KVM: SEV: Enable AMD SEV-ES DebugSwap
v2 cover KVM: SEV: Enable AMD SEV-ES DebugSwap
This is to use another AMD SEV-ES hardware assisted register swap, more detail in 2/3. The patches are fairly independend but required in this order.
This is based on sha1 0d1409e4ff08 and requires something like this for X86_FEATURE_NO_NESTED_DATA_BP:
https://lkml.org/lkml/2022/11/29/1229
This patchset is pushed out at
https://github.com/aik/linux/commits/debugswap
Please comment. Thanks.
btw the enormous cc: list came from get_maintainer.pl, keep it like this or tell the script something to reduce the list?
Alexey Kardashevskiy (3):
x86/amd: Cache values in percpu variables
KVM: SEV: Enable data breakpoints in SEV-ES
x86/sev: Do not handle #VC for DR7 read/write
arch/x86/include/asm/debugreg.h | 9 +++- arch/x86/include/asm/msr-index.h | 1 + arch/x86/include/asm/svm.h | 1 + arch/x86/kvm/svm/svm.h | 16 +++++-- tools/arch/x86/include/asm/msr-index.h | 1 + arch/x86/kernel/cpu/amd.c | 45 ++++++++++++++------ arch/x86/kernel/sev.c | 6 +++ arch/x86/kvm/svm/sev.c | 29 +++++++++++++ arch/x86/kvm/svm/svm.c | 3 +- 9 files changed, 90 insertions(+), 21 deletions(-)
Signature · 1 line
2.38.1
PATCH v2 1/3 x86/amd: Cache values in percpu variables
Reading DR[0-3]_ADDR_MASK MSRs takes about 250 cycles which is going to be noticeable with the AMD KVM SEV-ES DebugSwap feature enabled. KVM is going to store host's DR[0-3] and DR[0-3]_ADDR_MASK before switching to a guest; the hardware is going to swap these on VMRUN and VMEXIT.
Store MSR values passsed to set_dr_addr_mask() in percpu values (when changed) and return them via new amd_get_dr_addr_mask(). The gain here is about 10x.
As amd_set_dr_addr_mask() uses the array too, change the @dr type to unsigned to avoid checking for <0.
While at it, replace deprecated boot_cpu_has() with cpu_feature_enabled() in set_dr_addr_mask().
---
Changes:
v2:
* reworked to use arrays
* set() skips wrmsr() when the mask is not changed
* added stub for get_dr_addr_mask()
* changed @dr type to unsigned
* s/boot_cpu_has/cpu_feature_enabled/
* added amd_ prefix
--- arch/x86/include/asm/debugreg.h | 9 +++- arch/x86/kernel/cpu/amd.c | 45 ++++++++++++++------ 2 files changed, 38 insertions(+), 16 deletions(-)
arch/x86/include/asm/debugreg.h +7 -2
@@ -126,9 +126,14 @@ static __always_inline void local_db_restore(unsigned long dr7) }#ifdef CONFIG_CPU_SUP_AMD -extern void set_dr_addr_mask (unsigned long mask,int dr); +extern void set_dr_addr_mask (unsigned long mask,unsigned int dr); +extern unsigned long amd_get_dr_addr_mask (unsigned int dr);#else -static inline void set_dr_addr_mask (unsigned long mask,int dr) { } +static inline void set_dr_addr_mask (unsigned long mask,unsigned int dr) { } +static inline unsigned long amd_get_dr_addr_mask (unsigned int dr) +{ +return 0 ; +}#endif #endif /* _ASM_X86_DEBUGREG_H */
arch/x86/kernel/cpu/amd.c +31 -14
55 lines not on this page · raw message
Signature · 1 line
2.38.1
Borislav Petkov
maintainerRe: [PATCH kernel v2 1/3] x86/amd: Cache values in percpu variables
On Fri, Dec 09, 2022 at 03:38:02PM +1100, Alexey Kardashevskiy wrote:
Make that Subject:
"x86/amd: Cache debug register values in percpu variables"
to make it less generic and more specific as to what you're doing.
Quoted 7 lines
Reading DR[0-3]_ADDR_MASK MSRs takes about 250 cycles which is going to be noticeable with the AMD KVM SEV-ES DebugSwap feature enabled. KVM is going to store host's DR[0-3] and DR[0-3]_ADDR_MASK before switching to a guest; the hardware is going to swap these on VMRUN and VMEXIT.
Store MSR values passsed to
set_dr_addr_mask()in percpu values
s/values/variables/
Unknown word [passsed] in commit message.
Use a spellchecker pls.
Quoted 2 lines
(when changed) and return them via new
amd_get_dr_addr_mask(). The gain here is about 10x.
10x when reading percpu vars vs MSR reads?
Oh well.
Quoted 2 lines
As
amd_set_dr_addr_mask()uses the array too, change the @dr type to unsigned to avoid checking for <0.
I feel ya but that function will warn once, return 0 when the @dr number is outta bounds and that 0 will still get used as an address mask.
I think you really wanna return negative on error and the caller should not continue in that case.
arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1161
Quoted 10 lines
arch/x86/kernel/cpu/amd.c +1
-1@@ -1158,24 +1158,41 @@ static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum)return false; } -void set_dr_addr_mask (unsigned long mask,int dr) +DEFINE_PER_CPU_READ_MOSTLY(unsigned long[4], amd_dr_addr_mask);
static
arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1165
Quoted 4 lines
arch/x86/kernel/cpu/amd.c +4
-0+ +static unsigned int amd_msr_dr_addr_masks[] = { +MSR_F16H_DR0_ADDR_MASK , +MSR_F16H_DR1_ADDR_MASK -1 +1 ,
- 1 + 1 ?
Why?
Because of the DR0 and then DR1 being in a different MSR range?
Who cares, make it simple:
MSR_F16H_DR0_ADDR_MASK, MSR_F16H_DR1_ADDR_MASK, MSR_F16H_DR1_ADDR_MASK + 1, MSR_F16H_DR1_ADDR_MASK + 2
and add a comment if you want to denote the non-contiguous range but meh.
Quoted 16 lines
arch/x86/kernel/cpu/amd.c +6
-7+MSR_F16H_DR1_ADDR_MASK -1 +2 , + MSR_F16H_DR1_ADDR_MASK - 1 + 3 +}; + +void set_dr_addr_mask (unsigned long mask,unsigned int dr) { -if (!boot_cpu_has (X86_FEATURE_BPEXT )) +if (!cpu_feature_enabled (X86_FEATURE_BPEXT ))return ; -switch (dr) { -case 0 : -wrmsr (MSR_F16H_DR0_ADDR_MASK , mask,0 ); -break ; -case 1 : -case 2 :
arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1178
Quoted 10 lines
arch/x86/kernel/cpu/amd.c +4
-6-case 3 : -wrmsr (MSR_F16H_DR1_ADDR_MASK -1 + dr, mask,0 ); -break ; -default : -break ; - } +if (WARN_ON_ONCE (dr >=ARRAY_SIZE (amd_msr_dr_addr_masks))) +return ; + +if (per_cpu (amd_dr_addr_mask,smp_processor_id ())[dr] == mask)
Do that at function entry:
int cpu = smp_processor_id();
and use cpu here.
arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1183
Quoted 5 lines
arch/x86/kernel/cpu/amd.c +5
-0+return ; + +wrmsr (amd_msr_dr_addr_masks[dr], mask,0 ); +per_cpu (amd_dr_addr_mask,smp_processor_id ())[dr] = mask; +}
Thx.
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquetteRe: [PATCH kernel v2 1/3] x86/amd: Cache values in percpu variables
Quoted 27 lines · Borislav Petkov
On Fri, Dec 09, 2022 at 03:38:02PM +1100, Alexey Kardashevskiy wrote:
Make that Subject:
"x86/amd: Cache debug register values in percpu variables"
to make it less generic and more specific as to what you're doing.
Quoted 7 lines
Reading DR[0-3]_ADDR_MASK MSRs takes about 250 cycles which is going to be noticeable with the AMD KVM SEV-ES DebugSwap feature enabled. KVM is going to store host's DR[0-3] and DR[0-3]_ADDR_MASK before switching to a guest; the hardware is going to swap these on VMRUN and VMEXIT.
Store MSR values passsed to
set_dr_addr_mask()in percpu valuess/values/variables/
Unknown word [passsed] in commit message.
Use a spellchecker pls.
Quoted 2 lines
(when changed) and return them via new
amd_get_dr_addr_mask(). The gain here is about 10x.10x when reading percpu vars vs MSR reads?
Quoted 7 lines
Oh well.
Quoted 2 lines
As
amd_set_dr_addr_mask()uses the array too, change the @dr type to unsigned to avoid checking for <0.I feel ya but that function will warn once, return 0 when the @dr number is outta bounds and that 0 will still get used as an address mask.
"that function" is set_dr_addr_mask() (btw should I rename it to start with amd_? the commit log uses the wrong&longer name) which does not return a mask, amd_get_dr_addr_mask() does.
Quoted 2 lines
I think you really wanna return negative on error and the caller should not continue in that case.
If it is out of bounds, it won't neither set or get. And these 2 helpers are used only by the kernel and the callers pass 0..3 and nothing else. BUG_ON() would do too, for example.
Quoted 20 lines
arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1161
Quoted 10 lines
arch/x86/kernel/cpu/amd.c +1
-1@@ -1158,24 +1158,41 @@ static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum)return false; } -void set_dr_addr_mask (unsigned long mask,int dr) +DEFINE_PER_CPU_READ_MOSTLY(unsigned long[4], amd_dr_addr_mask);static
arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1165
Quoted 4 lines
arch/x86/kernel/cpu/amd.c +4
-0+ +static unsigned int amd_msr_dr_addr_masks[] = { +MSR_F16H_DR0_ADDR_MASK , +MSR_F16H_DR1_ADDR_MASK -1 +1 ,- 1 + 1 ?
Quoted 3 lines
Why?
Because of the DR0 and then DR1 being in a different MSR range?
Yup.
Quoted 9 lines
Who cares, make it simple:
MSR_F16H_DR0_ADDR_MASK, MSR_F16H_DR1_ADDR_MASK, MSR_F16H_DR1_ADDR_MASK + 1, MSR_F16H_DR1_ADDR_MASK + 2and add a comment if you want to denote the non-contiguous range but meh.
imho having 1,2,3 in the code eliminates the need in a comment and produces the exact same end result. But since nobody cares, I'll do it the shorter way with just +1 and +2.
Quoted 29 lines
Quoted 16 lines
arch/x86/kernel/cpu/amd.c +5
-7> + MSR_F16H_DR1_ADDR_MASK - 1 + 2, + MSR_F16H_DR1_ADDR_MASK - 1 + 3 +}; + +void set_dr_addr_mask (unsigned long mask,unsigned int dr) { -if (!boot_cpu_has (X86_FEATURE_BPEXT )) +if (!cpu_feature_enabled (X86_FEATURE_BPEXT ))return ; -switch (dr) { -case 0 : -wrmsr (MSR_F16H_DR0_ADDR_MASK , mask,0 ); -break ; -case 1 : -case 2 :arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1178
Quoted 10 lines
arch/x86/kernel/cpu/amd.c +4
-6-case 3 : -wrmsr (MSR_F16H_DR1_ADDR_MASK -1 + dr, mask,0 ); -break ; -default : -break ; - } +if (WARN_ON_ONCE (dr >=ARRAY_SIZE (amd_msr_dr_addr_masks))) +return ; + +if (per_cpu (amd_dr_addr_mask,smp_processor_id ())[dr] == mask)Do that at function entry:
Quoted 3 lines
int cpu = smp_processor_id();and use cpu here.
Sure. Out of curiosity - why?^w^w^w^w^ it reduced the vmlinux size by 48 bytes, nice.
Thanks for the review!
Quoted 9 lines
arch/x86/kernel/cpu/amd.c static bool cpu_has_amd_erratum(struct cpuinfo_x86 *cpu, const int *erratum) · line 1183
Quoted 6 lines
arch/x86/kernel/cpu/amd.c +5
-0+return ; + +wrmsr (amd_msr_dr_addr_masks[dr], mask,0 ); +per_cpu (amd_dr_addr_mask,smp_processor_id ())[dr] = mask; +}Thx.
Signature · 1 line
Alexey
Borislav Petkov
maintainerRe: [PATCH kernel v2 1/3] x86/amd: Cache values in percpu variables
Quoted 2 lines · Alexey Kardashevskiy
"that function" is
set_dr_addr_mask()(btw should I rename it to start withamd_?
If you really wanna... I mean the code is already doing AMD-specific handling but sure, it'll be more obvious then that arch_install_hw_breakpoint() does only AMD-specific masking there under the info->mask test.
Quoted 3 lines
If it is out of bounds, it won't neither set or get. And these 2 helpers are used only by the kernel and the callers pass 0..3 and nothing else.
BUG_ON()would do too, for example.
Yeah, we don't do BUG_ON - look for Linus' colorful explanations why. :)
In any case, I think we should always aim for proper recovery from errors but this case is not that important so let's leave it at the WARN_ON_ONCE.
Quoted 3 lines
imho having 1,2,3 in the code eliminates the need in a comment and produces the exact same end result. But since nobody cares, I'll do it the shorter way with just +1 and +2.
Yeah, the more important goal is simplicity. And that pays off when you have to revisit that code and figure out what it does, later.
Quoted 2 lines
Sure. Out of curiosity - why?^w^w^w^w^ it reduced the vmlinux size by 48 bytes, nice.
The same answer - simplicity and speed when reading it.
That
if (per_cpu(amd_dr_addr_mask, smp_processor_id())[dr] == mask)
is a bit harder to parse than
if (per_cpu(amd_dr_addr_mask, cpu)[dr] == mask)
Thx.
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquettePATCH v2 2/3 KVM: SEV: Enable data breakpoints in SEV-ES
In reply to Alexey KardashevskiyAMD Milan (Fam 19h) introduces support for the swapping, as type 'B', of DR[0-3] and DR[0-3]_ADDR_MASK registers. Software enables this by setting SEV_FEATURES[5] (called "DebugSwap") in the VMSA which makes data breakpoints work in SEV-ES VMs.
For type 'B' swaps the hardware saves/restores the VM state on VMEXIT/VMRUN in VMSA, and restores the host state on VMEXIT.
Enable DebugSwap in VMSA but only if CPUID Fn80000021_EAX[0] ("NoNestedDataBp", "Processor ignores nested data breakpoints") is supported by the SOC as otherwise a malicious guest can cause the infinite #DB loop DoS.
Save DR[0-3] / DR[0-3]_ADDR_MASK in the host save area before VMRUN as type 'B' swap does not do this part.
Eliminate DR7 and #DB intercepts as:
- they are not needed when DebugSwap is supported;
- #VC for these intercepts is most likely not supported anyway and kills the VM.
Keep DR7 intercepted unless DebugSwap enabled to prevent the infinite #DB loop DoS.
---
Changes:
v2:
* debug_swap moved from vcpu to module_param
* rewrote commit log
---
"DR7 access must remain intercepted for an SEV-ES guest" - I could not figure out the exact reasoning why it is there in the first place, IIUC this is to prevent loop of #DBs in the VM.
---
Tested with:
===
int x;
int main(int argc, char *argv[])
{
x = 1;
return 0;
}===
gcc -g a.c
rsync a.out ruby-954vm:~/
ssh -t ruby-954vm 'gdb -ex "file a.out" -ex "watch x" -ex r'
where ruby-954vm is a VM.
With "/sys/module/kvm_amd/parameters/debug_swap = 0", gdb does not stop on the watchpoint, with "= 1" - gdb does.
--- arch/x86/include/asm/svm.h | 1 + arch/x86/kvm/svm/svm.h | 16 ++++++++--- arch/x86/kvm/svm/sev.c | 29 ++++++++++++++++++++ arch/x86/kvm/svm/svm.c | 3 +- 4 files changed, 44 insertions(+), 5 deletions(-)
arch/x86/include/asm/svm.h +1 -0
@@ -273,6 +273,7 @@ enum avic_ipi_failure_cause {#define AVIC_HPA_MASK ~((0xFFFULL << 52) | 0xFFF)#define VMCB_AVIC_APIC_BAR_MASK 0xFFFFFFFFFF000ULL +#define SVM_SEV_FEAT_DEBUG_SWAP BIT(5)struct vmcb_seg {u16 selector;
arch/x86/kvm/svm/svm.h +12 -4
@@ -386,6 +386,8 @@ static inline bool vmcb12_is_intercept(struct vmcb_ctrl_area_cached *control, u3return test_bit (bit, (unsigned long *)&control->intercepts); } +extern bool sev_es_is_debug_swap_enabled (void ); +static inline void set_dr_intercepts (struct vcpu_svm *svm) {struct vmcb *vmcb = svm->vmcb01.ptr;
@@ -407,8 +409,10 @@ static inline void set_dr_intercepts(struct vcpu_svm *svm)vmcb_set_intercept (&vmcb->control,INTERCEPT_DR6_WRITE ); } -vmcb_set_intercept (&vmcb->control,INTERCEPT_DR7_READ ); -vmcb_set_intercept (&vmcb->control,INTERCEPT_DR7_WRITE ); +if (!sev_es_guest (svm->vcpu.kvm) || !sev_es_is_debug_swap_enabled ()) { +vmcb_set_intercept (&vmcb->control,INTERCEPT_DR7_READ ); +vmcb_set_intercept (&vmcb->control,INTERCEPT_DR7_WRITE ); + }recalc_intercepts (svm); }
@@ -419,8 +423,12 @@ static inline void clr_dr_intercepts(struct vcpu_svm *svm) vmcb->control.intercepts[INTERCEPT_DR ] =0 ; -/* DR7 access must remain intercepted for an SEV-ES guest */ -if (sev_es_guest (svm->vcpu.kvm)) { +/* +* DR7 access must remain intercepted for an SEV-ES guest unless DebugSwap +* (depends on NO_NESTED_DATA_BP) is enabled as otherwise a VM writing to DR7 +* from the #DB handler may trigger infinite loop of #DB's. +*/ +if (sev_es_guest (svm->vcpu.kvm) && !sev_es_is_debug_swap_enabled ()) {vmcb_set_intercept (&vmcb->control,INTERCEPT_DR7_READ );vmcb_set_intercept (&vmcb->control,INTERCEPT_DR7_WRITE ); }
arch/x86/kvm/svm/sev.c +29 -0
@@ -21,6 +21,7 @@#include <asm/pkru.h> #include <asm/trapnr.h> #include <asm/fpu/xcr.h> +#include <asm/debugreg.h> #include "mmu.h" #include "x86.h"
@@ -52,11 +53,21 @@ module_param_named(sev, sev_enabled, bool, 0444);/* enable/disable SEV-ES support */ static bool sev_es_enabled = true;module_param_named (sev_es, sev_es_enabled, bool,0444 ); + +/* enable/disable SEV-ES DebugSwap support */ +static bool sev_es_debug_swap_enabled = true; +module_param_named (debug_swap, sev_es_debug_swap_enabled, bool,0644 );#else #define sev_enabled false#define sev_es_enabled false +#define sev_es_debug_swap false#endif /* CONFIG_KVM_AMD_SEV */ +bool sev_es_is_debug_swap_enabled (void ) +{ +return sev_es_debug_swap_enabled; +} +static u8 sev_enc_bit;static DECLARE_RWSEM (sev_deactivate_lock);static DEFINE_MUTEX (sev_bitmap_lock);
@@ -604,6 +615,9 @@ static int sev_es_sync_vmsa(struct vcpu_svm *svm) save->xss = svm->vcpu.arch.ia32_xss; save->dr6 = svm->vcpu.arch.dr6; +if (sev_es_is_debug_swap_enabled ()) + save->sev_features |=SVM_SEV_FEAT_DEBUG_SWAP ; +pr_debug ("Virtual Machine Save Area (VMSA):\n" );print_hex_dump_debug ("" ,DUMP_PREFIX_NONE ,16 ,1 , save,sizeof (*save), false);
@@ -2249,6 +2263,9 @@ void __init sev_hardware_setup(void) out: sev_enabled = sev_supported; sev_es_enabled = sev_es_supported; +if (sev_es_debug_swap_enabled) + sev_es_debug_swap_enabled = sev_es_enabled && +boot_cpu_has (X86_FEATURE_NO_NESTED_DATA_BP );#endif }
@@ -3027,6 +3044,18 @@ void sev_es_prepare_switch_to_guest(struct sev_es_save_area *hostsa)/* MSR_IA32_XSS is restored on VMEXIT, save the currnet host value */ hostsa->xss = host_xss; + +/* The DebugSwap SEV feature does Type B swaps of DR[0-3] */ +if (sev_es_is_debug_swap_enabled ()) { + hostsa->dr0 =native_get_debugreg (0 ); + hostsa->dr1 =native_get_debugreg (1 ); + hostsa->dr2 =native_get_debugreg (2 ); + hostsa->dr3 =native_get_debugreg (3 ); + hostsa->dr0_addr_mask =amd_get_dr_addr_mask (0 ); + hostsa->dr1_addr_mask =amd_get_dr_addr_mask (1 ); + hostsa->dr2_addr_mask =amd_get_dr_addr_mask (2 ); + hostsa->dr3_addr_mask =amd_get_dr_addr_mask (3 ); + } }void sev_vcpu_deliver_sipi_vector (struct kvm_vcpu *vcpu,u8 vector)
arch/x86/kvm/svm/svm.c +2 -1
@@ -1189,7 +1189,8 @@ static void init_vmcb(struct kvm_vcpu *vcpu)set_exception_intercept (svm,UD_VECTOR );set_exception_intercept (svm,MC_VECTOR );set_exception_intercept (svm,AC_VECTOR ); -set_exception_intercept (svm,DB_VECTOR ); +if (!sev_es_is_debug_swap_enabled ()) +set_exception_intercept (svm,DB_VECTOR );/* * Guest access to VMware backdoor ports could legitimately * trigger #GP because of TSS I/O permission bitmap.
Signature · 1 line
2.38.1
Borislav Petkov
maintainerRe: [PATCH kernel v2 2/3] KVM: SEV: Enable data breakpoints in SEV-ES
Quoted 1 line · Alexey Kardashevskiy
AMD Milan (Fam 19h) introduces support for the swapping, as type 'B',
"type B" means nothing to people who don't have an intimate APM knowledge.
Let's try again, this time with a more accessible formulation:
"The debug registers are handled a bit differently when doing a world switch of a SEV-ES guest: the guest debug registers values are saved and restored as usual and as one would expect.
The *host* debug registers are not saved to the host save area so if the host is doing any debug activity, that host should take care to stash its debug registers values into the host save area before running guests.
See Table B-3. Swap Types and the AMD APM volume 2."
And now you can go into detail explaining which regs exactly and so on.
Quoted 6 lines
of DR[0-3] and DR[0-3]_ADDR_MASK registers. Software enables this by setting SEV_FEATURES[5] (called "DebugSwap") in the VMSA which makes data breakpoints work in SEV-ES VMs.
For type 'B' swaps the hardware saves/restores the VM state on VMEXIT/VMRUN in VMSA, and restores the host state on VMEXIT.
Yeah, close but I'd prefer a more detailed explanation and a reference to the APM so that people can follow and read more info if needed.
Quoted 23 lines
Enable DebugSwap in VMSA but only if CPUID Fn80000021_EAX[0] ("NoNestedDataBp", "Processor ignores nested data breakpoints") is supported by the SOC as otherwise a malicious guest can cause the infinite #DB loop DoS.
Save DR[0-3] / DR[0-3]_ADDR_MASK in the host save area before VMRUN as type 'B' swap does not do this part.
Eliminate DR7 and #DB intercepts as:
- they are not needed when DebugSwap is supported;
- #VC for these intercepts is most likely not supported anyway and kills the VM.
Keep DR7 intercepted unless DebugSwap enabled to prevent the infinite #DB loop DoS.Signed-off-by Alexey Kardashevskiy <[email protected]>---
Changes:
v2:
*debug_swapmoved from vcpu tomodule_param
* rewrote commit log
Quoted 5 lines
---
"DR7 access must remain intercepted for an SEV-ES guest" - I could not figure out the exact reasoning why it is there in the first place, IIUC this is to prevent loop of #DBs in the VM.
Let's ask Mr. Lendacky:
8d4846b9b150 ("KVM: SVM: Prevent debugging under SEV-ES")
Quoted 45 lines
arch/x86/kvm/svm/sev.c +14
-0@@ -21,6 +21,7 @@#include <asm/pkru.h> #include <asm/trapnr.h> #include <asm/fpu/xcr.h> +#include <asm/debugreg.h> #include "mmu.h" #include "x86.h" @@ -52,11 +53,21 @@ module_param_named(sev, sev_enabled, bool, 0444);/* enable/disable SEV-ES support */ static bool sev_es_enabled = true;module_param_named (sev_es, sev_es_enabled, bool,0444 ); + +/* enable/disable SEV-ES DebugSwap support */ +static bool sev_es_debug_swap_enabled = true; +module_param_named (debug_swap, sev_es_debug_swap_enabled, bool,0644 );#else #define sev_enabled false#define sev_es_enabled false +#define sev_es_debug_swap false#endif /* CONFIG_KVM_AMD_SEV */ +bool sev_es_is_debug_swap_enabled (void ) +{ +return sev_es_debug_swap_enabled; +} +static u8 sev_enc_bit;static DECLARE_RWSEM (sev_deactivate_lock);static DEFINE_MUTEX (sev_bitmap_lock);@@ -604,6 +615,9 @@ static int sev_es_sync_vmsa(struct vcpu_svm *svm) save->xss = svm->vcpu.arch.ia32_xss; save->dr6 = svm->vcpu.arch.dr6; +if (sev_es_is_debug_swap_enabled ()) + save->sev_features |=SVM_SEV_FEAT_DEBUG_SWAP ; +pr_debug ("Virtual Machine Save Area (VMSA):\n" );print_hex_dump_debug ("" ,DUMP_PREFIX_NONE ,16 ,1 , save,sizeof (*save), false);
arch/x86/kvm/svm/sev.c void __init sev_hardware_setup(void) · line 2268
Quoted 6 lines
arch/x86/kvm/svm/sev.c +3
-0@@ -2249,6 +2263,9 @@ void __init sev_hardware_setup(void) out: sev_enabled = sev_supported; sev_es_enabled = sev_es_supported; +if (sev_es_debug_swap_enabled) + sev_es_debug_swap_enabled = sev_es_enabled && +boot_cpu_has (X86_FEATURE_NO_NESTED_DATA_BP );
check_for_deprecated_apis: WARNING: arch/x86/kvm/svm/sev.c:2268: Do not use boot_cpu_has() - use cpu_feature_enabled() instead.
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquetteRe: [PATCH kernel v2 2/3] KVM: SEV: Enable data breakpoints in SEV-ES
Quoted 10 lines · Borislav Petkov
Quoted 5 lines · Alexey Kardashevskiy
"DR7 access must remain intercepted for an SEV-ES guest" - I could not figure out the exact reasoning why it is there in the first place, IIUC this is to prevent loop of #DBs in the VM.
Let's ask Mr. Lendacky:
8d4846b9b150("KVM: SVM: Prevent debugging under SEV-ES")
The DR7 requirements were to prevent a malicious SEV-ES guest from setting up data breakpoints on the #VC IDT entry/stack and causing an infinite loop.
Thanks,
Tom
Quoted 55 lines
Quoted 46 lines
arch/x86/kvm/svm/sev.c +14
-0@@ -21,6 +21,7 @@#include <asm/pkru.h> #include <asm/trapnr.h> #include <asm/fpu/xcr.h> +#include <asm/debugreg.h> #include "mmu.h" #include "x86.h" @@ -52,11 +53,21 @@ module_param_named(sev, sev_enabled, bool, 0444);/* enable/disable SEV-ES support */ static bool sev_es_enabled = true;module_param_named (sev_es, sev_es_enabled, bool,0444 ); + +/* enable/disable SEV-ES DebugSwap support */ +static bool sev_es_debug_swap_enabled = true; +module_param_named (debug_swap, sev_es_debug_swap_enabled, bool,0644 );#else #define sev_enabled false#define sev_es_enabled false +#define sev_es_debug_swap false#endif /* CONFIG_KVM_AMD_SEV */ +bool sev_es_is_debug_swap_enabled (void ) +{ +return sev_es_debug_swap_enabled; +} +static u8 sev_enc_bit;static DECLARE_RWSEM (sev_deactivate_lock);static DEFINE_MUTEX (sev_bitmap_lock);@@ -604,6 +615,9 @@ static int sev_es_sync_vmsa(struct vcpu_svm *svm) save->xss = svm->vcpu.arch.ia32_xss; save->dr6 = svm->vcpu.arch.dr6; +if (sev_es_is_debug_swap_enabled ()) + save->sev_features |=SVM_SEV_FEAT_DEBUG_SWAP ; +pr_debug ("Virtual Machine Save Area (VMSA):\n" );print_hex_dump_debug ("" ,DUMP_PREFIX_NONE ,16 ,1 , save,sizeof (*save), false);arch/x86/kvm/svm/sev.c void __init sev_hardware_setup(void) · line 2268
Quoted 6 lines
arch/x86/kvm/svm/sev.c +3
-0@@ -2249,6 +2263,9 @@ void __init sev_hardware_setup(void) out: sev_enabled = sev_supported; sev_es_enabled = sev_es_supported; +if (sev_es_debug_swap_enabled) + sev_es_debug_swap_enabled = sev_es_enabled && +boot_cpu_has (X86_FEATURE_NO_NESTED_DATA_BP );
check_for_deprecated_apis: WARNING:arch/x86/kvm/svm/sev.c:2268: Do not useboot_cpu_has()- usecpu_feature_enabled()instead.
Re: [PATCH kernel v2 2/3] KVM: SEV: Enable data breakpoints in SEV-ES
In reply to Borislav PetkovQuoted 10 lines · Borislav Petkov
Quoted 1 line · Alexey Kardashevskiy
AMD Milan (Fam 19h) introduces support for the swapping, as type 'B',
"type B" means nothing to people who don't have an intimate APM knowledge.
Let's try again, this time with a more accessible formulation:
"The debug registers are handled a bit differently when doing a world switch of a SEV-ES guest: the guest debug registers values are saved and restored as usual and as one would expect.
Well, SEV-ES KVM (ES == Encrypted State) does not save/restore them for the guest (well, as I would expect) as the guest registers are not visible to host to save, they are intercepted and the VM does this GHCB dance with VMGEXIT(SVM_EXIT_WRITE_DR7).
Quoted 12 lines
The *host* debug registers are not saved to the host save area so if the host is doing any debug activity, that host should take care to stash its debug registers values into the host save area before running guests.
See Table B-3. Swap Types and the AMD APM volume 2."
And now you can go into detail explaining which regs exactly and so on.
Quoted 4 lines
of DR[0-3] and DR[0-3]_ADDR_MASK registers. Software enables this by setting SEV_FEATURES[5] (called "DebugSwap") in the VMSA which makes data breakpoints work in SEV-ES VMs.
Quoted 5 lines
Quoted 2 lines
For type 'B' swaps the hardware saves/restores the VM state on VMEXIT/VMRUN in VMSA, and restores the host state on VMEXIT.
Yeah, close but I'd prefer a more detailed explanation and a reference to the APM so that people can follow and read more info if needed.
Well, the only place in APM is that "Table B-3. Swap Types and the AMD APM volume 2", and it is pretty brief, do I miss something? Thanks,
Quoted 86 lines
Quoted 22 lines
Quoted 1 line
Enable DebugSwap in VMSA but only if CPUID Fn80000021_EAX[0]
("NoNestedDataBp", "Processor ignores nested data breakpoints") is supported by the SOC as otherwise a malicious guest can cause the infinite #DB loop DoS.
Save DR[0-3] / DR[0-3]_ADDR_MASK in the host save area before VMRUN as type 'B' swap does not do this part.
Eliminate DR7 and #DB intercepts as:
- they are not needed when DebugSwap is supported;
- #VC for these intercepts is most likely not supported anyway and kills the VM.
Keep DR7 intercepted unless DebugSwap enabled to prevent the infinite #DB loop DoS.Signed-off-by Alexey Kardashevskiy <[email protected]>---
Changes:
v2:
*debug_swapmoved from vcpu tomodule_param
* rewrote commit logQuoted 5 lines
---
"DR7 access must remain intercepted for an SEV-ES guest" - I could not figure out the exact reasoning why it is there in the first place, IIUC this is to prevent loop of #DBs in the VM.
Let's ask Mr. Lendacky:
8d4846b9b150("KVM: SVM: Prevent debugging under SEV-ES")Quoted 45 lines
arch/x86/kvm/svm/sev.c +14
-0@@ -21,6 +21,7 @@#include <asm/pkru.h> #include <asm/trapnr.h> #include <asm/fpu/xcr.h> +#include <asm/debugreg.h> #include "mmu.h" #include "x86.h" @@ -52,11 +53,21 @@ module_param_named(sev, sev_enabled, bool, 0444);/* enable/disable SEV-ES support */ static bool sev_es_enabled = true;module_param_named (sev_es, sev_es_enabled, bool,0444 ); + +/* enable/disable SEV-ES DebugSwap support */ +static bool sev_es_debug_swap_enabled = true; +module_param_named (debug_swap, sev_es_debug_swap_enabled, bool,0644 );#else #define sev_enabled false#define sev_es_enabled false +#define sev_es_debug_swap false#endif /* CONFIG_KVM_AMD_SEV */ +bool sev_es_is_debug_swap_enabled (void ) +{ +return sev_es_debug_swap_enabled; +} +static u8 sev_enc_bit;static DECLARE_RWSEM (sev_deactivate_lock);static DEFINE_MUTEX (sev_bitmap_lock);@@ -604,6 +615,9 @@ static int sev_es_sync_vmsa(struct vcpu_svm *svm) save->xss = svm->vcpu.arch.ia32_xss; save->dr6 = svm->vcpu.arch.dr6; +if (sev_es_is_debug_swap_enabled ()) + save->sev_features |=SVM_SEV_FEAT_DEBUG_SWAP ; +pr_debug ("Virtual Machine Save Area (VMSA):\n" );print_hex_dump_debug ("" ,DUMP_PREFIX_NONE ,16 ,1 , save,sizeof (*save), false);arch/x86/kvm/svm/sev.c void __init sev_hardware_setup(void) · line 2268
Quoted 6 lines
arch/x86/kvm/svm/sev.c +3
-0@@ -2249,6 +2263,9 @@ void __init sev_hardware_setup(void) out: sev_enabled = sev_supported; sev_es_enabled = sev_es_supported; +if (sev_es_debug_swap_enabled) + sev_es_debug_swap_enabled = sev_es_enabled && +boot_cpu_has (X86_FEATURE_NO_NESTED_DATA_BP );
check_for_deprecated_apis: WARNING:arch/x86/kvm/svm/sev.c:2268: Do not useboot_cpu_has()- usecpu_feature_enabled()instead.
Signature · 1 line
Alexey
Borislav Petkov
maintainerRe: [PATCH kernel v2 2/3] KVM: SEV: Enable data breakpoints in SEV-ES
Quoted 4 lines · Alexey Kardashevskiy
Well, SEV-ES KVM (ES == Encrypted State) does not save/restore them for the guest (well, as I would expect) as the guest registers are not visible to host to save, they are intercepted and the VM does this GHCB dance with
VMGEXIT(SVM_EXIT_WRITE_DR7).
But they're saved in the VMSA, as Table B-3 says.
Quoted 2 lines
Well, the only place in APM is that "Table B-3. Swap Types and the AMD APM volume 2", and it is pretty brief, do I miss something?
I don't understand that question - please elaborate.
Thx.
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquetteRe: [PATCH kernel v2 2/3] KVM: SEV: Enable data breakpoints in SEV-ES
Quoted 7 lines · Borislav Petkov
Quoted 4 lines · Alexey Kardashevskiy
Well, SEV-ES KVM (ES == Encrypted State) does not save/restore them for the guest (well, as I would expect) as the guest registers are not visible to host to save, they are intercepted and the VM does this GHCB dance with
VMGEXIT(SVM_EXIT_WRITE_DR7).But they're saved in the VMSA, as Table B-3 says.
Correct, when this feature is enabled, the VMRUN execution will restore the guest debug registers on guest entry and save them on guest exit.
Thanks,
Tom
Quoted 8 lines
Quoted 3 lines
Well, the only place in APM is that "Table B-3. Swap Types and the AMD APM volume 2", and it is pretty brief, do I miss something?
I don't understand that question - please elaborate.
Thx.
PATCH v2 3/3 x86/sev: Do not handle #VC for DR7 read/write
In reply to Alexey KardashevskiyWith MSR_AMD64_SEV_DEBUG_SWAP enabled, the VM should not get #VC events for DR7 read/write which it rather avoided.
---
Changes:
v2:
* use new bit definition
--- arch/x86/include/asm/msr-index.h | 1 + tools/arch/x86/include/asm/msr-index.h | 1 + arch/x86/kernel/sev.c | 6 ++++++ 3 files changed, 8 insertions(+)
arch/x86/include/asm/msr-index.h +1 -0
@@ -570,6 +570,7 @@#define MSR_AMD64_SEV_ENABLED BIT_ULL(MSR_AMD64_SEV_ENABLED_BIT)#define MSR_AMD64_SEV_ES_ENABLED BIT_ULL(MSR_AMD64_SEV_ES_ENABLED_BIT)#define MSR_AMD64_SEV_SNP_ENABLED BIT_ULL(MSR_AMD64_SEV_SNP_ENABLED_BIT) +#define MSR_AMD64_SEV_DEBUG_SWAP BIT_ULL(7)#define MSR_AMD64_VIRT_SPEC_CTRL 0xc001011f
tools/arch/x86/include/asm/msr-index.h +1 -0
@@ -570,6 +570,7 @@#define MSR_AMD64_SEV_ENABLED BIT_ULL(MSR_AMD64_SEV_ENABLED_BIT)#define MSR_AMD64_SEV_ES_ENABLED BIT_ULL(MSR_AMD64_SEV_ES_ENABLED_BIT)#define MSR_AMD64_SEV_SNP_ENABLED BIT_ULL(MSR_AMD64_SEV_SNP_ENABLED_BIT) +#define MSR_AMD64_SEV_DEBUG_SWAP BIT_ULL(7)#define MSR_AMD64_VIRT_SPEC_CTRL 0xc001011f
arch/x86/kernel/sev.c +6 -0
@@ -1618,6 +1618,9 @@ static enum es_result vc_handle_dr7_write(struct ghcb *ghcb,long val, *reg =vc_insn_get_rm (ctxt);enum es_result ret; +if (sev_status &MSR_AMD64_SEV_DEBUG_SWAP ) +return ES_VMM_ERROR ; +if (!reg)return ES_DECODE_FAILED ;
@@ -1655,6 +1658,9 @@ static enum es_result vc_handle_dr7_read(struct ghcb *ghcb,struct sev_es_runtime_data *data =this_cpu_read (runtime_data);long *reg =vc_insn_get_rm (ctxt); +if (sev_status &MSR_AMD64_SEV_DEBUG_SWAP ) +return ES_VMM_ERROR ; +if (!reg)return ES_DECODE_FAILED ;
Signature · 1 line
2.38.1
Re: [PATCH kernel v2 3/3] x86/sev: Do not handle #VC for DR7 read/write
Quoted 8 lines · Alexey Kardashevskiy
With
MSR_AMD64_SEV_DEBUG_SWAPenabled, the VM should not get #VC events for DR7 read/write which it rather avoided.Signed-off-by Alexey Kardashevskiy <[email protected]>---
Changes:
v2:
* use new bit definition
Quoted 5 lines
--- arch/x86/include/asm/msr-index.h | 1 + tools/arch/x86/include/asm/msr-index.h | 1 + arch/x86/kernel/sev.c | 6 ++++++ 3 files changed, 8 insertions(+)
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquetteQuoted 8 lines · Alexey Kardashevskiy
This is to use another AMD SEV-ES hardware assisted register swap, more detail in 2/3. The patches are fairly independend but required in this order.
This is based on sha1
0d1409e4ff08and requires something like this forX86_FEATURE_NO_NESTED_DATA_BP:
https://lkml.org/lkml/2022/11/29/1229
Quoted 5 lines
This patchset is pushed out at
https://github.com/aik/linux/commits/debugswapPlease comment. Thanks.
Ping? Thanks,
Quoted 21 lines
btw the enormous cc: list came from get_maintainer.pl, keep it like this or tell the script something to reduce the list?
Alexey Kardashevskiy (3):
x86/amd: Cache values in percpu variables
KVM: SEV: Enable data breakpoints in SEV-ES
x86/sev: Do not handle #VC for DR7 read/writearch/x86/include/asm/debugreg.h | 9 +++- arch/x86/include/asm/msr-index.h | 1 + arch/x86/include/asm/svm.h | 1 + arch/x86/kvm/svm/svm.h | 16 +++++-- tools/arch/x86/include/asm/msr-index.h | 1 + arch/x86/kernel/cpu/amd.c | 45 ++++++++++++++------ arch/x86/kernel/sev.c | 6 +++ arch/x86/kvm/svm/sev.c | 29 +++++++++++++ arch/x86/kvm/svm/svm.c | 3 +- 9 files changed, 90 insertions(+), 21 deletions(-)
Signature · 1 line
Alexey
Borislav Petkov
maintainerQuoted 7 lines · Alexey Kardashevskiy
This is to use another AMD SEV-ES hardware assisted register swap, more detail in 2/3. The patches are fairly independend but required in this order.
This is based on sha1
0d1409e4ff08and requires something like this forX86_FEATURE_NO_NESTED_DATA_BP:
https://lkml.org/lkml/2022/11/29/1229
Please use
lore.kernel.org/r/<Message-ID>
when you want to refer to a mail on a public ML.
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquette