Re: xeniface: assertion failure on hibernation

[Paul Durrant <xadimgnik@xxxxxxxxx>]

On 15/07/2024 16:59, Rafał Wojdyła wrote:
> Hi,
>
> I was doing some tests with the default system shutdown that actually 
> hibernates the OS and I'm hitting the following assertion:
> xen|SystemPowerStateCallback: <- S0
> xeniface|__FdoQuerySystemPower: ====> (Hibernate:Hibernate)
> xeniface|FdoRequestQueryDevicePower: D3
> xeniface|__FdoQueryDevicePower: ====> (D3:Hibernate)
> xeniface|__FdoRequestQueryDevicePower: ASSERTION FAILED: 
> (((NTSTATUS)(IoStatus->Status)) >= 0)
> Assertion xeniface\src\xeniface\fdo.c(1944): 
> (((NTSTATUS)(IoStatus->Status)) >= 0)
> It's reproducible on Win 10 and 11 with the latest xenbus and xeniface 
> (debug build), no other pv drivers or tools. A bit more info below.
This is exactly the kind of reason why I feared changing the power IRP 
handling code. Anything to do with PnP or power handling is just so 
fragile. Any chance you could dig into this further?
> 0: kd> k
>   # Child-SP          RetAddr               Call Site
> 00 ffff930b`bf235750 fffff802`34f7acc4 
> xeniface!__FdoRequestQueryDevicePower+0x57 [xeniface\src\xeniface\fdo.c 
> @ 1944]
> 01 ffff930b`bf235790 fffff802`34e54975     nt!PopRequestCompletion+0x64
> 02 ffff930b`bf2357e0 fffff802`34e547b7     nt!IopfCompleteRequest+0x1a5
> 03 ffff930b`bf2358c0 fffff802`380a81be     nt!IofCompleteRequest+0x17
> 04 ffff930b`bf2358f0 fffff802`380a7dfd     xenbus!PdoDispatchPower+0x6e 
> [xenbus\src\xenbus\pdo.c @ 1934]
> 05 ffff930b`bf235940 fffff802`3808a2e8     xenbus!PdoDispatch+0x5d 
> [xenbus\src\xenbus\pdo.c @ 1970]
> 06 ffff930b`bf235980 fffff802`34f986cf     xenbus!DriverDispatch+0x228 
> [xenbus\src\xenbus\driver.c @ 342]
> 07 ffff930b`bf235a00 fffff802`34e2d40d     nt!IopPoHandleIrp+0x3b
> 08 ffff930b`bf235a30 fffff802`3b1c1029     nt!IofCallDriver+0x6d
> 09 ffff930b`bf235a70 fffff802`3b1c0eb7 
> xeniface!__FdoQueryDevicePowerDown+0x119 [xeniface\src\xeniface\fdo.c @ 
> 1882]
> 0a ffff930b`bf235ad0 fffff802`3b1bcd1c 
> xeniface!__FdoQueryDevicePower+0x1b7 [xeniface\src\xeniface\fdo.c @ 1915]
> 0b ffff930b`bf235b50 fffff802`3b1c2b69     xeniface!FdoDevicePower+0xfc 
> [xeniface\src\xeniface\fdo.c @ 2126]
> 0c ffff930b`bf235bc0 fffff802`34f48da5     xeniface!ThreadFunction+0x39 
> [xeniface\src\xeniface\thread.c @ 110]
> 0d ffff930b`bf235c10 fffff802`35006b58     nt!PspSystemThreadStartup+0x55
> 0e ffff930b`bf235c60 00000000`00000000     nt!KiStartSystemThread+0x28
> 0: kd> ?? (DWORD) IoStatus->Status
> DWORD 0xc00000bb
> 0: kd> !error 0xc00000bb
> Error code: (NTSTATUS) 0xc00000bb (3221225659) - The request is not 
> supported.
> 0: kd> .frame 0n4;dv /t /v
> 04 ffff930b`bf2358f0 fffff802`380a7dfd     xenbus!PdoDispatchPower+0x6e 
> [xenbus\src\xenbus\pdo.c @ 1934]
> ffff930b`bf235940 struct _XENBUS_PDO * Pdo = 0xffffc78e`45efe230
> ffff930b`bf235948 struct _IRP * Irp = 0xffffc78e`4c29ade0
> ffff930b`bf235910 unsigned char MinorFunction = 0x03 ''
> ffff930b`bf235920 struct _IO_STACK_LOCATION * StackLocation = 
> 0xffffc78e`4c29aef8 IRP_MJ_POWER / IRP_MN_WAIT_WAKE for Device for 
> "\Driver\xenbus"
> ffff930b`bf235914 long status = 0n-1073741637
> 0: kd> ?? (DWORD)Irp->IoStatus.Status
> DWORD 0xc00000bb
> 0: kd> .frame 0n9;dv /t /v
> 09 ffff930b`bf235a70 fffff802`3b1c0eb7 
> xeniface!__FdoQueryDevicePowerDown+0x119 [xeniface\src\xeniface\fdo.c @ 
> 1882]
> ffff930b`bf235ad0 struct _XENIFACE_FDO * Fdo = 0xffffc78e`465154a0
> ffff930b`bf235ad8 struct _IRP * Irp = 0xffffc78e`4c29ade0
> ffff930b`bf235ab0 struct _IO_STACK_LOCATION * StackLocation = 
> 0xffffc78e`4c29aef8 IRP_MJ_POWER / IRP_MN_WAIT_WAKE for Device for 
> "\Driver\xenbus"
> ffff930b`bf235a90 _DEVICE_POWER_STATE DeviceState = PowerDeviceD3 (0n4)
> ffff930b`bf235a9c long status = 0n0
> 0: kd> ?? (DWORD)Irp->IoStatus.Status
> DWORD 0xc00000bb
>
> It's probably harmless here, but I also found some probable bugs in the 
> gnttab code seemingly related to hibernation so thought I'd share this 
> first.
Ah. maybe you're already on it.

  Paul