crypto: x86/cast6 - Fix RBP usage

Using RBP as a temporary register breaks frame pointer convention and
breaks stack traces when unwinding from an interrupt in the crypto code.

Use R15 instead of RBP.  R15 can't be used as the RID1 register because
of x86 instruction encoding limitations.  So use R15 for CTX and RDI for
CTX.  This means that CTX is no longer an implicit function argument.
Instead it needs to be explicitly copied from RDI.
Reported-by: NEric Biggers <ebiggers@google.com>
Reported-by: NPeter Zijlstra <peterz@infradead.org>
Tested-by: NEric Biggers <ebiggers@google.com>
Acked-by: NEric Biggers <ebiggers@google.com>
Signed-off-by: NJosh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: NHerbert Xu <herbert@gondor.apana.org.au>

arch/x86/crypto/cast6-avx-x86_64-asm_64.S

@@ -47,7 +47,7 @@
 /**********************************************************************
   8-way AVX cast6
  **********************************************************************/
-#define CTX %rdi
+#define CTX %r15
 
 #define RA1 %xmm0
 #define RB1 %xmm1
@@ -70,8 +70,8 @@
 
 #define RTMP %xmm15
 
-#define RID1  %rbp
-#define RID1d %ebp
+#define RID1  %rdi
+#define RID1d %edi
 #define RID2  %rsi
 #define RID2d %esi
 
@@ -264,15 +264,17 @@
 .align 8
 __cast6_enc_blk8:
 	/* input:
-	 *	%rdi: ctx, CTX
+	 *	%rdi: ctx
 	 *	RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: blocks
 	 * output:
 	 *	RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: encrypted blocks
 	 */
 
-	pushq %rbp;
+	pushq %r15;
 	pushq %rbx;
 
+	movq %rdi, CTX;
+
 	vmovdqa .Lbswap_mask, RKM;
 	vmovd .Lfirst_mask, R1ST;
 	vmovd .L32_mask, R32;
@@ -297,7 +299,7 @@ __cast6_enc_blk8:
 	QBAR(11);
 
 	popq %rbx;
-	popq %rbp;
+	popq %r15;
 
 	vmovdqa .Lbswap_mask, RKM;
 
@@ -310,15 +312,17 @@ ENDPROC(__cast6_enc_blk8)
 .align 8
 __cast6_dec_blk8:
 	/* input:
-	 *	%rdi: ctx, CTX
+	 *	%rdi: ctx
 	 *	RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: encrypted blocks
 	 * output:
 	 *	RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: decrypted blocks
 	 */
 
-	pushq %rbp;
+	pushq %r15;
 	pushq %rbx;
 
+	movq %rdi, CTX;
+
 	vmovdqa .Lbswap_mask, RKM;
 	vmovd .Lfirst_mask, R1ST;
 	vmovd .L32_mask, R32;
@@ -343,7 +347,7 @@ __cast6_dec_blk8:
 	QBAR(0);
 
 	popq %rbx;
-	popq %rbp;
+	popq %r15;
 
 	vmovdqa .Lbswap_mask, RKM;
 	outunpack_blocks(RA1, RB1, RC1, RD1, RTMP, RX, RKRF, RKM);
@@ -354,12 +358,14 @@ ENDPROC(__cast6_dec_blk8)
 
 ENTRY(cast6_ecb_enc_8way)
 	/* input:
-	 *	%rdi: ctx, CTX
+	 *	%rdi: ctx
 	 *	%rsi: dst
 	 *	%rdx: src
 	 */
 	FRAME_BEGIN
+	pushq %r15;
 
+	movq %rdi, CTX;
 	movq %rsi, %r11;
 
 	load_8way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
@@ -368,18 +374,21 @@ ENTRY(cast6_ecb_enc_8way)
 
 	store_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
 
+	popq %r15;
 	FRAME_END
 	ret;
 ENDPROC(cast6_ecb_enc_8way)
 
 ENTRY(cast6_ecb_dec_8way)
 	/* input:
-	 *	%rdi: ctx, CTX
+	 *	%rdi: ctx
 	 *	%rsi: dst
 	 *	%rdx: src
 	 */
 	FRAME_BEGIN
+	pushq %r15;
 
+	movq %rdi, CTX;
 	movq %rsi, %r11;
 
 	load_8way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
@@ -388,20 +397,22 @@ ENTRY(cast6_ecb_dec_8way)
 
 	store_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
 
+	popq %r15;
 	FRAME_END
 	ret;
 ENDPROC(cast6_ecb_dec_8way)
 
 ENTRY(cast6_cbc_dec_8way)
 	/* input:
-	 *	%rdi: ctx, CTX
+	 *	%rdi: ctx
 	 *	%rsi: dst
 	 *	%rdx: src
 	 */
 	FRAME_BEGIN
-
 	pushq %r12;
+	pushq %r15;
 
+	movq %rdi, CTX;
 	movq %rsi, %r11;
 	movq %rdx, %r12;
 
@@ -411,8 +422,8 @@ ENTRY(cast6_cbc_dec_8way)
 
 	store_cbc_8way(%r12, %r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
 
+	popq %r15;
 	popq %r12;
-
 	FRAME_END
 	ret;
 ENDPROC(cast6_cbc_dec_8way)
@@ -425,9 +436,10 @@ ENTRY(cast6_ctr_8way)
 	 *	%rcx: iv (little endian, 128bit)
 	 */
 	FRAME_BEGIN
-
 	pushq %r12;
+	pushq %r15
 
+	movq %rdi, CTX;
 	movq %rsi, %r11;
 	movq %rdx, %r12;
 
@@ -438,8 +450,8 @@ ENTRY(cast6_ctr_8way)
 
 	store_ctr_8way(%r12, %r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
 
+	popq %r15;
 	popq %r12;
-
 	FRAME_END
 	ret;
 ENDPROC(cast6_ctr_8way)
@@ -452,7 +464,9 @@ ENTRY(cast6_xts_enc_8way)
 	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
 	 */
 	FRAME_BEGIN
+	pushq %r15;
 
+	movq %rdi, CTX
 	movq %rsi, %r11;
 
 	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
@@ -464,6 +478,7 @@ ENTRY(cast6_xts_enc_8way)
 	/* dst <= regs xor IVs(in dst) */
 	store_xts_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
 
+	popq %r15;
 	FRAME_END
 	ret;
 ENDPROC(cast6_xts_enc_8way)
@@ -476,7 +491,9 @@ ENTRY(cast6_xts_dec_8way)
 	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
 	 */
 	FRAME_BEGIN
+	pushq %r15;
 
+	movq %rdi, CTX
 	movq %rsi, %r11;
 
 	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
@@ -488,6 +505,7 @@ ENTRY(cast6_xts_dec_8way)
 	/* dst <= regs xor IVs(in dst) */
 	store_xts_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
 
+	popq %r15;
 	FRAME_END
 	ret;
 ENDPROC(cast6_xts_dec_8way)