[PATCH v7 08/14] x86: add new features for paravirt patching

[Juergen Gross <jgross@xxxxxxxx>]

For being able to switch paravirt patching from special cased custom
code sequences to ALTERNATIVE handling some X86_FEATURE_* are needed
as new features. This enables to have the standard indirect pv call
as the default code and to patch that with the non-Xen custom code
sequence via ALTERNATIVE patching later.

Make sure paravirt patching is performed before alternative patching.

Signed-off-by: Juergen Gross <jgross@xxxxxxxx>
Acked-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
---
V3:
- add comment (Boris Petkov)
- no negative features (Boris Petkov)
V4:
- move paravirt_set_cap() to paravirt-spinlocks.c
---
 arch/x86/include/asm/cpufeatures.h   |  2 ++
 arch/x86/include/asm/paravirt.h      | 10 ++++++++++
 arch/x86/kernel/alternative.c        | 30 ++++++++++++++++++++++++++--
 arch/x86/kernel/paravirt-spinlocks.c |  9 +++++++++
 4 files changed, 49 insertions(+), 2 deletions(-)

diff --git a/arch/x86/include/asm/cpufeatures.h 
b/arch/x86/include/asm/cpufeatures.h
index cc96e26d69f7..b440c950246d 100644
--- a/arch/x86/include/asm/cpufeatures.h
+++ b/arch/x86/include/asm/cpufeatures.h
@@ -236,6 +236,8 @@
 #define X86_FEATURE_EPT_AD             ( 8*32+17) /* Intel Extended Page Table 
access-dirty bit */
 #define X86_FEATURE_VMCALL             ( 8*32+18) /* "" Hypervisor supports 
the VMCALL instruction */
 #define X86_FEATURE_VMW_VMMCALL                ( 8*32+19) /* "" VMware prefers 
VMMCALL hypercall instruction */
+#define X86_FEATURE_PVUNLOCK           ( 8*32+20) /* "" PV unlock function */
+#define X86_FEATURE_VCPUPREEMPT                ( 8*32+21) /* "" PV 
vcpu_is_preempted function */
 
 /* Intel-defined CPU features, CPUID level 0x00000007:0 (EBX), word 9 */
 #define X86_FEATURE_FSGSBASE           ( 9*32+ 0) /* RDFSBASE, WRFSBASE, 
RDGSBASE, WRGSBASE instructions*/
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 6408fd0f55ab..def450f46097 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -45,6 +45,10 @@ static inline u64 paravirt_steal_clock(int cpu)
        return static_call(pv_steal_clock)(cpu);
 }
 
+#ifdef CONFIG_PARAVIRT_SPINLOCKS
+void __init paravirt_set_cap(void);
+#endif
+
 /* The paravirtualized I/O functions */
 static inline void slow_down_io(void)
 {
@@ -809,5 +813,11 @@ static inline void paravirt_arch_exit_mmap(struct 
mm_struct *mm)
 {
 }
 #endif
+
+#ifndef CONFIG_PARAVIRT_SPINLOCKS
+static inline void paravirt_set_cap(void)
+{
+}
+#endif
 #endif /* __ASSEMBLY__ */
 #endif /* _ASM_X86_PARAVIRT_H */
diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
index 133b549dc091..76ad4ce454c0 100644
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -28,6 +28,7 @@
 #include <asm/insn.h>
 #include <asm/io.h>
 #include <asm/fixmap.h>
+#include <asm/paravirt.h>
 
 int __read_mostly alternatives_patched;
 
@@ -733,6 +734,33 @@ void __init alternative_instructions(void)
         * patching.
         */
 
+       /*
+        * Paravirt patching and alternative patching can be combined to
+        * replace a function call with a short direct code sequence (e.g.
+        * by setting a constant return value instead of doing that in an
+        * external function).
+        * In order to make this work the following sequence is required:
+        * 1. set (artificial) features depending on used paravirt
+        *    functions which can later influence alternative patching
+        * 2. apply paravirt patching (generally replacing an indirect
+        *    function call with a direct one)
+        * 3. apply alternative patching (e.g. replacing a direct function
+        *    call with a custom code sequence)
+        * Doing paravirt patching after alternative patching would clobber
+        * the optimization of the custom code with a function call again.
+        */
+       paravirt_set_cap();
+
+       /*
+        * First patch paravirt functions, such that we overwrite the indirect
+        * call with the direct call.
+        */
+       apply_paravirt(__parainstructions, __parainstructions_end);
+
+       /*
+        * Then patch alternatives, such that those paravirt calls that are in
+        * alternatives can be overwritten by their immediate fragments.
+        */
        apply_alternatives(__alt_instructions, __alt_instructions_end);
 
 #ifdef CONFIG_SMP
@@ -751,8 +779,6 @@ void __init alternative_instructions(void)
        }
 #endif
 
-       apply_paravirt(__parainstructions, __parainstructions_end);
-
        restart_nmi();
        alternatives_patched = 1;
 }
diff --git a/arch/x86/kernel/paravirt-spinlocks.c 
b/arch/x86/kernel/paravirt-spinlocks.c
index 4f75d0cf6305..9e1ea99ad9df 100644
--- a/arch/x86/kernel/paravirt-spinlocks.c
+++ b/arch/x86/kernel/paravirt-spinlocks.c
@@ -32,3 +32,12 @@ bool pv_is_native_vcpu_is_preempted(void)
        return pv_ops.lock.vcpu_is_preempted.func ==
                __raw_callee_save___native_vcpu_is_preempted;
 }
+
+void __init paravirt_set_cap(void)
+{
+       if (!pv_is_native_spin_unlock())
+               setup_force_cpu_cap(X86_FEATURE_PVUNLOCK);
+
+       if (!pv_is_native_vcpu_is_preempted())
+               setup_force_cpu_cap(X86_FEATURE_VCPUPREEMPT);
+}
-- 
2.26.2