Note: This bug is displayed in read-only format because
the product is no longer active in Red Hat Bugzilla.
RHEL Engineering is moving the tracking of its product development work on RHEL 6 through RHEL 9 to Red Hat Jira (issues.redhat.com). If you're a Red Hat customer, please continue to file support cases via the Red Hat customer portal. If you're not, please head to the "RHEL project" in Red Hat Jira and file new tickets here. Individual Bugzilla bugs in the statuses "NEW", "ASSIGNED", and "POST" are being migrated throughout September 2023. Bugs of Red Hat partners with an assigned Engineering Partner Manager (EPM) are migrated in late September as per pre-agreed dates. Bugs against components "kernel", "kernel-rt", and "kpatch" are only migrated if still in "NEW" or "ASSIGNED". If you cannot log in to RH Jira, please consult article #7032570. That failing, please send an e-mail to the RH Jira admins at rh-issues@redhat.com to troubleshoot your issue as a user management inquiry. The email creates a ServiceNow ticket with Red Hat. Individual Bugzilla bugs that are migrated will be moved to status "CLOSED", resolution "MIGRATED", and set with "MigratedToJIRA" in "Keywords". The link to the successor Jira issue will be found under "Links", have a little "two-footprint" icon next to it, and direct you to the "RHEL project" in Red Hat Jira (issue links are of type "https://issues.redhat.com/browse/RHEL-XXXX", where "X" is a digit). This same link will be available in a blue banner at the top of the page informing you that that bug has been migrated.
DescriptionYvugenfi@redhat.com
2012-09-24 09:39:20 UTC
Description of problem:
Current default offload setting on Windows include both IP checksum and TCP\UDP checksum offload. It is done in order to pass HCK tests on variety of OSes (this default is a requirement for HCK).
Such setting wasn't required on the previous test kit - WLK 1.6. Also in this case IP checksum is calculated by the driver anyway as virtio header cannot contain information need to calculate two checksums for one packet.
In some cases Windows will not only set checksum in IP header but also will not indicate packet size in the header. In those cases driver will fail to calculate the checksum and the performance will be badly effected.
Version-Release number of selected component (if applicable):
How reproducible:
Steps to Reproduce:
1.
2.
3.
Actual results:
Expected results:
Additional info:
Comment 1Yvugenfi@redhat.com
2012-09-24 09:44:11 UTC
This is a regression from previous versions.
Comment 2Yvugenfi@redhat.com
2012-09-27 16:21:15 UTC
http://msdn.microsoft.com/en-us/library/windows/hardware/ff568840(v=vs.85).aspx
Although the MSDN clearly states that the TCPIP stack fills the length in IP header, it seems the MSFT decided to ignore this statement (possible, for LSO V2) because the packet will be fragmented anyway, so they think the HW does not need this field and they expect us to use the length value in NET_BUFFER struct.
We are going to populate the length anyway.
Comment 3Yvugenfi@redhat.com
2012-09-30 12:02:47 UTC
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.
http://rhn.redhat.com/errata/RHBA-2013-0441.html
The drivers included in the virtio-win ISO don't appear to have been updated. Where can we get the current version of the driver?
/var/lib/libvirt/images/virtio-win-0.1-49.iso
57240 57240 0 100% /mnt/disc1
/var/lib/libvirt/images/virtio-win-1.6.4.iso
85418 85418 0 100% /mnt/disc2
51f50a8f1967ae50c279d7c4d18ba5fe disc1/win7/amd64/netkvm.sys
51f50a8f1967ae50c279d7c4d18ba5fe disc2/NetKVM/2k8R2/amd64/netkvm.sys
51f50a8f1967ae50c279d7c4d18ba5fe disc2/NetKVM/w7/amd64/netkvm.sys
I apologize, I was working on the impression that the former ISO was obtained through RHEL channels, but I may be mistaken.
Using driver NetKVM driver 61.64.104.4900 on 2k8R2, we see gigabit throughput on UDP (e.g. CIFS) but only about 3Mbps on TCP flows. I thought that might have been due to this bug. I'll continue to gather information and submit a new bug.