RE: [PATCH v4 2/5] x86/hyper-v: Add synthetic debugger definitions

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



From: Jon Doron <arilou@xxxxxxxxx>  Sent: Monday, March 9, 2020 8:25 PM
> 
> On 09/03/2020, Michael Kelley wrote:
> >From: Jon Doron <arilou@xxxxxxxxx> Sent: Monday, March 9, 2020 11:20 AM
> >>
> >> Hyper-V synthetic debugger has two modes, one that uses MSRs and
> >> the other that use Hypercalls.
> >>
> >> Add all the required definitions to both types of synthetic debugger
> >> interface.
> >>
> >> Signed-off-by: Jon Doron <arilou@xxxxxxxxx>
> >
> >I got some additional details from the Hyper-V team about the Hyper-V
> >synthetic debugger functionality.  Starting with Windows 10 and Windows
> >Server 2016, KDNET is built-in to the Windows OS.  So when these and later
> >Windows versions are running as a guest on KVM, the synthetic debugger
> >support should not be needed.  It would only be needed for older Windows
> >versions (Windows 8.1, Windows Server 2012 R2, and earlier) that lack a
> >built-in KDNET.  Given the age of these Windows versions, I'm wondering
> >whether having KVM try to emulate Hyper-V's synthetic debugging support
> >is worthwhile.  While the synthetic debugger support is still present in
> >current Windows releases along with the built-in KDNET, it is a legacy feature
> >that is subject to being removed at any time in a future release.  Also, the
> >debug hypercalls are only offered to the parent partition, so they are
> >undocumented in the TLFS and the interfaces are subject to change at any
> >time.
> >
> >Given the situation, I would rather not have the MSR and CPUID leaf definitions
> >added to hyperv-tlfs.h.  But maybe I'm misunderstanding what you are trying
> >to accomplish.  Is there a bigger picture of what the goals are for adding the
> >synthetic debugger support?
> >
> >Michael
> >
> 
> Hi Michael, thank you for getting back on this, the goal of this is to
> allow fast kernel debugging mainly for the older OSs, without the
> synthetic kernel debugger you must opt-in to serial which is pretty
> "painful" when it comes to debugging.

OK, so you really are targeting the older Windows OS versions.

> 
> With that said it sounds like I need to look into setting up KDNet with
> KVM, might be trivial I have not tried it yet, but in general KDNet
> support only a small set of network adapters.
> (https://docs.microsoft.com/en-us/windows-hardware/drivers/debugger/supported-ethernet-nics-for-network-kernel-debugging-in-windows-8-1)
> 

I don't know my vendor IDs and device IDs very well.  Are there
commonly used network adapters that aren't on the list?  I'm asking
just out of curiosity ....

> As for the hypercalls, it's just something I ran into, I dont mind
> dropping off the hypercall interface (since the MSRs are way faster and
> simpler anyway, as I dont need to deal with UDP encapsulation.
> 
> In case you end up insisting I'll remove the MSR and CPUID leaf
> definitions what would be the workaround to implement the syndbg functionality?

I'm flexible, and trying to not be a pain-in-the-neck. :-)  What would
the KVM guys think about putting the definitions in a KVM specific
#include file, and clearly marking them as deprecated, mostly
undocumented, and used only to support debugging old Windows
versions?

Michael

> 
> Thanks,
> -- Jon.
> 
> >> ---
> >>  arch/x86/include/asm/hyperv-tlfs.h | 26 ++++++++++++++++++++++++++
> >>  1 file changed, 26 insertions(+)
> >>
> >> diff --git a/arch/x86/include/asm/hyperv-tlfs.h b/arch/x86/include/asm/hyperv-tlfs.h
> >> index 92abc1e42bfc..12596da95a53 100644
> >> --- a/arch/x86/include/asm/hyperv-tlfs.h
> >> +++ b/arch/x86/include/asm/hyperv-tlfs.h
> >> @@ -33,6 +33,9 @@
> >>  #define HYPERV_CPUID_ENLIGHTMENT_INFO		0x40000004
> >>  #define HYPERV_CPUID_IMPLEMENT_LIMITS		0x40000005
> >>  #define HYPERV_CPUID_NESTED_FEATURES		0x4000000A
> >> +#define HYPERV_CPUID_SYNDBG_VENDOR_AND_MAX_FUNCTIONS	0x40000080
> >> +#define HYPERV_CPUID_SYNDBG_INTERFACE			0x40000081
> >> +#define HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES	0x40000082
> >>
> >>  #define HYPERV_HYPERVISOR_PRESENT_BIT		0x80000000
> >>  #define HYPERV_CPUID_MIN			0x40000005
> >> @@ -131,6 +134,8 @@
> >>  #define HV_FEATURE_FREQUENCY_MSRS_AVAILABLE		BIT(8)
> >>  /* Crash MSR available */
> >>  #define HV_FEATURE_GUEST_CRASH_MSR_AVAILABLE		BIT(10)
> >> +/* Support for debug MSRs available */
> >> +#define HV_FEATURE_DEBUG_MSRS_AVAILABLE			BIT(11)
> >>  /* stimer Direct Mode is available */
> >>  #define HV_STIMER_DIRECT_MODE_AVAILABLE			BIT(19)
> >>
> >> @@ -194,6 +199,12 @@
> >>  #define HV_X64_NESTED_GUEST_MAPPING_FLUSH		BIT(18)
> >>  #define HV_X64_NESTED_MSR_BITMAP			BIT(19)
> >>
> >> +/*
> >> + * Hyper-V synthetic debugger platform capabilities
> >> + * These are HYPERV_CPUID_SYNDBG_PLATFORM_CAPABILITIES.EAX bits.
> >> + */
> >> +#define HV_X64_SYNDBG_CAP_ALLOW_KERNEL_DEBUGGING	BIT(1)
> >> +
> >>  /* Hyper-V specific model specific registers (MSRs) */
> >>
> >>  /* MSR used to identify the guest OS. */
> >> @@ -267,6 +278,17 @@
> >>  /* Hyper-V guest idle MSR */
> >>  #define HV_X64_MSR_GUEST_IDLE			0x400000F0
> >>
> >> +/* Hyper-V Synthetic debug options MSR */
> >> +#define HV_X64_MSR_SYNDBG_CONTROL		0x400000F1
> >> +#define HV_X64_MSR_SYNDBG_STATUS		0x400000F2
> >> +#define HV_X64_MSR_SYNDBG_SEND_BUFFER		0x400000F3
> >> +#define HV_X64_MSR_SYNDBG_RECV_BUFFER		0x400000F4
> >> +#define HV_X64_MSR_SYNDBG_PENDING_BUFFER	0x400000F5
> >> +#define HV_X64_MSR_SYNDBG_OPTIONS		0x400000FF
> >> +
> >> +/* Hyper-V HV_X64_MSR_SYNDBG_OPTIONS bits */
> >> +#define HV_X64_SYNDBG_OPTION_USE_HCALLS		BIT(2)
> >> +
> >>  /* Hyper-V guest crash notification MSR's */
> >>  #define HV_X64_MSR_CRASH_P0			0x40000100
> >>  #define HV_X64_MSR_CRASH_P1			0x40000101
> >> @@ -376,6 +398,9 @@ struct hv_tsc_emulation_status {
> >>  #define HVCALL_SEND_IPI_EX			0x0015
> >>  #define HVCALL_POST_MESSAGE			0x005c
> >>  #define HVCALL_SIGNAL_EVENT			0x005d
> >> +#define HVCALL_POST_DEBUG_DATA			0x0069
> >> +#define HVCALL_RETRIEVE_DEBUG_DATA		0x006a
> >> +#define HVCALL_RESET_DEBUG_SESSION		0x006b
> >>  #define HVCALL_FLUSH_GUEST_PHYSICAL_ADDRESS_SPACE 0x00af
> >>  #define HVCALL_FLUSH_GUEST_PHYSICAL_ADDRESS_LIST 0x00b0
> >>
> >> @@ -419,6 +444,7 @@ enum HV_GENERIC_SET_FORMAT {
> >>  #define HV_STATUS_INVALID_HYPERCALL_INPUT	3
> >>  #define HV_STATUS_INVALID_ALIGNMENT		4
> >>  #define HV_STATUS_INVALID_PARAMETER		5
> >> +#define HV_STATUS_OPERATION_DENIED		8
> >>  #define HV_STATUS_INSUFFICIENT_MEMORY		11
> >>  #define HV_STATUS_INVALID_PORT_ID		17
> >>  #define HV_STATUS_INVALID_CONNECTION_ID		18
> >> --
> >> 2.24.1
> >




[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux