xen/pv: sanitize xen pv guest msr accesses
Subsystems (3)
Juergen Gross
maintainerv3 cover xen/pv: sanitize xen pv guest msr accesses
Historically when running as Xen PV guest all MSR accesses have been silently swallowing any GP faults, even when the kernel was using not the *msr_safe() access functions.
Change that by making the behavior controllable via kernel config and via a boot parameter.
This will help finding paths where MSRs are being accessed under Xen which are not emulated by the hypervisor.
Changes in V3:
- new patch 2
- addressed comments
Juergen Gross (4):
xen/pv: add fault recovery control to pmu msr accesses xen/pv: fix vendor checks for pmu emulation
xen/pv: refactor msr access functions to support safe and unsafe
accesses
xen/pv: support selecting safe/unsafe msr accesses
.../admin-guide/kernel-parameters.txt | 6 ++ arch/x86/xen/Kconfig | 9 ++ arch/x86/xen/enlighten_pv.c | 99 +++++++++++++------ arch/x86/xen/pmu.c | 71 +++++++------ 4 files changed, 127 insertions(+), 58 deletions(-)
Signature · 1 line
2.35.3
Juergen Gross
maintainerPATCH v3 1/4 xen/pv: add fault recovery control to pmu msr accesses
Today pmu_msr_read() and pmu_msr_write() fall back to the safe variants of read/write MSR in case the MSR access isn't emulated via Xen. Allow the caller to select that faults should not be recovered from by passing NULL for the error pointer.
Restructure the code to make it more readable.
---
V2:
- do some restructuring (Jan Beulich, Boris Ostrovsky) V3:
- commit message rephrasing (Jan Beulich)
- more restructuring (Boris Ostrovsky)
--- arch/x86/xen/pmu.c | 66 ++++++++++++++++++++++++++-------------------- 1 file changed, 38 insertions(+), 28 deletions(-)
arch/x86/xen/pmu.c +38 -28
@@ -131,6 +131,9 @@ static inline uint32_t get_fam15h_addr(u32 addr)static inline bool is_amd_pmu_msr (unsigned int msr) { +if (boot_cpu_data.x86_vendor ==X86_VENDOR_INTEL ) +return false; + if ((msr >= MSR_F15H_PERF_CTL && msr < MSR_F15H_PERF_CTR + (amd_num_counters * 2)) || (msr >= MSR_K7_EVNTSEL0 &&
@@ -144,6 +147,9 @@ static int is_intel_pmu_msr(u32 msr_index, int *type, int *index) {u32 msr_index_pmc; +if (boot_cpu_data.x86_vendor !=X86_VENDOR_INTEL ) +return false; +switch (msr_index) {case MSR_CORE_PERF_FIXED_CTR_CTRL :case MSR_IA32_DS_AREA :
@@ -290,48 +296,52 @@ static bool xen_amd_pmu_emulate(unsigned int msr, u64 *val, bool is_read)return false; } +static bool pmu_msr_chk_emulated (unsigned int msr,uint64_t *val,bool is_read, +bool *emul) +{ +int type, index; + +if (is_amd_pmu_msr (msr)) + *emul =xen_amd_pmu_emulate (msr, val, is_read); +else if (is_intel_pmu_msr (msr, &type, &index)) + *emul =xen_intel_pmu_emulate (msr, val, type, index, is_read); +else +return false; + +return true; +} +bool pmu_msr_read (unsigned int msr,uint64_t *val,int *err) { -if (boot_cpu_data.x86_vendor !=X86_VENDOR_INTEL ) { -if (is_amd_pmu_msr (msr)) { -if (!xen_amd_pmu_emulate (msr, val,1 )) - *val =native_read_msr_safe (msr, err); -return true; - } - }else { -int type, index; +bool emulated; -if (is_intel_pmu_msr (msr, &type, &index)) { -if (!xen_intel_pmu_emulate (msr, val, type, index,1 )) - *val =native_read_msr_safe (msr, err); -return true; - } +if (!pmu_msr_chk_emulated (msr, val, true, &emulated)) +return false; + +if (!emulated) { + *val = err ?native_read_msr_safe (msr, err) + :native_read_msr (msr); } -return false; +return true; }bool pmu_msr_write (unsigned int msr,uint32_t low,uint32_t high,int *err) {uint64_t val = ((uint64_t )high <<32 ) | low; +bool emulated; -if (boot_cpu_data.x86_vendor !=X86_VENDOR_INTEL ) { -if (is_amd_pmu_msr (msr)) { -if (!xen_amd_pmu_emulate (msr, &val,0 )) - *err =native_write_msr_safe (msr, low, high); -return true; - } - }else { -int type, index; +if (!pmu_msr_chk_emulated (msr, &val, false, &emulated)) +return false; -if (is_intel_pmu_msr (msr, &type, &index)) { -if (!xen_intel_pmu_emulate (msr, &val, type, index,0 )) - *err =native_write_msr_safe (msr, low, high); -return true; - } +if (!emulated) { +if (err) + *err =native_write_msr_safe (msr, low, high); +else +native_write_msr (msr, low, high); } -return false; +return true; }static unsigned long long xen_amd_read_pmc (int counter)
Signature · 1 line
2.35.3
Re: [PATCH v3 1/4] xen/pv: add fault recovery control to pmu msr accesses
Quoted 8 lines · Juergen Gross
Today
pmu_msr_read()andpmu_msr_write()fall back to the safe variants of read/write MSR in case the MSR access isn't emulated via Xen. Allow the caller to select that faults should not be recovered from by passing NULL for the error pointer.Restructure the code to make it more readable.
Signed-off-by Juergen Gross <[email protected]>
Juergen Gross
maintainerPATCH v3 2/4 xen/pv: fix vendor checks for pmu emulation
In reply to Juergen GrossThe CPU vendor checks for pmu emulation are rather limited today, as the assumption seems to be that only Intel and AMD are existing and/or supported vendors.
Fix that by handling Centaur and Zhaoxin CPUs the same way as Intel, and Hygon the same way as AMD.
While at it fix the return type of is_intel_pmu_msr().
---
V3:
- new patch
--- arch/x86/xen/pmu.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
arch/x86/xen/pmu.c +6 -3
@@ -131,7 +131,8 @@ static inline uint32_t get_fam15h_addr(u32 addr)static inline bool is_amd_pmu_msr (unsigned int msr) { -if (boot_cpu_data.x86_vendor ==X86_VENDOR_INTEL ) +if (boot_cpu_data.x86_vendor !=X86_VENDOR_AMD && + boot_cpu_data.x86_vendor !=X86_VENDOR_HYGON )return false; if ((msr >= MSR_F15H_PERF_CTL &&
@@ -143,11 +144,13 @@ static inline bool is_amd_pmu_msr(unsigned int msr)return false; } -static int is_intel_pmu_msr (u32 msr_index,int *type,int *index) +static bool is_intel_pmu_msr (u32 msr_index,int *type,int *index) {u32 msr_index_pmc; -if (boot_cpu_data.x86_vendor !=X86_VENDOR_INTEL ) +if (boot_cpu_data.x86_vendor !=X86_VENDOR_INTEL && + boot_cpu_data.x86_vendor !=X86_VENDOR_CENTAUR && + boot_cpu_data.x86_vendor !=X86_VENDOR_ZHAOXIN )return false;switch (msr_index) {
Signature · 1 line
2.35.3
Re: [PATCH v3 2/4] xen/pv: fix vendor checks for pmu emulation
Quoted 11 lines · Juergen Gross
The CPU vendor checks for pmu emulation are rather limited today, as the assumption seems to be that only Intel and AMD are existing and/or supported vendors.
Fix that by handling Centaur and Zhaoxin CPUs the same way as Intel, and Hygon the same way as AMD.
While at it fix the return type of
is_intel_pmu_msr().Suggested-by Jan Beulich <[email protected]>Signed-off-by Juergen Gross <[email protected]>
Juergen Gross
maintainerPATCH v3 3/4 xen/pv: refactor msr access functions to support safe and unsafe accesses
In reply to Juergen GrossRefactor and rename xen_read_msr_safe() and xen_write_msr_safe() to support both cases of MSR accesses, safe ones and potentially GP-fault generating ones.
This will prepare to no longer swallow GPs silently in xen_read_msr() and xen_write_msr().
---
V2:
- init val in xen_do_read_msr() to 0 (Jan Beulich)
--- arch/x86/xen/enlighten_pv.c | 75 +++++++++++++++++++++++++++---------- 1 file changed, 56 insertions(+), 19 deletions(-)
arch/x86/xen/enlighten_pv.c +56 -19
@@ -917,14 +917,18 @@ static void xen_write_cr4(unsigned long cr4)native_write_cr4 (cr4); } -static u64 xen_read_msr_safe (unsigned int msr,int *err) +static u64 xen_do_read_msr (unsigned int msr,int *err) { -u64 val; +u64 val =0 ;/* Avoid uninitialized value for safe variant. */ if (pmu_msr_read (msr, &val, err))return val; - val =native_read_msr_safe (msr, err); +if (err) + val =native_read_msr_safe (msr, err); +else + val =native_read_msr (msr); +switch (msr) {case MSR_IA32_APICBASE : val &= ~X2APIC_ENABLE ;
@@ -933,23 +937,39 @@ static u64 xen_read_msr_safe(unsigned int msr, int *err)return val; } -static int xen_write_msr_safe (unsigned int msr,unsigned low,unsigned high) +static void set_seg (unsigned int which,unsigned int low,unsigned int high, +int *err) { -int ret; -unsigned int which; -u64 base; +u64 base = ((u64 )high <<32 ) | low; + +if (HYPERVISOR_set_segment_base (which, base) ==0 ) +return ; - ret =0 ; +if (err) + *err = -EIO ; +else +WARN (1 ,"Xen set_segment_base(%u, %llx) failed\n" , which, base); +} +/* +* Support write_msr_safe() and write_msr() semantics. +* With err == NULL write_msr() semantics are selected. +* Supplying an err pointer requires err to be pre-initialized with 0. +*/ +static void xen_do_write_msr (unsigned int msr,unsigned int low, +unsigned int high,int *err) +{switch (msr) { -case MSR_FS_BASE : which =SEGBASE_FS ; goto set; -case MSR_KERNEL_GS_BASE : which =SEGBASE_GS_USER ; goto set; -case MSR_GS_BASE : which =SEGBASE_GS_KERNEL ; goto set; - - set: - base = ((u64 )high <<32 ) | low; -if (HYPERVISOR_set_segment_base (which, base) !=0 ) - ret = -EIO ; +case MSR_FS_BASE : +set_seg (SEGBASE_FS , low, high, err); +break ; + +case MSR_KERNEL_GS_BASE : +set_seg (SEGBASE_GS_USER , low, high, err); +break ; + +case MSR_GS_BASE : +set_seg (SEGBASE_GS_KERNEL , low, high, err);break ;case MSR_STAR :
@@ -965,11 +985,28 @@ static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)break ;default : -if (!pmu_msr_write (msr, low, high, &ret)) - ret =native_write_msr_safe (msr, low, high); +if (!pmu_msr_write (msr, low, high, err)) { +if (err) + *err =native_write_msr_safe (msr, low, high); +else +native_write_msr (msr, low, high); + } } +} + +static u64 xen_read_msr_safe (unsigned int msr,int *err) +{ +return xen_do_read_msr (msr, err); +} + +static int xen_write_msr_safe (unsigned int msr,unsigned int low, +unsigned int high) +{ +int err =0 ; + +xen_do_write_msr (msr, low, high, &err); -return ret; +return err; }static u64 xen_read_msr (unsigned int msr)
Signature · 1 line
2.35.3
Juergen Gross
maintainerPATCH v3 4/4 xen/pv: support selecting safe/unsafe msr accesses
In reply to Juergen GrossInstead of always doing the safe variants for reading and writing MSRs in Xen PV guests, make the behavior controllable via Kconfig option and a boot parameter.
The default will be the current behavior, which is to always use the safe variant.
--- .../admin-guide/kernel-parameters.txt | 6 +++++ arch/x86/xen/Kconfig | 9 +++++++ arch/x86/xen/enlighten_pv.c | 24 +++++++++++-------- 3 files changed, 29 insertions(+), 10 deletions(-)
Documentation/admin-guide/kernel-parameters.txt +6 -0
@@ -6836,6 +6836,12 @@ Crash from Xen panic notifier, without executing late panic() code such as dumping handler. + xen_msr_safe= [X86,XEN] + Format: <bool> + Select whether to always use non-faulting (safe) MSR + access functions when running as Xen PV guest. The + default value is controlled by CONFIG_XEN_PV_MSR_SAFE. + xen_nopvspin [X86,XEN] Disables the qspinlock slowpath using Xen PV optimizations. This parameter is obsoleted by "nopvspin" parameter, which
arch/x86/xen/Kconfig +9 -0
@@ -92,3 +92,12 @@ config XEN_DOM0 select X86_X2APIC if XEN_PVH && X86_64 help Support running as a Xen Dom0 guest. + +config XEN_PV_MSR_SAFE + bool "Always use safe MSR accesses in PV guests" + default y + depends on XEN_PV + help + Use safe (not faulting) MSR access functions even if the MSR access + should not fault anyway. + The default can be changed by using the "xen_msr_safe" boot parameter.
arch/x86/xen/enlighten_pv.c +14 -10
@@ -108,6 +108,16 @@ struct tls_descs {*/ static DEFINE_PER_CPU(struct tls_descs, shadow_tls_desc); +static __read_mostly bool xen_msr_safe =IS_ENABLED (CONFIG_XEN_PV_MSR_SAFE ); + +static int __init parse_xen_msr_safe (char *str) +{ +if (str) +return strtobool (str, &xen_msr_safe); +return -EINVAL ; +} +early_param ("xen_msr_safe" , parse_xen_msr_safe); +static void __init xen_pv_init_platform (void ) {/* PV guests can't operate virtio devices without grants. */
@@ -1011,22 +1021,16 @@ static int xen_write_msr_safe(unsigned int msr, unsigned int low,static u64 xen_read_msr (unsigned int msr) { -/* -* This will silently swallow a #GP from RDMSR. It may be worth -* changing that. -*/ int err; -return xen_read_msr_safe (msr, &err); +return xen_do_read_msr (msr, xen_msr_safe ? &err :NULL ); }static void xen_write_msr (unsigned int msr,unsigned low,unsigned high) { -/* -* This will silently swallow a #GP from WRMSR. It may be worth -* changing that. -*/ -xen_write_msr_safe (msr, low, high); +int err; + +xen_do_write_msr (msr, low, high, xen_msr_safe ? &err :NULL ); }/* This is called once we have the cpu_possible_mask */
Signature · 1 line
2.35.3