powerpc/uaccess: Switch __put_user_size_allowed() to __put_user_asm_goto()

__put_user_asm_goto() provides more flexibility to GCC and avoids using
a local variable to tell if the write succeeded or not.
GCC can then avoid implementing a cmp in the fast path.

See the difference for a small function like the PPC64 version of
save_general_regs() in arch/powerpc/kernel/signal_32.c:

Before the patch (unreachable nop removed):

0000000000000c10 <.save_general_regs>:
     c10:	39 20 00 2c 	li      r9,44
     c14:	39 40 00 00 	li      r10,0
     c18:	7d 29 03 a6 	mtctr   r9
     c1c:	38 c0 00 00 	li      r6,0
     c20:	48 00 00 14 	b       c34 <.save_general_regs+0x24>
     c30:	42 40 00 40 	bdz     c70 <.save_general_regs+0x60>
     c34:	28 2a 00 27 	cmpldi  r10,39
     c38:	7c c8 33 78 	mr      r8,r6
     c3c:	79 47 1f 24 	rldicr  r7,r10,3,60
     c40:	39 20 00 01 	li      r9,1
     c44:	41 82 00 0c 	beq     c50 <.save_general_regs+0x40>
     c48:	7d 23 38 2a 	ldx     r9,r3,r7
     c4c:	79 29 00 20 	clrldi  r9,r9,32
     c50:	91 24 00 00 	stw     r9,0(r4)
     c54:	2c 28 00 00 	cmpdi   r8,0
     c58:	39 4a 00 01 	addi    r10,r10,1
     c5c:	38 84 00 04 	addi    r4,r4,4
     c60:	41 82 ff d0 	beq     c30 <.save_general_regs+0x20>
     c64:	38 60 ff f2 	li      r3,-14
     c68:	4e 80 00 20 	blr
     c70:	38 60 00 00 	li      r3,0
     c74:	4e 80 00 20 	blr

0000000000000000 <.fixup>:
  cc:	39 00 ff f2 	li      r8,-14
  d0:	48 00 00 00 	b       d0 <.fixup+0xd0>
			d0: R_PPC64_REL24	.text+0xc54

After the patch:

0000000000001490 <.save_general_regs>:
    1490:	39 20 00 2c 	li      r9,44
    1494:	39 40 00 00 	li      r10,0
    1498:	7d 29 03 a6 	mtctr   r9
    149c:	60 00 00 00 	nop
    14a0:	28 2a 00 27 	cmpldi  r10,39
    14a4:	79 48 1f 24 	rldicr  r8,r10,3,60
    14a8:	39 20 00 01 	li      r9,1
    14ac:	41 82 00 0c 	beq     14b8 <.save_general_regs+0x28>
    14b0:	7d 23 40 2a 	ldx     r9,r3,r8
    14b4:	79 29 00 20 	clrldi  r9,r9,32
    14b8:	91 24 00 00 	stw     r9,0(r4)
    14bc:	39 4a 00 01 	addi    r10,r10,1
    14c0:	38 84 00 04 	addi    r4,r4,4
    14c4:	42 00 ff dc 	bdnz    14a0 <.save_general_regs+0x10>
    14c8:	38 60 00 00 	li      r3,0
    14cc:	4e 80 00 20 	blr
    14d0:	38 60 ff f2 	li      r3,-14
    14d4:	4e 80 00 20 	blr

Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/94ba5a5138f99522e1562dbcdb38d31aa790dc89.1599216721.git.christophe.leroy@csgroup.eu