virtio: handle non contigious s/g entries

hw/virtio/virtio.c +2 -1

+    for (i = 0; i < *num_sg; i++) {
         len = sg[i].iov_len;
         sg[i].iov_base = cpu_physical_memory_map(addr[i], &len, is_write);
-        if (sg[i].iov_base == NULL || len != sg[i].iov_len) {
+        if (!sg[i].iov_base) {
             error_report("virtio: error trying to map MMIO memory");
             exit(1);

Quoted 7 lines · Michael S. Tsirkin

hw/virtio/virtio.c +2 -1

+    for (i = 0; i < *num_sg; i++) {
         len = sg[i].iov_len;
         sg[i].iov_base = cpu_physical_memory_map(addr[i], &len, is_write);
-        if (sg[i].iov_base == NULL || len != sg[i].iov_len) {
+        if (!sg[i].iov_base) {
             error_report("virtio: error trying to map MMIO memory");
             exit(1);

hw/virtio/virtio.c int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes, · line 485

Quoted 10 lines

hw/virtio/virtio.c +9 -0

         }
+        if (len == sg[i].iov_len) {
+            continue;
+        }
+        if (*num_sg >= max_size) {
+            error_report("virtio: memory split makes iovec too large");
+            exit(1);
+        }
+        memcpy(sg + i + 1, sg + i, sizeof(*sg) * (*num_sg - i));
+        memcpy(addr + i + 1, addr + i, sizeof(*addr) * (*num_sg - i));

These should be memmove() since memcpy() arguments are not allowed to overlap.

This will still fail if there's no space left in the sg, but luckily max queue size in use is currently 256, while max sg size is 1024, so we should be OK even is all entries happen to cross a single DIMM boundary.

Quoted 3 lines · Michael S. Tsirkin

This will still fail if there's no space left in the sg, but luckily max queue size in use is currently 256, while max sg size is 1024, so we should be OK even is all entries happen to cross a single DIMM boundary.

Don't forget about indirect descriptors. They can use all 1024 iovecs, regardless of virtqueue size, so virtqueue size of 256 isn't the true maximum.

I'm worried that we could now see failures due to non-contiguous HVAs.

Quoted 8 lines · Stefan Hajnoczi

Quoted 3 lines · Michael S. Tsirkin

This will still fail if there's no space left in the sg, but luckily max queue size in use is currently 256, while max sg size is 1024, so we should be OK even is all entries happen to cross a single DIMM boundary.

Don't forget about indirect descriptors. They can use all 1024 iovecs, regardless of virtqueue size, so virtqueue size of 256 isn't the true maximum.

Not according to the spec - virtio spec says vq size is the maximum size of a chain.

Quoted 1 line

I'm worried that we could now see failures due to non-contiguous HVAs.

Does linux guest create chains > vq size then? Does it actually have 1024 hardcoded somewhere?

virtio_map_sg currently fails if one of the entries it's mapping is contigious in GPA but not HVA address space. Introduce virtio_map which handles this by splitting sg entries.

This new API generally turns out to be a good idea since it's harder to
misuse: at least in one case the existing one was used incorrectly.

This will still fail if there's no space left in the sg, but luckily max queue size in use is currently 256, while max sg size is 1024, so we should be OK even is all entries happen to cross a single DIMM boundary.

Won't work well with very small DIMM sizes, unfortunately:
e.g. this will fail with 4K DIMMs where a single request might span a large number of DIMMs.

Let's hope these are uncommon - at least we are not breaking things.

Note: virtio-scsi calls virtio_map_sg on data loaded from network, and validates input, asserting on failure. Copy the validating code here - it will be dropped from virtio-scsi in a follow-up patch.

Reported-by Igor Mammedov <[email protected]>

Signed-off-by Michael S. Tsirkin <[email protected]>

---
 include/hw/virtio/virtio.h |  1 +
 hw/virtio/virtio.c         | 56 ++++++++++++++++++++++++++++++++++++++--------
 2 files changed, 48 insertions(+), 9 deletions(-)

include/hw/virtio/virtio.h +1 -0

index 9d09115..9d9abb4 100644
@@ -153,6 +153,7 @@ void virtqueue_fill(VirtQueue *vq, const VirtQueueElement *elem,
 
 void virtqueue_map_sg(struct iovec *sg, hwaddr *addr,
     size_t num_sg, int is_write);
+void virtqueue_map(VirtQueueElement *elem);
 int virtqueue_pop(VirtQueue *vq, VirtQueueElement *elem);
 int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes,
                           unsigned int out_bytes);

hw/virtio/virtio.c +0 -0

index d0bc72e..a6878c0 100644

hw/virtio/virtio.c +44 -7

 {
     unsigned int i;
     hwaddr len;
 
-    if (num_sg > VIRTQUEUE_MAX_SIZE) {
-        error_report("virtio: map attempt out of bounds: %zd > %d",
-                     num_sg, VIRTQUEUE_MAX_SIZE);
-        exit(1);
-    }
+    /* Note: this function MUST validate input, some callers
+     * are passing in num_sg values received over the network.
+     */
+    /* TODO: teach all callers that this can fail, and return failure instead
+     * of asserting here.
+     * When we do, we might be able to re-enable NDEBUG below.
+     */
+#ifdef NDEBUG
+#error building with NDEBUG is not supported
+#endif
+    assert(*num_sg <= max_size);
 
-    for (i = 0; i < num_sg; i++) {
+    for (i = 0; i < *num_sg; i++) {
         len = sg[i].iov_len;
         sg[i].iov_base = cpu_physical_memory_map(addr[i], &len, is_write);
-        if (sg[i].iov_base == NULL || len != sg[i].iov_len) {
+        if (!sg[i].iov_base) {
             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");
+            exit(1);
+        }
+        memcpy(sg + i + 1, sg + i, sizeof(*sg) * (*num_sg - i));
+        memcpy(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;
     }
 }
 
+/* Deprecated: don't use in new code */
+void virtqueue_map_sg(struct iovec *sg, hwaddr *addr,
+                      size_t num_sg, int is_write)
+{
+    virtqueue_map_iovec(sg, addr, &num_sg, num_sg, is_write);
+}
+
+void virtqueue_map(VirtQueueElement *elem)
+{
+    virtqueue_map_iovec(elem->in_sg, elem->in_addr, &elem->in_num,
+                        MIN(ARRAY_SIZE(elem->in_sg), ARRAY_SIZE(elem->in_addr)),
+                        1);
+    virtqueue_map_iovec(elem->out_sg, elem->out_addr, &elem->out_num,
+                        MIN(ARRAY_SIZE(elem->out_sg), ARRAY_SIZE(elem->out_addr)),
+                        0);
+}
+
 int virtqueue_pop(VirtQueue *vq, VirtQueueElement *elem)
 {
     unsigned int i, head, max;

virtio_map_sg currently fails if one of the entries it's mapping is contigious in GPA but not HVA address space. Introduce virtio_map which handles this by splitting sg entries.

This new API generally turns out to be a good idea since it's harder to
misuse: at least in one case the existing one was used incorrectly.

This will still fail if there's no space left in the sg, but luckily max queue size in use is currently 256, while max sg size is 1024, so we should be OK even is all entries happen to cross a single DIMM boundary.

Won't work well with very small DIMM sizes, unfortunately:
e.g. this will fail with 4K DIMMs where a single request might span a large number of DIMMs.

Let's hope these are uncommon - at least we are not breaking things.

Note: virtio-scsi calls virtio_map_sg on data loaded from network, and validates input, asserting on failure. Copy the validating code here - it will be dropped from virtio-scsi in a follow-up patch.

Reported-by Igor Mammedov <[email protected]>

Signed-off-by Michael S. Tsirkin <[email protected]>

---
 include/hw/virtio/virtio.h |  1 +
 hw/virtio/virtio.c         | 56 ++++++++++++++++++++++++++++++++++++++--------
 2 files changed, 48 insertions(+), 9 deletions(-)

include/hw/virtio/virtio.h +1 -0

index 9d09115..9d9abb4 100644
@@ -153,6 +153,7 @@ void virtqueue_fill(VirtQueue *vq, const VirtQueueElement *elem,
 
 void virtqueue_map_sg(struct iovec *sg, hwaddr *addr,
     size_t num_sg, int is_write);
+void virtqueue_map(VirtQueueElement *elem);
 int virtqueue_pop(VirtQueue *vq, VirtQueueElement *elem);
 int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes,
                           unsigned int out_bytes);

hw/virtio/virtio.c +47 -9

index d0bc72e..a6878c0 100644
@@ -448,28 +448,66 @@ int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes,
     return in_bytes <= in_total && out_bytes <= out_total;
 }
 
-void virtqueue_map_sg(struct iovec *sg, hwaddr *addr,
-    size_t num_sg, int is_write)
+static void virtqueue_map_iovec(struct iovec *sg, hwaddr *addr,
+                                size_t *num_sg, size_t max_size,
+                                int is_write)
 {
     unsigned int i;
     hwaddr len;
 
-    if (num_sg > VIRTQUEUE_MAX_SIZE) {
-        error_report("virtio: map attempt out of bounds: %zd > %d",
-                     num_sg, VIRTQUEUE_MAX_SIZE);
-        exit(1);
-    }
+    /* Note: this function MUST validate input, some callers
+     * are passing in num_sg values received over the network.
+     */
+    /* TODO: teach all callers that this can fail, and return failure instead
+     * of asserting here.
+     * When we do, we might be able to re-enable NDEBUG below.
+     */
+#ifdef NDEBUG
+#error building with NDEBUG is not supported
+#endif
+    assert(*num_sg <= max_size);
 
-    for (i = 0; i < num_sg; i++) {
+    for (i = 0; i < *num_sg; i++) {
         len = sg[i].iov_len;
         sg[i].iov_base = cpu_physical_memory_map(addr[i], &len, is_write);
-        if (sg[i].iov_base == NULL || len != sg[i].iov_len) {
+        if (!sg[i].iov_base) {
             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");
+            exit(1);
+        }
+        memcpy(sg + i + 1, sg + i, sizeof(*sg) * (*num_sg - i));
+        memcpy(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;
     }
 }
 
+/* Deprecated: don't use in new code */
+void virtqueue_map_sg(struct iovec *sg, hwaddr *addr,
+                      size_t num_sg, int is_write)
+{
+    virtqueue_map_iovec(sg, addr, &num_sg, num_sg, is_write);
+}
+
+void virtqueue_map(VirtQueueElement *elem)
+{
+    virtqueue_map_iovec(elem->in_sg, elem->in_addr, &elem->in_num,
+                        MIN(ARRAY_SIZE(elem->in_sg), ARRAY_SIZE(elem->in_addr)),
+                        1);
+    virtqueue_map_iovec(elem->out_sg, elem->out_addr, &elem->out_num,
+                        MIN(ARRAY_SIZE(elem->out_sg), ARRAY_SIZE(elem->out_addr)),
+                        0);
+}
+
 int virtqueue_pop(VirtQueue *vq, VirtQueueElement *elem)
 {
     unsigned int i, head, max;

Drop use of the deprecated virtio_map_sg in virtio core.

Signed-off-by Michael S. Tsirkin <[email protected]>

---
 hw/virtio/virtio.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

hw/virtio/virtio.c +1 -2

index a6878c0..84e2320 100644
@@ -569,8 +569,7 @@ int virtqueue_pop(VirtQueue *vq, VirtQueueElement *elem)
     } while ((i = virtqueue_next_desc(vdev, desc_pa, i, max)) != max);
 
     /* Now map what we have collected */
-    virtqueue_map_sg(elem->in_sg, elem->in_addr, elem->in_num, 1);
-    virtqueue_map_sg(elem->out_sg, elem->out_addr, elem->out_num, 0);
+    virtqueue_map(elem);
 
     elem->index = head;
 

Drop deprecated use of virtqueue_map_sg.

Signed-off-by Michael S. Tsirkin <[email protected]>

---
 hw/block/virtio-blk.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

hw/block/virtio-blk.c +1 -4

index 8beb26b..3e230de 100644
@@ -839,10 +839,7 @@ static int virtio_blk_load_device(VirtIODevice *vdev, QEMUFile *f,
         req->next = s->rq;
         s->rq = req;
 
-        virtqueue_map_sg(req->elem.in_sg, req->elem.in_addr,
-            req->elem.in_num, 1);
-        virtqueue_map_sg(req->elem.out_sg, req->elem.out_addr,
-            req->elem.out_num, 0);
+        virtqueue_map(&req->elem);
     }
 
     return 0;

This also fixes a minor bug:

hw/char/virtio-serial-bus.c +0 -2

-                virtqueue_map_sg(port->elem.out_sg, port->elem.out_addr,
-                                 port->elem.out_num, 1);

is wrong: out_sg is not written so should not be marked dirty.

Signed-off-by Michael S. Tsirkin <[email protected]>

---
 hw/char/virtio-serial-bus.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

hw/char/virtio-serial-bus.c +1 -4

index be97058..497b0af 100644
@@ -705,10 +705,7 @@ static int fetch_active_ports_list(QEMUFile *f, int version_id,
 
                 qemu_get_buffer(f, (unsigned char *)&port->elem,
                                 sizeof(port->elem));
-                virtqueue_map_sg(port->elem.in_sg, port->elem.in_addr,
-                                 port->elem.in_num, 1);
-                virtqueue_map_sg(port->elem.out_sg, port->elem.out_addr,
-                                 port->elem.out_num, 1);
+                virtqueue_map(&port->elem);
 
                 /*
                  *  Port was throttled on source machine.  Let's

Note: virtqueue_map already validates input
so virtio-scsi does not have to.

Signed-off-by Michael S. Tsirkin <[email protected]>

---
 hw/scsi/virtio-scsi.c | 16 ++--------------
 1 file changed, 2 insertions(+), 14 deletions(-)

hw/scsi/virtio-scsi.c +2 -14

index 1c33f14..33bd25a 100644
@@ -207,20 +207,8 @@ static void *virtio_scsi_load_request(QEMUFile *f, SCSIRequest *sreq)
     assert(n < vs->conf.num_queues);
     req = virtio_scsi_init_req(s, vs->cmd_vqs[n]);
     qemu_get_buffer(f, (unsigned char *)&req->elem, sizeof(req->elem));
-    /* TODO: add a way for SCSIBusInfo's load_request to fail,
-     * and fail migration instead of asserting here.
-     * When we do, we might be able to re-enable NDEBUG below.
-     */
-#ifdef NDEBUG
-#error building with NDEBUG is not supported
-#endif
-    assert(req->elem.in_num <= ARRAY_SIZE(req->elem.in_sg));
-    assert(req->elem.out_num <= ARRAY_SIZE(req->elem.out_sg));
-
-    virtqueue_map_sg(req->elem.in_sg, req->elem.in_addr,
-                     req->elem.in_num, 1);
-    virtqueue_map_sg(req->elem.out_sg, req->elem.out_addr,
-                     req->elem.out_num, 0);
+
+    virtqueue_map(&req->elem);
 
     if (virtio_scsi_parse_req(req, sizeof(VirtIOSCSICmdReq) + vs->cdb_size,
                               sizeof(VirtIOSCSICmdResp) + vs->sense_size) < 0) {

Deprecated in favor of virtqueue_map.

Signed-off-by Michael S. Tsirkin <[email protected]>

---
 include/hw/virtio/virtio.h | 2 --
 hw/virtio/virtio.c         | 7 -------
 2 files changed, 9 deletions(-)

include/hw/virtio/virtio.h +0 -2

index 9d9abb4..205fadf 100644
@@ -151,8 +151,6 @@ void virtqueue_discard(VirtQueue *vq, const VirtQueueElement *elem,
 void virtqueue_fill(VirtQueue *vq, const VirtQueueElement *elem,
                     unsigned int len, unsigned int idx);
 
-void virtqueue_map_sg(struct iovec *sg, hwaddr *addr,
-    size_t num_sg, int is_write);
 void virtqueue_map(VirtQueueElement *elem);
 int virtqueue_pop(VirtQueue *vq, VirtQueueElement *elem);
 int virtqueue_avail_bytes(VirtQueue *vq, unsigned int in_bytes,

hw/virtio/virtio.c +0 -7

index 84e2320..be32145 100644
@@ -491,13 +491,6 @@ static void virtqueue_map_iovec(struct iovec *sg, hwaddr *addr,
     }
 }
 
-/* Deprecated: don't use in new code */
-void virtqueue_map_sg(struct iovec *sg, hwaddr *addr,
-                      size_t num_sg, int is_write)
-{
-    virtqueue_map_iovec(sg, addr, &num_sg, num_sg, is_write);
-}
-
 void virtqueue_map(VirtQueueElement *elem)
 {
     virtqueue_map_iovec(elem->in_sg, elem->in_addr, &elem->in_num,

TL;DR:
This fixes virtio in a way transparent to guest.
We should now be able to revert commits aa8580cd and df0acded19ec which worked around it in a way that's not transparent.

-----

commit aa8580cd "pc: memhp: force gaps between DIMM's GPA" introduced gaps in GPA space to work around a bug in virtio, originally reported by Igor, and I quote:

------

    QEMU aborts during guest reboot with following backtrace:

    Breakpoint 1, virtqueue_map_sg (sg=0x555557cbc6c0, addr=0x555557cb86c0, 
    num_sg=0x12, is_write=0x1) at hw/virtio/virtio.c:453
    453                 error_report("virtio: error trying to map MMIO memory");
    (gdb) bt
    #0  virtqueue_map_sg (sg=0x555557cbc6c0, addr=0x555557cb86c0, num_sg=0x12, 
    is_write=0x1) at hw/virtio/virtio.c:453
    #1  0x000055555569b3ef in virtqueue_pop (vq=0x555558a3fab0, 
    elem=0x555557cb86b0) at hw/virtio/virtio.c:520
    #2  0x0000555555666611 in virtio_blk_get_request (s=0x5555588c7a00) at 
    hw/block/virtio-blk.c:194
    #3  0x00005555556676ec in virtio_blk_handle_output (vdev=0x5555588c7a00, 
    vq=0x555558a3fab0) at hw/block/virtio-blk.c:603
    #4  0x000055555569c5c8 in virtio_queue_notify_vq (vq=0x555558a3fab0) at 
    hw/virtio/virtio.c:921
    #5  0x000055555569e009 in virtio_queue_host_notifier_read (n=0x555558a3faf8) at 
    hw/virtio/virtio.c:1480
    #6  0x000055555591b062 in qemu_iohandler_poll (pollfds=0x555556363800, ret=0x1) 
    at iohandler.c:126
    #7  0x000055555591ad33 in main_loop_wait (nonblocking=0x0) at main-loop.c:503
    #8  0x00005555557466b5 in main_loop () at vl.c:1902
    #9  0x000055555574e69b in main (argc=0x4d, argv=0x7fffffffda18, 
    envp=0x7fffffffdc88) at vl.c:4653

    could be reproduced with following options:

    -enable-kvm  -m 1G,slots=250,maxmem=32G  -drive if=virtio,file=rhel72 -netdev 
    tap,id=foo,ifname=tap0,script=./qemu-ifup -device 
    virtio-net-pci,id=n1,netdev=foo `for i in $(seq 0 15); do echo -n "-object 
    memory-backend-ram,id=m$i,size=10M -device pc-dimm,id=dimm$i,memdev=m$i "; 
    done` -snapshot -monitor unix:/tmp/m,server,nowait

    boot and login to guest shell and execute 'reboot' command
    on the reboot when guest kernel boots, QEMU will abort in virtio.
    Reproducible in about 80% cases. If QEMU doesn't crash on reboot
    then try again whit freshly started QEMU.


    Reason for crashing is that guest allocates buffer that crosses
    boundary between 2 different memory regions and as result
    cpu_physical_memory_map() maps GPA to HVA for only head of buffer
    that belongs to the first region which makes conditon
     len != sg[i].iov_len
    true since declared buffer size (sg[i].iov_len) isn't what
    cpu_physical_memory_map() has been able to map (len),
    which leads to abort:

    virtqueue_map_sg() {
      ...
      sg[i].iov_base = cpu_physical_memory_map(addr[i], &len, is_write);
      if (sg[i].iov_base == NULL || len != sg[i].iov_len) {
           abort()

------

However, the work-around regressed memory hot-unplug for linux guests triggering the following within the guest:

------
=====
kernel BUG at mm/memory_hotplug.c:703!
...

 [<ffffffff81385fa7>] acpi_memory_device_remove+0x79/0xa5
 [<ffffffff81357818>] acpi_bus_trim+0x5a/0x8d
 [<ffffffff81359026>] acpi_device_hotplug+0x1b7/0x418
 ===
    BUG_ON(phys_start_pfn & ~PAGE_SECTION_MASK);
 ===

------

The reason for the crash is that x86-64 linux guest supports memory hotplug in chunks of 128Mb and assumes that memory sections are also 128Mb aligned. However gaps forced between 128Mb DIMMs with backend's natural alignment of 2Mb make the 2nd and following DIMMs not being aligned on 128Mb boundary.

------

Michael S. Tsirkin (6):
virtio: introduce virtio_map
virtio: switch to virtio_map
virtio-blk: convert to virtqueue_map
virtio-serial: convert to virtio_map
virtio-scsi: convert to virtqueue_map
virtio: drop virtqueue_map_sg

 include/hw/virtio/virtio.h  |  3 +--
 hw/block/virtio-blk.c       |  5 +----
 hw/char/virtio-serial-bus.c |  5 +----
 hw/scsi/virtio-scsi.c       | 16 ++------------
 hw/virtio/virtio.c          | 52 +++++++++++++++++++++++++++++++++++----------
 5 files changed, 46 insertions(+), 35 deletions(-)

Signature · 1 line

MST

TL;DR:
This fixes virtio in a way transparent to guest.
We should now be able to revert commits aa8580cd and df0acded19ec which worked around it in a way that's not transparent.

-----

commit aa8580cd "pc: memhp: force gaps between DIMM's GPA" introduced gaps in GPA space to work around a bug in virtio, originally reported by Igor, and I quote:

------

    QEMU aborts during guest reboot with following backtrace:

    Breakpoint 1, virtqueue_map_sg (sg=0x555557cbc6c0, addr=0x555557cb86c0, 
    num_sg=0x12, is_write=0x1) at hw/virtio/virtio.c:453
    453                 error_report("virtio: error trying to map MMIO memory");
    (gdb) bt
    #0  virtqueue_map_sg (sg=0x555557cbc6c0, addr=0x555557cb86c0, num_sg=0x12, 
    is_write=0x1) at hw/virtio/virtio.c:453
    #1  0x000055555569b3ef in virtqueue_pop (vq=0x555558a3fab0, 
    elem=0x555557cb86b0) at hw/virtio/virtio.c:520
    #2  0x0000555555666611 in virtio_blk_get_request (s=0x5555588c7a00) at 
    hw/block/virtio-blk.c:194
    #3  0x00005555556676ec in virtio_blk_handle_output (vdev=0x5555588c7a00, 
    vq=0x555558a3fab0) at hw/block/virtio-blk.c:603
    #4  0x000055555569c5c8 in virtio_queue_notify_vq (vq=0x555558a3fab0) at 
    hw/virtio/virtio.c:921
    #5  0x000055555569e009 in virtio_queue_host_notifier_read (n=0x555558a3faf8) at 
    hw/virtio/virtio.c:1480
    #6  0x000055555591b062 in qemu_iohandler_poll (pollfds=0x555556363800, ret=0x1) 
    at iohandler.c:126
    #7  0x000055555591ad33 in main_loop_wait (nonblocking=0x0) at main-loop.c:503
    #8  0x00005555557466b5 in main_loop () at vl.c:1902
    #9  0x000055555574e69b in main (argc=0x4d, argv=0x7fffffffda18, 
    envp=0x7fffffffdc88) at vl.c:4653

    could be reproduced with following options:

    -enable-kvm  -m 1G,slots=250,maxmem=32G  -drive if=virtio,file=rhel72 -netdev 
    tap,id=foo,ifname=tap0,script=./qemu-ifup -device 
    virtio-net-pci,id=n1,netdev=foo `for i in $(seq 0 15); do echo -n "-object 
    memory-backend-ram,id=m$i,size=10M -device pc-dimm,id=dimm$i,memdev=m$i "; 
    done` -snapshot -monitor unix:/tmp/m,server,nowait

    boot and login to guest shell and execute 'reboot' command
    on the reboot when guest kernel boots, QEMU will abort in virtio.
    Reproducible in about 80% cases. If QEMU doesn't crash on reboot
    then try again whit freshly started QEMU.


    Reason for crashing is that guest allocates buffer that crosses
    boundary between 2 different memory regions and as result
    cpu_physical_memory_map() maps GPA to HVA for only head of buffer
    that belongs to the first region which makes conditon
     len != sg[i].iov_len
    true since declared buffer size (sg[i].iov_len) isn't what
    cpu_physical_memory_map() has been able to map (len),
    which leads to abort:

    virtqueue_map_sg() {
      ...
      sg[i].iov_base = cpu_physical_memory_map(addr[i], &len, is_write);
      if (sg[i].iov_base == NULL || len != sg[i].iov_len) {
           abort()

------

However, the work-around regressed memory hot-unplug for linux guests triggering the following within the guest:

------
=====
kernel BUG at mm/memory_hotplug.c:703!
...

 [<ffffffff81385fa7>] acpi_memory_device_remove+0x79/0xa5
 [<ffffffff81357818>] acpi_bus_trim+0x5a/0x8d
 [<ffffffff81359026>] acpi_device_hotplug+0x1b7/0x418
 ===
    BUG_ON(phys_start_pfn & ~PAGE_SECTION_MASK);
 ===

------

The reason for the crash is that x86-64 linux guest supports memory hotplug in chunks of 128Mb and assumes that memory sections are also 128Mb aligned. However gaps forced between 128Mb DIMMs with backend's natural alignment of 2Mb make the 2nd and following DIMMs not being aligned on 128Mb boundary.

------

Michael S. Tsirkin (6):
virtio: introduce virtio_map
virtio: switch to virtio_map
virtio-blk: convert to virtqueue_map
virtio-serial: convert to virtio_map
virtio-scsi: convert to virtqueue_map
virtio: drop virtqueue_map_sg

 include/hw/virtio/virtio.h  |  3 +--
 hw/block/virtio-blk.c       |  5 +----
 hw/char/virtio-serial-bus.c |  5 +----
 hw/scsi/virtio-scsi.c       | 16 ++------------
 hw/virtio/virtio.c          | 52 +++++++++++++++++++++++++++++++++++----------
 5 files changed, 46 insertions(+), 35 deletions(-)

Signature · 1 line

MST

Quoted 4 lines · Michael S. Tsirkin

TL;DR:
This fixes virtio in a way transparent to guest.
We should now be able to revert commits aa8580cd and df0acded19ec which worked around it in a way that's not transparent.

I didn't check dataplane BTW. Igor? Stefan?

Quoted 4 lines · Michael S. Tsirkin

TL;DR:
This fixes virtio in a way transparent to guest.
We should now be able to revert commits aa8580cd and df0acded19ec which worked around it in a way that's not transparent.

I didn't check dataplane BTW. Igor? Stefan?

TL;DR:
This fixes virtio in a way transparent to guest.
We should now be able to revert commits aa8580cd and df0acded19ec which worked around it in a way that's not transparent.

Michael S. Tsirkin (6):
virtio: introduce virtio_map
virtio: switch to virtio_map
virtio-blk: convert to virtqueue_map
virtio-serial: convert to virtio_map
virtio-scsi: convert to virtqueue_map
virtio: drop virtqueue_map_sg

 include/hw/virtio/virtio.h  |  3 +--
 hw/block/virtio-blk.c       |  5 +----
 hw/char/virtio-serial-bus.c |  5 +----
 hw/scsi/virtio-scsi.c       | 16 ++------------
 hw/virtio/virtio.c          | 52 +++++++++++++++++++++++++++++++++++----------
 5 files changed, 46 insertions(+), 35 deletions(-)