Opened 4 years ago

Closed 3 years ago

#16041 closed bug (no change required)

[XHCI] transfer error: Length invalid

Reported by: Paradoxon Owned by: waddlesplash
Priority: normal Milestone: Unscheduled
Component: Drivers/USB/XHCI Version: R1/Development
Keywords: Cc: tqh
Blocked By: Blocking:
Platform: All

Description (last modified by Paradoxon)

Trying to boot (legacy) on Schenker 14 slim causes KDL usb error XHCI. The system is on a usb 2.0 Stick on a usb 3.0 Port. UEFI disabled. Enabling nealry all Safemodes bring me to desktop but horrible slow (10 minutes or more to boot opening apps last for ever :-) ).

System: I5-10210U UHDC Graphics

Attachments (9)

KDL.jpg​ (2.1 MB ) - added by Paradoxon 4 years ago.
listdev.txt​ (3.0 KB ) - added by Paradoxon 4 years ago.
Listdev avter safeboot
sysinfo.txt​ (1.4 KB ) - added by Paradoxon 4 years ago.
sysinfo after safebook
syslog​ (93.4 KB ) - added by Paradoxon 4 years ago.
Syslog after safeboot
syslog_ACPI_ENABLED​ (121.5 KB ) - added by Paradoxon 4 years ago.
All Safemodes enabled except ACPI
syslog_only_ACIP_Disabled​ (260.9 KB ) - added by Paradoxon 4 years ago.
Only safemode ACPCI_Disabled
kdl_ints.JPG​ (1.2 MB ) - added by Paradoxon 4 years ago.
ints after kdl during boot up
kdl_sc.png​ (3.1 MB ) - added by Paradoxon 4 years ago.
sc after kdl on boot .. in case it helps
kdl_hrev54226.jpg​ (419.3 KB ) - added by Paradoxon 4 years ago.

Change History (26)

by Paradoxon, 4 years ago

Attachment: KDL.jpg​ added

comment:1 by Paradoxon, 4 years ago

Description: modified (diff)

by Paradoxon, 4 years ago

Attachment: listdev.txt​ added

Listdev avter safeboot

by Paradoxon, 4 years ago

Attachment: sysinfo.txt​ added

sysinfo after safebook

by Paradoxon, 4 years ago

Attachment: syslog​ added

Syslog after safeboot

comment:2 by waddlesplash, 4 years ago

Component: - General → Drivers/USB/XHCI
Owner: changed from nobody to waddlesplash
Platform: x86 → All
Summary: Haiku KDL on bootup → [XHCI] transfer error: Length invalid

comment:3 by waddlesplash, 4 years ago 

device Serial bus controller (USB controller, XHCI) [c|3|30]
	  vendor 8086: Intel Corporation
	  device 02ed: Unknown

This is apparently a Comet Lake device, I guess it is too new to be in our pci.ids: ​https://linux-hardware.org/index.php?id=pci:8086-02ed-1558-1404

Strangely, there are no XHCI errors in the syslog. Can you try enabling all safe mode options *except* "Disable ACPI" and see if that still gets you to the desktop; and if it does not, try enabling *only* "Disable ACPI"?

comment:4 by Paradoxon, 4 years ago

Enabling all safe mode options *except* "Disable ACPI" boot only on shutdown there is a crash... in deskbar but i guess its not realated.

by Paradoxon, 4 years ago

Attachment: syslog_ACPI_ENABLED​ added

All Safemodes enabled except ACPI

comment:5 by Paradoxon, 4 years ago

Also booting with all safemodes not checked and onl "Disable ACPI" works ... and its also way faster...

by Paradoxon, 4 years ago

Attachment: syslog_only_ACIP_Disabled​ added

Only safemode ACPCI_Disabled

comment:6 by tqh, 4 years ago

Sounds like interrupt storm as disabling ACPI will also map IRQ's differently. Going into kernel debugger and run the ints command would be interesting with ACPI enabled.

Unlikely but some machines have had 'creative' ways to handle Embedded Controller talking to hw, which means EC code runs non-stop. I would not think this is the case here though.

by Paradoxon, 4 years ago

Attachment: kdl_ints.JPG​ added

ints after kdl during boot up

by Paradoxon, 4 years ago

Attachment: kdl_sc.png​ added

sc after kdl on boot .. in case it helps

comment:7 by waddlesplash, 4 years ago

No unhandled interrupts in there, which is interesting...

Can you retest with hrev54214 (or later) and see if anything is different?

in reply to:  7 comment:8 by Paradoxon, 4 years ago

Replying to waddlesplash:

No unhandled interrupts in there, which is interesting...

Can you retest with hrev54214 (or later) and see if anything is different?

Got again a kdl - from my point of view there is no difference

by Paradoxon, 4 years ago

Attachment: kdl_hrev54226.jpg​ added

comment:9 by waddlesplash, 4 years ago

Cc: tqh added

Disabling ACPI makes everything work, though, correct?

CC'ing tqh: I'm not sure what the issue is here. It seems strange that XHCI would "work", but the controller would return errors, when ACPI was enabled but not when it was disabled.

comment:10 by tqh, 4 years ago

ACPI flag is most likely only involved in mapping IRQ's and such. Someone that knows the IRQ mapping and setup probably can understand this.

comment:11 by waddlesplash, 4 years ago

I do note that this is a "Comet Lake" CPU, released only last year; perhaps we need to update our ACPICA?

in reply to:  11 comment:12 by Paradoxon, 4 years ago

Replying to waddlesplash:

I do note that this is a "Comet Lake" CPU, released only last year; perhaps we need to update our ACPICA?

Actually the crazy thing is... All Safemodes enabled except ACPI boots to desktop and also Only safemode ACPCI_Disabled

I am really would help with this lowlevel stuff.. but i am not really into this :(

Are there Any more logs i can provide? ... or postponed this one after Beta ;-)

Last edited 4 years ago by Paradoxon (previous) (diff)

comment:13 by waddlesplash, 4 years ago

Can you get clean syslogs from both the boot with only ACPI disabled, and also from everything but ACPI disabled?

comment:14 by Paradoxon, 4 years ago

Already there: by Paradoxon, 5 days ago Attachment: syslog_ACPI_ENABLED​added All Safemodes enabled except ACPI

comment:5 by Paradoxon, 5 days ago Also booting with all safemodes not checked and onl "Disable ACPI" works ... and its also way faster...

by Paradoxon, 5 days ago Attachment: syslog_only_ACIP_Disabled​added Only safemode ACPCI_Disabled

comment:15 by waddlesplash, 4 years ago

Well, there aren't any errors in either, so I have no idea what's going on then :/

Some kind of race condition, I guess; or maybe ACPI is doing something unexpected, but either way, I have no idea.

comment:16 by Paradoxon, 3 years ago

With the new 55079 the error is gone when booting in legacy mode. This can be closed

comment:17 by waddlesplash, 3 years ago

Resolution: → no change required
Status: new → closed
Note: See TracTickets for help on using tickets.