[vbox-dev] build issues of 5.0.22

Hans-Peter Jansen hpj at urpla.net
Mon Jun 27 14:35:26 UTC 2016


just in case, somebody stumbles upon some similar build issue:

/home/abuild/rpmbuild/BUILD/VirtualBox-5.0.22/kmk_sed "s/<NL>/\n/g" --output /home/abuild/rpmbuild/BUILD/VirtualBox-5.0.22/out/linux.amd64/release/obj/VBoxDD/vboxssdt-cpuhotplug.hex.pre1 /home/abuild/rpmbuild/BUILD/VirtualBox-5.0.22/out/linux.amd64/release/obj/VBoxDD/vboxssdt-cpuhotplug.hex.pre
/home/abuild/rpmbuild/BUILD/VirtualBox-5.0.22/kmk_sed: line 7: NL: file or directory not found

it might be fixed using the attached patch, which is MIT licensed just in 

I wonder, how this could creep into the distribution, but it could also be
related to the openSUSE build, that I based my build upon (but the applied 
patches doesn't appear to be guilty, AFAICS). 

openSUSE had a 5.0.18 in the stores, that is carrying a vbox-4.7.patch 
around. After merging with 5.0.22, just one hunk was left, and that's 
looking rather fishy:

Index: b/src/VBox/HostDrivers/VBoxNetAdp/linux/VBoxNetAdp-linux.c
--- a/src/VBox/HostDrivers/VBoxNetAdp/linux/VBoxNetAdp-linux.c
+++ b/src/VBox/HostDrivers/VBoxNetAdp/linux/VBoxNetAdp-linux.c
@@ -152,8 +152,10 @@ static int vboxNetAdpLinuxXmit(struct sk
     /* Update the stats. */
     pPriv->Stats.tx_bytes += pSkb->len;
     /* Update transmission time stamp. */
+    netif_trans_update(pNetDev);
     pNetDev->trans_start = jiffies;
     /* Nothing else to do, just free the sk_buff. */

Since this is not a transparent patch for kernel versions below 4.7, I 
refrained from applying this at all. It bears the question though, how this is 
done correctly, e.g. does 4.7 and above needs the netif_trans_update call, 
2.6.31 and below the jiffies tagging, and what to do with the kernel versions 
in between.

In case, you want to peak into the build, here it is:


The reason to build this package myself is due to suffering from an issue with 
5.0.18 and a bridged nic setup, that is flooding my logs with:

	VBoxNetFlt: Failed to allocate packet buffer, dropping the packet.
    Last message 'VBoxNetFlt: Failed t' repeated 196 times, suppressed by syslog-ng

That's produced from a w2k8R2 VM on a Linux 4.2.25 host. The VM seems to suffer
under high host load. Let's see, how 5.0.22 behaves in this respect. In case, this 
rings a bell for you, let me know.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: vbox-fix-build.diff
Type: text/x-patch
Size: 1031 bytes
Desc: not available
URL: <http://www.virtualbox.org/pipermail/vbox-dev/attachments/20160627/003fd6c9/attachment.bin>

More information about the vbox-dev mailing list