c62bf522d 100644
--- a/hw/fw_cfg.c
+++ b/hw/fw_cfg.c
@@ -179,7 +179,7 @@ static int get_uint32_as_uint16(QEMUFile *f, void *pv, size_t size)
 
 static void put_unused(QEMUFile *f, void *pv, size_t size)
 {
-    fprintf(stderr, "uint32_as_uint16 is only used for backward compatibilty.\n");
+    fprintf(stderr, "uint32_as_uint16 is only used for backward compatibility.\n");
     fprintf(stderr, "This functions shouldn't be called.\n");
 }
 
diff --git a/hw/ide/cmd646.c b/hw/ide/cmd646.c
index e1e626e2a..835c98d72 100644
--- a/hw/ide/cmd646.c
+++ b/hw/ide/cmd646.c
@@ -70,11 +70,7 @@ static void ide_map(PCIDevice *pci_dev, int region_num,
 
 static PCIIDEState *pci_from_bm(BMDMAState *bm)
 {
-    if (bm->unit == 0) {
-        return container_of(bm, PCIIDEState, bmdma[0]);
-    } else {
-        return container_of(bm, PCIIDEState, bmdma[1]);
-    }
+    return bm->pci_dev;
 }
 
 static uint32_t bmdma_readb(void *opaque, uint32_t addr)
@@ -145,6 +141,7 @@ static void bmdma_map(PCIDevice *pci_dev, int region_num,
         BMDMAState *bm = &d->bmdma[i];
         d->bus[i].bmdma = bm;
         bm->bus = d->bus+i;
+        bm->pci_dev = d;
         qemu_add_vm_change_state_handler(ide_dma_restart_cb, bm);
 
         register_ioport_write(addr, 1, 1, bmdma_cmd_writeb, bm);
diff --git a/hw/ide/internal.h b/hw/ide/internal.h
index 8869a0834..8615d1a1f 100644
--- a/hw/ide/internal.h
+++ b/hw/ide/internal.h
@@ -481,6 +481,7 @@ struct BMDMAState {
     uint8_t status;
     uint32_t addr;
 
+    struct PCIIDEState *pci_dev;
     IDEBus *bus;
     /* current transfer state */
     uint32_t cur_addr;
diff --git a/hw/ide/piix.c b/hw/ide/piix.c
index de3648023..2776ac365 100644
--- a/hw/ide/piix.c
+++ b/hw/ide/piix.c
@@ -78,6 +78,7 @@ static void bmdma_map(PCIDevice *pci_dev, int region_num,
         BMDMAState *bm = &d->bmdma[i];
         d->bus[i].bmdma = bm;
         bm->bus = d->bus+i;
+        bm->pci_dev = d;
         qemu_add_vm_change_state_handler(ide_dma_restart_cb, bm);
 
         register_ioport_write(addr, 1, 1, bmdma_cmd_writeb, bm);
diff --git a/hw/lsi53c895a.c b/hw/lsi53c895a.c
index 014c85dbd..e0ade1e54 100644
--- a/hw/lsi53c895a.c
+++ b/hw/lsi53c895a.c
@@ -173,11 +173,15 @@ do { fprintf(stderr, "lsi_scsi: error: " fmt , ## __VA_ARGS__);} while (0)
 /* Flag set if this is a tagged command.  */
 #define LSI_TAG_VALID     (1 << 16)
 
-typedef struct {
+typedef struct lsi_request {
     uint32_t tag;
+    SCSIDevice *dev;
+    uint32_t dma_len;
+    uint8_t *dma_buf;
     uint32_t pending;
     int out;
-} lsi_queue;
+    QTAILQ_ENTRY(lsi_request) next;
+} lsi_request;
 
 typedef struct {
     PCIDevice dev;
@@ -198,16 +202,13 @@ typedef struct {
      * 3 if a DMA operation is in progress.  */
     int waiting;
     SCSIBus bus;
-    SCSIDevice *current_dev;
+    SCSIDevice *select_dev;
     int current_lun;
     /* The tag is a combination of the device ID and the SCSI tag.  */
-    uint32_t current_tag;
-    uint32_t current_dma_len;
+    uint32_t select_tag;
     int command_complete;
-    uint8_t *dma_buf;
-    lsi_queue *queue;
-    int queue_len;
-    int active_commands;
+    QTAILQ_HEAD(, lsi_request) queue;
+    lsi_request *current;
 
     uint32_t dsa;
     uint32_t temp;
@@ -370,7 +371,7 @@ static int lsi_dma_64bit(LSIState *s)
 static uint8_t lsi_reg_readb(LSIState *s, int offset);
 static void lsi_reg_writeb(LSIState *s, int offset, uint8_t val);
 static void lsi_execute_script(LSIState *s);
-static void lsi_reselect(LSIState *s, uint32_t tag);
+static void lsi_reselect(LSIState *s, lsi_request *p);
 
 static inline uint32_t read_dword(LSIState *s, uint32_t addr)
 {
@@ -391,9 +392,9 @@ static void lsi_stop_script(LSIState *s)
 
 static void lsi_update_irq(LSIState *s)
 {
-    int i;
     int level;
     static int last_level;
+    lsi_request *p;
 
     /* It's unclear whether the DIP/SIP bits should be cleared when the
        Interrupt Status Registers are cleared or when istat0 is read.
@@ -427,9 +428,9 @@ static void lsi_update_irq(LSIState *s)
     if (!level && lsi_irq_on_rsl(s) && !(s->scntl1 & LSI_SCNTL1_CON)) {
         DPRINTF("Handled IRQs & disconnected, looking for pending "
                 "processes\n");
-        for (i = 0; i < s->active_commands; i++) {
-            if (s->queue[i].pending) {
-                lsi_reselect(s, s->queue[i].tag);
+        QTAILQ_FOREACH(p, &s->queue, next) {
+            if (p->pending) {
+                lsi_reselect(s, p);
                 break;
             }
         }
@@ -508,15 +509,16 @@ static void lsi_do_dma(LSIState *s, int out)
     uint32_t count;
     target_phys_addr_t addr;
 
-    if (!s->current_dma_len) {
+    assert(s->current);
+    if (!s->current->dma_len) {
         /* Wait until data is available.  */
         DPRINTF("DMA no data available\n");
         return;
     }
 
     count = s->dbc;
-    if (count > s->current_dma_len)
-        count = s->current_dma_len;
+    if (count > s->current->dma_len)
+        count = s->current->dma_len;
 
     addr = s->dnad;
     /* both 40 and Table Indirect 64-bit DMAs store upper bits in dnad64 */
@@ -532,29 +534,29 @@ static void lsi_do_dma(LSIState *s, int out)
     s->dnad += count;
     s->dbc -= count;
 
-    if (s->dma_buf == NULL) {
-        s->dma_buf = s->current_dev->info->get_buf(s->current_dev,
-                                                   s->current_tag);
+    if (s->current->dma_buf == NULL) {
+        s->current->dma_buf = s->current->dev->info->get_buf(s->current->dev,
+                                                             s->current->tag);
     }
 
     /* ??? Set SFBR to first data byte.  */
     if (out) {
-        cpu_physical_memory_read(addr, s->dma_buf, count);
+        cpu_physical_memory_read(addr, s->current->dma_buf, count);
     } else {
-        cpu_physical_memory_write(addr, s->dma_buf, count);
+        cpu_physical_memory_write(addr, s->current->dma_buf, count);
     }
-    s->current_dma_len -= count;
-    if (s->current_dma_len == 0) {
-        s->dma_buf = NULL;
+    s->current->dma_len -= count;
+    if (s->current->dma_len == 0) {
+        s->current->dma_buf = NULL;
         if (out) {
             /* Write the data.  */
-            s->current_dev->info->write_data(s->current_dev, s->current_tag);
+            s->current->dev->info->write_data(s->current->dev, s->current->tag);
         } else {
             /* Request any remaining data.  */
-            s->current_dev->info->read_data(s->current_dev, s->current_tag);
+            s->current->dev->info->read_data(s->current->dev, s->current->tag);
         }
     } else {
-        s->dma_buf += count;
+        s->current->dma_buf += count;
         lsi_resume_script(s);
     }
 }
@@ -563,15 +565,14 @@ static void lsi_do_dma(LSIState *s, int out)
 /* Add a command to the queue.  */
 static void lsi_queue_command(LSIState *s)
 {
-    lsi_queue *p;
+    lsi_request *p = s->current;
 
     DPRINTF("Queueing tag=0x%x\n", s->current_tag);
-    if (s->queue_len == s->active_commands) {
-        s->queue_len++;
-        s->queue = qemu_realloc(s->queue, s->queue_len * sizeof(lsi_queue));
-    }
-    p = &s->queue[s->active_commands++];
-    p->tag = s->current_tag;
+    assert(s->current != NULL);
+    assert(s->current->dma_len == 0);
+    QTAILQ_INSERT_TAIL(&s->queue, s->current, next);
+    s->current = NULL;
+
     p->pending = 0;
     p->out = (s->sstat1 & PHASE_MASK) == PHASE_DO;
 }
@@ -588,45 +589,29 @@ static void lsi_add_msg_byte(LSIState *s, uint8_t data)
 }
 
 /* Perform reselection to continue a command.  */
-static void lsi_reselect(LSIState *s, uint32_t tag)
+static void lsi_reselect(LSIState *s, lsi_request *p)
 {
-    lsi_queue *p;
-    int n;
     int id;
 
-    p = NULL;
-    for (n = 0; n < s->active_commands; n++) {
-        p = &s->queue[n];
-        if (p->tag == tag)
-            break;
-    }
-    if (n == s->active_commands) {
-        BADF("Reselected non-existant command tag=0x%x\n", tag);
-        return;
-    }
-    id = (tag >> 8) & 0xf;
+    assert(s->current == NULL);
+    QTAILQ_REMOVE(&s->queue, p, next);
+    s->current = p;
+
+    id = (p->tag >> 8) & 0xf;
     s->ssid = id | 0x80;
     /* LSI53C700 Family Compatibility, see LSI53C895A 4-73 */
     if (!s->dcntl & LSI_DCNTL_COM) {
         s->sfbr = 1 << (id & 0x7);
     }
     DPRINTF("Reselected target %d\n", id);
-    s->current_dev = s->bus.devs[id];
-    s->current_tag = tag;
     s->scntl1 |= LSI_SCNTL1_CON;
     lsi_set_phase(s, PHASE_MI);
     s->msg_action = p->out ? 2 : 3;
-    s->current_dma_len = p->pending;
-    s->dma_buf = NULL;
+    s->current->dma_len = p->pending;
     lsi_add_msg_byte(s, 0x80);
-    if (s->current_tag & LSI_TAG_VALID) {
+    if (s->current->tag & LSI_TAG_VALID) {
         lsi_add_msg_byte(s, 0x20);
-        lsi_add_msg_byte(s, tag & 0xff);
-    }
-
-    s->active_commands--;
-    if (n != s->active_commands) {
-        s->queue[n] = s->queue[s->active_commands];
+        lsi_add_msg_byte(s, p->tag & 0xff);
     }
 
     if (lsi_irq_on_rsl(s)) {
@@ -638,10 +623,9 @@ static void lsi_reselect(LSIState *s, uint32_t tag)
    the device was reselected, nonzero if the IO is deferred.  */
 static int lsi_queue_tag(LSIState *s, uint32_t tag, uint32_t arg)
 {
-    lsi_queue *p;
-    int i;
-    for (i = 0; i < s->active_commands; i++) {
-        p = &s->queue[i];
+    lsi_request *p;
+
+    QTAILQ_FOREACH(p, &s->queue, next) {
         if (p->tag == tag) {
             if (p->pending) {
                 BADF("Multiple IO pending for tag %d\n", tag);
@@ -656,10 +640,10 @@ static int lsi_queue_tag(LSIState *s, uint32_t tag, uint32_t arg)
                 (lsi_irq_on_rsl(s) && !(s->scntl1 & LSI_SCNTL1_CON) &&
                  !(s->istat0 & (LSI_ISTAT0_SIP | LSI_ISTAT0_DIP)))) {
                 /* Reselect device.  */
-                lsi_reselect(s, tag);
+                lsi_reselect(s, p);
                 return 0;
             } else {
-               DPRINTF("Queueing IO tag=0x%x\n", tag);
+                DPRINTF("Queueing IO tag=0x%x\n", tag);
                 p->pending = arg;
                 return 1;
             }
@@ -687,11 +671,15 @@ static void lsi_command_complete(SCSIBus *bus, int reason, uint32_t tag,
         } else {
             lsi_set_phase(s, PHASE_ST);
         }
+
+        qemu_free(s->current);
+        s->current = NULL;
+
         lsi_resume_script(s);
         return;
     }
 
-    if (s->waiting == 1 || tag != s->current_tag ||
+    if (s->waiting == 1 || !s->current || tag != s->current->tag ||
         (lsi_irq_on_rsl(s) && !(s->scntl1 & LSI_SCNTL1_CON))) {
         if (lsi_queue_tag(s, tag, arg))
             return;
@@ -699,7 +687,7 @@ static void lsi_command_complete(SCSIBus *bus, int reason, uint32_t tag,
 
     /* host adapter (re)connected */
     DPRINTF("Data ready tag=0x%x len=%d\n", tag, arg);
-    s->current_dma_len = arg;
+    s->current->dma_len = arg;
     s->command_complete = 1;
     if (!s->waiting)
         return;
@@ -721,14 +709,20 @@ static void lsi_do_command(LSIState *s)
     cpu_physical_memory_read(s->dnad, buf, s->dbc);
     s->sfbr = buf[0];
     s->command_complete = 0;
-    n = s->current_dev->info->send_command(s->current_dev, s->current_tag, buf,
-                                           s->current_lun);
+
+    assert(s->current == NULL);
+    s->current = qemu_mallocz(sizeof(lsi_request));
+    s->current->tag = s->select_tag;
+    s->current->dev = s->select_dev;
+
+    n = s->current->dev->info->send_command(s->current->dev, s->current->tag, buf,
+                                            s->current_lun);
     if (n > 0) {
         lsi_set_phase(s, PHASE_DI);
-        s->current_dev->info->read_data(s->current_dev, s->current_tag);
+        s->current->dev->info->read_data(s->current->dev, s->current->tag);
     } else if (n < 0) {
         lsi_set_phase(s, PHASE_DO);
-        s->current_dev->info->write_data(s->current_dev, s->current_tag);
+        s->current->dev->info->write_data(s->current->dev, s->current->tag);
     }
 
     if (!s->command_complete) {
@@ -851,16 +845,16 @@ static void lsi_do_msgout(LSIState *s)
             }
             break;
         case 0x20: /* SIMPLE queue */
-            s->current_tag |= lsi_get_msgbyte(s) | LSI_TAG_VALID;
+            s->select_tag |= lsi_get_msgbyte(s) | LSI_TAG_VALID;
             DPRINTF("SIMPLE queue tag=0x%x\n", s->current_tag & 0xff);
             break;
         case 0x21: /* HEAD of queue */
             BADF("HEAD queue not implemented\n");
-            s->current_tag |= lsi_get_msgbyte(s) | LSI_TAG_VALID;
+            s->select_tag |= lsi_get_msgbyte(s) | LSI_TAG_VALID;
             break;
         case 0x22: /* ORDERED queue */
             BADF("ORDERED queue not implemented\n");
-            s->current_tag |= lsi_get_msgbyte(s) | LSI_TAG_VALID;
+            s->select_tag |= lsi_get_msgbyte(s) | LSI_TAG_VALID;
             break;
         default:
             if ((msg & 0x80) == 0) {
@@ -905,17 +899,17 @@ static void lsi_memcpy(LSIState *s, uint32_t dest, uint32_t src, int count)
 
 static void lsi_wait_reselect(LSIState *s)
 {
-    int i;
+    lsi_request *p;
+
     DPRINTF("Wait Reselect\n");
-    if (s->current_dma_len)
-        BADF("Reselect with pending DMA\n");
-    for (i = 0; i < s->active_commands; i++) {
-        if (s->queue[i].pending) {
-            lsi_reselect(s, s->queue[i].tag);
+
+    QTAILQ_FOREACH(p, &s->queue, next) {
+        if (p->pending) {
+            lsi_reselect(s, p);
             break;
         }
     }
-    if (s->current_dma_len == 0) {
+    if (s->current == NULL) {
         s->waiting = 1;
     }
 }
@@ -1093,8 +1087,8 @@ again:
                 /* ??? Linux drivers compain when this is set.  Maybe
                    it only applies in low-level mode (unimplemented).
                 lsi_script_scsi_interrupt(s, LSI_SIST0_CMP, 0); */
-                s->current_dev = s->bus.devs[id];
-                s->current_tag = id << 8;
+                s->select_dev = s->bus.devs[id];
+                s->select_tag = id << 8;
                 s->scntl1 |= LSI_SCNTL1_CON;
                 if (insn & (1 << 3)) {
                     s->socl |= LSI_SOCL_ATN;
@@ -2006,9 +2000,11 @@ static void lsi_pre_save(void *opaque)
 {
     LSIState *s = opaque;
 
-    assert(s->dma_buf == NULL);
-    assert(s->current_dma_len == 0);
-    assert(s->active_commands == 0);
+    if (s->current) {
+        assert(s->current->dma_buf == NULL);
+        assert(s->current->dma_len == 0);
+    }
+    assert(QTAILQ_EMPTY(&s->queue));
 }
 
 static const VMStateDescription vmstate_lsi_scsi = {
@@ -2101,8 +2097,6 @@ static int lsi_scsi_uninit(PCIDevice *d)
     cpu_unregister_io_memory(s->mmio_io_addr);
     cpu_unregister_io_memory(s->ram_io_addr);
 
-    qemu_free(s->queue);
-
     return 0;
 }
 
@@ -2138,9 +2132,7 @@ static int lsi_scsi_init(PCIDevice *dev)
                            PCI_BASE_ADDRESS_SPACE_MEMORY, lsi_mmio_mapfunc);
     pci_register_bar((struct PCIDevice *)s, 2, 0x2000,
                            PCI_BASE_ADDRESS_SPACE_MEMORY, lsi_ram_mapfunc);
-    s->queue = qemu_malloc(sizeof(lsi_queue));
-    s->queue_len = 1;
-    s->active_commands = 0;
+    QTAILQ_INIT(&s->queue);
 
     lsi_soft_reset(s);
 
diff --git a/hw/scsi-disk.c b/hw/scsi-disk.c
index b34fbaa67..a79201213 100644
--- a/hw/scsi-disk.c
+++ b/hw/scsi-disk.c
@@ -434,7 +434,9 @@ static int scsi_disk_emulate_inquiry(SCSIRequest *req, uint8_t *outbuf)
         memcpy(&outbuf[16], "QEMU HARDDISK   ", 16);
     }
     memcpy(&outbuf[8], "QEMU    ", 8);
-    memcpy(&outbuf[32], s->version ? s->version : QEMU_VERSION, 4);
+    memset(&outbuf[32], 0, 4);
+    memcpy(&outbuf[32], s->version ? s->version : QEMU_VERSION,
+           MIN(4, strlen(s->version ? s->version : QEMU_VERSION)));
     /* Identify device as SCSI-3 rev 1.
        Some later commands are also implemented. */
     outbuf[2] = 3;
diff --git a/hw/sh7750.c b/hw/sh7750.c
index 933bbc0c7..9c39f4b68 100644
--- a/hw/sh7750.c
+++ b/hw/sh7750.c
@@ -396,8 +396,11 @@ static void sh7750_mem_writel(void *opaque, target_phys_addr_t addr,
 	portb_changed(s, temp);
 	return;
     case SH7750_MMUCR_A7:
-	s->cpu->mmucr = mem_value;
-	return;
+        if (mem_value & MMUCR_TI) {
+            cpu_sh4_invalidate_tlb(s->cpu);
+        }
+        s->cpu->mmucr = mem_value & ~MMUCR_TI;
+        return;
     case SH7750_PTEH_A7:
         /* If asid changes, clear all registered tlb entries. */
 	if ((s->cpu->pteh & 0xff) != (mem_value & 0xff))
diff --git a/hw/sh_pci.c b/hw/sh_pci.c
index abe4c7568..441879a9d 100644
--- a/hw/sh_pci.c
+++ b/hw/sh_pci.c
@@ -47,10 +47,15 @@ static void sh_pci_reg_write (void *p, target_phys_addr_t addr, uint32_t val)
         pcic->par = val;
         break;
     case 0x1c4:
-        pcic->mbr = val;
+        pcic->mbr = val & 0xff000001;
         break;
     case 0x1c8:
-        pcic->iobr = val;
+        if ((val & 0xfffc0000) != (pcic->iobr & 0xfffc0000)) {
+            cpu_register_physical_memory(pcic->iobr & 0xfffc0000, 0x40000,
+                                         IO_MEM_UNASSIGNED);
+            pcic->iobr = val & 0xfffc0001;
+            isa_mmio_init(pcic->iobr & 0xfffc0000, 0x40000);
+        }
         break;
     case 0x220:
         pci_data_write(pcic->bus, pcic->par, val, 4);
@@ -66,89 +71,16 @@ static uint32_t sh_pci_reg_read (void *p, target_phys_addr_t addr)
         return le32_to_cpup((uint32_t*)(pcic->dev->config + addr));
     case 0x1c0:
         return pcic->par;
+    case 0x1c4:
+        return pcic->mbr;
+    case 0x1c8:
+        return pcic->iobr;
     case 0x220:
         return pci_data_read(pcic->bus, pcic->par, 4);
     }
     return 0;
 }
 
-static void sh_pci_data_write (SHPCIC *pcic, target_phys_addr_t addr,
-                               uint32_t val, int size)
-{
-    pci_data_write(pcic->bus, addr + pcic->mbr, val, size);
-}
-
-static uint32_t sh_pci_mem_read (SHPCIC *pcic, target_phys_addr_t addr,
-                                 int size)
-{
-    return pci_data_read(pcic->bus, addr + pcic->mbr, size);
-}
-
-static void sh_pci_writeb (void *p, target_phys_addr_t addr, uint32_t val)
-{
-    sh_pci_data_write(p, addr, val, 1);
-}
-
-static void sh_pci_writew (void *p, target_phys_addr_t addr, uint32_t val)
-{
-    sh_pci_data_write(p, addr, val, 2);
-}
-
-static void sh_pci_writel (void *p, target_phys_addr_t addr, uint32_t val)
-{
-    sh_pci_data_write(p, addr, val, 4);
-}
-
-static uint32_t sh_pci_readb (void *p, target_phys_addr_t addr)
-{
-    return sh_pci_mem_read(p, addr, 1);
-}
-
-static uint32_t sh_pci_readw (void *p, target_phys_addr_t addr)
-{
-    return sh_pci_mem_read(p, addr, 2);
-}
-
-static uint32_t sh_pci_readl (void *p, target_phys_addr_t addr)
-{
-    return sh_pci_mem_read(p, addr, 4);
-}
-
-static int sh_pci_addr2port(SHPCIC *pcic, target_phys_addr_t addr)
-{
-    return addr + pcic->iobr;
-}
-
-static void sh_pci_outb (void *p, target_phys_addr_t addr, uint32_t val)
-{
-    cpu_outb(sh_pci_addr2port(p, addr), val);
-}
-
-static void sh_pci_outw (void *p, target_phys_addr_t addr, uint32_t val)
-{
-    cpu_outw(sh_pci_addr2port(p, addr), val);
-}
-
-static void sh_pci_outl (void *p, target_phys_addr_t addr, uint32_t val)
-{
-    cpu_outl(sh_pci_addr2port(p, addr), val);
-}
-
-static uint32_t sh_pci_inb (void *p, target_phys_addr_t addr)
-{
-    return cpu_inb(sh_pci_addr2port(p, addr));
-}
-
-static uint32_t sh_pci_inw (void *p, target_phys_addr_t addr)
-{
-    return cpu_inw(sh_pci_addr2port(p, addr));
-}
-
-static uint32_t sh_pci_inl (void *p, target_phys_addr_t addr)
-{
-    return cpu_inl(sh_pci_addr2port(p, addr));
-}
-
 typedef struct {
     CPUReadMemoryFunc * const r[3];
     CPUWriteMemoryFunc * const w[3];
@@ -159,21 +91,11 @@ static MemOp sh_pci_reg = {
     { NULL, NULL, sh_pci_reg_write },
 };
 
-static MemOp sh_pci_mem = {
-    { sh_pci_readb, sh_pci_readw, sh_pci_readl },
-    { sh_pci_writeb, sh_pci_writew, sh_pci_writel },
-};
-
-static MemOp sh_pci_iop = {
-    { sh_pci_inb, sh_pci_inw, sh_pci_inl },
-    { sh_pci_outb, sh_pci_outw, sh_pci_outl },
-};
-
 PCIBus *sh_pci_register_bus(pci_set_irq_fn set_irq, pci_map_irq_fn map_irq,
                             void *opaque, int devfn_min, int nirq)
 {
     SHPCIC *p;
-    int mem, reg, iop;
+    int reg;
 
     p = qemu_mallocz(sizeof(SHPCIC));
     p->bus = pci_register_bus(NULL, "pci",
@@ -182,14 +104,11 @@ PCIBus *sh_pci_register_bus(pci_set_irq_fn set_irq, pci_map_irq_fn map_irq,
     p->dev = pci_register_device(p->bus, "SH PCIC", sizeof(PCIDevice),
                                  -1, NULL, NULL);
     reg = cpu_register_io_memory(sh_pci_reg.r, sh_pci_reg.w, p);
-    iop = cpu_register_io_memory(sh_pci_iop.r, sh_pci_iop.w, p);
-    mem = cpu_register_io_memory(sh_pci_mem.r, sh_pci_mem.w, p);
     cpu_register_physical_memory(0x1e200000, 0x224, reg);
-    cpu_register_physical_memory(0x1e240000, 0x40000, iop);
-    cpu_register_physical_memory(0x1d000000, 0x1000000, mem);
     cpu_register_physical_memory(0xfe200000, 0x224, reg);
-    cpu_register_physical_memory(0xfe240000, 0x40000, iop);
-    cpu_register_physical_memory(0xfd000000, 0x1000000, mem);
+
+    p->iobr = 0xfe240000;
+    isa_mmio_init(p->iobr, 0x40000);
 
     pci_config_set_vendor_id(p->dev->config, PCI_VENDOR_ID_HITACHI);
     pci_config_set_device_id(p->dev->config, PCI_DEVICE_ID_HITACHI_SH7751R);
diff --git a/hw/usb-uhci.c b/hw/usb-uhci.c
index dc52737ae..cfd77ebed 100644
--- a/hw/usb-uhci.c
+++ b/hw/usb-uhci.c
@@ -677,9 +677,6 @@ static int uhci_complete_td(UHCIState *s, UHCI_TD *td, UHCIAsync *async, uint32_
 
     ret = async->packet.len;
 
-    if (td->ctrl & TD_CTRL_IOC)
-        *int_mask |= 0x01;
-
     if (td->ctrl & TD_CTRL_IOS)
         td->ctrl &= ~TD_CTRL_ACTIVE;
 
@@ -693,6 +690,8 @@ static int uhci_complete_td(UHCIState *s, UHCI_TD *td, UHCIAsync *async, uint32_
        here.  The docs are somewhat unclear, but win2k relies on this
        behavior.  */
     td->ctrl &= ~(TD_CTRL_ACTIVE | TD_CTRL_NAK);
+    if (td->ctrl & TD_CTRL_IOC)
+        *int_mask |= 0x01;
 
     if (pid == USB_TOKEN_IN) {
         if (len > max_len) {
@@ -750,6 +749,8 @@ out:
         if (err == 0) {
             td->ctrl &= ~TD_CTRL_ACTIVE;
             s->status |= UHCI_STS_USBERR;
+            if (td->ctrl & TD_CTRL_IOC)
+                *int_mask |= 0x01;
             uhci_update_irq(s);
         }
     }
diff --git a/json-parser.c b/json-parser.c
index 2ab6f6c11..3497cd365 100644
--- a/json-parser.c
+++ b/json-parser.c
@@ -266,7 +266,7 @@ static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_l
 
     peek = qlist_peek(working);
     key = parse_value(ctxt, &working, ap);
-    if (qobject_type(key) != QTYPE_QSTRING) {
+    if (!key || qobject_type(key) != QTYPE_QSTRING) {
         parse_error(ctxt, peek, "key is not a string in object");
         goto out;
     }
diff --git a/linux-user/main.c b/linux-user/main.c
index e51539eff..802bd8852 100644
--- a/linux-user/main.c
+++ b/linux-user/main.c
@@ -2684,7 +2684,7 @@ int main(int argc, char **argv, char **envp)
 #endif
 #elif defined(TARGET_PPC)
 #ifdef TARGET_PPC64
-        cpu_model = "970";
+        cpu_model = "970fx";
 #else
         cpu_model = "750";
 #endif
diff --git a/net.c b/net.c
index 029b0d79b..28b6b43a9 100644
--- a/net.c
+++ b/net.c
@@ -1187,26 +1187,6 @@ void net_host_device_remove(Monitor *mon, const QDict *qdict)
     qemu_del_vlan_client(vc);
 }
 
-void net_set_boot_mask(int net_boot_mask)
-{
-    int i;
-
-    /* Only the first four NICs may be bootable */
-    net_boot_mask = net_boot_mask & 0xF;
-
-    for (i = 0; i < nb_nics; i++) {
-        if (net_boot_mask & (1 << i)) {
-            nd_table[i].bootable = 1;
-            net_boot_mask &= ~(1 << i);
-        }
-    }
-
-    if (net_boot_mask) {
-        fprintf(stderr, "Cannot boot from non-existent NIC\n");
-        exit(1);
-    }
-}
-
 void do_info_network(Monitor *mon)
 {
     VLANState *vlan;
diff --git a/net.h b/net.h
index 673b3554d..209431dd0 100644
--- a/net.h
+++ b/net.h
@@ -163,7 +163,6 @@ int net_client_parse(QemuOptsList *opts_list, const char *str);
 int net_init_clients(void);
 void net_check_clients(void);
 void net_cleanup(void);
-void net_set_boot_mask(int boot_mask);
 void net_host_device_add(Monitor *mon, const QDict *qdict);
 void net_host_device_remove(Monitor *mon, const QDict *qdict);
 
diff --git a/qemu-char.c b/qemu-char.c
index 5a1b535bd..30f39caae 100644
--- a/qemu-char.c
+++ b/qemu-char.c
@@ -2317,8 +2317,7 @@ QemuOpts *qemu_chr_parse_compat(const char *label, const char *filename)
         qemu_opt_set(opts, "backend", "udp");
         if (sscanf(p, "%64[^:]:%32[^@,]%n", host, port, &pos) < 2) {
             host[0] = 0;
-            if (sscanf(p, ":%32[^,]%n", port, &pos) < 1) {
-                fprintf(stderr, "udp #1\n");
+            if (sscanf(p, ":%32[^@,]%n", port, &pos) < 1) {
                 goto fail;
             }
         }
@@ -2329,7 +2328,6 @@ QemuOpts *qemu_chr_parse_compat(const char *label, const char *filename)
             if (sscanf(p, "%64[^:]:%32[^,]%n", host, port, &pos) < 2) {
                 host[0] = 0;
                 if (sscanf(p, ":%32[^,]%n", port, &pos) < 1) {
-                    fprintf(stderr, "udp #2\n");
                     goto fail;
                 }
             }
@@ -2357,7 +2355,6 @@ QemuOpts *qemu_chr_parse_compat(const char *label, const char *filename)
     }
 
 fail:
-    fprintf(stderr, "%s: fail on \"%s\"\n", __FUNCTION__, filename);
     qemu_opts_del(opts);
     return NULL;
 }
diff --git a/qemu-common.h b/qemu-common.h
index 1c5c0b222..b604ddfa0 100644
--- a/qemu-common.h
+++ b/qemu-common.h
@@ -54,6 +54,10 @@ struct iovec {
     void *iov_base;
     size_t iov_len;
 };
+/*
+ * Use the same value as Linux for now.
+ */
+#define IOV_MAX		1024
 #else
 #include <sys/uio.h>
 #endif
diff --git a/qemu-options.hx b/qemu-options.hx
index e2fbebd9b..65bb11115 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -969,7 +969,7 @@ the guest, use the following:
 
 @example
 # on the host
-qemu -net user,hostfwd=tcp:5555::23 [...]
+qemu -net user,hostfwd=tcp::5555-:23 [...]
 telnet localhost 5555
 @end example
 
diff --git a/target-i386/translate.c b/target-i386/translate.c
index 511a4eae9..3de65bd1d 100644
--- a/target-i386/translate.c
+++ b/target-i386/translate.c
@@ -2047,8 +2047,8 @@ static void gen_lea_modrm(DisasContext *s, int modrm, int *reg_ptr, int *offset_
                 gen_op_movl_A0_im(disp);
             }
         }
-        /* XXX: index == 4 is always invalid */
-        if (havesib && (index != 4 || scale != 0)) {
+        /* index == 4 means no index */
+        if (havesib && (index != 4)) {
 #ifdef TARGET_X86_64
             if (s->aflag == 2) {
                 gen_op_addq_A0_reg_sN(scale, index);
@@ -3165,12 +3165,17 @@ static void gen_sse(DisasContext *s, int b, target_ulong pc_start, int rex_r)
         case 0x1e7: /* movntdq */
         case 0x02b: /* movntps */
         case 0x12b: /* movntps */
-        case 0x3f0: /* lddqu */
             if (mod == 3)
                 goto illegal_op;
             gen_lea_modrm(s, modrm, &reg_addr, &offset_addr);
             gen_sto_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg]));
             break;
+        case 0x3f0: /* lddqu */
+            if (mod == 3)
+                goto illegal_op;
+            gen_lea_modrm(s, modrm, &reg_addr, &offset_addr);
+            gen_ldo_env_A0(s->mem_index, offsetof(CPUX86State,xmm_regs[reg]));
+            break;
         case 0x22b: /* movntss */
         case 0x32b: /* movntsd */
             if (mod == 3)
@@ -4591,9 +4596,7 @@ static target_ulong disas_insn(DisasContext *s, target_ulong pc_start)
                 /* operand size for jumps is 64 bit */
                 ot = OT_QUAD;
             } else if (op == 3 || op == 5) {
-                /* for call calls, the operand is 16 or 32 bit, even
-                   in long mode */
-                ot = dflag ? OT_LONG : OT_WORD;
+                ot = dflag ? OT_LONG + (rex_w == 1) : OT_WORD;
             } else if (op == 6) {
                 /* default push size is 64 bit */
                 ot = dflag ? OT_QUAD : OT_WORD;
diff --git a/target-sh4/cpu.h b/target-sh4/cpu.h
index 366e7986e..015d59845 100644
--- a/target-sh4/cpu.h
+++ b/target-sh4/cpu.h
@@ -167,6 +167,7 @@ int cpu_sh4_handle_mmu_fault(CPUSH4State * env, target_ulong address, int rw,
 void do_interrupt(CPUSH4State * env);
 
 void sh4_cpu_list(FILE *f, int (*cpu_fprintf)(FILE *f, const char *fmt, ...));
+void cpu_sh4_invalidate_tlb(CPUSH4State *s);
 void cpu_sh4_write_mmaped_utlb_addr(CPUSH4State *s, target_phys_addr_t addr,
 				    uint32_t mem_value);
 
@@ -222,6 +223,7 @@ enum {
 /* MMU control register */
 #define MMUCR    0x1F000010
 #define MMUCR_AT (1<<0)
+#define MMUCR_TI (1<<2)
 #define MMUCR_SV (1<<8)
 #define MMUCR_URC_BITS (6)
 #define MMUCR_URC_OFFSET (10)
diff --git a/target-sh4/helper.c b/target-sh4/helper.c
index 088d36a5f..29b681398 100644
--- a/target-sh4/helper.c
+++ b/target-sh4/helper.c
@@ -377,7 +377,7 @@ static int get_mmu_address(CPUState * env, target_ulong * physical,
 	n = find_itlb_entry(env, address, use_asid, 1);
 	if (n >= 0) {
 	    matching = &env->itlb[n];
-	    if ((env->sr & SR_MD) & !(matching->pr & 2))
+	    if (!(env->sr & SR_MD) && !(matching->pr & 2))
 		n = MMU_ITLB_VIOLATION;
 	    else
 		*prot = PAGE_READ;
@@ -430,7 +430,7 @@ static int get_physical_address(CPUState * env, target_ulong * physical,
     if ((address >= 0x80000000 && address < 0xc0000000) ||
 	address >= 0xe0000000) {
 	if (!(env->sr & SR_MD)
-	    && (address < 0xe0000000 || address > 0xe4000000)) {
+	    && (address < 0xe0000000 || address >= 0xe4000000)) {
 	    /* Unauthorized access in user mode (only store queues are available) */
 	    fprintf(stderr, "Unauthorized access\n");
 	    if (rw == 0)
@@ -574,6 +574,24 @@ void cpu_load_tlb(CPUSH4State * env)
     entry->tc   = (uint8_t)cpu_ptea_tc(env->ptea);
 }
 
+ void cpu_sh4_invalidate_tlb(CPUSH4State *s)
+{
+    int i;
+
+    /* UTLB */
+    for (i = 0; i < UTLB_SIZE; i++) {
+        tlb_t * entry = &s->utlb[i];
+        entry->v = 0;
+    }
+    /* ITLB */
+    for (i = 0; i < UTLB_SIZE; i++) {
+        tlb_t * entry = &s->utlb[i];
+        entry->v = 0;
+    }
+
+    tlb_flush(s, 1);
+}
+
 void cpu_sh4_write_mmaped_utlb_addr(CPUSH4State *s, target_phys_addr_t addr,
 				    uint32_t mem_value)
 {
diff --git a/target-sh4/translate.c b/target-sh4/translate.c
index 895b978d3..7f9527a83 100644
--- a/target-sh4/translate.c
+++ b/target-sh4/translate.c
@@ -1905,7 +1905,7 @@ gen_intermediate_code_internal(CPUState * env, TranslationBlock * tb,
     ctx.bstate = BS_NONE;
     ctx.sr = env->sr;
     ctx.fpscr = env->fpscr;
-    ctx.memidx = (env->sr & SR_MD) ? 1 : 0;
+    ctx.memidx = (env->sr & SR_MD) == 0 ? 1 : 0;
     /* We don't know if the delayed pc came from a dynamic or static branch,
        so assume it is a dynamic branch.  */
     ctx.delayed_pc = -1; /* use delayed pc from env pointer */
diff --git a/tcg/arm/tcg-target.c b/tcg/arm/tcg-target.c
index f8d626d02..81f533f73 100644
--- a/tcg/arm/tcg-target.c
+++ b/tcg/arm/tcg-target.c
@@ -1697,12 +1697,15 @@ static inline void tcg_out_movi(TCGContext *s, TCGType type,
 
 void tcg_target_qemu_prologue(TCGContext *s)
 {
-    /* stmdb sp!, { r9 - r11, lr } */
-    tcg_out32(s, (COND_AL << 28) | 0x092d4e00);
+    /* Theoretically there is no need to save r12, but an
+       even number of registers to be saved as per EABI */
+
+    /* stmdb sp!, { r4 - r12, lr } */
+    tcg_out32(s, (COND_AL << 28) | 0x092d5ff0);
 
     tcg_out_bx(s, COND_AL, TCG_REG_R0);
     tb_ret_addr = s->code_ptr;
 
-    /* ldmia sp!, { r9 - r11, pc } */
-    tcg_out32(s, (COND_AL << 28) | 0x08bd8e00);
+    /* ldmia sp!, { r4 - r12, pc } */
+    tcg_out32(s, (COND_AL << 28) | 0x08bd9ff0);
 }
diff --git a/tcg/mips/tcg-target.c b/tcg/mips/tcg-target.c
index 807b8fdfa..79ba9a662 100644
--- a/tcg/mips/tcg-target.c
+++ b/tcg/mips/tcg-target.c
@@ -340,6 +340,17 @@ static inline void tcg_out_opc_imm(TCGContext *s, int opc, int rt, int rs, int i
 }
 
 /*
+ * Type branch
+ */
+static inline void tcg_out_opc_br(TCGContext *s, int opc, int rt, int rs)
+{
+    /* We need to keep the offset unchanged for retranslation */
+    uint16_t offset = (uint16_t)(*(uint32_t *) &s->code_ptr);
+
+    tcg_out_opc_imm(s, opc, rt, rs, offset);
+}
+
+/*
  * Type sa
  */
 static inline void tcg_out_opc_sa(TCGContext *s, int opc, int rd, int rt, int sa)
@@ -469,42 +480,42 @@ static void tcg_out_brcond(TCGContext *s, int cond, int arg1,
 
     switch (cond) {
     case TCG_COND_EQ:
-        tcg_out_opc_imm(s, OPC_BEQ, arg1, arg2, 0);
+        tcg_out_opc_br(s, OPC_BEQ, arg1, arg2);
         break;
     case TCG_COND_NE:
-        tcg_out_opc_imm(s, OPC_BNE, arg1, arg2, 0);
+        tcg_out_opc_br(s, OPC_BNE, arg1, arg2);
         break;
     case TCG_COND_LT:
         tcg_out_opc_reg(s, OPC_SLT, TCG_REG_AT, arg1, arg2);
-        tcg_out_opc_imm(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO);
         break;
     case TCG_COND_LTU:
         tcg_out_opc_reg(s, OPC_SLTU, TCG_REG_AT, arg1, arg2);
-        tcg_out_opc_imm(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO);
         break;
     case TCG_COND_GE:
         tcg_out_opc_reg(s, OPC_SLT, TCG_REG_AT, arg1, arg2);
-        tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO);
         break;
     case TCG_COND_GEU:
         tcg_out_opc_reg(s, OPC_SLTU, TCG_REG_AT, arg1, arg2);
-        tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO);
         break;
     case TCG_COND_LE:
         tcg_out_opc_reg(s, OPC_SLT, TCG_REG_AT, arg2, arg1);
-        tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO);
         break;
     case TCG_COND_LEU:
         tcg_out_opc_reg(s, OPC_SLTU, TCG_REG_AT, arg2, arg1);
-        tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BEQ, TCG_REG_AT, TCG_REG_ZERO);
         break;
     case TCG_COND_GT:
         tcg_out_opc_reg(s, OPC_SLT, TCG_REG_AT, arg2, arg1);
-        tcg_out_opc_imm(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO);
         break;
     case TCG_COND_GTU:
         tcg_out_opc_reg(s, OPC_SLTU, TCG_REG_AT, arg2, arg1);
-        tcg_out_opc_imm(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO, 0);
+        tcg_out_opc_br(s, OPC_BNE, TCG_REG_AT, TCG_REG_ZERO);
         break;
     default:
         tcg_abort();
@@ -553,7 +564,7 @@ static void tcg_out_brcond2(TCGContext *s, int cond, int arg1,
     }
 
     label_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BNE, arg2, arg4, 0);
+    tcg_out_opc_br(s, OPC_BNE, arg2, arg4);
     tcg_out_nop(s);
 
     switch(cond) {
@@ -670,20 +681,20 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args,
 
 # if TARGET_LONG_BITS == 64
     label3_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BNE, TCG_REG_T0, TCG_REG_AT, 0);
+    tcg_out_opc_br(s, OPC_BNE, TCG_REG_T0, TCG_REG_AT);
     tcg_out_nop(s);
 
     tcg_out_opc_imm(s, OPC_LW, TCG_REG_AT, TCG_REG_A0,
                     offsetof(CPUState, tlb_table[mem_index][0].addr_read) + addr_memh);
 
     label1_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BEQ, addr_regh, TCG_REG_AT, 0);
+    tcg_out_opc_br(s, OPC_BEQ, addr_regh, TCG_REG_AT);
     tcg_out_nop(s);
 
     reloc_pc16(label3_ptr, (tcg_target_long) s->code_ptr);
 # else
     label1_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_T0, TCG_REG_AT, 0);
+    tcg_out_opc_br(s, OPC_BEQ, TCG_REG_T0, TCG_REG_AT);
     tcg_out_nop(s);
 # endif
 
@@ -725,7 +736,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg *args,
     }
 
     label2_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_ZERO, TCG_REG_ZERO, 0);
+    tcg_out_opc_br(s, OPC_BEQ, TCG_REG_ZERO, TCG_REG_ZERO);
     tcg_out_nop(s);
 
     /* label1: fast path */
@@ -857,20 +868,20 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args,
 
 # if TARGET_LONG_BITS == 64
     label3_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BNE, TCG_REG_T0, TCG_REG_AT, 0);
+    tcg_out_opc_br(s, OPC_BNE, TCG_REG_T0, TCG_REG_AT);
     tcg_out_nop(s);
 
     tcg_out_opc_imm(s, OPC_LW, TCG_REG_AT, TCG_REG_A0,
                     offsetof(CPUState, tlb_table[mem_index][0].addr_write) + addr_memh);
 
     label1_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BEQ, addr_regh, TCG_REG_AT, 0);
+    tcg_out_opc_br(s, OPC_BEQ, addr_regh, TCG_REG_AT);
     tcg_out_nop(s);
 
     reloc_pc16(label3_ptr, (tcg_target_long) s->code_ptr);
 # else
     label1_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_T0, TCG_REG_AT, 0);
+    tcg_out_opc_br(s, OPC_BEQ, TCG_REG_T0, TCG_REG_AT);
     tcg_out_nop(s);
 # endif
 
@@ -911,7 +922,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args,
     tcg_out_nop(s);
 
     label2_ptr = s->code_ptr;
-    tcg_out_opc_imm(s, OPC_BEQ, TCG_REG_ZERO, TCG_REG_ZERO, 0);
+    tcg_out_opc_br(s, OPC_BEQ, TCG_REG_ZERO, TCG_REG_ZERO);
     tcg_out_nop(s);
 
     /* label1: fast path */
@@ -1262,7 +1273,9 @@ static const TCGTargetOpDef mips_op_defs[] = {
 };
 
 static int tcg_target_callee_save_regs[] = {
+#if 0 /* used for the global env (TCG_AREG0), so no need to save */
     TCG_REG_S0,
+#endif
     TCG_REG_S1,
     TCG_REG_S2,
     TCG_REG_S3,
@@ -1271,8 +1284,7 @@ static int tcg_target_callee_save_regs[] = {
     TCG_REG_S6,
     TCG_REG_S7,
     TCG_REG_GP,
-    /* TCG_REG_FP, */ /* currently used for the global env, so np
-                         need to save */
+    TCG_REG_FP,
     TCG_REG_RA,       /* should be last for ABI compliance */
 };
 
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index 46760a50e..baae6d85c 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -92,9 +92,9 @@ enum {
 #undef TCG_TARGET_HAS_ext16u_i32   /* andi rt, rs, 0xffff */
 
 /* Note: must be synced with dyngen-exec.h */
-#define TCG_AREG0 TCG_REG_FP
-#define TCG_AREG1 TCG_REG_S0
-#define TCG_AREG2 TCG_REG_S1
+#define TCG_AREG0 TCG_REG_S0
+#define TCG_AREG1 TCG_REG_S1
+#define TCG_AREG2 TCG_REG_FP
 
 #include <sys/cachectl.h>
 
diff --git a/vl.c b/vl.c
index 59f8084fa..fdee28916 100644
--- a/vl.c
+++ b/vl.c
@@ -2597,9 +2597,9 @@ static void numa_add(const char *optarg)
                         fprintf(stderr,
                             "only 63 CPUs in NUMA mode supported.\n");
                     }
-                    value = (1 << (endvalue + 1)) - (1 << value);
+                    value = (2ULL << endvalue) - (1ULL << value);
                 } else {
-                    value = 1 << value;
+                    value = 1ULL << value;
                 }
             }
             node_cpumask[nodenr] = value;
@@ -3515,6 +3515,8 @@ static int cpu_can_run(CPUState *env)
         return 0;
     if (env->stopped)
         return 0;
+    if (!vm_running)
+        return 0;
     return 1;
 }
 
@@ -4909,7 +4911,7 @@ int main(int argc, char **argv, char **envp)
     const char *gdbstub_dev = NULL;
     uint32_t boot_devices_bitmap = 0;
     int i;
-    int snapshot, linux_boot, net_boot;
+    int snapshot, linux_boot;
     const char *initrd_filename;
     const char *kernel_filename, *kernel_cmdline;
     char boot_devices[33] = "cad"; /* default to HD->floppy->CD-ROM */
@@ -5428,6 +5430,9 @@ int main(int argc, char **argv, char **envp)
             case QEMU_OPTION_serial:
                 add_device_config(DEV_SERIAL, optarg);
                 default_serial = 0;
+                if (strncmp(optarg, "mon:", 4) == 0) {
+                    default_monitor = 0;
+                }
                 break;
             case QEMU_OPTION_watchdog:
                 if (watchdog) {
@@ -5446,10 +5451,16 @@ int main(int argc, char **argv, char **envp)
             case QEMU_OPTION_virtiocon:
                 add_device_config(DEV_VIRTCON, optarg);
                 default_virtcon = 0;
+                if (strncmp(optarg, "mon:", 4) == 0) {
+                    default_monitor = 0;
+                }
                 break;
             case QEMU_OPTION_parallel:
                 add_device_config(DEV_PARALLEL, optarg);
                 default_parallel = 0;
+                if (strncmp(optarg, "mon:", 4) == 0) {
+                    default_monitor = 0;
+                }
                 break;
 	    case QEMU_OPTION_loadvm:
 		loadvm = optarg;
@@ -5905,6 +5916,12 @@ int main(int argc, char **argv, char **envp)
             fprintf(stderr, "Could not initialize KVM, will disable KVM support\n");
             kvm_allowed = 0;
         }
+    } else {
+        /* without kvm enabled, we can only support 4095 MB RAM */
+        if (ram_size > (4095UL << 20)) {
+            fprintf(stderr, "qemu: without kvm support at most 4095 MB RAM can be simulated\n");
+            exit(1);
+        }
     }
 
     if (qemu_init_main_loop()) {
@@ -5948,9 +5965,6 @@ int main(int argc, char **argv, char **envp)
         exit(1);
     }
 
-    net_boot = (boot_devices_bitmap >> ('n' - 'a')) & 0xF;
-    net_set_boot_mask(net_boot);
-
     /* init the bluetooth world */
     if (foreach_device_config(DEV_BT, bt_parse))
         exit(1);
diff --git a/vnc.c b/vnc.c
index a1f9c9293..8566a4ba1 100644
--- a/vnc.c
+++ b/vnc.c
@@ -1421,8 +1421,10 @@ static void pointer_event(VncState *vs, int button_mask, int x, int y)
         dz = 1;
 
     if (vs->absolute) {
-        kbd_mouse_event(x * 0x7FFF / (ds_get_width(vs->ds) - 1),
-                        y * 0x7FFF / (ds_get_height(vs->ds) - 1),
+        kbd_mouse_event(ds_get_width(vs->ds) > 1 ?
+                          x * 0x7FFF / (ds_get_width(vs->ds) - 1) : 0x4000,
+                        ds_get_height(vs->ds) > 1 ?
+                          y * 0x7FFF / (ds_get_height(vs->ds) - 1) : 0x4000,
                         dz, buttons);
     } else if (vnc_has_feature(vs, VNC_FEATURE_POINTER_TYPE_CHANGE)) {
         x -= 0x7FFF;