x64: avoid stack instructions in context switch code

Due to the "red zone", stack operations may corrupt local variables.
Use ordinary moves and jumps instead.

arch/x64/arch-switch.hh

@@ -24,15 +24,18 @@ void thread::switch_to()
     barrier();
     _cpu->arch.set_exception_stack(&_arch);
     asm volatile
-        ("push %%rbp \n\t"
-         "pushq $1f \n\t"
-         "mov %%rsp, (%0) \n\t"
-         "mov %1, %%rsp \n\t"
-         "ret \n\t"
+        ("mov %%rbp, %c[rbp](%0) \n\t"
+         "movq $1f, %c[rip](%0) \n\t"
+         "mov %%rsp, %c[rsp](%0) \n\t"
+         "mov %c[rsp](%1), %%rsp \n\t"
+         "mov %c[rbp](%1), %%rbp \n\t"
+         "jmpq *%c[rip](%1) \n\t"
          "1: \n\t"
-         "pop %%rbp"
          :
-         : "a"(&old->_state.rsp), "c"(this->_state.rsp)
+         : "a"(&old->_state), "c"(&this->_state),
+           [rsp]"i"(offsetof(thread_state, rsp)),
+           [rbp]"i"(offsetof(thread_state, rbp)),
+           [rip]"i"(offsetof(thread_state, rip))
          : "rbx", "rdx", "rsi", "rdi", "r8", "r9",
            "r10", "r11", "r12", "r13", "r14", "r15", "memory");
 }
@@ -45,10 +48,14 @@ void thread::switch_to_first()
     s_current = this;
     _cpu->arch.set_exception_stack(&_arch);
     asm volatile
-        ("mov %0, %%rsp \n\t"
-         "ret"
+        ("mov %c[rsp](%0), %%rsp \n\t"
+         "mov %c[rbp](%0), %%rbp \n\t"
+         "jmp *%c[rip](%0)"
          :
-         : "c"(this->_state.rsp)
+         : "c"(&this->_state),
+           [rsp]"i"(offsetof(thread_state, rsp)),
+           [rbp]"i"(offsetof(thread_state, rbp)),
+           [rip]"i"(offsetof(thread_state, rip))
          : "rbx", "rdx", "rsi", "rdi", "r8", "r9",
            "r10", "r11", "r12", "r13", "r14", "r15", "memory");
 }
@@ -64,8 +71,8 @@ void thread::init_stack()
         stack.deleter = free;
     }
     void** stacktop = reinterpret_cast<void**>(stack.begin + stack.size);
-    *--stacktop = this;
-    *--stacktop = reinterpret_cast<void*>(thread_main);
+    _state.rbp = this;
+    _state.rip = reinterpret_cast<void*>(thread_main);
     _state.rsp = stacktop;
 }
 

arch/x64/arch-thread-state.hh

@@ -3,6 +3,8 @@
 
 struct thread_state {
     void* rsp;
+    void* rbp;
+    void* rip;
 };
 
 #endif /* ARCH_THREAD_STATE_HH_ */

arch/x64/entry.S

@@ -97,7 +97,7 @@ exception_error_entry interrupt_entry_common, interrupt
 thread_main:
 	.cfi_startproc simple
 	.cfi_def_cfa %rsp, 0
-	pop %rdi
+	mov %rbp, %rdi
 	call thread_main_c
 	.cfi_endproc