Bug 1874740

Created attachment 1715667 [details]
tcpdump trace

Something is fishy on the virt-viewer side.  Open the trace (comment #2) in wireshark, filter for spice packets.  There are two usbredir streams (right-click on a packet -> colorize conversation -> tcp -> pick some color will highlight the streams).  Initial handshake happens.  Then two PING messages, followed by two PONG responses.  Then silence on the usbredir streams, even though virt-viewer should send a "usb device connect" message at this point.  On the display stream one can see the boot display updates.

Reassigning to virt-viewer for investigation.

It looks like the client is not redirecting the device. Maybe is was locked by the kernel? Can you enable and post client logs?
Can you also post kernel logs?

I can reproduce bug issue with packages:

libvirt-6.6.0-5.module+el8.3.0+8092+f9e72d7e.x86_64
qemu-kvm-5.1.0-6.module+el8.3.0+8041+42ff16b8.x86_64
virt-viewer-9.0-4.el8.x86_64
spice-gtk3-0.38-3.el8.x86_64
libusbx-1.0.23-4.el8.x86_64

Steps:
1. Plug usb device to your host, then use # lsusb -vvv to find the information of your usb device.
# lsblk 
NAME          MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT
sda             8:0    0 238.5G  0 disk 
...
sdb             8:16   1  28.9G  0 disk 
├─sdb1          8:17   1   4.2G  0 part /run/media/juzhou/RHEL-7.8 Server.x86_6
└─sdb2          8:18   1   8.6M  0 part 

Note: sdb is an usb device.

# lsusb
Bus 002 Device 002: ID 0951:1666 Kingston Technology DataTraveler 100 G3/G4/SE9 G2
...


2. Prepare a health spice vm and start it.

#virsh dumpxml spice
...

    <graphics type='spice' port='5900' autoport='yes' listen='127.0.0.1'>
      <listen type='address' address='127.0.0.1'/>
      <image compression='off'/>
    </graphics>
...
    <video>
      <model type='qxl' ram='65536' vram='65536' vgamem='16384' heads='1' primary='yes'/>
      <alias name='video0'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x01' function='0x0'/>
    </video>
    <redirdev bus='usb' type='spicevmc'>
      <alias name='redir0'/>
      <address type='usb' bus='0' port='2'/>
    </redirdev>
    <redirdev bus='usb' type='spicevmc'>
      <alias name='redir1'/>
      <address type='usb' bus='0' port='3'/>
    </redirdev>

3. Run virt-viewer to connect to the vm with option: --spice-usbredir-redirect-on-connect=<filter-string>                     Filter selecting USB devices to redirect on connect

# virt-viewer  --spice-usbredir-redirect-on-connect="0x8,0x0951,0x1666,0x100,1" spice

Result:
3.1 Check on the host where "virt-viewer" command runs, we can't the usb device.
]# lsblk 
NAME          MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT
sda             8:0    0 238.5G  0 disk 
├─sda1          8:1    0     1G  0 part /boot
└─sda2          8:2    0 237.5G  0 part 
  ├─rhel-root 253:0    0    70G  0 lvm  /
  ├─rhel-swap 253:1    0   5.7G  0 lvm  [SWAP]
  └─rhel-home 253:2    0 161.9G  0 lvm  /home

3.2 Check on the vm
3.2.1 Click 'File'->'USB device selection', then 'Select USB devices for redirection' window pops up, we can find 'Kingston DataTraveler 3.0' is checked already, then click 'Close' button.

3.2.2 Run #lsblk in the vm, but we find there is no usb device in guest.

# lsblk 
NAME          MAJ:MIN RM  SIZE RO TYPE MOUNTPOINT
vda           252:0    0    8G  0 disk 
├─vda1        252:1    0    1G  0 part /boot
└─vda2        252:2    0    7G  0 part 
  ├─rhel-root 253:0    0  6.2G  0 lvm  /
  └─rhel-swap 253:1    0  820M  0 lvm  [SWAP]

4. Close the virt-viewer window, then finding the usb is released to the host again.
# lsblk 
NAME          MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT
sda             8:0    0 238.5G  0 disk 
├─sda1          8:1    0     1G  0 part /boot
└─sda2          8:2    0 237.5G  0 part 
  ├─rhel-root 253:0    0    70G  0 lvm  /
  ├─rhel-swap 253:1    0   5.7G  0 lvm  [SWAP]
  └─rhel-home 253:2    0 161.9G  0 lvm  /home
sdb             8:16   1  28.9G  0 disk 
├─sdb1          8:17   1   4.2G  0 part /run/media/root/RHEL-7.8 Server.x86_64
└─sdb2          8:18   1   8.6M  0 part 

I will attached the debug log, named usb.log.

Created attachment 1715695 [details]
usb.log

Created attachment 1715697 [details]
usb_replug.log

Replug the usb device into the host when running # virt-viewer  --spice-usbredir-redirect-on-connect="0x8,0x0951,0x1666,0x100,1" spice --debug --spice-debug |&  tee>usb_replug.log

Then we can see the usb device in the vm, thanks.

I can reproduce on Fedora 32 client/host/guest

spice-gtk3-0.38-1.fc32.x86_64
virt-viewer-9.0-1.fc32.x86_64

$ lsusb | grep -i disk
Bus 003 Device 006: ID 0781:5567 SanDisk Corp. Cruzer Blade

$ remote-viewer --spice-usbredir-redirect-on-connect="-1,0x0781,0x5567,-1,1" spice://localhost:5932

[guest] $ lsusb # does not show the SanDisk device

remote-viewer menu File -> Usb device selection shows the device
is marked as shared with the guest.


If on the USB device selection menu I unshare the device and
share it again then it works. Running lsusb on the guest shows it.

Created attachment 1716056 [details]
usb_work_well.log

Created attachment 1716067 [details]
failed__spice_gtk3.log

Created attachment 1716076 [details]
client_dmesg

Created attachment 1716077 [details]
client_messages

Created attachment 1716078 [details]
client_rv_spice_debug

(In reply to Frediano Ziglio from comment #4)
> It looks like the client is not redirecting the device. Maybe is was locked
> by the kernel?
> Can you enable and post client logs?

Do you mean to add '--debug' and '--spice-debug' options to remote-viewer?
If yes, please check client_rv_spice_debug.

> Can you also post kernel logs?
Client dmesg and /var/log/messages are attached.

Thanks

The patch causing the regression is

commit 3e20f17b90598e740c4e274b81d99f28187da800
Author: Yuri Benditovich <yuri.benditovich>
Date:   Thu Sep 19 16:11:21 2019 +0200

    usb-redir: extend USB backend to support emulated devices
    
    Redirection of emulated devices requires special approach, as
    usbredirhost can't be used for that (it works only with libusb
    devices). For emulated devices we create instance of usbredirparser
    that implements USB redirection protocol.  In order to work with the
    same set of protocol capabilities that usbredirhost sets up with
    remote side, the parser shall: - not send its own 'hello' to the
    server
    - initialize the same capabilities that usbredirhost
    - receive the same 'hello' response
    For that we analyze 'hello' sent by USB redir parser and extract set
    of capabilities from it and upon receive of 'hello' response we
    provide it also to the parser.  When libusb device is redirected via a
    channel, instance of usbredirhost serves it.  When emulated device is
    redirected via a channel, instance of usbredirparser does the job.
    
    Signed-off-by: Yuri Benditovich <yuri.benditovich>
    Signed-off-by: Frediano Ziglio <fziglio>

Proposed a fix at https://gitlab.freedesktop.org/spice/spice-gtk/-/merge_requests/75

I can reproduce with packages on rhel 8.4 client:
spice-gtk3-0.38-3.el8.x86_64
virt-viewer-9.0-4.el8.x86_64
usbredir-0.8.0-1.el8.x86_64


Bus 002 Device 003: ID 0951:1666 Kingston Technology DataTraveler 100 G3/G4/SE9 G2

remote-viewer --spice-usbredir-redirect-on-connect="-1,0x0951,0x1666,-1,1" spice://localhost:5900

After USB redirection USB device is no available in guest nor client.

Patch accepted upstream, https://gitlab.freedesktop.org/spice/spice-gtk/-/commit/afc5872bff1eb327dd299bacdc4eec5e26d37a10

*** Bug 1858655 has been marked as a duplicate of this bug. ***

*** Bug 1876714 has been marked as a duplicate of this bug. ***

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (spice-gtk bug fix and enhancement update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2021:1821