selftests/powerpc: Test the preservation of FPU and VMX regs across syscall

Test that the non volatile floating point and Altivec registers get
correctly preserved across the fork() syscall.

fork() works nicely for this purpose, the registers should be the same for
both parent and child

Signed-off-by: Cyril Bur <cyrilbur@gmail.com>
[mpe: Add include guards to basic_asm.h, minor formatting]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

1 files changed, 161 insertions, 0 deletions

diff --git a/tools/testing/selftests/powerpc/math/fpu_asm.S b/tools/testing/selftests/powerpc/math/fpu_asm.S
new file mode 100644
index 000000000000..7617da3117ba
--- /dev/null
+++ b/tools/testing/selftests/powerpc/math/fpu_asm.S
@@ -0,0 +1,161 @@
+/*
+ * Copyright 2015, Cyril Bur, IBM Corp.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include "../basic_asm.h"
+
+#define PUSH_FPU(pos) \
+	stfd	f14,pos(sp); \
+	stfd	f15,pos+8(sp); \
+	stfd	f16,pos+16(sp); \
+	stfd	f17,pos+24(sp); \
+	stfd	f18,pos+32(sp); \
+	stfd	f19,pos+40(sp); \
+	stfd	f20,pos+48(sp); \
+	stfd	f21,pos+56(sp); \
+	stfd	f22,pos+64(sp); \
+	stfd	f23,pos+72(sp); \
+	stfd	f24,pos+80(sp); \
+	stfd	f25,pos+88(sp); \
+	stfd	f26,pos+96(sp); \
+	stfd	f27,pos+104(sp); \
+	stfd	f28,pos+112(sp); \
+	stfd	f29,pos+120(sp); \
+	stfd	f30,pos+128(sp); \
+	stfd	f31,pos+136(sp);
+
+#define POP_FPU(pos) \
+	lfd	f14,pos(sp); \
+	lfd	f15,pos+8(sp); \
+	lfd	f16,pos+16(sp); \
+	lfd	f17,pos+24(sp); \
+	lfd	f18,pos+32(sp); \
+	lfd	f19,pos+40(sp); \
+	lfd	f20,pos+48(sp); \
+	lfd	f21,pos+56(sp); \
+	lfd	f22,pos+64(sp); \
+	lfd	f23,pos+72(sp); \
+	lfd	f24,pos+80(sp); \
+	lfd	f25,pos+88(sp); \
+	lfd	f26,pos+96(sp); \
+	lfd	f27,pos+104(sp); \
+	lfd	f28,pos+112(sp); \
+	lfd	f29,pos+120(sp); \
+	lfd	f30,pos+128(sp); \
+	lfd	f31,pos+136(sp);
+
+# Careful calling this, it will 'clobber' fpu (by design)
+# Don't call this from C
+FUNC_START(load_fpu)
+	lfd	f14,0(r3)
+	lfd	f15,8(r3)
+	lfd	f16,16(r3)
+	lfd	f17,24(r3)
+	lfd	f18,32(r3)
+	lfd	f19,40(r3)
+	lfd	f20,48(r3)
+	lfd	f21,56(r3)
+	lfd	f22,64(r3)
+	lfd	f23,72(r3)
+	lfd	f24,80(r3)
+	lfd	f25,88(r3)
+	lfd	f26,96(r3)
+	lfd	f27,104(r3)
+	lfd	f28,112(r3)
+	lfd	f29,120(r3)
+	lfd	f30,128(r3)
+	lfd	f31,136(r3)
+	blr
+FUNC_END(load_fpu)
+
+FUNC_START(check_fpu)
+	mr r4,r3
+	li	r3,1 # assume a bad result
+	lfd	f0,0(r4)
+	fcmpu	cr1,f0,f14
+	bne	cr1,1f
+	lfd	f0,8(r4)
+	fcmpu	cr1,f0,f15
+	bne	cr1,1f
+	lfd	f0,16(r4)
+	fcmpu	cr1,f0,f16
+	bne	cr1,1f
+	lfd	f0,24(r4)
+	fcmpu	cr1,f0,f17
+	bne	cr1,1f
+	lfd	f0,32(r4)
+	fcmpu	cr1,f0,f18
+	bne	cr1,1f
+	lfd	f0,40(r4)
+	fcmpu	cr1,f0,f19
+	bne	cr1,1f
+	lfd	f0,48(r4)
+	fcmpu	cr1,f0,f20
+	bne	cr1,1f
+	lfd	f0,56(r4)
+	fcmpu	cr1,f0,f21
+	bne	cr1,1f
+	lfd	f0,64(r4)
+	fcmpu	cr1,f0,f22
+	bne	cr1,1f
+	lfd	f0,72(r4)
+	fcmpu	cr1,f0,f23
+	bne	cr1,1f
+	lfd	f0,80(r4)
+	fcmpu	cr1,f0,f24
+	bne	cr1,1f
+	lfd	f0,88(r4)
+	fcmpu	cr1,f0,f25
+	bne	cr1,1f
+	lfd	f0,96(r4)
+	fcmpu	cr1,f0,f26
+	bne	cr1,1f
+	lfd	f0,104(r4)
+	fcmpu	cr1,f0,f27
+	bne	cr1,1f
+	lfd	f0,112(r4)
+	fcmpu	cr1,f0,f28
+	bne	cr1,1f
+	lfd	f0,120(r4)
+	fcmpu	cr1,f0,f29
+	bne	cr1,1f
+	lfd	f0,128(r4)
+	fcmpu	cr1,f0,f30
+	bne	cr1,1f
+	lfd	f0,136(r4)
+	fcmpu	cr1,f0,f31
+	bne	cr1,1f
+	li	r3,0 # Success!!!
+1:	blr
+
+FUNC_START(test_fpu)
+	# r3 holds pointer to where to put the result of fork
+	# r4 holds pointer to the pid
+	# f14-f31 are non volatiles
+	PUSH_BASIC_STACK(256)
+	std	r3,STACK_FRAME_PARAM(0)(sp) # Address of darray
+	std r4,STACK_FRAME_PARAM(1)(sp) # Address of pid
+	PUSH_FPU(STACK_FRAME_LOCAL(2,0))
+
+	bl load_fpu
+	nop
+	li	r0,__NR_fork
+	sc
+
+	# pass the result of the fork to the caller
+	ld	r9,STACK_FRAME_PARAM(1)(sp)
+	std	r3,0(r9)
+
+	ld r3,STACK_FRAME_PARAM(0)(sp)
+	bl check_fpu
+	nop
+
+	POP_FPU(STACK_FRAME_LOCAL(2,0))
+	POP_BASIC_STACK(256)
+	blr
+FUNC_END(test_fpu)