1
2
3
4
5
6
7
8
9
10
11
12
13
|
<think>
Okay, so the user is reporting a bug in QEMU related to failing qemu-iotests. The specific test that's failing is 040, along with others like 041, 127, 256, and 267. They're using a RISC-V64 target and cross-compiled QEMU based on the latest code.
Looking at the error message from the log, it says that there's no 'PCI' bus found for the device 'virtio-scsi-pci'. The user tried running the test without specifying a machine type, which defaults to 'spike', but 'spike' might not have a PCI bus. They then attempted using the 'virt' machine by setting QEMU_DEFAULT_MACHINE and adding -machine virt, but it didn't fix the issue. However, when they modified testenv.py to include ('riscv64', 'virt') in the machine map, all tests passed.
So, the problem seems to be that the default machine for RISC-V isn't suitable for these tests because it lacks the necessary PCI bus. The solution was to set the correct machine type (virt) which includes the required components.
Since the issue is related to the machine configuration and device setup rather than translation or assembly issues, it's not a mistranslation or assembly lowering problem. It falls under how the QEMU machine is configured, particularly regarding devices and hardware simulation.
</think>
The bug report indicates an issue with the QEMU machine type and device configuration for RISC-V64 targets. The default 'spike' machine doesn't support PCI buses needed by certain tests, but switching to 'virt' resolved this. This relates to how devices are handled in the hypervisor/machine setup.
**hypervisor**
|