[Bug 220181] New: Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[Bug 220181] New: Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

            Bug ID: 220181
           Summary: Users across distributions see “config failed, hub
                    doesn’t have any ports! (err -19)” from xhci_hcd at
                    boot.
           Product: Drivers
           Version: 2.5
          Hardware: AMD
                OS: Linux
            Status: NEW
          Severity: low
          Priority: P3
         Component: USB
          Assignee: drivers_usb@kernel-bugs.kernel.org
          Reporter: 6svcyk03@rokejulianlockhart.addy.io
        Regression: No

At boot, I see the undermentioned in red in `dmesg`:

> ~~~log
> config failed, hub doesn't have any ports! (err -19)
> ~~~

I've one entry for each boot in `journalctl`.

Using the first section of the error:

> ~~~YAML
> hub 12-0:1.0
> ~~~

...I believe that I've ascertained what the cause is. Using `lsusb -t`, I find:

> ~~~YAML
> /:  Bus 012.Port 001: Dev 001, Class=root_hub, Driver=xhci_hcd/0p, 5000M
> ~~~

`Driver=xhci_hcd/0p` means it has 0 ports, which is invalid, hence the error.

I tried to `readlink /sys/class/usb_host/usb12` it to ascertain what the cause
is, but that fails, because the device unsuccessfully enumerated. Consequently,
I used `readlink /sys/bus/usb/devices/usb12` to verify the device ID:

> ~~~log
> ../../../devices/pci0000:00/0000:00:08.3/0000:5a:00.0/usb12
> ~~~

When known, I used `5a:00.0` to locate the cause with `lspci -s 5a:00.0`:

> ~~~YAML
> 5a:00.0 USB controller: Advanced Micro Devices, Inc. [AMD] Raphael/Granite Ridge USB 2.0 xHCI
> ~~~

It's the virtual "Advanced Micro Devices, Inc. [AMD] Raphael/Granite Ridge USB
2.0 xHCI" USB controller, not exposing any ports despite the driver purportedly
requiring it. However, there obviously aren't ports on a virtual device.

At
https://discussion.fedoraproject.org/t/what-does-config-failed-hub-doesnt-have-any-ports-err-19-mean/153954,
I cite corroborations from Kali Linux, Ubuntu, Debian and Fedora. However, I've
also received a personal e-mail from an interested party who claims to
reproduce this on openSUSE.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

Michał Pecio (michal.pecio@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |michal.pecio@gmail.com

--- Comment #1 from Michał Pecio (michal.pecio@gmail.com) ---
I recall seeing a thread about it on ArchLinux forum. Posted logs indicated
that the xHCI chips in those cases only exposed USB 2.0 ports and hence the USB
3.x bus registered by xhci_hcd has no ports and USB core freaks out.

Such controllers are legal and supported, perhaps the driver could be patched
not to bother registering 3.x buses for them.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

Mathias Nyman (mathias.nyman@linux.intel.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mathias.nyman@linux.intel.c
                   |                            |om

--- Comment #2 from Mathias Nyman (mathias.nyman@linux.intel.com) ---
xhci "core" driver support hosts with only usb2 ports, (0 ports on usb3
roothub).

It's just missing some minor pieces for PCI xHC hosts in xhci-pci.c

Patform and MediaTek xHC have the support ready, as can be seen by the
"xhci->allow_single_roothub = 1" setting in xhci-plat.c and xhci-mtk.c

Or does this xHC host have 0 ports on usb2 roothub as well?

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

--- Comment #3 from Mr. Beedell, Roke Julian Lockhart (RJLB) (6svcyk03@rokejulianlockhart.addy.io) ---
(In reply to Mathias Nyman from comment #2)

> Or does this xHC host have 0 ports on usb2 roothub as well?

I dare say that I'm operating at the boundaries of my lay man's knowledge as it
is. How do you suggest that I ascertain this?

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

--- Comment #4 from Mathias Nyman (mathias.nyman@linux.intel.com) ---
(In reply to Mr. Beedell, Roke Julian Lockhart (RJLB) from comment #3)
> (In reply to Mathias Nyman from comment #2)
> 
> > Or does this xHC host have 0 ports on usb2 roothub as well?
> 
> I dare say that I'm operating at the boundaries of my lay man's knowledge as
> it is. How do you suggest that I ascertain this?

Probably best to take a closer at the xHC port information of this host

The reg-ext-protocol:00 and reg-ext-protocol:01 files in debugfs should show
all info about the ports the host provides:

Example, this machine has only one xHC at PCI address 0000:00:14.0:

cat /sys/kernel/debug/usb/xhci/0000:00:14.0/reg-ext-protocol*
EXTCAP_REVISION = 0x02000802
EXTCAP_NAME = 0x20425355
EXTCAP_PORTINFO = 0x30010e01
EXTCAP_PORTTYPE = 0x00000000
EXTCAP_MANTISSA1 = 0x000c0021
EXTCAP_MANTISSA2 = 0x05dc0012
EXTCAP_MANTISSA3 = 0x01e00023

EXTCAP_REVISION = 0x03000802
EXTCAP_NAME = 0x20425355
EXTCAP_PORTINFO = 0x10000610
EXTCAP_PORTTYPE = 0x00000000
EXTCAP_MANTISSA1 = 0x00050134


Above shows:
EXTCAP_REVISION tells if ports are USB3 or USB2,
EXTCAP_PORTNINFO = 0x****AA**, where AA tells us how many ports

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

Mr. Beedell, Roke Julian Lockhart (RJLB) (6svcyk03@rokejulianlockhart.addy.io) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Kernel Version|                            |https://koji.fedoraproject.
                   |                            |org/koji/rpminfo?rpmID=4290
                   |                            |0892

--- Comment #5 from Mr. Beedell, Roke Julian Lockhart (RJLB) (6svcyk03@rokejulianlockhart.addy.io) ---
"`sudo ls /sys/kernel/debug/usb/xhci/`" returns "`0000:5a:00.0`":

```
0000:08:00.0  0000:54:00.0  0000:56:00.0  0000:59:00.3  0000:59:00.4 
0000:5a:00.0
```

...so I tried the problematic port (and a random alternative):

```
#!/usr/bin/env sh
sudo cat /sys/kernel/debug/usb/xhci/0000:08:00.0/reg-ext-protocol:00
sudo cat /sys/kernel/debug/usb/xhci/0000:08:00.0/reg-ext-protocol:01
```

For the problematic port, solely the first command returns:

```
EXTCAP_REVISION = 0x02000402
EXTCAP_NAME = 0x20425355
EXTCAP_PORTINFO = 0x00180101
EXTCAP_PORTTYPE = 0x00000000
```

For the second command, the file supposedly doesn't exist:

> cat: '/sys/kernel/debug/usb/xhci/0000:5a:00.0/reg-ext-protocol:01': No such file or directory

This somewhat perplexes me, considering that the random alternative at "`sudo
cat /sys/kernel/debug/usb/xhci/0000:08:00.0/reg-ext-protocol:01`" quite happily
works:

```
EXTCAP_REVISION = 0x03101402
EXTCAP_NAME = 0x20425355
EXTCAP_PORTINFO = 0x80000203
EXTCAP_PORTTYPE = 0x00000000
EXTCAP_MANTISSA1 = 0x00050134
EXTCAP_MANTISSA2 = 0x000a4135
EXTCAP_MANTISSA3 = 0x04e00126
EXTCAP_MANTISSA4 = 0x09c00127
EXTCAP_MANTISSA5 = 0x13800128
EXTCAP_MANTISSA6 = 0x05b10129
```

At least, this appears to demonstrate that one port exists, although apologies
if I've misunderstood. For reference, I'm using
`kernel-6.14.9-300.fc42.x86_64`.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

--- Comment #6 from Mathias Nyman (mathias.nyman@linux.intel.com) ---
This means the xHC controller only has one entry for "supported protocol
capability", which lists just one USB2 port.

There is no entry at all for USB3 ports

The xhci driver needs some minor changes in xhci-pci.c code to make this work
properly.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

niklas.neronin@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |niklas.neronin@gmail.com

--- Comment #7 from niklas.neronin@gmail.com ---
Created attachment 308204
  --> https://bugzilla.kernel.org/attachment.cgi?id=308204&action=edit
Proposed patch to add support for zero USB3 port systems

I don't have a system with zero USB3 ports to fully test this patch.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

George N. White III (gnw3@acm.org) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gnw3@acm.org

--- Comment #8 from George N. White III (gnw3@acm.org) ---
A Fedora user with a system with USB4 ports is seeing  
`fedora kernel: hub 6-0:1.0: config failed, hub doesn't have any ports! (err
-19)`.
I wonder if the "the zero USB3 ports" issue is at fault.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

Nick Nielsen (nick.kainielsen@free.fr) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nick.kainielsen@free.fr

--- Comment #9 from Nick Nielsen (nick.kainielsen@free.fr) ---
I'm in Tumbleweed - or I would be 
-
-
My error message reads
"[1.361031][T442] hub 6-0:1.3 : config failed hub doesn't have any ports (error
-19)" - there's a difference of numbering of the hub.   
-
-
My set up : an AMD Ryzen 5 8600G processor mounted on an ASRock A620M Pro RS
mother board which uses the AMD A620 chipset.
-
-
Willing to help/test bearing in mind my skills are very limited... which is why
I can't format this text

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

--- Comment #10 from Mr. Beedell, Roke Julian Lockhart (RJLB) (6svcyk03@rokejulianlockhart.addy.io) ---
> I can't format this text

It's merely CommonMark. You can read the standard online, at
https://github.com/commonmark/commonmark-spec/blob/90c06837975098002cc89d3da1be30529919a8a2/spec.txt#L1934-L2356.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

--- Comment #11 from Nick Nielsen (nick.kainielsen@free.fr) ---
The patch seems to work.

A member of the openSuse community prepared a KMP (kernel module package) for
me containing @Niklas Neronin 's patch and I no longer get the error message.

Also, looking at journalctl, the USB ports are enumerated and described as USB3
or 2 where appropriate.

It's not the end of my worries - boot hangs again afterwards, whether this is
kernel related is yet to ascertain.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

--- Comment #12 from niklas.neronin@gmail.com ---
(In reply to Nick Nielsen from comment #11)
> The patch seems to work.
> 
> A member of the openSuse community prepared a KMP (kernel module package)
> for me containing @Niklas Neronin 's patch and I no longer get the error
> message.
> 
> Also, looking at journalctl, the USB ports are enumerated and described as
> USB3 or 2 where appropriate.
> 
> It's not the end of my worries - boot hangs again afterwards, whether this
> is kernel related is yet to ascertain.

Thank you for testing my patch.

If you suspect that the boot hanging issue is related to my patch or the xhci
driver, please send over any logs you have, and I'll try to fix it.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

Artem S. Tashkinov (aros@gmx.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |CODE_FIX

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

grm1@mailbox.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |grm1@mailbox.org

--- Comment #13 from grm1@mailbox.org ---
I found this bug report here through
https://bbs.archlinux.org/viewtopic.php?id=302115&p=2 after lookup my journald
error "archlinux kernel: hub 6-0:1.0: config failed, hub doesn't have any
ports! (err -19)"

I tested the patch
0001-usb-xhci-pci-add-support-for-hosts-with-zero-usb3-po.patch 2025-06-04
13:14 UTC, Niklas Neronin.

Running archlinux with custom linux 6.16.2-arch1 PKGBUILD to just add the
patch.

The message disappear and i don't notice any crash.

System is "ASRock B650M PG Lightning/B650M PG Lightning, BIOS 3.30 06/16/2025."

Thank you Niklas Neronin.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

Dana Goyette (danagoyette@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |danagoyette@gmail.com

--- Comment #14 from Dana Goyette (danagoyette@gmail.com) ---
I filed a bug a while ago on the Ubuntu bug tracker with information about the
device from a few utilities on Windows, on an X870E Taichi.

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2097466

I should test out the patch sometime this week.  What kernel version is the
patch for?

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

--- Comment #15 from Niklas Neronin (niklas.neronin@gmail.com) ---
I believe it was built for kernel version 6.15.
But I tested and it cleanly applies to 6.16.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.

[Bug 220181] Users across distributions see “config failed, hub doesn’t have any ports! (err -19)” from xhci_hcd at boot.

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=220181

Anthony (manya3084@gmail.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |manya3084@gmail.com

--- Comment #16 from Anthony (manya3084@gmail.com) ---
I applied the patch and issues i was having with my internal USB 2 header on my
MSI MEG x870e seem to be cleared up. 

Previously i couldn't get access to my internal USB 2 header, which has a LED
segment display connected. 

The patch applied cleanly to 6.17RC2.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.