xen/x86: use flag byte for decision whether xen_cr3 is valid

Today cpu_info->xen_cr3 is either 0 to indicate %cr3 doesn't need to
be switched on entry to Xen, or negative for keeping the value while
indicating not to restore %cr3, or positive in case %cr3 is to be
restored.

Switch to use a flag byte instead of a negative xen_cr3 value in order
to allow %cr3 values with the high bit set in case we want to keep TLB
entries when using the PCID feature.

This reduces the number of branches in interrupt handling and results
in better performance (e.g. parallel make of the Xen hypervisor on my
system was using about 3% less system time).

Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>

xen/arch/x86/domain.c

@@ -1696,6 +1696,7 @@ void context_switch(struct vcpu *prev, struct vcpu *next)
 
     ASSERT(local_irq_is_enabled());
 
+    get_cpu_info()->use_pv_cr3 = false;
     get_cpu_info()->xen_cr3 = 0;
 
     if ( unlikely(dirty_cpu != cpu) && dirty_cpu != VCPU_CPU_CLEAN )

xen/arch/x86/mm.c

@@ -517,7 +517,8 @@ void write_ptbase(struct vcpu *v)
     }
     else
     {
-        /* Make sure to clear xen_cr3 before pv_cr3. */
+        /* Make sure to clear use_pv_cr3 and xen_cr3 before pv_cr3. */
+        cpu_info->use_pv_cr3 = false;
         cpu_info->xen_cr3 = 0;
         /* switch_cr3_cr4() serializes. */
         switch_cr3_cr4(v->arch.cr3, new_cr4);

xen/arch/x86/smpboot.c

@@ -336,6 +336,7 @@ void start_secondary(void *unused)
      */
     spin_debug_disable();
 
+    get_cpu_info()->use_pv_cr3 = false;
     get_cpu_info()->xen_cr3 = 0;
     get_cpu_info()->pv_cr3 = 0;
 
@@ -1135,6 +1136,7 @@ void __init smp_prepare_boot_cpu(void)
     per_cpu(scratch_cpumask, cpu) = &scratch_cpu0mask;
 #endif
 
+    get_cpu_info()->use_pv_cr3 = false;
     get_cpu_info()->xen_cr3 = 0;
     get_cpu_info()->pv_cr3 = 0;
 }

xen/arch/x86/x86_64/asm-offsets.c

@@ -144,6 +144,7 @@ void __dummy__(void)
     OFFSET(CPUINFO_use_shadow_spec_ctrl, struct cpu_info, use_shadow_spec_ctrl);
     OFFSET(CPUINFO_bti_ist_info, struct cpu_info, bti_ist_info);
     OFFSET(CPUINFO_root_pgt_changed, struct cpu_info, root_pgt_changed);
+    OFFSET(CPUINFO_use_pv_cr3, struct cpu_info, use_pv_cr3);
     DEFINE(CPUINFO_sizeof, sizeof(struct cpu_info));
     BLANK();
 

xen/arch/x86/x86_64/compat/entry.S

@@ -210,10 +210,9 @@ ENTRY(cstar_enter)
 
         GET_STACK_END(bx)
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%rbx), %rcx
-        neg   %rcx
+        test  %rcx, %rcx
         jz    .Lcstar_cr3_okay
-        mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
-        neg   %rcx
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%rbx)
         mov   %rcx, %cr3
         movq  $0, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
 .Lcstar_cr3_okay:

xen/arch/x86/x86_64/entry.S

@@ -154,6 +154,7 @@ restore_all_guest:
         rep movsq
 .Lrag_copy_done:
         mov   %r9, STACK_CPUINFO_FIELD(xen_cr3)(%rdx)
+        movb  $1, STACK_CPUINFO_FIELD(use_pv_cr3)(%rdx)
         mov   %rax, %cr3
 .Lrag_keep_cr3:
 
@@ -202,14 +203,9 @@ restore_all_xen:
          * case we return to late PV exit code (from an NMI or #MC).
          */
         GET_STACK_END(bx)
-        mov   STACK_CPUINFO_FIELD(xen_cr3)(%rbx), %rdx
+        cmpb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%rbx)
+UNLIKELY_START(ne, exit_cr3)
         mov   STACK_CPUINFO_FIELD(pv_cr3)(%rbx), %rax
-        test  %rdx, %rdx
-        /*
-         * Ideally the condition would be "nsz", but such doesn't exist,
-         * so "g" will have to do.
-         */
-UNLIKELY_START(g, exit_cr3)
         mov   %rax, %cr3
 UNLIKELY_END(exit_cr3)
 
@@ -251,10 +247,9 @@ ENTRY(lstar_enter)
 
         GET_STACK_END(bx)
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%rbx), %rcx
-        neg   %rcx
+        test  %rcx, %rcx
         jz    .Llstar_cr3_okay
-        mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
-        neg   %rcx
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%rbx)
         mov   %rcx, %cr3
         movq  $0, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
 .Llstar_cr3_okay:
@@ -288,10 +283,9 @@ GLOBAL(sysenter_eflags_saved)
         /* PUSHF above has saved EFLAGS.IF clear (the caller had it set). */
         orl   $X86_EFLAGS_IF, UREGS_eflags(%rsp)
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%rbx), %rcx
-        neg   %rcx
+        test  %rcx, %rcx
         jz    .Lsyse_cr3_okay
-        mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
-        neg   %rcx
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%rbx)
         mov   %rcx, %cr3
         movq  $0, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
 .Lsyse_cr3_okay:
@@ -338,10 +332,9 @@ ENTRY(int80_direct_trap)
 
         GET_STACK_END(bx)
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%rbx), %rcx
-        neg   %rcx
+        test  %rcx, %rcx
         jz    .Lint80_cr3_okay
-        mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
-        neg   %rcx
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%rbx)
         mov   %rcx, %cr3
         movq  $0, STACK_CPUINFO_FIELD(xen_cr3)(%rbx)
 .Lint80_cr3_okay:
@@ -552,24 +545,24 @@ ENTRY(common_interrupt)
         /* WARNING! `ret`, `call *`, `jmp *` not safe before this point. */
 
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%r14), %rcx
+        mov   STACK_CPUINFO_FIELD(use_pv_cr3)(%r14), %bl
         mov   %rcx, %r15
-        neg   %rcx
+        test  %rcx, %rcx
         jz    .Lintr_cr3_okay
-        jns   .Lintr_cr3_load
-        mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
-        neg   %rcx
-.Lintr_cr3_load:
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         mov   %rcx, %cr3
         xor   %ecx, %ecx
         mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
         testb $3, UREGS_cs(%rsp)
         cmovnz %rcx, %r15
+        cmovnz %rcx, %rbx
 .Lintr_cr3_okay:
 
         CR4_PV32_RESTORE
         movq %rsp,%rdi
         callq do_IRQ
         mov   %r15, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
+        mov   %bl, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         jmp ret_from_intr
 
 ENTRY(page_fault)
@@ -584,18 +577,17 @@ GLOBAL(handle_exception)
         /* WARNING! `ret`, `call *`, `jmp *` not safe before this point. */
 
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%r14), %rcx
+        mov   STACK_CPUINFO_FIELD(use_pv_cr3)(%r14), %r13b
         mov   %rcx, %r15
-        neg   %rcx
+        test  %rcx, %rcx
         jz    .Lxcpt_cr3_okay
-        jns   .Lxcpt_cr3_load
-        mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
-        neg   %rcx
-.Lxcpt_cr3_load:
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         mov   %rcx, %cr3
         xor   %ecx, %ecx
         mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
         testb $3, UREGS_cs(%rsp)
         cmovnz %rcx, %r15
+        cmovnz %rcx, %r13
 .Lxcpt_cr3_okay:
 
 handle_exception_saved:
@@ -650,6 +642,7 @@ handle_exception_saved:
         mov   (%rdx, %rax, 8), %rdx
         INDIRECT_CALL %rdx
         mov   %r15, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
+        mov   %r13b, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         testb $3,UREGS_cs(%rsp)
         jz    restore_all_xen
         leaq  VCPU_trap_bounce(%rbx),%rdx
@@ -683,6 +676,7 @@ exception_with_ints_disabled:
 1:      movq  UREGS_error_code(%rsp),%rax # ec/ev
         movq  %rax,UREGS_kernel_sizeof(%rsp)
         mov   %r15, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
+        mov   %r13b, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         jmp   restore_all_xen           # return to fixup code
 
 /* No special register assumptions. */
@@ -768,12 +762,9 @@ ENTRY(double_fault)
         /* WARNING! `ret`, `call *`, `jmp *` not safe before this point. */
 
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%r14), %rbx
-        neg   %rbx
+        test  %rbx, %rbx
         jz    .Ldblf_cr3_okay
-        jns   .Ldblf_cr3_load
-        mov   %rbx, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
-        neg   %rbx
-.Ldblf_cr3_load:
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         mov   %rbx, %cr3
         movq $0, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
 .Ldblf_cr3_okay:
@@ -803,13 +794,11 @@ handle_ist_exception:
         /* WARNING! `ret`, `call *`, `jmp *` not safe before this point. */
 
         mov   STACK_CPUINFO_FIELD(xen_cr3)(%r14), %rcx
+        mov   STACK_CPUINFO_FIELD(use_pv_cr3)(%r14), %bl
         mov   %rcx, %r15
-        neg   %rcx
+        test  %rcx, %rcx
         jz    .List_cr3_okay
-        jns   .List_cr3_load
-        mov   %rcx, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
-        neg   %rcx
-.List_cr3_load:
+        movb  $0, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         mov   %rcx, %cr3
         movq  $0, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
 .List_cr3_okay:
@@ -822,6 +811,7 @@ handle_ist_exception:
          * and copy the context to stack bottom.
          */
         xor   %r15, %r15
+        xor   %ebx, %ebx
         GET_CPUINFO_FIELD(guest_cpu_user_regs,di)
         movq  %rsp,%rsi
         movl  $UREGS_kernel_sizeof/8,%ecx
@@ -833,6 +823,7 @@ handle_ist_exception:
         mov   (%rdx, %rax, 8), %rdx
         INDIRECT_CALL %rdx
         mov   %r15, STACK_CPUINFO_FIELD(xen_cr3)(%r14)
+        mov   %bl, STACK_CPUINFO_FIELD(use_pv_cr3)(%r14)
         cmpb  $TRAP_nmi,UREGS_entry_vector(%rsp)
         jne   ret_from_intr
 

xen/include/asm-x86/current.h

@@ -47,10 +47,7 @@ struct cpu_info {
      * context is being entered. A value of zero indicates no setting of CR3
      * is to be performed.
      * The former is the value to restore when re-entering Xen, if any. IOW
-     * its value being zero means there's nothing to restore. However, its
-     * value can also be negative, indicating to the exit-to-Xen code that
-     * restoring is not necessary, but allowing any nested entry code paths
-     * to still know the value to put back into CR3.
+     * its value being zero means there's nothing to restore.
      */
     unsigned long xen_cr3;
     unsigned long pv_cr3;
@@ -68,6 +65,13 @@ struct cpu_info {
      */
     bool         root_pgt_changed;
 
+    /*
+     * use_pv_cr3 is set in case the value of pv_cr3 is to be written into
+     * CR3 when returning from an interrupt. The main use is when returning
+     * from a NMI or MCE to hypervisor code where pv_cr3 was active.
+     */
+    bool         use_pv_cr3;
+
     unsigned long __pad;
     /* get_stack_bottom() must be 16-byte aligned */
 };