blob: adbb05c80657df611aae47689a3d71951f37bca4 (
plain) (
blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
boot: 0.873
virtual: 0.856
performance: 0.853
device: 0.846
architecture: 0.783
graphic: 0.705
hypervisor: 0.696
semantic: 0.678
vnc: 0.662
PID: 0.619
ppc: 0.579
network: 0.568
permissions: 0.561
peripherals: 0.556
kernel: 0.530
risc-v: 0.528
TCG: 0.524
files: 0.485
register: 0.477
x86: 0.469
VMM: 0.465
arm: 0.448
debug: 0.448
mistranslation: 0.425
i386: 0.402
socket: 0.372
user-level: 0.322
assembly: 0.304
KVM: 0.204
Iothread is stuck at 100% CPU usage with virtio-scsi on QEMU 7.0.0
Description of problem:
Starting with QEMU 7.0.0, the iothread associated attached to a virtio-scsi controller is stuck at 100% CPU usage. Bisected to: https://gitlab.com/qemu-project/qemu/-/commit/826cc32423db2a99d184dbf4f507c737d7e7a4ae
- Works as expected without the iothread
- No issue with virtio-blk + iothread
- Same behavior regardless of io=threads/native/io_uring
- Same behavior with default vs increased queue count
- The issue is triggered when the guest OS initializes the virtio driver
Steps to reproduce:
1. Add virtio-scsi controller with iothread
2. Boot VM
3. Check per-thread CPU usage such as in htop
Additional information:
[fedora.log](/uploads/776fbf8e5b823d0ab326946684ef9022/fedora.log)
[fedora.xml](/uploads/54879e5adfb227ddef79d382e86fc608/fedora.xml)
|