Re: Network driver domain broken

[Jan Beulich <jbeulich@xxxxxxxx>]

On 07.03.2022 15:20, Andrea Stevanato wrote:
> On 3/7/2022 12:46 PM, Roger Pau Monné wrote:
>> On Mon, Mar 07, 2022 at 12:39:22PM +0100, Andrea Stevanato wrote:
>>> /local/domain/2 = ""   (n0,r2)
>>> /local/domain/2/vm = "/vm/f6dca20a-54bb-43af-9a62-67c55cb75708"   (n0,r2)
>>> /local/domain/2/name = "guest1"   (n0,r2)
>>> /local/domain/2/cpu = ""   (n0,r2)
>>> /local/domain/2/cpu/0 = ""   (n0,r2)
>>> /local/domain/2/cpu/0/availability = "online"   (n0,r2)
>>> /local/domain/2/cpu/1 = ""   (n0,r2)
>>> /local/domain/2/cpu/1/availability = "online"   (n0,r2)
>>> /local/domain/2/memory = ""   (n0,r2)
>>> /local/domain/2/memory/static-max = "1048576"   (n0,r2)
>>> /local/domain/2/memory/target = "1048577"   (n0,r2)
>>> /local/domain/2/memory/videoram = "-1"   (n0,r2)
>>> /local/domain/2/device = ""   (n0,r2)
>>> /local/domain/2/device/suspend = ""   (n0,r2)
>>> /local/domain/2/device/suspend/event-channel = ""   (n2)
>>> /local/domain/2/device/vif = ""   (n0,r2)
>>> /local/domain/2/device/vif/0 = ""   (n2,r1)
>>> /local/domain/2/device/vif/0/backend = "/local/domain/1/backend/vif/2/0"
>>> (n2,r1)
>>> /local/domain/2/device/vif/0/backend-id = "1"   (n2,r1)
>>> /local/domain/2/device/vif/0/state = "6"   (n2,r1)
>>> /local/domain/2/device/vif/0/handle = "0"   (n2,r1)
>>> /local/domain/2/device/vif/0/mac = "00:16:3e:07:df:91"   (n2,r1)
>>> /local/domain/2/device/vif/0/xdp-headroom = "0"   (n2,r1)
>>> /local/domain/2/control = ""   (n0,r2)
>>> /local/domain/2/control/shutdown = ""   (n2)
>>> /local/domain/2/control/feature-poweroff = "1"   (n2)
>>> /local/domain/2/control/feature-reboot = "1"   (n2)
>>> /local/domain/2/control/feature-suspend = ""   (n2)
>>> /local/domain/2/control/sysrq = ""   (n2)
>>> /local/domain/2/control/platform-feature-multiprocessor-suspend = "1"
>>> (n0,r2)
>>> /local/domain/2/control/platform-feature-xs_reset_watches = "1"   (n0,r2)
>>> /local/domain/2/data = ""   (n2)
>>> /local/domain/2/drivers = ""   (n2)
>>> /local/domain/2/feature = ""   (n2)
>>> /local/domain/2/attr = ""   (n2)
>>> /local/domain/2/error = ""   (n2)
>>> /local/domain/2/error/device = ""   (n2)
>>> /local/domain/2/error/device/vif = ""   (n2)
>>> /local/domain/2/error/device/vif/0 = ""   (n2)
>>> /local/domain/2/error/device/vif/0/error = "1 allocating event channel"
>>> (n2)
>>
>> That's the real error. Your guest netfront fails to allocate the event
>> channel. Do you get any messages in the guest dmesg after trying to
>> attach the network interface?
> 
> Just these two lines:
> 
> [  389.453390] vif vif-0: 1 allocating event channel
> [  389.804135] vif vif-0: 1 allocating event channel

Well, these are the error messages, from xenbus_alloc_evtchn().
What's a little odd is that the error code is positive, but that's
how -EPERM is logged. Is there perhaps a strange or broken XSM
policy in use? I ask because evtchn_alloc_unbound() itself
wouldn't return -EPERM afaics.

Jan