x86/fpu/math-emu: Add support for F[U]COMI[P] insns

Run-tested by booting with "no387 nofxsr" and running test
program:

  [RUN]   Testing f[u]comi[p] instructions

  [OK]    f[u]comi[p]

Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/1442588010-20055-2-git-send-email-dvlasenk@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

arch/x86/math-emu/fpu_entry.c

@@ -40,6 +40,8 @@
 
 #define __BAD__ FPU_illegal	/* Illegal on an 80486, causes SIGILL */
 
+/* f(u)comi(p) are enabled if CPUID(1).EDX(15) "cmov" is set */
+
 /* WARNING: "u" entries are not documented by Intel in their 80486 manual
    and may not work on FPU clones or later Intel FPUs.
    Changes to support them provided by Linus Torvalds. */
@@ -57,10 +59,10 @@ static FUNC const st_instr_table[64] = {
 /* d8..f */	fcompst,/*u*/	fstp_i,		fcompp,		fstp_i,/*u*/
 /* e0..7 */	fsub__,		FPU_etc,	__BAD__,	finit_,
 /* e0..7 */	fsubri,		fucom_,		fsubrp,		fstsw_,
-/* e8..f */	fsubr_,		fconst,		fucompp,	__BAD__,
-/* e8..f */	fsub_i,		fucomp,		fsubp_,		__BAD__,
-/* f0..7 */	fdiv__,		FPU_triga,	__BAD__,	__BAD__,
-/* f0..7 */	fdivri,		__BAD__,	fdivrp,		__BAD__,
+/* e8..f */	fsubr_,		fconst,		fucompp,	fucomi_,
+/* e8..f */	fsub_i,		fucomp,		fsubp_,		fucomip,
+/* f0..7 */	fdiv__,		FPU_triga,	__BAD__,	fcomi_,
+/* f0..7 */	fdivri,		__BAD__,	fdivrp,		fcomip,
 /* f8..f */	fdivr_,		FPU_trigb,	__BAD__,	__BAD__,
 /* f8..f */	fdiv_i,		__BAD__,	fdivp_,		__BAD__,
 };
@@ -77,14 +79,15 @@ static FUNC const st_instr_table[64] = {
 #define _REGIn 0		/* Uses st(0) and st(rm), but handle checks later */
 
 static u_char const type_table[64] = {
-	_REGI_, _NONE_, _null_, _null_, _REGIi, _REGi_, _REGIp, _REGi_,
-	_REGI_, _REGIn, _null_, _null_, _REGIi, _REGI_, _REGIp, _REGI_,
-	_REGIc, _NONE_, _null_, _null_, _REGIc, _REG0_, _REGIc, _REG0_,
-	_REGIc, _REG0_, _null_, _null_, _REGIc, _REG0_, _REGIc, _REG0_,
-	_REGI_, _NONE_, _null_, _NONE_, _REGIi, _REGIc, _REGIp, _NONE_,
-	_REGI_, _NONE_, _REGIc, _null_, _REGIi, _REGIc, _REGIp, _null_,
-	_REGI_, _NONE_, _null_, _null_, _REGIi, _null_, _REGIp, _null_,
-	_REGI_, _NONE_, _null_, _null_, _REGIi, _null_, _REGIp, _null_
+/* Opcode:	d8	d9	da	db	dc	dd	de	df */
+/* c0..7 */	_REGI_, _NONE_, _null_, _null_, _REGIi, _REGi_, _REGIp, _REGi_,
+/* c8..f */	_REGI_, _REGIn, _null_, _null_, _REGIi, _REGI_, _REGIp, _REGI_,
+/* d0..7 */	_REGIc, _NONE_, _null_, _null_, _REGIc, _REG0_, _REGIc, _REG0_,
+/* d8..f */	_REGIc, _REG0_, _null_, _null_, _REGIc, _REG0_, _REGIc, _REG0_,
+/* e0..7 */	_REGI_, _NONE_, _null_, _NONE_, _REGIi, _REGIc, _REGIp, _NONE_,
+/* e8..f */	_REGI_, _NONE_, _REGIc, _REGIc, _REGIi, _REGIc, _REGIp, _REGIc,
+/* f0..7 */	_REGI_, _NONE_, _null_, _REGIc, _REGIi, _null_, _REGIp, _REGIc,
+/* f8..f */	_REGI_, _NONE_, _null_, _null_, _REGIi, _null_, _REGIp, _null_,
 };
 
 #ifdef RE_ENTRANT_CHECKING

arch/x86/math-emu/fpu_proto.h

@@ -108,6 +108,10 @@ extern void fcompp(void);
 extern void fucom_(void);
 extern void fucomp(void);
 extern void fucompp(void);
+extern void fcomi_(void);
+extern void fcomip(void);
+extern void fucomi_(void);
+extern void fucomip(void);
 /* reg_constant.c */
 extern void fconst(void);
 /* reg_ld_str.c */

arch/x86/math-emu/reg_compare.c

@@ -249,6 +249,54 @@ static int compare_st_st(int nr)
 	return 0;
 }
 
+static int compare_i_st_st(int nr)
+{
+	int f, c;
+	FPU_REG *st_ptr;
+
+	if (!NOT_EMPTY(0) || !NOT_EMPTY(nr)) {
+		FPU_EFLAGS |= (X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF);
+		/* Stack fault */
+		EXCEPTION(EX_StackUnder);
+		return !(control_word & CW_Invalid);
+	}
+
+	partial_status &= ~SW_C0;
+	st_ptr = &st(nr);
+	c = compare(st_ptr, FPU_gettagi(nr));
+	if (c & COMP_NaN) {
+		FPU_EFLAGS |= (X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF);
+		EXCEPTION(EX_Invalid);
+		return !(control_word & CW_Invalid);
+	}
+
+	switch (c & 7) {
+	case COMP_A_lt_B:
+		f = X86_EFLAGS_CF;
+		break;
+	case COMP_A_eq_B:
+		f = X86_EFLAGS_ZF;
+		break;
+	case COMP_A_gt_B:
+		f = 0;
+		break;
+	case COMP_No_Comp:
+		f = X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF;
+		break;
+#ifdef PARANOID
+	default:
+		EXCEPTION(EX_INTERNAL | 0x122);
+		f = 0;
+		break;
+#endif /* PARANOID */
+	}
+	FPU_EFLAGS = (FPU_EFLAGS & ~(X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF)) | f;
+	if (c & COMP_Denormal) {
+		return denormal_operand() < 0;
+	}
+	return 0;
+}
+
 static int compare_u_st_st(int nr)
 {
 	int f = 0, c;
@@ -299,6 +347,58 @@ static int compare_u_st_st(int nr)
 	return 0;
 }
 
+static int compare_ui_st_st(int nr)
+{
+	int f = 0, c;
+	FPU_REG *st_ptr;
+
+	if (!NOT_EMPTY(0) || !NOT_EMPTY(nr)) {
+		FPU_EFLAGS |= (X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF);
+		/* Stack fault */
+		EXCEPTION(EX_StackUnder);
+		return !(control_word & CW_Invalid);
+	}
+
+	partial_status &= ~SW_C0;
+	st_ptr = &st(nr);
+	c = compare(st_ptr, FPU_gettagi(nr));
+	if (c & COMP_NaN) {
+		FPU_EFLAGS |= (X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF);
+		if (c & COMP_SNaN) {	/* This is the only difference between
+					   un-ordered and ordinary comparisons */
+			EXCEPTION(EX_Invalid);
+			return !(control_word & CW_Invalid);
+		}
+		return 0;
+	}
+
+	switch (c & 7) {
+	case COMP_A_lt_B:
+		f = X86_EFLAGS_CF;
+		break;
+	case COMP_A_eq_B:
+		f = X86_EFLAGS_ZF;
+		break;
+	case COMP_A_gt_B:
+		f = 0;
+		break;
+	case COMP_No_Comp:
+		f = X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF;
+		break;
+#ifdef PARANOID
+	default:
+		EXCEPTION(EX_INTERNAL | 0x123);
+		f = 0;
+		break;
+#endif /* PARANOID */
+	}
+	FPU_EFLAGS = (FPU_EFLAGS & ~(X86_EFLAGS_ZF | X86_EFLAGS_PF | X86_EFLAGS_CF)) | f;
+	if (c & COMP_Denormal) {
+		return denormal_operand() < 0;
+	}
+	return 0;
+}
+
 /*---------------------------------------------------------------------------*/
 
 void fcom_st(void)
@@ -348,3 +448,31 @@ void fucompp(void)
 	} else
 		FPU_illegal();
 }
+
+/* P6+ compare-to-EFLAGS ops */
+
+void fcomi_(void)
+{
+	/* fcomi st(i) */
+	compare_i_st_st(FPU_rm);
+}
+
+void fcomip(void)
+{
+	/* fcomip st(i) */
+	if (!compare_i_st_st(FPU_rm))
+		FPU_pop();
+}
+
+void fucomi_(void)
+{
+	/* fucomi st(i) */
+	compare_ui_st_st(FPU_rm);
+}
+
+void fucomip(void)
+{
+	/* fucomip st(i) */
+	if (!compare_ui_st_st(FPU_rm))
+		FPU_pop();
+}