diff mbox

[v2,6/7] virtio-blk: Unify {non-, }dataplane's request handlings

Message ID 1402986730-25642-7-git-send-email-famz@redhat.com
State New
Headers show

Commit Message

Fam Zheng June 17, 2014, 6:32 a.m. UTC
This drops request handling code from dataplane, and uses code from
hw/block/virtio-blk.c.

It starts to use multiwrite as non-dataplane does.

Dataplane sets VirtIOBlock.complete_request to vring version, and calls
into non-dataplane's process handling. In complete_request_early,
qiov.size is added to vring push length, because it's also called in rw
completion now.

Signed-off-by: Fam Zheng <famz@redhat.com>
---
 hw/block/dataplane/virtio-blk.c | 183 +++++-----------------------------------
 1 file changed, 19 insertions(+), 164 deletions(-)

Comments

Max Reitz July 1, 2014, 11:29 a.m. UTC | #1
On 17.06.2014 08:32, Fam Zheng wrote:
> This drops request handling code from dataplane, and uses code from
> hw/block/virtio-blk.c.
>
> It starts to use multiwrite as non-dataplane does.
>
> Dataplane sets VirtIOBlock.complete_request to vring version, and calls
> into non-dataplane's process handling. In complete_request_early,
> qiov.size is added to vring push length, because it's also called in rw
> completion now.
>
> Signed-off-by: Fam Zheng <famz@redhat.com>
> ---
>   hw/block/dataplane/virtio-blk.c | 183 +++++-----------------------------------
>   1 file changed, 19 insertions(+), 164 deletions(-)

Is it intended that qemu segfaults after this commit with x-data-plane=on?

$ ./qemu-img create -f qcow2 test.qcow2 64M
$ x86_64-softmmu/qemu-system-x86_64 -drive 
if=none,file=test.qcow2,id=drv0 -device 
virtio-blk-pci,drive=drv0,x-data-plane=on
[1]    4604 segmentation fault  x86_64-softmmu/qemu-system-x86_64 -drive 
if=none,file=test.qcow2,id=drv0

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffd75ff700 (LWP 5217)]
0x00005555556373af in virtio_blk_rw_complete (opaque=0x5555565ff5e0, 
ret=0) at hw/block/virtio-blk.c:99
99          bdrv_acct_done(req->dev->bs, &req->acct);
(gdb) print req
$1 = (VirtIOBlockReq *) 0x5555565ff5e0
(gdb) print req->dev
$2 = (VirtIOBlock *) 0x0
(gdb) bt
#0  0x00005555556373af in virtio_blk_rw_complete (opaque=0x5555565ff5e0, 
ret=0) at hw/block/virtio-blk.c:99
#1  0x0000555555840ebe in bdrv_co_em_bh (opaque=0x5555566152d0) at 
block.c:4675
#2  0x000055555583de77 in aio_bh_poll (ctx=ctx@entry=0x5555563a8150) at 
async.c:81
#3  0x000055555584b7a7 in aio_poll (ctx=0x5555563a8150, 
blocking=blocking@entry=true) at aio-posix.c:188
#4  0x00005555556e520e in iothread_run (opaque=0x5555563a7fd8) at 
iothread.c:41
#5  0x00007ffff42ba124 in start_thread () from /usr/lib/libpthread.so.0
#6  0x00007ffff16d14bd in clone () from /usr/lib/libc.so.6

Max
Stefan Hajnoczi July 1, 2014, 1:22 p.m. UTC | #2
On Tue, Jul 01, 2014 at 01:29:36PM +0200, Max Reitz wrote:
> On 17.06.2014 08:32, Fam Zheng wrote:
> >This drops request handling code from dataplane, and uses code from
> >hw/block/virtio-blk.c.
> >
> >It starts to use multiwrite as non-dataplane does.
> >
> >Dataplane sets VirtIOBlock.complete_request to vring version, and calls
> >into non-dataplane's process handling. In complete_request_early,
> >qiov.size is added to vring push length, because it's also called in rw
> >completion now.
> >
> >Signed-off-by: Fam Zheng <famz@redhat.com>
> >---
> >  hw/block/dataplane/virtio-blk.c | 183 +++++-----------------------------------
> >  1 file changed, 19 insertions(+), 164 deletions(-)
> 
> Is it intended that qemu segfaults after this commit with x-data-plane=on?
> 
> $ ./qemu-img create -f qcow2 test.qcow2 64M
> $ x86_64-softmmu/qemu-system-x86_64 -drive if=none,file=test.qcow2,id=drv0
> -device virtio-blk-pci,drive=drv0,x-data-plane=on
> [1]    4604 segmentation fault  x86_64-softmmu/qemu-system-x86_64 -drive
> if=none,file=test.qcow2,id=drv0
> 
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0x7fffd75ff700 (LWP 5217)]
> 0x00005555556373af in virtio_blk_rw_complete (opaque=0x5555565ff5e0, ret=0)
> at hw/block/virtio-blk.c:99
> 99          bdrv_acct_done(req->dev->bs, &req->acct);
> (gdb) print req
> $1 = (VirtIOBlockReq *) 0x5555565ff5e0
> (gdb) print req->dev
> $2 = (VirtIOBlock *) 0x0
> (gdb) bt
> #0  0x00005555556373af in virtio_blk_rw_complete (opaque=0x5555565ff5e0,
> ret=0) at hw/block/virtio-blk.c:99
> #1  0x0000555555840ebe in bdrv_co_em_bh (opaque=0x5555566152d0) at
> block.c:4675
> #2  0x000055555583de77 in aio_bh_poll (ctx=ctx@entry=0x5555563a8150) at
> async.c:81
> #3  0x000055555584b7a7 in aio_poll (ctx=0x5555563a8150,
> blocking=blocking@entry=true) at aio-posix.c:188
> #4  0x00005555556e520e in iothread_run (opaque=0x5555563a7fd8) at
> iothread.c:41
> #5  0x00007ffff42ba124 in start_thread () from /usr/lib/libpthread.so.0
> #6  0x00007ffff16d14bd in clone () from /usr/lib/libc.so.6

I'm looking into this and will send a patch to fix it.

Stefan
diff mbox

Patch

diff --git a/hw/block/dataplane/virtio-blk.c b/hw/block/dataplane/virtio-blk.c
index 29e7ec3..b6afa55 100644
--- a/hw/block/dataplane/virtio-blk.c
+++ b/hw/block/dataplane/virtio-blk.c
@@ -47,6 +47,8 @@  struct VirtIOBlockDataPlane {
 
     /* Operation blocker on BDS */
     Error *blocker;
+    void (*saved_complete_request)(struct VirtIOBlockReq *req,
+                                   unsigned char status);
 };
 
 /* Raise an interrupt to signal guest, if necessary */
@@ -59,179 +61,27 @@  static void notify_guest(VirtIOBlockDataPlane *s)
     event_notifier_set(s->guest_notifier);
 }
 
-static void complete_rdwr(void *opaque, int ret)
-{
-    VirtIOBlockReq *req = opaque;
-    struct virtio_blk_inhdr hdr;
-    int len;
-
-    if (likely(ret == 0)) {
-        hdr.status = VIRTIO_BLK_S_OK;
-        len = req->qiov.size;
-    } else {
-        hdr.status = VIRTIO_BLK_S_IOERR;
-        len = 0;
-    }
-
-    trace_virtio_blk_data_plane_complete_request(req->dev->dataplane,
-                                                 req->elem->index, ret);
-
-    stb_p(&req->in->status, hdr.status);
-
-    /* According to the virtio specification len should be the number of bytes
-     * written to, but for virtio-blk it seems to be the number of bytes
-     * transferred plus the status bytes.
-     */
-    vring_push(&req->dev->dataplane->vring, req->elem, len + sizeof(hdr));
-    notify_guest(req->dev->dataplane);
-    g_slice_free(VirtIOBlockReq, req);
-}
-
 static void complete_request_early(VirtIOBlockReq *req, unsigned char status)
 {
     stb_p(&req->in->status, status);
 
-    vring_push(&req->dev->dataplane->vring, req->elem, sizeof(*req->in));
+    vring_push(&req->dev->dataplane->vring, req->elem,
+               req->qiov.size + sizeof(*req->in));
     notify_guest(req->dev->dataplane);
     g_slice_free(VirtIOBlockReq, req);
 }
 
-/* Get disk serial number */
-static void do_get_id_cmd(VirtIOBlockReq *req,
-                          struct iovec *iov, unsigned int iov_cnt)
-{
-    char id[VIRTIO_BLK_ID_BYTES];
-
-    /* Serial number not NUL-terminated when longer than buffer */
-    strncpy(id, req->dev->blk.serial ? req->dev->blk.serial : "", sizeof(id));
-    iov_from_buf(iov, iov_cnt, 0, id, sizeof(id));
-    complete_request_early(req, VIRTIO_BLK_S_OK);
-}
-
-
-static void do_rdwr_cmd(bool read, VirtIOBlockReq *req,
-                        struct iovec *iov, unsigned iov_cnt)
-{
-    QEMUIOVector *qiov;
-    int nb_sectors;
-    int64_t sector_num;
-
-    qemu_iovec_init_external(&req->qiov, iov, iov_cnt);
-
-    qiov = &req->qiov;
-
-    sector_num = req->out.sector * 512 / BDRV_SECTOR_SIZE;
-    nb_sectors = qiov->size / BDRV_SECTOR_SIZE;
-
-    if (read) {
-        bdrv_aio_readv(req->dev->blk.conf.bs,
-                       sector_num, qiov, nb_sectors,
-                       complete_rdwr, req);
-    } else {
-        bdrv_aio_writev(req->dev->blk.conf.bs,
-                        sector_num, qiov, nb_sectors,
-                        complete_rdwr, req);
-    }
-}
-
-static void complete_flush(void *opaque, int ret)
-{
-    VirtIOBlockReq *req = opaque;
-    unsigned char status;
-
-    if (ret == 0) {
-        status = VIRTIO_BLK_S_OK;
-    } else {
-        status = VIRTIO_BLK_S_IOERR;
-    }
-
-    complete_request_early(req, status);
-}
-
-static void do_flush_cmd(VirtIOBlockReq *req)
-{
-
-    bdrv_aio_flush(req->dev->blk.conf.bs, complete_flush, req);
-}
-
-static void do_scsi_cmd(VirtIOBlockReq *req)
-{
-    int status;
-
-    status = virtio_blk_handle_scsi_req(req->dev, req->elem);
-    complete_request_early(req, status);
-}
-
-static int process_request(VirtIOBlockDataPlane *s, VirtQueueElement *elem)
-{
-    struct iovec *iov = elem->out_sg;
-    struct iovec *in_iov = elem->in_sg;
-    unsigned out_num = elem->out_num;
-    unsigned in_num = elem->in_num;
-    VirtIOBlockReq *req;
-
-    req = g_slice_new(VirtIOBlockReq);
-    req->dev = VIRTIO_BLK(s->vdev);
-    req->elem = elem;
-    /* Copy in outhdr */
-    if (unlikely(iov_to_buf(iov, out_num, 0, &req->out,
-                            sizeof(req->out)) != sizeof(req->out))) {
-        error_report("virtio-blk request outhdr too short");
-        g_slice_free(VirtIOBlockReq, req);
-        return -EFAULT;
-    }
-    iov_discard_front(&iov, &out_num, sizeof(req->out));
-
-    /* We are likely safe with the iov_len check, because inhdr is only 1 byte,
-     * but checking here in case the header gets bigger in the future. */
-    if (in_num < 1 || in_iov[in_num - 1].iov_len < sizeof(*req->in)) {
-        error_report("virtio-blk request inhdr too short");
-        return -EFAULT;
-    }
-
-    /* Grab inhdr for later */
-    req->in = (void *)in_iov[in_num - 1].iov_base
-            + in_iov[in_num - 1].iov_len - sizeof(*req->in);
-    iov_discard_back(in_iov, &in_num, sizeof(struct virtio_blk_inhdr));
-
-    /* TODO Linux sets the barrier bit even when not advertised! */
-    req->out.type &= ~VIRTIO_BLK_T_BARRIER;
-
-    switch (req->out.type) {
-    case VIRTIO_BLK_T_IN:
-        do_rdwr_cmd(true, req, in_iov, in_num);
-        return 0;
-
-    case VIRTIO_BLK_T_OUT:
-        do_rdwr_cmd(false, req, iov, out_num);
-        return 0;
-
-    case VIRTIO_BLK_T_SCSI_CMD:
-        do_scsi_cmd(req);
-        return 0;
-
-    case VIRTIO_BLK_T_FLUSH:
-        do_flush_cmd(req);
-        return 0;
-
-    case VIRTIO_BLK_T_GET_ID:
-        do_get_id_cmd(req, in_iov, in_num);
-        return 0;
-
-    default:
-        error_report("virtio-blk unsupported request type %#x", req->out.type);
-        g_slice_free(VirtIOBlockReq, req);
-        return -EFAULT;
-    }
-}
-
 static void handle_notify(EventNotifier *e)
 {
     VirtIOBlockDataPlane *s = container_of(e, VirtIOBlockDataPlane,
                                            host_notifier);
 
     VirtQueueElement *elem;
+    VirtIOBlockReq *req;
     int ret;
+    MultiReqBuffer mrb = {
+        .num_writes = 0,
+    };
 
     event_notifier_test_and_clear(&s->host_notifier);
     for (;;) {
@@ -248,14 +98,14 @@  static void handle_notify(EventNotifier *e)
             trace_virtio_blk_data_plane_process_request(s, elem->out_num,
                                                         elem->in_num, elem->index);
 
-            if (process_request(s, elem) < 0) {
-                vring_set_broken(&s->vring);
-                vring_free_element(elem);
-                ret = -EFAULT;
-                break;
-            }
+            req = g_slice_new(VirtIOBlockReq);
+            req->dev = VIRTIO_BLK(s->vdev);
+            req->elem = elem;
+            virtio_blk_handle_request(req, &mrb);
         }
 
+        virtio_submit_multiwrite(s->blk->conf.bs, &mrb);
+
         if (likely(ret == -EAGAIN)) { /* vring emptied */
             /* Re-enable guest->host notifies and stop processing the vring.
              * But if the guest has snuck in more descriptors, keep processing.
@@ -275,6 +125,7 @@  void virtio_blk_data_plane_create(VirtIODevice *vdev, VirtIOBlkConf *blk,
                                   Error **errp)
 {
     VirtIOBlockDataPlane *s;
+    VirtIOBlock *vblk = VIRTIO_BLK(vdev);
     Error *local_err = NULL;
 
     *dataplane = NULL;
@@ -317,6 +168,8 @@  void virtio_blk_data_plane_create(VirtIODevice *vdev, VirtIOBlkConf *blk,
     bdrv_op_block_all(blk->conf.bs, s->blocker);
 
     *dataplane = s;
+    s->saved_complete_request = vblk->complete_request;
+    vblk->complete_request = complete_request_early;
 }
 
 /* Context: QEMU global mutex held */
@@ -391,10 +244,12 @@  void virtio_blk_data_plane_stop(VirtIOBlockDataPlane *s)
 {
     BusState *qbus = BUS(qdev_get_parent_bus(DEVICE(s->vdev)));
     VirtioBusClass *k = VIRTIO_BUS_GET_CLASS(qbus);
+    VirtIOBlock *vblk = VIRTIO_BLK(s->vdev);
     if (!s->started || s->stopping) {
         return;
     }
     s->stopping = true;
+    vblk->complete_request = s->saved_complete_request;
     trace_virtio_blk_data_plane_stop(s);
 
     aio_context_acquire(s->ctx);