qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH v4 27/30] vmxnet3: validate interrupt indices co


From: Dmitry Fleytman
Subject: Re: [Qemu-devel] [PATCH v4 27/30] vmxnet3: validate interrupt indices coming from guest
Date: Fri, 4 Apr 2014 12:47:08 +0300


On Apr 3, 2014, at 19:07 PM, Michael S. Tsirkin <address@hidden> wrote:

On Tue, Apr 01, 2014 at 02:07:52PM +0100, Dr. David Alan Gilbert wrote:
* Dmitry Fleytman (address@hidden) wrote:

On Apr 1, 2014, at 14:33 PM, Dr. David Alan Gilbert <address@hidden> wrote:

* Michael S. Tsirkin (address@hidden) wrote:
From: Dmitry Fleytman <address@hidden>

CVE-2013-4544

Signed-off-by: Dmitry Fleytman <address@hidden>
Reported-by: Michael S. Tsirkin <address@hidden>
Signed-off-by: Michael S. Tsirkin <address@hidden>
---
hw/net/vmxnet3.c | 36 ++++++++++++++++++++++++++++++++++--
1 file changed, 34 insertions(+), 2 deletions(-)

diff --git a/hw/net/vmxnet3.c b/hw/net/vmxnet3.c
index 5be807c..a4b5c11 100644
--- a/hw/net/vmxnet3.c
+++ b/hw/net/vmxnet3.c
@@ -52,6 +52,9 @@
#define VMXNET3_DEVICE_VERSION    0x1
#define VMXNET3_DEVICE_REVISION   0x1

+/* Number of interrupt vectors for INTx/MSI */
+#define VMXNET3_MAX_NMSIX_INTRS   (1)

As per Dmitry's reply this is apparently the number of non-MSIX
interrupts; can we change the comment to make this clear.

Not sure how to change it.
There are three modes of operation:
1. INTx - 1 interrupt is used
2. MSI - 1 interrupt is used
3. MSIx - up to 25 interrupts are used.

This define covers 2 first modes of operation.
Would something like

/* Number of interrupt vectors for non-MSIx modes */

be better?

Yes - that's fine.

Dave


OK Dmitry, can you please resend patches with suggested comment changes?

Done. Sent to the list.





+
/* Macros for rings descriptors access */
#define VMXNET3_READ_TX_QUEUE_DESCR8(dpa, field) \
   (vmw_shmem_ld8(dpa + offsetof(struct Vmxnet3_TxQueueDesc, field)))
@@ -1305,6 +1308,34 @@ static bool vmxnet3_verify_intx(VMXNET3State *s, int intx)
          (pci_get_byte(s->parent_obj.config + PCI_INTERRUPT_PIN) - 1));
}

+static void vmxnet3_validate_interrupt_idx(bool is_msix, int idx)
+{
+    int max_ints = is_msix ? VMXNET3_MAX_INTRS : VMXNET3_MAX_NMSIX_INTRS;
+    if (idx >= max_ints) {
+        hw_error("Bad interrupt index: %d\n", idx);
+    }

Can we avoid hw_error here, we're using in this in state load, so I'm
OK at the thought of error_report, but then we should make
this return a bool to say it's failed and bubble this up so
that it just fails the post_load test    like any other bad state load.

Dave

+}
+
+static void vmxnet3_validate_interrupts(VMXNET3State *s)
+{
+    int i;
+
+    VMW_CFPRN("Verifying event interrupt index (%d)", s->event_int_idx);
+    vmxnet3_validate_interrupt_idx(s->msix_used, s->event_int_idx);
+
+    for (i = 0; i < s->txq_num; i++) {
+        int idx = s->txq_descr[i].intr_idx;
+        VMW_CFPRN("Verifying TX queue %d interrupt index (%d)", i, idx);
+        vmxnet3_validate_interrupt_idx(s->msix_used, idx);
+    }
+
+    for (i = 0; i < s->rxq_num; i++) {
+        int idx = s->rxq_descr[i].intr_idx;
+        VMW_CFPRN("Verifying RX queue %d interrupt index (%d)", i, idx);
+        vmxnet3_validate_interrupt_idx(s->msix_used, idx);
+    }
+}
+
static void vmxnet3_activate_device(VMXNET3State *s)
{
   int i;
@@ -1447,6 +1478,8 @@ static void vmxnet3_activate_device(VMXNET3State *s)
              sizeof(s->rxq_descr[i].rxq_stats));
   }

+    vmxnet3_validate_interrupts(s);
+
   /* Make sure everything is in place before device activation */
   smp_wmb();

@@ -2005,7 +2038,6 @@ vmxnet3_cleanup_msix(VMXNET3State *s)
   }
}

-#define VMXNET3_MSI_NUM_VECTORS   (1)
#define VMXNET3_MSI_OFFSET        (0x50)
#define VMXNET3_USE_64BIT         (true)
#define VMXNET3_PER_VECTOR_MASK   (false)
@@ -2016,7 +2048,7 @@ vmxnet3_init_msi(VMXNET3State *s)
   PCIDevice *d = PCI_DEVICE(s);
   int res;

-    res = msi_init(d, VMXNET3_MSI_OFFSET, VMXNET3_MSI_NUM_VECTORS,
+    res = msi_init(d, VMXNET3_MSI_OFFSET, VMXNET3_MAX_NMSIX_INTRS,
                  VMXNET3_USE_64BIT, VMXNET3_PER_VECTOR_MASK);
   if (0 > res) {
       VMW_WRPRN("Failed to initialize MSI, error %d", res);
-- 
MST

--
Dr. David Alan Gilbert / address@hidden / Manchester, UK

--
Dr. David Alan Gilbert / address@hidden / Manchester, UK


reply via email to

[Prev in Thread] Current Thread [Next in Thread]