On Thu, May 19, 2016 at 11:45 PM, Laine Stump <laine@laine.org> wrote:
On 05/18/2016 05:30 PM, Shivaprasad G Bhat wrote:
This function was not used so far. Now, that we begin to use it, make sure to
check the address before actually releasing the address.

Signed-off-by: Shivaprasad G Bhat <sbhat@linux.vnet.ibm.com>
---
  src/conf/domain_addr.c |   18 +++++++++++++++++-
  1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/src/conf/domain_addr.c b/src/conf/domain_addr.c
index 35c7cd4..7ea9e4d 100644
--- a/src/conf/domain_addr.c
+++ b/src/conf/domain_addr.c
@@ -497,8 +497,24 @@ int
  virDomainPCIAddressReleaseAddr(virDomainPCIAddressSetPtr addrs,
                                 virPCIDeviceAddressPtr addr)
  {
+    /* permit any kind of connection type in validation, since we
+     * already had it, and are giving it back.
+     */
+    virDomainPCIConnectFlags flags = VIR_PCI_CONNECT_TYPES_MASK;
+    int ret = -1;
+    char *addrStr = NULL;
+
+    if (!(addrStr = virDomainPCIAddressAsString(addr)))
+        goto cleanup;
+
+    if (!virDomainPCIAddressValidate(addrs, addr, addrStr, flags, false))
+        goto cleanup;

I don't think this is necessary. virDomainPCIAddressValidate() is used to verify that it's okay to plug a particular device into a particular slot. But since the device is already in the slot, that's a moot point. The only thing that needs to be verified is that libvirt has information that this device is in the given slot (which I'm assuming is done at a higher level).


virDomainPCIAddressReleaseSlot valiadates the address before releasing today. I didnt want to skip the checks which used to happen before. 

  

+
      addrs->buses[addr->bus].slots[addr->slot] &= ~(1 << addr->function);
-    return 0;
+    ret = 0;
+ cleanup:
+    VIR_FREE(addrStr);
+    return ret;
  }
    int

--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list