#18425 closed bug (fixed)
System hangs on 3rd icon after upgrade
Reported by: | Starcrasher | Owned by: | waddlesplash |
---|---|---|---|
Priority: | normal | Milestone: | R1/beta5 |
Component: | Drivers/USB/XHCI | Version: | R1/beta4 |
Keywords: | Cc: | ||
Blocked By: | Blocking: | ||
Platform: | All |
Description
The system is a nightly x64 (hrv57033) running in KVM/QEmu. It is running fine with previous state. Joined are 2 syslogs first boot and after forcing reinit.
usb error xhci -1: Host Controller halted
Will check again after today upgrade.
Attachments (3)
Change History (13)
by , 18 months ago
Attachment: | syslog.txt added |
---|
comment:1 by , 18 months ago
Component: | - General → Drivers/USB/XHCI |
---|---|
Owner: | changed from | to
Appears to be a problem with MSI-X interrupts.
I received a report from X512 about a problem like this, too. I have tested in QEMU without any issues. What is your exact QEMU/KVM configuration?
comment:2 by , 18 months ago
If it can help this is QEMU emulator version 5.2.0 (qemu-5.2.0-4.mga8) on 'regular' Mageia x64.
comment:3 by , 18 months ago
After updating to hrev57038, the VM still hangs with same messages about unhandled interrupt.
As a side note, I have also a nightly on an USB key which is booting without hanging. Perhaps is it a VM driver or configuration problem?
comment:4 by , 18 months ago
In fact, the message could be about real interrupts not MSI ones. I was looking at syslog and I think that I found out why interrupts above 15 are not handled. It seems that APICs are not working and PIC mode only manages up to 15 interrupts. Now the question is: Why APIC failed?
found io-apic with address 0xfec00000, global interrupt base 0, apic-id 0 mapping io-apic 0 at physical address 0xfec00000 io-apic 0 has range 0-23, 24 entries, version 0x00170011, apic-id 0 setting ACPI interrupt model to APIC failed to put ACPI into APIC interrupt model, ignoring IRQ routing preparation failed, not configuring io-apics setting ACPI interrupt model to PIC setting up x2apic for CPU 1: apic id 1, version 327700
comment:5 by , 18 months ago
Reverting XHCI MSI-X patch fix infinite reset problem for me. But even with MSI-X patch most USB devices are working fine.
comment:6 by , 18 months ago
The same problem for me, fall back to legacy PIC:
KERN: ACPI: Enabled 7 GPEs in block 00 to 1F KERN: found io-apic with address 0xfec00000, global interrupt base 0, apic-id 13 KERN: mapping io-apic 0 at physical address 0xfec00000 KERN: io-apic 0 has range 0-23, 24 entries, version 0x00178021, apic-id 13 KERN: found io-apic with address 0xfec01000, global interrupt base 24, apic-id 14 KERN: mapping io-apic 1 at physical address 0xfec01000 KERN: io-apic 1 has range 24-55, 32 entries, version 0x001f8021, apic-id 14 KERN: setting ACPI interrupt model to APIC KERN: IRQ routing preparation failed, not configuring io-apics KERN: setting ACPI interrupt model to PIC
comment:9 by , 18 months ago
Milestone: | Unscheduled → R1/beta5 |
---|---|
Resolution: | → fixed |
Status: | new → closed |
comment:10 by , 18 months ago
Thanks for testing! I guess there is some problem in our MSI-X implementation. We should investigate that later, probably.
Syslog from first boot