[Qemu-devel] [PATCH V4 for 3.1 0/4] Fix possible OOB during queuing packets

[Qemu-devel] [PATCH V4 for 3.1 0/4] Fix possible OOB during queuing packets

[Jason Wang]

Hi:

This series tries to fix a possible OOB during queueing packets
through qemu_net_queue_append_iov(). This could happen when it tries
to queue a packet whose size is larger than INT_MAX which may lead
integer overflow. We've fixed similar issue in the past during
qemu_net_queue_deliver_iov() by ignoring large packets there. Let's
just move the check earlier to qemu_sendv_packet_async() and reduce
the limitation to NET_BUFSIZE. A simple qtest were also added this.

Please review.

Thanks

Changes from V1:
- slient compiling warnings
Changes from V2:
- annotate pci_test_start() with GCC_FMT_ATTR()
- drop intermediate cmd string variable

Jason Wang (4):
  net: drop too large packet early
  virtio-net-test: accept variable length argument in pci_test_start()
  virtio-net-test: remove unused macro
  virtio-net-test: add large tx buffer test

 net/net.c               | 13 +++++----
 tests/virtio-net-test.c | 62 ++++++++++++++++++++++++++++++++++++-----
 2 files changed, 62 insertions(+), 13 deletions(-)

-- 
2.17.1

[Qemu-devel] [PATCH V4 for 3.1 1/4] net: drop too large packet early

[Jason Wang]

We try to detect and drop too large packet (>INT_MAX) in 1592a9947036
("net: ignore packet size greater than INT_MAX") during packet
delivering. Unfortunately, this is not sufficient as we may hit
another integer overflow when trying to queue such large packet in
qemu_net_queue_append_iov():

- size of the allocation may overflow on 32bit
- packet->size is integer which may overflow even on 64bit

Fixing this by move the check to qemu_sendv_packet_async() which is
the entrance of all networking codes and reduce the limit to
NET_BUFSIZE to be more conservative.

Cc: qemu-stable@nongnu.org
Cc: Li Qiang <liq3ea@163.com>
Reported-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Li Qiang <liq3ea@gmail.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 net/net.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/net/net.c b/net/net.c
index 07c194a8f6..affe1877cf 100644
--- a/net/net.c
+++ b/net/net.c
@@ -712,15 +712,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *sender,
                                 void *opaque)
 {
     NetClientState *nc = opaque;
-    size_t size = iov_size(iov, iovcnt);
     int ret;
 
-    if (size > INT_MAX) {
-        return size;
-    }
 
     if (nc->link_down) {
-        return size;
+        return iov_size(iov, iovcnt);
     }
 
     if (nc->receive_disabled) {
@@ -745,10 +741,15 @@ ssize_t qemu_sendv_packet_async(NetClientState *sender,
                                 NetPacketSent *sent_cb)
 {
     NetQueue *queue;
+    size_t size = iov_size(iov, iovcnt);
     int ret;
 
+    if (size > NET_BUFSIZE) {
+        return size;
+    }
+
     if (sender->link_down || !sender->peer) {
-        return iov_size(iov, iovcnt);
+        return size;
     }
 
     /* Let filters handle the packet first */
-- 
2.17.1

[Qemu-devel] [PATCH V4 for 3.1 2/4] virtio-net-test: accept variable length argument in pci_test_start()

[Jason Wang]

This allows flexibility to be reused for all kinds of command line
used by other tests.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 tests/virtio-net-test.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/tests/virtio-net-test.c b/tests/virtio-net-test.c
index dcb87a8b6e..587a043e67 100644
--- a/tests/virtio-net-test.c
+++ b/tests/virtio-net-test.c
@@ -52,17 +52,21 @@ static QVirtioPCIDevice *virtio_net_pci_init(QPCIBus *bus, int slot)
     return dev;
 }
 
-static QOSState *pci_test_start(int socket)
+GCC_FMT_ATTR(1, 2)
+static QOSState *pci_test_start(const char *cmd, ...)
 {
     QOSState *qs;
+    va_list ap;
     const char *arch = qtest_get_arch();
-    const char *cmd = "-netdev socket,fd=%d,id=hs0 -device "
-                      "virtio-net-pci,netdev=hs0";
 
     if (strcmp(arch, "i386") == 0 || strcmp(arch, "x86_64") == 0) {
-        qs = qtest_pc_boot(cmd, socket);
+        va_start(ap, cmd);
+        qs = qtest_pc_vboot(cmd, ap);
+        va_end(ap);
     } else if (strcmp(arch, "ppc64") == 0) {
-        qs = qtest_spapr_boot(cmd, socket);
+        va_start(ap, cmd);
+        qs = qtest_spapr_vboot(cmd, ap);
+        va_end(ap);
     } else {
         g_printerr("virtio-net tests are only available on x86 or ppc64\n");
         exit(EXIT_FAILURE);
@@ -223,7 +227,8 @@ static void pci_basic(gconstpointer data)
     ret = socketpair(PF_UNIX, SOCK_STREAM, 0, sv);
     g_assert_cmpint(ret, !=, -1);
 
-    qs = pci_test_start(sv[1]);
+    qs = pci_test_start("-netdev socket,fd=%d,id=hs0 -device "
+                        "virtio-net-pci,netdev=hs0", sv[1]);
     dev = virtio_net_pci_init(qs->pcibus, PCI_SLOT);
 
     rx = (QVirtQueuePCI *)qvirtqueue_setup(&dev->vdev, qs->alloc, 0);
-- 
2.17.1

[Qemu-devel] [PATCH V4 for 3.1 3/4] virtio-net-test: remove unused macro

[Jason Wang]

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 tests/virtio-net-test.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/tests/virtio-net-test.c b/tests/virtio-net-test.c
index 587a043e67..bdd6af9999 100644
--- a/tests/virtio-net-test.c
+++ b/tests/virtio-net-test.c
@@ -24,7 +24,6 @@
 
 #define PCI_SLOT_HP             0x06
 #define PCI_SLOT                0x04
-#define PCI_FN                  0x00
 
 #define QVIRTIO_NET_TIMEOUT_US (30 * 1000 * 1000)
 #define VNET_HDR_SIZE sizeof(struct virtio_net_hdr_mrg_rxbuf)
-- 
2.17.1

[Qemu-devel] [PATCH V4 for 3.1 4/4] virtio-net-test: add large tx buffer test

[Jason Wang]

This test tries to build a packet whose size is greater than INT_MAX
which tries to trigger integer overflow in qemu_net_queue_append_iov()
which may result OOB.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 tests/virtio-net-test.c | 44 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)

diff --git a/tests/virtio-net-test.c b/tests/virtio-net-test.c
index bdd6af9999..566596a397 100644
--- a/tests/virtio-net-test.c
+++ b/tests/virtio-net-test.c
@@ -245,6 +245,49 @@ static void pci_basic(gconstpointer data)
     g_free(dev);
     qtest_shutdown(qs);
 }
+
+static void large_tx(gconstpointer data)
+{
+    QVirtioPCIDevice *dev;
+    QOSState *qs;
+    QVirtQueuePCI *tx, *rx;
+    QVirtQueue *vq;
+    uint64_t req_addr;
+    uint32_t free_head;
+    size_t alloc_size = UINT_MAX / 64;
+    int i;
+
+    qs = pci_test_start("-netdev hubport,id=hp0,hubid=0 "
+                        "-device virtio-net-pci,netdev=hp0 ");
+    dev = virtio_net_pci_init(qs->pcibus, PCI_SLOT);
+
+    rx = (QVirtQueuePCI *)qvirtqueue_setup(&dev->vdev, qs->alloc, 0);
+    tx = (QVirtQueuePCI *)qvirtqueue_setup(&dev->vdev, qs->alloc, 1);
+
+    driver_init(&dev->vdev);
+    vq = &tx->vq;
+
+    /* Bypass the limitation by pointing several descriptors to a single
+     * smaller area */
+    req_addr = guest_alloc(qs->alloc, alloc_size);
+    free_head = qvirtqueue_add(vq, req_addr, alloc_size, false, true);
+
+    for (i = 0; i < 64; i++) {
+        qvirtqueue_add(vq, req_addr, alloc_size, false, i == 63 ?
+                       false : true);
+    }
+    qvirtqueue_kick(&dev->vdev, vq, free_head);
+
+    qvirtio_wait_used_elem(&dev->vdev, vq, free_head, NULL,
+                           QVIRTIO_NET_TIMEOUT_US);
+
+    qvirtqueue_cleanup(dev->vdev.bus, &tx->vq, qs->alloc);
+    qvirtqueue_cleanup(dev->vdev.bus, &rx->vq, qs->alloc);
+    qvirtio_pci_device_disable(dev);
+    g_free(dev->pdev);
+    g_free(dev);
+    qtest_shutdown(qs);
+}
 #endif
 
 static void hotplug(void)
@@ -270,6 +313,7 @@ int main(int argc, char **argv)
     qtest_add_data_func("/virtio/net/pci/basic", send_recv_test, pci_basic);
     qtest_add_data_func("/virtio/net/pci/rx_stop_cont",
                         stop_cont_test, pci_basic);
+    qtest_add_data_func("/virtio/net/pci/large_tx", NULL, large_tx);
 #endif
     qtest_add_func("/virtio/net/pci/hotplug", hotplug);
 
-- 
2.17.1

Re: [Qemu-devel] [PATCH V4 for 3.1 1/4] net: drop too large packet early

[Eric Blake]

On 12/3/18 4:06 AM, Jason Wang wrote:
> We try to detect and drop too large packet (>INT_MAX) in 1592a9947036
> ("net: ignore packet size greater than INT_MAX") during packet
> delivering. Unfortunately, this is not sufficient as we may hit
> another integer overflow when trying to queue such large packet in
> qemu_net_queue_append_iov():
> 
> - size of the allocation may overflow on 32bit
> - packet->size is integer which may overflow even on 64bit
> 
> Fixing this by move the check to qemu_sendv_packet_async() which is

s/move/moving/

> the entrance of all networking codes and reduce the limit to
> NET_BUFSIZE to be more conservative.

Please mention commit 1592a994 in the commit message (since you are 
effectively reverting that with this as its replacement), and if this is 
(as I suspect) an additional patch required for the complete fix to 
CVE-2018-10839, also mention that.

> 
> Cc: qemu-stable@nongnu.org
> Cc: Li Qiang <liq3ea@163.com>
> Reported-by: Li Qiang <liq3ea@gmail.com>
> Reviewed-by: Li Qiang <liq3ea@gmail.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>   net/net.c | 13 +++++++------
>   1 file changed, 7 insertions(+), 6 deletions(-)
> 
> diff --git a/net/net.c b/net/net.c
> index 07c194a8f6..affe1877cf 100644
> --- a/net/net.c
> +++ b/net/net.c
> @@ -712,15 +712,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *sender,
>                                   void *opaque)
>   {
>       NetClientState *nc = opaque;
> -    size_t size = iov_size(iov, iovcnt);
>       int ret;
>   
> -    if (size > INT_MAX) {
> -        return size;
> -    }
>   
>       if (nc->link_down) {
> -        return size;
> +        return iov_size(iov, iovcnt);

Reverts 1592a994, and...

>       }
>   
>       if (nc->receive_disabled) {
> @@ -745,10 +741,15 @@ ssize_t qemu_sendv_packet_async(NetClientState *sender,
>                                   NetPacketSent *sent_cb)
>   {
>       NetQueue *queue;
> +    size_t size = iov_size(iov, iovcnt);
>       int ret;
>   
> +    if (size > NET_BUFSIZE) {
> +        return size;
> +    }

...returns early for packets larger than 68k (a much smaller limit than 
INT_MAX, which makes analysis for int overflow a lot easier) at a saner 
point in the code.  Returning a large value is weird, but auditing all 
callers:

hw/net/virtio-net.c:        ret = 
qemu_sendv_packet_async(qemu_get_subqueue(n->nic, queue_index),
  - only checks if ret is 0 (returns -EBUSY) or not (assumes packet was 
sent)
net/netmap.c:        iovsize = qemu_sendv_packet_async(&s->nc, s->iov, 
iovcnt,
  - only checks if ret is 0 (starts polling) or not (assumes packet was 
sent)
net/net.c:    return qemu_sendv_packet_async(nc, iov, iovcnt, NULL);
  - implementation of qemu_sendv_packet() - so we have to audit those 
callers as well:

hw/net/net_tx_pkt.c:        qemu_sendv_packet(nc, iov, iov_cnt);
hw/net/rocker/rocker_fp.c:        qemu_sendv_packet(nc, iov, iovcnt);
hw/net/rtl8139.c:            qemu_sendv_packet(qemu_get_queue(s->nic), 
iov, 3);
net/hub.c:        qemu_sendv_packet(&port->nc, iov, iovcnt);
- all four of these do not check the return status

So, it looks like none of the callers cares if the return value is 
overlarge (no further math on the values), just that it is non-zero 
(where the callers then presumably assume the packet was sent). 
However, I am not familiar enough with the code to know if skipping the 
packet by returning a non-zero value is going to have knock-on effects - 
that is, my audit shows what the callers do, but not whether it was sane.

> +
>       if (sender->link_down || !sender->peer) {
> -        return iov_size(iov, iovcnt);
> +        return size;
>       }

If this is indeed CVE fixing, then we want it in -rc4, but I don't know 
if the patch is correctly secure yet without answers to my questions. 
Especially on a CVE fix for -rc4, you want to make the reviewer's life 
as easy as possible by providing a commit message that includes enough 
details to make analysis easy.

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

Re: [Qemu-devel] [PATCH V4 for 3.1 0/4] Fix possible OOB during queuing packets

[Peter Maydell]

On Mon, 3 Dec 2018 at 10:06, Jason Wang <jasowang@redhat.com> wrote:
>
> Hi:
>
> This series tries to fix a possible OOB during queueing packets
> through qemu_net_queue_append_iov(). This could happen when it tries
> to queue a packet whose size is larger than INT_MAX which may lead
> integer overflow. We've fixed similar issue in the past during
> qemu_net_queue_deliver_iov() by ignoring large packets there. Let's
> just move the check earlier to qemu_sendv_packet_async() and reduce
> the limitation to NET_BUFSIZE. A simple qtest were also added this.
>
> Please review.

I did a test build and run, and the new test generates warning
messages during "make check":
  /ppc64/virtio/net/pci/large_tx:
qemu-system-ppc64: warning: hub 0 is not connected to host network
(similarly for /i386/ and /x86_64/).

thank
-- PMM

Re: [Qemu-devel] [PATCH V4 for 3.1 2/4] virtio-net-test: accept variable length argument in pci_test_start()

[Eric Blake]

On 12/3/18 4:06 AM, Jason Wang wrote:
> This allows flexibility to be reused for all kinds of command line
> used by other tests.
> 
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>   tests/virtio-net-test.c | 17 +++++++++++------
>   1 file changed, 11 insertions(+), 6 deletions(-)
> 

Reviewed-by: Eric Blake <eblake@redhat.com>

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

Re: [Qemu-devel] [PATCH V4 for 3.1 3/4] virtio-net-test: remove unused macro

[Eric Blake]

On 12/3/18 4:06 AM, Jason Wang wrote:
> Reviewed-by: Thomas Huth <thuth@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>   tests/virtio-net-test.c | 1 -
>   1 file changed, 1 deletion(-)
> 

Reviewed-by: Eric Blake <eblake@redhat.com>

> diff --git a/tests/virtio-net-test.c b/tests/virtio-net-test.c
> index 587a043e67..bdd6af9999 100644
> --- a/tests/virtio-net-test.c
> +++ b/tests/virtio-net-test.c
> @@ -24,7 +24,6 @@
>   
>   #define PCI_SLOT_HP             0x06
>   #define PCI_SLOT                0x04
> -#define PCI_FN                  0x00
>   
>   #define QVIRTIO_NET_TIMEOUT_US (30 * 1000 * 1000)
>   #define VNET_HDR_SIZE sizeof(struct virtio_net_hdr_mrg_rxbuf)
> 

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

Re: [Qemu-devel] [PATCH V4 for 3.1 4/4] virtio-net-test: add large tx buffer test

[Eric Blake]

On 12/3/18 4:06 AM, Jason Wang wrote:
> This test tries to build a packet whose size is greater than INT_MAX
> which tries to trigger integer overflow in qemu_net_queue_append_iov()
> which may result OOB.

Can you also add a packet just slightly larger than NET_BUFSIZE (68k) to 
show that we aren't having any further issues at even smaller (and more 
likely) values of oversized packets?

> 
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>   tests/virtio-net-test.c | 44 +++++++++++++++++++++++++++++++++++++++++
>   1 file changed, 44 insertions(+)
> 

> +++ b/tests/virtio-net-test.c
> @@ -245,6 +245,49 @@ static void pci_basic(gconstpointer data)
>       g_free(dev);
>       qtest_shutdown(qs);
>   }
> +
> +static void large_tx(gconstpointer data)
> +{
> +    QVirtioPCIDevice *dev;
> +    QOSState *qs;
> +    QVirtQueuePCI *tx, *rx;
> +    QVirtQueue *vq;
> +    uint64_t req_addr;
> +    uint32_t free_head;
> +    size_t alloc_size = UINT_MAX / 64;
> +    int i;
> +
> +    qs = pci_test_start("-netdev hubport,id=hp0,hubid=0 "
> +                        "-device virtio-net-pci,netdev=hp0 ");

Why the trailing space?

> +    dev = virtio_net_pci_init(qs->pcibus, PCI_SLOT);
> +
> +    rx = (QVirtQueuePCI *)qvirtqueue_setup(&dev->vdev, qs->alloc, 0);
> +    tx = (QVirtQueuePCI *)qvirtqueue_setup(&dev->vdev, qs->alloc, 1);
> +
> +    driver_init(&dev->vdev);
> +    vq = &tx->vq;
> +
> +    /* Bypass the limitation by pointing several descriptors to a single
> +     * smaller area */
> +    req_addr = guest_alloc(qs->alloc, alloc_size);
> +    free_head = qvirtqueue_add(vq, req_addr, alloc_size, false, true);
> +
> +    for (i = 0; i < 64; i++) {
> +        qvirtqueue_add(vq, req_addr, alloc_size, false, i == 63 ?
> +                       false : true);

Any time I see both 'true' and 'false' in a ?: operator, I have to 
wonder why you didn't just write the simpler version directly on the 
condition.  This is the same as  'i != 63'.

> +    }
> +    qvirtqueue_kick(&dev->vdev, vq, free_head);
> +
> +    qvirtio_wait_used_elem(&dev->vdev, vq, free_head, NULL,
> +                           QVIRTIO_NET_TIMEOUT_US);
> +
> +    qvirtqueue_cleanup(dev->vdev.bus, &tx->vq, qs->alloc);
> +    qvirtqueue_cleanup(dev->vdev.bus, &rx->vq, qs->alloc);
> +    qvirtio_pci_device_disable(dev);
> +    g_free(dev->pdev);
> +    g_free(dev);
> +    qtest_shutdown(qs);
> +}
>   #endif
>   
>   static void hotplug(void)
> @@ -270,6 +313,7 @@ int main(int argc, char **argv)
>       qtest_add_data_func("/virtio/net/pci/basic", send_recv_test, pci_basic);
>       qtest_add_data_func("/virtio/net/pci/rx_stop_cont",
>                           stop_cont_test, pci_basic);
> +    qtest_add_data_func("/virtio/net/pci/large_tx", NULL, large_tx);
>   #endif
>       qtest_add_func("/virtio/net/pci/hotplug", hotplug);
>   

I can reproduce Peter's complaints of this introducing noise to 'make 
check':
qemu-system-x86_64: warning: hub 0 is not connected to host network

With patches 2-4 applied but patch 1 omitted, 'make check' fails with:

Broken pipe
tests/libqtest.c:125: kill_qemu() detected QEMU death from signal 11 
(Segmentation fault) (core dumped)
GTester: last random seed: R02S9569aabcb2d834a9dadcce272c5588db

so the test is definitely triggering the problem as patched by part 1. 
Although I'm not confident enough of what the test is doing for an R-b, 
and would like it to be less noisy, I can at least add:

Tested-by: Eric Blake <eblake@redhat.com>

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

Re: [Qemu-devel] [PATCH V4 for 3.1 1/4] net: drop too large packet early

[Thomas Huth]

On 2018-12-03 11:06, Jason Wang wrote:
> We try to detect and drop too large packet (>INT_MAX) in 1592a9947036
> ("net: ignore packet size greater than INT_MAX") during packet
> delivering. Unfortunately, this is not sufficient as we may hit
> another integer overflow when trying to queue such large packet in
> qemu_net_queue_append_iov():
> 
> - size of the allocation may overflow on 32bit
> - packet->size is integer which may overflow even on 64bit
> 
> Fixing this by move the check to qemu_sendv_packet_async() which is
> the entrance of all networking codes and reduce the limit to
> NET_BUFSIZE to be more conservative.
> 
> Cc: qemu-stable@nongnu.org
> Cc: Li Qiang <liq3ea@163.com>
> Reported-by: Li Qiang <liq3ea@gmail.com>
> Reviewed-by: Li Qiang <liq3ea@gmail.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  net/net.c | 13 +++++++------
>  1 file changed, 7 insertions(+), 6 deletions(-)

Since this is a critical patch for rc4, here's a verbose review...

> diff --git a/net/net.c b/net/net.c
> index 07c194a8f6..affe1877cf 100644
> --- a/net/net.c
> +++ b/net/net.c
> @@ -712,15 +712,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *sender,
>                                  void *opaque)
>  {
>      NetClientState *nc = opaque;
> -    size_t size = iov_size(iov, iovcnt);
>      int ret;
>  
> -    if (size > INT_MAX) {
> -        return size;
> -    }
>  
>      if (nc->link_down) {
> -        return size;
> +        return iov_size(iov, iovcnt);
>      }

In case you respin this patch again, please make
qemu_deliver_packet_iov() "static", so that it is clear that it can not
be called directly from the outside anymore. And in case there is no
need to respin, please consider to send a separate patch for 4.0 instead.

Ok, thinking now load about the call chain:

Anyway, qemu_deliver_packet_iov is not directly called from any other
file currently, so this is ok ...

So let's see how it is used in net.c ... it's only used as paramter
here: qemu_new_net_queue(qemu_deliver_packet_iov, nc) ...
qemu_new_net_queue() assigns it to NetQueue->deliver which is only used
within queue.c.

Functions using that ->deliver function pointer are the static functions
qemu_net_queue_deliver() and qemu_net_queue_deliver_iov(). First one
only uses one iov, so I don't think we can overflow the size here.
Second one is used in turn are used by the public function
qemu_net_queue_send_iov(). This has two callers, one in net.c in
qemu_sendv_packet_async() which you guard below ==> OK.
The other caller is in qemu_netfilter_pass_to_next() in filter.c - and
this function is called from many more other places ... but as far as I
can see, these don't call it in a way where the size could overflow.

==> Removing the check in qemu_deliver_packet_iov() sounds ok to me, if
it is checked in qemu_sendv_packet_async() instead.

>      if (nc->receive_disabled) {
> @@ -745,10 +741,15 @@ ssize_t qemu_sendv_packet_async(NetClientState *sender,
>                                  NetPacketSent *sent_cb)
>  {
>      NetQueue *queue;
> +    size_t size = iov_size(iov, iovcnt);
>      int ret;
>  
> +    if (size > NET_BUFSIZE) {
> +        return size;
> +    }

It's a little bit unfortunate that the unsigned size will be cast to
ssize_t, so a very large size could suddenly change sign. But as Eric
already wrote in his mail, it seems like the callers are either ignoring
the return value, or just checking for != 0, so it should be ok for now.

To be more consistent, maybe it would be better to always return an
negative error code here instead?

>      if (sender->link_down || !sender->peer) {
> -        return iov_size(iov, iovcnt);
> +        return size;
>      }
>  
>      /* Let filters handle the packet first */
> 

I think I'm fine if this patch is applied in its current shape for rc4, so:

Reviewed-by: Thomas Huth <thuth@redhat.com>

... but please consider some follow-up patches to make
qemu_deliver_packet_iov() static, and maybe to return an error code in
qemu_sendv_packet_async() instead.

Re: [Qemu-devel] [PATCH V4 for 3.1 2/4] virtio-net-test: accept variable length argument in pci_test_start()

[Thomas Huth]

On 2018-12-03 11:06, Jason Wang wrote:
> This allows flexibility to be reused for all kinds of command line
> used by other tests.
> 
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  tests/virtio-net-test.c | 17 +++++++++++------
>  1 file changed, 11 insertions(+), 6 deletions(-)

Reviewed-by: Thomas Huth <thuth@redhat.com>

Re: [Qemu-devel] [PATCH V4 for 3.1 0/4] Fix possible OOB during queuing packets

[Jason Wang]

On 2018/12/4 上午12:18, Peter Maydell wrote:
> On Mon, 3 Dec 2018 at 10:06, Jason Wang <jasowang@redhat.com> wrote:
>> Hi:
>>
>> This series tries to fix a possible OOB during queueing packets
>> through qemu_net_queue_append_iov(). This could happen when it tries
>> to queue a packet whose size is larger than INT_MAX which may lead
>> integer overflow. We've fixed similar issue in the past during
>> qemu_net_queue_deliver_iov() by ignoring large packets there. Let's
>> just move the check earlier to qemu_sendv_packet_async() and reduce
>> the limitation to NET_BUFSIZE. A simple qtest were also added this.
>>
>> Please review.
> I did a test build and run, and the new test generates warning
> messages during "make check":
>    /ppc64/virtio/net/pci/large_tx:
> qemu-system-ppc64: warning: hub 0 is not connected to host network
> (similarly for /i386/ and /x86_64/).
>
> thank
> -- PMM


This is intended, we don't need any host networking device. This makes 
it very simpler and easier to trigger the queuing at hub. I can add a 
patch to suppress this warning if qtest is enabled.

Thanks

Re: [Qemu-devel] [PATCH V4 for 3.1 1/4] net: drop too large packet early

[Jason Wang]

On 2018/12/4 上午12:18, Eric Blake wrote:
> On 12/3/18 4:06 AM, Jason Wang wrote:
>> We try to detect and drop too large packet (>INT_MAX) in 1592a9947036
>> ("net: ignore packet size greater than INT_MAX") during packet
>> delivering. Unfortunately, this is not sufficient as we may hit
>> another integer overflow when trying to queue such large packet in
>> qemu_net_queue_append_iov():
>>
>> - size of the allocation may overflow on 32bit
>> - packet->size is integer which may overflow even on 64bit
>>
>> Fixing this by move the check to qemu_sendv_packet_async() which is
>
> s/move/moving/


Ok.


>
>> the entrance of all networking codes and reduce the limit to
>> NET_BUFSIZE to be more conservative.
>
> Please mention commit 1592a994 in the commit message (since you are 
> effectively reverting that with this as its replacement),


I think I did it?


> and if this is (as I suspect) an additional patch required for the 
> complete fix to CVE-2018-10839, also mention that.


Ok.


>
>>
>> Cc: qemu-stable@nongnu.org
>> Cc: Li Qiang <liq3ea@163.com>
>> Reported-by: Li Qiang <liq3ea@gmail.com>
>> Reviewed-by: Li Qiang <liq3ea@gmail.com>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   net/net.c | 13 +++++++------
>>   1 file changed, 7 insertions(+), 6 deletions(-)
>>
>> diff --git a/net/net.c b/net/net.c
>> index 07c194a8f6..affe1877cf 100644
>> --- a/net/net.c
>> +++ b/net/net.c
>> @@ -712,15 +712,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState 
>> *sender,
>>                                   void *opaque)
>>   {
>>       NetClientState *nc = opaque;
>> -    size_t size = iov_size(iov, iovcnt);
>>       int ret;
>>   -    if (size > INT_MAX) {
>> -        return size;
>> -    }
>>         if (nc->link_down) {
>> -        return size;
>> +        return iov_size(iov, iovcnt);
>
> Reverts 1592a994, and...
>
>>       }
>>         if (nc->receive_disabled) {
>> @@ -745,10 +741,15 @@ ssize_t qemu_sendv_packet_async(NetClientState 
>> *sender,
>>                                   NetPacketSent *sent_cb)
>>   {
>>       NetQueue *queue;
>> +    size_t size = iov_size(iov, iovcnt);
>>       int ret;
>>   +    if (size > NET_BUFSIZE) {
>> +        return size;
>> +    }
>
> ...returns early for packets larger than 68k (a much smaller limit 
> than INT_MAX, which makes analysis for int overflow a lot easier) at a 
> saner point in the code.  Returning a large value is weird, 


Might be, but we did this for years, see the following return value when 
link is down.


> but auditing all callers:
>
> hw/net/virtio-net.c:        ret = 
> qemu_sendv_packet_async(qemu_get_subqueue(n->nic, queue_index),
>  - only checks if ret is 0 (returns -EBUSY) or not (assumes packet was 
> sent)
> net/netmap.c:        iovsize = qemu_sendv_packet_async(&s->nc, s->iov, 
> iovcnt,
>  - only checks if ret is 0 (starts polling) or not (assumes packet was 
> sent)
> net/net.c:    return qemu_sendv_packet_async(nc, iov, iovcnt, NULL);
>  - implementation of qemu_sendv_packet() - so we have to audit those 
> callers as well:
>
> hw/net/net_tx_pkt.c:        qemu_sendv_packet(nc, iov, iov_cnt);
> hw/net/rocker/rocker_fp.c:        qemu_sendv_packet(nc, iov, iovcnt);
> hw/net/rtl8139.c: qemu_sendv_packet(qemu_get_queue(s->nic), iov, 3);
> net/hub.c:        qemu_sendv_packet(&port->nc, iov, iovcnt);
> - all four of these do not check the return status
>
> So, it looks like none of the callers cares if the return value is 
> overlarge (no further math on the values), just that it is non-zero 
> (where the callers then presumably assume the packet was sent). 


Yes, the caller may only care if it returns zero.


> However, I am not familiar enough with the code to know if skipping 
> the packet by returning a non-zero value is going to have knock-on 
> effects - that is, my audit shows what the callers do, but not whether 
> it was sane.
>

The difference between qemu_sendv_packet() and qemu_sendv_packet_async() 
is that the latter can trigger a callback (sent_cb) when peer can accept 
more packets. This could be used by some high speed networking 
implementation to prevent the source from producing more packets and 
wasting cpu cycles in dropping packets. After peer can accept more, 
sent_cb usually enable the source to produce packets. Those who use 
qemu_sendv_packet() will just waste some cpu in dropping the packets.

Consider we are emulating ethernet and packet will be copied if queued, 
it's safe to assume that the packet was sent.


>> +
>>       if (sender->link_down || !sender->peer) {
>> -        return iov_size(iov, iovcnt);
>> +        return size;
>>       }
>
> If this is indeed CVE fixing, then we want it in -rc4, but I don't 
> know if the patch is correctly secure yet without answers to my 
> questions. Especially on a CVE fix for -rc4, you want to make the 
> reviewer's life as easy as possible by providing a commit message that 
> includes enough details to make analysis easy.


Hope my answer help. If it is, I will add them to the commit log.

Thanks for the reviewing.

Re: [Qemu-devel] [PATCH V4 for 3.1 4/4] virtio-net-test: add large tx buffer test

[Jason Wang]

On 2018/12/4 上午12:46, Eric Blake wrote:
> On 12/3/18 4:06 AM, Jason Wang wrote:
>> This test tries to build a packet whose size is greater than INT_MAX
>> which tries to trigger integer overflow in qemu_net_queue_append_iov()
>> which may result OOB.
>
> Can you also add a packet just slightly larger than NET_BUFSIZE (68k) 
> to show that we aren't having any further issues at even smaller (and 
> more likely) values of oversized packets?


Ok.


>
>>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   tests/virtio-net-test.c | 44 +++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 44 insertions(+)
>>
>
>> +++ b/tests/virtio-net-test.c
>> @@ -245,6 +245,49 @@ static void pci_basic(gconstpointer data)
>>       g_free(dev);
>>       qtest_shutdown(qs);
>>   }
>> +
>> +static void large_tx(gconstpointer data)
>> +{
>> +    QVirtioPCIDevice *dev;
>> +    QOSState *qs;
>> +    QVirtQueuePCI *tx, *rx;
>> +    QVirtQueue *vq;
>> +    uint64_t req_addr;
>> +    uint32_t free_head;
>> +    size_t alloc_size = UINT_MAX / 64;
>> +    int i;
>> +
>> +    qs = pci_test_start("-netdev hubport,id=hp0,hubid=0 "
>> +                        "-device virtio-net-pci,netdev=hp0 ");
>
> Why the trailing space?


I guess this is a cut and paste error.


>
>> +    dev = virtio_net_pci_init(qs->pcibus, PCI_SLOT);
>> +
>> +    rx = (QVirtQueuePCI *)qvirtqueue_setup(&dev->vdev, qs->alloc, 0);
>> +    tx = (QVirtQueuePCI *)qvirtqueue_setup(&dev->vdev, qs->alloc, 1);
>> +
>> +    driver_init(&dev->vdev);
>> +    vq = &tx->vq;
>> +
>> +    /* Bypass the limitation by pointing several descriptors to a 
>> single
>> +     * smaller area */
>> +    req_addr = guest_alloc(qs->alloc, alloc_size);
>> +    free_head = qvirtqueue_add(vq, req_addr, alloc_size, false, true);
>> +
>> +    for (i = 0; i < 64; i++) {
>> +        qvirtqueue_add(vq, req_addr, alloc_size, false, i == 63 ?
>> +                       false : true);
>
> Any time I see both 'true' and 'false' in a ?: operator, I have to 
> wonder why you didn't just write the simpler version directly on the 
> condition.  This is the same as  'i != 63'.
>

Ok.


>> +    }
>> +    qvirtqueue_kick(&dev->vdev, vq, free_head);
>> +
>> +    qvirtio_wait_used_elem(&dev->vdev, vq, free_head, NULL,
>> +                           QVIRTIO_NET_TIMEOUT_US);
>> +
>> +    qvirtqueue_cleanup(dev->vdev.bus, &tx->vq, qs->alloc);
>> +    qvirtqueue_cleanup(dev->vdev.bus, &rx->vq, qs->alloc);
>> +    qvirtio_pci_device_disable(dev);
>> +    g_free(dev->pdev);
>> +    g_free(dev);
>> +    qtest_shutdown(qs);
>> +}
>>   #endif
>>     static void hotplug(void)
>> @@ -270,6 +313,7 @@ int main(int argc, char **argv)
>>       qtest_add_data_func("/virtio/net/pci/basic", send_recv_test, 
>> pci_basic);
>>       qtest_add_data_func("/virtio/net/pci/rx_stop_cont",
>>                           stop_cont_test, pci_basic);
>> +    qtest_add_data_func("/virtio/net/pci/large_tx", NULL, large_tx);
>>   #endif
>>       qtest_add_func("/virtio/net/pci/hotplug", hotplug);
>
> I can reproduce Peter's complaints of this introducing noise to 'make 
> check':
> qemu-system-x86_64: warning: hub 0 is not connected to host network


Yes, this is intended since the test does not require any host network. 
I will add a patch to suppress this warning if qtest is enabled.


>
> With patches 2-4 applied but patch 1 omitted, 'make check' fails with:
>
> Broken pipe
> tests/libqtest.c:125: kill_qemu() detected QEMU death from signal 11 
> (Segmentation fault) (core dumped)
> GTester: last random seed: R02S9569aabcb2d834a9dadcce272c5588db
>
> so the test is definitely triggering the problem as patched by part 1. 
> Although I'm not confident enough of what the test is doing for an 
> R-b, and would like it to be less noisy, I can at least add:
>
> Tested-by: Eric Blake <eblake@redhat.com>
>

Thanks

Re: [Qemu-devel] [PATCH V4 for 3.1 1/4] net: drop too large packet early

[Jason Wang]

On 2018/12/4 上午2:13, Thomas Huth wrote:
> On 2018-12-03 11:06, Jason Wang wrote:
>> We try to detect and drop too large packet (>INT_MAX) in 1592a9947036
>> ("net: ignore packet size greater than INT_MAX") during packet
>> delivering. Unfortunately, this is not sufficient as we may hit
>> another integer overflow when trying to queue such large packet in
>> qemu_net_queue_append_iov():
>>
>> - size of the allocation may overflow on 32bit
>> - packet->size is integer which may overflow even on 64bit
>>
>> Fixing this by move the check to qemu_sendv_packet_async() which is
>> the entrance of all networking codes and reduce the limit to
>> NET_BUFSIZE to be more conservative.
>>
>> Cc: qemu-stable@nongnu.org
>> Cc: Li Qiang <liq3ea@163.com>
>> Reported-by: Li Qiang <liq3ea@gmail.com>
>> Reviewed-by: Li Qiang <liq3ea@gmail.com>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   net/net.c | 13 +++++++------
>>   1 file changed, 7 insertions(+), 6 deletions(-)
> Since this is a critical patch for rc4, here's a verbose review...
>
>> diff --git a/net/net.c b/net/net.c
>> index 07c194a8f6..affe1877cf 100644
>> --- a/net/net.c
>> +++ b/net/net.c
>> @@ -712,15 +712,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *sender,
>>                                   void *opaque)
>>   {
>>       NetClientState *nc = opaque;
>> -    size_t size = iov_size(iov, iovcnt);
>>       int ret;
>>   
>> -    if (size > INT_MAX) {
>> -        return size;
>> -    }
>>   
>>       if (nc->link_down) {
>> -        return size;
>> +        return iov_size(iov, iovcnt);
>>       }
> In case you respin this patch again, please make
> qemu_deliver_packet_iov() "static", so that it is clear that it can not
> be called directly from the outside anymore. And in case there is no
> need to respin, please consider to send a separate patch for 4.0 instead.


Yes, will try to make it for the next version.


>
> Ok, thinking now load about the call chain:
>
> Anyway, qemu_deliver_packet_iov is not directly called from any other
> file currently, so this is ok ...
>
> So let's see how it is used in net.c ... it's only used as paramter
> here: qemu_new_net_queue(qemu_deliver_packet_iov, nc) ...
> qemu_new_net_queue() assigns it to NetQueue->deliver which is only used
> within queue.c.
>
> Functions using that ->deliver function pointer are the static functions
> qemu_net_queue_deliver() and qemu_net_queue_deliver_iov(). First one
> only uses one iov, so I don't think we can overflow the size here.
> Second one is used in turn are used by the public function
> qemu_net_queue_send_iov(). This has two callers, one in net.c in
> qemu_sendv_packet_async() which you guard below ==> OK.
> The other caller is in qemu_netfilter_pass_to_next() in filter.c - and
> this function is called from many more other places ... but as far as I
> can see, these don't call it in a way where the size could overflow.
>
> ==> Removing the check in qemu_deliver_packet_iov() sounds ok to me, if
> it is checked in qemu_sendv_packet_async() instead.


Yes. Let me add more in the commit message.


>
>>       if (nc->receive_disabled) {
>> @@ -745,10 +741,15 @@ ssize_t qemu_sendv_packet_async(NetClientState *sender,
>>                                   NetPacketSent *sent_cb)
>>   {
>>       NetQueue *queue;
>> +    size_t size = iov_size(iov, iovcnt);
>>       int ret;
>>   
>> +    if (size > NET_BUFSIZE) {
>> +        return size;
>> +    }
> It's a little bit unfortunate that the unsigned size will be cast to
> ssize_t, so a very large size could suddenly change sign. But as Eric
> already wrote in his mail, it seems like the callers are either ignoring
> the return value, or just checking for != 0, so it should be ok for now.
>
> To be more consistent, maybe it would be better to always return an
> negative error code here instead?


Yes, and we can do it for 4.0.


>
>>       if (sender->link_down || !sender->peer) {
>> -        return iov_size(iov, iovcnt);
>> +        return size;
>>       }
>>   
>>       /* Let filters handle the packet first */
>>
> I think I'm fine if this patch is applied in its current shape for rc4, so:
>
> Reviewed-by: Thomas Huth <thuth@redhat.com>
>
> ... but please consider some follow-up patches to make
> qemu_deliver_packet_iov() static, and maybe to return an error code in
> qemu_sendv_packet_async() instead.


Yes.


Thanks for the reviewing.