Bug 1012804

Summary:	IRQs of passthru MSI nic device are not correctly distributed on REHL5.10 (32 bits) guest
Product:	Red Hat Enterprise Linux 7	Reporter:	huiqingding <huding>
Component:	qemu-kvm	Assignee:	Alex Williamson <alex.williamson>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Virtualization Bugs <virt-bugs>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	7.0	CC:	acathrow, chayang, hhuang, juzhang, michen, mrezanin, virt-maint, xfu
Target Milestone:	rc
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:	qemu-kvm-1.5.3-17.el7	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2014-06-13 10:55:15 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description huiqingding 2013-09-27 07:44:43 UTC

Description of problem:
Passthru a nic card, which only supports MSI, ping external box, modify smp_affinity and found IRQ only distributed to CPU0.

Version-Release number of selected component (if applicable):
The host kernel if kernel-3.10.0-23.el7.x86_64
The version of qemu-kvm is qemu-kvm-rhev-1.5.3-6.el7.x86_64
The guest kernel is kernel-2.6.18-371.el5PAE

How reproducible:
100%

Steps to Reproduce:
1. Boot a guest and passthru a 82579LM Gigabit nic card which only supports MSI.
/usr/libexec/qemu-kvm -M pc -cpu SandyBridge -enable-kvm -m 2048 -smp 4,sockets=2,cores=2,threads=1 -name rhel5.10-32 -uuid 6afa5f93-2d4f-420f-81c6-e5fdddbd1c83 -drive file=/home/RHEL-Server-5-10.qcow2,if=none,id=drive-virtio-disk0,format=qcow2,serial=40c061dd-5d60-4fc5-865f-55db700407f0,cache=none,werror=stop,rerror=stop,aio=threads -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x5,drive=drive-virtio-disk0,id=virtio-disk0 -device pci-assign,host=00:19.0 -vnc :1  -monitor stdio

2. Stop irqbalance on guest
service irqbalance stop

3. On guest, flood ping remote box
#ping -f 10.66.5.4

4. On remote box, flood ping guest
#ping -f 10.66.110.242

5. Set smp_affinity to be 1
#echo 1 > /proc/irq/XX/smp_affinity
Check the interrupts number of each vcpu.
#cat /proc/interrupts

6. Set smp_affinity to be 2
#echo 2 > /proc/irq/XX/smp_affinity
Check the interrupts number of each vcpu.
#cat /proc/interrupts

7. Set smp_affinity to be 3
#echo 4 > /proc/irq/XX/smp_affinity
Check the interrupts number of each vcpu.
#cat /proc/interrupts


Actual results:
Step6 and Step7, only the IRQs are only distributed to CPU0 

Expected results:
Step6, the IRQs should be distributed to CPU1.
Step7, the IRQs should be distributed to CPU2.

Additional info:

Comment 2 huiqingding 2013-09-27 07:57:41 UTC

I found Bug 919761 is similar with this problem. So set the component of this bug to be "qemu-kvm". If the component is not correct, please fix me. thanks.

Comment 3 Alex Williamson 2013-12-19 21:54:35 UTC

This appears to be a duplicate of bug 1025477, the only difference is the RHEL5 guest.  Moving to ON_QA for testing.  Note that pci-assign is not supported on RHEL7, vfio-pci should be used for verification.

Comment 5 Chao Yang 2014-01-21 09:07:58 UTC

Verified as passed in a fresh installed rhel5.10 i686 guest with qemu-kvm-1.5.3-38.el7.x86_64.

Actual Result:

# cat /proc/interrupts 
           CPU0       CPU1       CPU2       CPU3       
  0:     114865      12074      12061      12060    IO-APIC-edge  timer
  1:        118         16          2         14    IO-APIC-edge  i8042
  6:          0          0          3          1    IO-APIC-edge  floppy
  8:          0          0          0          1    IO-APIC-edge  rtc
  9:          0          0          0          0   IO-APIC-level  acpi
 12:        739         26         29         29    IO-APIC-edge  i8042
 15:        618         54        418         62    IO-APIC-edge  ide1
177:          0          0          0          0       PCI-MSI-X  virtio0-config
185:       6399       1555          0          0       PCI-MSI-X  virtio0-requests
193:        215          0          0        143         PCI-MSI  eth0

# cat /proc/irq/193/smp_affinity 
00000001

# cat /proc/interrupts | grep -i 'cpu\|eth0'
           CPU0       CPU1       CPU2       CPU3       
193:      14859          0          0        143         PCI-MSI  eth0

# echo 2 > /proc/irq/193/smp_affinity 
# cat /proc/irq/193/smp_affinity
00000002
# cat /proc/interrupts | grep -i 'cpu\|eth0'
           CPU0       CPU1       CPU2       CPU3       
193:      69089      26035          0        143         PCI-MSI  eth0
# cat /proc/interrupts | grep -i 'cpu\|eth0'
           CPU0       CPU1       CPU2       CPU3       
193:      69089      30103          0        143         PCI-MSI  eth0


# echo 4 > /proc/irq/193/smp_affinity 
# cat /proc/irq/193/smp_affinity
00000004
# cat /proc/interrupts | grep -i 'cpu\|eth0'
           CPU0       CPU1       CPU2       CPU3       
193:     215344      67496      40597        143         PCI-MSI  eth0
# cat /proc/interrupts | grep -i 'cpu\|eth0'
           CPU0       CPU1       CPU2       CPU3       
193:     215344      67496      46630        143         PCI-MSI  eth0


So, this issue has been fixed.

Comment 7 Ludek Smid 2014-06-13 10:55:15 UTC

This request was resolved in Red Hat Enterprise Linux 7.0.

Contact your manager or support representative in case you have further questions about the request.