qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Michael S. Tsirkin" <mst@redhat.com>
To: qemu-devel@nongnu.org
Cc: Cornelia Huck <cornelia.huck@de.ibm.com>,
	Peter Maydell <peter.maydell@linaro.org>,
	Eduardo Habkost <ehabkost@redhat.com>,
	Paolo Bonzini <pbonzini@redhat.com>
Subject: [Qemu-devel] [PULL 09/49] virtio: slim down allocation of VirtQueueElements
Date: Thu, 4 Feb 2016 23:51:32 +0200	[thread overview]
Message-ID: <1454612376-7072-10-git-send-email-mst@redhat.com> (raw)
In-Reply-To: <1454612376-7072-1-git-send-email-mst@redhat.com>

From: Paolo Bonzini <pbonzini@redhat.com>

Build the addresses and s/g lists on the stack, and then copy them
to a VirtQueueElement that is just as big as required to contain this
particular s/g list.  The cost of the copy is minimal compared to that
of a large malloc.

When virtqueue_map is used on the destination side of migration or on
loadvm, the iovecs have already been split at memory region boundary,
so we can just reuse the out_num/in_num we find in the file.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
 hw/virtio/virtio.c | 82 +++++++++++++++++++++++++++++++++---------------------
 1 file changed, 51 insertions(+), 31 deletions(-)

diff --git a/hw/virtio/virtio.c b/hw/virtio/virtio.c
index 661a1e1..c3e00a8 100644
--- a/hw/virtio/virtio.c
+++ b/hw/virtio/virtio.c
@@ -448,6 +448,32 @@ int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes,
     return in_bytes <= in_total && out_bytes <= out_total;
 }
 
+static void virtqueue_map_desc(unsigned int *p_num_sg, hwaddr *addr, struct iovec *iov,
+                               unsigned int max_num_sg, bool is_write,
+                               hwaddr pa, size_t sz)
+{
+    unsigned num_sg = *p_num_sg;
+    assert(num_sg <= max_num_sg);
+
+    while (sz) {
+        hwaddr len = sz;
+
+        if (num_sg == max_num_sg) {
+            error_report("virtio: too many write descriptors in indirect table");
+            exit(1);
+        }
+
+        iov[num_sg].iov_base = cpu_physical_memory_map(pa, &len, is_write);
+        iov[num_sg].iov_len = len;
+        addr[num_sg] = pa;
+
+        sz -= len;
+        pa += len;
+        num_sg++;
+    }
+    *p_num_sg = num_sg;
+}
+
 static void virtqueue_map_iovec(struct iovec *sg, hwaddr *addr,
                                 unsigned int *num_sg, unsigned int max_size,
                                 int is_write)
@@ -474,20 +500,10 @@ static void virtqueue_map_iovec(struct iovec *sg, hwaddr *addr,
             error_report("virtio: error trying to map MMIO memory");
             exit(1);
         }
-        if (len == sg[i].iov_len) {
-            continue;
-        }
-        if (*num_sg >= max_size) {
-            error_report("virtio: memory split makes iovec too large");
+        if (len != sg[i].iov_len) {
+            error_report("virtio: unexpected memory split");
             exit(1);
         }
-        memmove(sg + i + 1, sg + i, sizeof(*sg) * (*num_sg - i));
-        memmove(addr + i + 1, addr + i, sizeof(*addr) * (*num_sg - i));
-        assert(len < sg[i + 1].iov_len);
-        sg[i].iov_len = len;
-        addr[i + 1] += len;
-        sg[i + 1].iov_len -= len;
-        ++*num_sg;
     }
 }
 
@@ -526,14 +542,16 @@ void *virtqueue_pop(VirtQueue *vq, size_t sz)
     hwaddr desc_pa = vq->vring.desc;
     VirtIODevice *vdev = vq->vdev;
     VirtQueueElement *elem;
+    unsigned out_num, in_num;
+    hwaddr addr[VIRTQUEUE_MAX_SIZE];
+    struct iovec iov[VIRTQUEUE_MAX_SIZE];
 
     if (!virtqueue_num_heads(vq, vq->last_avail_idx)) {
         return NULL;
     }
 
     /* When we start there are none of either input nor output. */
-    elem = virtqueue_alloc_element(sz, VIRTQUEUE_MAX_SIZE, VIRTQUEUE_MAX_SIZE);
-    elem->out_num = elem->in_num = 0;
+    out_num = in_num = 0;
 
     max = vq->vring.num;
 
@@ -556,37 +574,39 @@ void *virtqueue_pop(VirtQueue *vq, size_t sz)
 
     /* Collect all the descriptors */
     do {
-        struct iovec *sg;
+        hwaddr pa = vring_desc_addr(vdev, desc_pa, i);
+        size_t len = vring_desc_len(vdev, desc_pa, i);
 
         if (vring_desc_flags(vdev, desc_pa, i) & VRING_DESC_F_WRITE) {
-            if (elem->in_num >= VIRTQUEUE_MAX_SIZE) {
-                error_report("Too many write descriptors in indirect table");
-                exit(1);
-            }
-            elem->in_addr[elem->in_num] = vring_desc_addr(vdev, desc_pa, i);
-            sg = &elem->in_sg[elem->in_num++];
+            virtqueue_map_desc(&in_num, addr + out_num, iov + out_num,
+                               VIRTQUEUE_MAX_SIZE - out_num, true, pa, len);
         } else {
-            if (elem->out_num >= VIRTQUEUE_MAX_SIZE) {
-                error_report("Too many read descriptors in indirect table");
+            if (in_num) {
+                error_report("Incorrect order for descriptors");
                 exit(1);
             }
-            elem->out_addr[elem->out_num] = vring_desc_addr(vdev, desc_pa, i);
-            sg = &elem->out_sg[elem->out_num++];
+            virtqueue_map_desc(&out_num, addr, iov,
+                               VIRTQUEUE_MAX_SIZE, false, pa, len);
         }
 
-        sg->iov_len = vring_desc_len(vdev, desc_pa, i);
-
         /* If we've got too many, that implies a descriptor loop. */
-        if ((elem->in_num + elem->out_num) > max) {
+        if ((in_num + out_num) > max) {
             error_report("Looped descriptor");
             exit(1);
         }
     } while ((i = virtqueue_next_desc(vdev, desc_pa, i, max)) != max);
 
-    /* Now map what we have collected */
-    virtqueue_map(elem);
-
+    /* Now copy what we have collected and mapped */
+    elem = virtqueue_alloc_element(sz, out_num, in_num);
     elem->index = head;
+    for (i = 0; i < out_num; i++) {
+        elem->out_addr[i] = addr[i];
+        elem->out_sg[i] = iov[i];
+    }
+    for (i = 0; i < in_num; i++) {
+        elem->in_addr[i] = addr[out_num + i];
+        elem->in_sg[i] = iov[out_num + i];
+    }
 
     vq->inuse++;
 
-- 
MST

  parent reply	other threads:[~2016-02-04 21:51 UTC|newest]

Thread overview: 75+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-04 21:50 [Qemu-devel] [PULL 00/49] pc and misc cleanups and fixes, virtio optimizations Michael S. Tsirkin
2016-02-04 21:50 ` [Qemu-devel] [PULL 14/49] virtio: combine write of an entry into used ring Michael S. Tsirkin
2016-02-04 21:50 ` [Qemu-devel] [PULL 01/49] Fix virtio migration Michael S. Tsirkin
2016-02-04 21:50 ` [Qemu-devel] [PULL 02/49] pc: acpi: merge SSDT into DSDT Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 03/49] tests: pc: acpi: drop not needed 'expected SSDT' blobs Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 04/49] tests: pc: acpi: add expected DSDT.bridge blobs and update DSDT blobs Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 05/49] virtio: move VirtQueueElement at the beginning of the structs Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 06/49] virtio: move allocation to virtqueue_pop/vring_pop Michael S. Tsirkin
2016-02-05 12:52   ` Peter Maydell
2016-02-06 18:10     ` Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 07/49] virtio: introduce qemu_get/put_virtqueue_element Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 08/49] virtio: introduce virtqueue_alloc_element Michael S. Tsirkin
2016-02-04 21:51 ` Michael S. Tsirkin [this message]
2016-02-04 21:51 ` [Qemu-devel] [PULL 10/49] vring: slim down allocation of VirtQueueElements Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 11/49] virtio: combine the read of a descriptor Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 12/49] virtio: cache used_idx in a VirtQueue field Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 13/49] virtio: read avail_idx from VQ only when necessary Michael S. Tsirkin
2016-02-04 21:51 ` [Qemu-devel] [PULL 15/49] hw/pxb: add pxb devices to the bridge category Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 16/49] vhost-user-test: use correct ROM to speed up and avoid spurious failures Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 17/49] hw/pci: ensure that only PCI/PCIe bridges can be attached to pxb/pxb-pcie devices Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 18/49] ipmi: replace goto by a return statement Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 19/49] ipmi: replace *_MAXCMD defines Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 20/49] ipmi: cleanup error_report messages Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 21/49] ipmi: fix SDR length value Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 22/49] ipmi: introduce a struct ipmi_sdr_compact Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 23/49] ipmi: add get and set SENSOR_TYPE commands Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 24/49] ipmi: add GET_SYS_RESTART_CAUSE chassis command Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 25/49] ipmi: add ACPI power and GUID commands Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 26/49] pc: Move PcGuestInfo declaration to top of file Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 27/49] pc: Eliminate struct PcGuestInfoState Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 28/49] pc: Simplify pc_memory_init() signature Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 29/49] pc: Simplify xen_load_linux() signature Michael S. Tsirkin
2016-02-04 21:52 ` [Qemu-devel] [PULL 30/49] acpi: Remove guest_info parameters from functions Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 31/49] acpi: Don't save PcGuestInfo on AcpiBuildState Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 32/49] pc: Remove compat fields from PcGuestInfo Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 33/49] pc: Remove RAM size " Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 34/49] pc: Remove PcGuestInfo.isapc_ram_fw field Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 35/49] pc: Move PcGuestInfo.fw_cfg to PCMachineState Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 36/49] pc: Move APIC and NUMA data from PcGuestInfo " Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 37/49] pc: Eliminate PcGuestInfo struct Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 38/49] acpi: take oem_id in build_header(), optionally Michael S. Tsirkin
2016-02-04 22:25   ` Laszlo Ersek
2016-02-04 21:53 ` [Qemu-devel] [PULL 39/49] acpi: expose oem_id and oem_table_id in build_rsdt() Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 40/49] acpi: add function to extract oem_id and oem_table_id from the user's SLIC Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 41/49] pc: set the OEM fields in the RSDT and the FADT from the SLIC Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 42/49] dimm: Correct type of MemoryHotplugState->base Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 43/49] intel_iommu: large page support Michael S. Tsirkin
2016-02-04 21:53 ` [Qemu-devel] [PULL 44/49] fix MSI injection on Xen Michael S. Tsirkin
2016-02-04 21:54 ` [Qemu-devel] [PULL 45/49] net: set endianness on all backend devices Michael S. Tsirkin
2016-02-05  8:54   ` Greg Kurz
2016-02-04 21:54 ` [Qemu-devel] [PULL 46/49] i386/acpi: make floppy controller object dynamic Michael S. Tsirkin
2016-02-04 21:54 ` [Qemu-devel] [PULL 47/49] expose floppy drive geometry and CMOS type Michael S. Tsirkin
2016-02-04 21:54 ` [Qemu-devel] [PULL 48/49] i386: populate floppy drive information in DSDT Michael S. Tsirkin
2016-02-05 18:25   ` Igor Mammedov
2016-02-08 13:14     ` Roman Kagan
2016-02-08 20:20       ` John Snow
2016-02-09 15:52         ` Roman Kagan
2016-02-09 16:22           ` John Snow
2016-02-09 18:36             ` Laszlo Ersek
2016-02-09 18:48               ` Michael S. Tsirkin
2016-02-10 16:14                 ` John Snow
2016-02-10 16:48                   ` Michael S. Tsirkin
2016-02-10 17:24                   ` Roman Kagan
2016-02-10 17:10               ` Roman Kagan
2016-02-10 17:16                 ` John Snow
2016-02-10 17:33                   ` Roman Kagan
2016-02-10 21:54                     ` John Snow
2016-02-13 17:26               ` Kevin O'Connor
2016-02-14  6:45                 ` Laszlo Ersek
2016-02-14 15:02                 ` Michael S. Tsirkin
2016-02-17 14:31                   ` Roman Kagan
2016-02-10 16:57             ` Roman Kagan
2016-02-04 21:54 ` [Qemu-devel] [PULL 49/49] acpi: update expected DSDT Michael S. Tsirkin
2016-02-05 15:03 ` [Qemu-devel] [PULL 00/49] pc and misc cleanups and fixes, virtio optimizations Peter Maydell
2016-02-05 18:19   ` Igor Mammedov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1454612376-7072-10-git-send-email-mst@redhat.com \
    --to=mst@redhat.com \
    --cc=cornelia.huck@de.ibm.com \
    --cc=ehabkost@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).