patch-2.1.80 linux/arch/arm/lib/checksum.S

Next file: linux/arch/arm/lib/delay.S
Previous file: linux/arch/arm/lib/bitops.S
Back to the patch index
Back to the overall index

diff -u --recursive --new-file v2.1.79/linux/arch/arm/lib/checksum.S linux/arch/arm/lib/checksum.S
@@ -0,0 +1,600 @@
+/*
+ * linux/arch/arm/lib/iputils.S
+ *
+ * Copyright (C) 1995, 1996, 1997, 1998 Russell King
+ */
+#include <linux/linkage.h>
+#include <asm/assembler.h>
+#include <asm/errno.h>
+
+		.text
+
+/* Function: __u32 csum_partial(const char *src, int len, __u32)
+ * Params  : r0 = buffer, r1 = len, r2 = checksum
+ * Returns : r0 = new checksum
+ */
+
+ENTRY(csum_partial)
+		tst	r0, #2
+		beq	1f
+		subs	r1, r1, #2
+		addmi	r1, r1, #2
+		bmi	3f
+		bic	r0, r0, #3
+		ldr	r3, [r0], #4
+		adds	r2, r2, r3, lsr #16
+		adcs	r2, r2, #0
+1:		adds	r2, r2, #0
+		bics	ip, r1, #31
+		beq	3f
+		stmfd	sp!, {r4 - r6}
+2:		ldmia	r0!, {r3 - r6}
+		adcs	r2, r2, r3
+		adcs	r2, r2, r4
+		adcs	r2, r2, r5
+		adcs	r2, r2, r6
+		ldmia	r0!, {r3 - r6}
+		adcs	r2, r2, r3
+		adcs	r2, r2, r4
+		adcs	r2, r2, r5
+		adcs	r2, r2, r6
+		sub	ip, ip, #32
+		teq	ip, #0
+		bne	2b
+		adcs	r2, r2, #0
+		ldmfd	sp!, {r4 - r6}
+3:		ands	ip, r1, #0x1c
+		beq	5f
+4:		ldr	r3, [r0], #4
+		adcs	r2, r2, r3
+		sub	ip, ip, #4
+		teq	ip, #0
+		bne	4b
+		adcs	r2, r2, #0
+5:		ands	ip, r1, #3
+		moveq	r0, r2
+		RETINSTR(moveq,pc,lr)
+		mov	ip, ip, lsl #3
+		rsb	ip, ip, #32
+		ldr	r3, [r0]
+		mov	r3, r3, lsl ip
+		adds	r2, r2, r3, lsr ip
+		adc	r0, r2, #0
+		RETINSTR(mov,pc,lr)
+
+/* Function: __u32 csum_partial_copy_from_user (const char *src, char *dst, int len, __u32 sum, int *err_ptr)
+ * Params  : r0 = src, r1 = dst, r2 = len, r3 = sum, [sp, #0] = &err
+ * Returns : r0 = checksum, [[sp, #0], #0] = 0 or -EFAULT
+ */
+
+#define USER_LDR(instr...)			\
+9999:		instr;				\
+		.section __ex_table, "a";	\
+		.align	3;			\
+		.long	9999b, 6001f;		\
+		.previous;
+
+ENTRY(csum_partial_copy_from_user)
+		mov	ip, sp
+		stmfd	sp!, {r4 - r8, fp, ip, lr, pc}
+		sub	fp, ip, #4
+		cmp	r2, #4
+		blt	.too_small_user
+		tst	r1, #2			@ Test destination alignment
+		beq	.dst_aligned_user
+		subs	r2, r2, #2		@ We dont know if SRC is aligned...
+USER_LDR(	ldrbt	ip, [r0], #1)
+USER_LDR(	ldrbt	r8, [r0], #1)
+		orr	ip, ip, r8, lsl #8
+		adds	r3, r3, ip
+		adcs	r3, r3, #0
+		strb	ip, [r1], #1
+		mov	ip, ip, lsr #8
+		strb	ip, [r1], #1		@ Destination now aligned
+.dst_aligned_user:
+		tst	r0, #3
+		bne	.src_not_aligned_user
+		adds	r3, r3, #0
+		bics	ip, r2, #15		@ Routine for src & dst aligned
+		beq	2f
+1:
+USER_LDR(	ldrt	r4, [r0], #4)
+USER_LDR(	ldrt	r5, [r0], #4)
+USER_LDR(	ldrt	r6, [r0], #4)
+USER_LDR(	ldrt	r7, [r0], #4)
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+2:		ands	ip, r2, #12
+		beq	4f
+		tst	ip, #8
+		beq	3f
+USER_LDR(	ldrt	r4, [r0], #4)
+USER_LDR(	ldrt	r5, [r0], #4)
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		tst	ip, #4
+		beq	4f
+3:
+USER_LDR(	ldrt	r4, [r0], #4)
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+4:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+USER_LDR(	ldrt	r4, [r0], #4)
+		tst	r2, #2
+		beq	.exit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+.exit:		tst	r2, #1
+		strneb	r4, [r1], #1
+		andne	r4, r4, #255
+		adcnes	r3, r3, r4
+		adcs	r0, r3, #0
+		LOADREGS(ea,fp,{r4 - r8, fp, sp, pc})
+
+.too_small_user:
+		teq	r2, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		cmp	r2, #2
+		blt	.too_small_user1
+USER_LDR(	ldrbt	ip, [r0], #1)
+USER_LDR(	ldrbt	r8, [r0], #1)
+		orr	ip, ip, r8, lsl #8
+		adds	r3, r3, ip
+		strb	ip, [r1], #1
+		strb	r8, [r1], #1
+		tst	r2, #1
+.too_small_user1:
+USER_LDR(	ldrnebt	ip, [r0], #1)
+		strneb	ip, [r1], #1
+		adcnes	r3, r3, ip
+		adcs	r0, r3, #0
+		LOADREGS(ea,fp,{r4 - r8, fp, sp, pc})
+
+.src_not_aligned_user:
+		cmp	r2, #4
+		blt	.too_small_user
+		and	ip, r0, #3
+		bic	r0, r0, #3
+USER_LDR(	ldrt	r4, [r0], #4)
+		cmp	ip, #2
+		beq	.src2_aligned_user
+		bhi	.src3_aligned_user
+		mov	r4, r4, lsr #8
+		adds	r3, r3, #0
+		bics	ip, r2, #15
+		beq	2f
+1:
+USER_LDR(	ldrt	r5, [r0], #4)
+USER_LDR(	ldrt	r6, [r0], #4)
+USER_LDR(	ldrt	r7, [r0], #4)
+USER_LDR(	ldrt	r8, [r0], #4)
+		orr	r4, r4, r5, lsl #24
+		mov	r5, r5, lsr #8
+		orr	r5, r5, r6, lsl #24
+		mov	r6, r6, lsr #8
+		orr	r6, r6, r7, lsl #24
+		mov	r7, r7, lsr #8
+		orr	r7, r7, r8, lsl #24
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		mov	r4, r8, lsr #8
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+2:		ands	ip, r2, #12
+		beq	4f
+		tst	ip, #8
+		beq	3f
+USER_LDR(	ldrt	r5, [r0], #4)
+USER_LDR(	ldrt	r6, [r0], #4)
+		orr	r4, r4, r5, lsl #24
+		mov	r5, r5, lsr #8
+		orr	r5, r5, r6, lsl #24
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		mov	r4, r6, lsr #8
+		tst	ip, #4
+		beq	4f
+3:
+USER_LDR(	ldrt	r5, [r0], #4)
+		orr	r4, r4, r5, lsl #24
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+		mov	r4, r5, lsr #8
+4:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		tst	r2, #2
+		beq	.exit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		b	.exit
+
+.src2_aligned_user:
+		mov	r4, r4, lsr #16
+		adds	r3, r3, #0
+		bics	ip, r2, #15
+		beq	2f
+1:
+USER_LDR(	ldrt	r5, [r0], #4)
+USER_LDR(	ldrt	r6, [r0], #4)
+USER_LDR(	ldrt	r7, [r0], #4)
+USER_LDR(	ldrt	r8, [r0], #4)
+		orr	r4, r4, r5, lsl #16
+		mov	r5, r5, lsr #16
+		orr	r5, r5, r6, lsl #16
+		mov	r6, r6, lsr #16
+		orr	r6, r6, r7, lsl #16
+		mov	r7, r7, lsr #16
+		orr	r7, r7, r8, lsl #16
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		mov	r4, r8, lsr #16
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+2:		ands	ip, r2, #12
+		beq	4f
+		tst	ip, #8
+		beq	3f
+USER_LDR(	ldrt	r5, [r0], #4)
+USER_LDR(	ldrt	r6, [r0], #4)
+		orr	r4, r4, r5, lsl #16
+		mov	r5, r5, lsr #16
+		orr	r5, r5, r6, lsl #16
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		mov	r4, r6, lsr #16
+		tst	ip, #4
+		beq	4f
+3:
+USER_LDR(	ldrt	r5, [r0], #4)
+		orr	r4, r4, r5, lsl #16
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+		mov	r4, r5, lsr #16
+4:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		tst	r2, #2
+		beq	.exit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		strb	r4, [r1], #1
+USER_LDR(	ldrb	r4, [r0], #1)
+		b	.exit
+
+.src3_aligned_user:
+		mov	r4, r4, lsr #24
+		adds	r3, r3, #0
+		bics	ip, r2, #15
+		beq	2f
+1:
+USER_LDR(	ldrt	r5, [r0], #4)
+USER_LDR(	ldrt	r6, [r0], #4)
+USER_LDR(	ldrt	r7, [r0], #4)
+USER_LDR(	ldrt	r8, [r0], #4)
+		orr	r4, r4, r5, lsl #8
+		mov	r5, r5, lsr #24
+		orr	r5, r5, r6, lsl #8
+		mov	r6, r6, lsr #24
+		orr	r6, r6, r7, lsl #8
+		mov	r7, r7, lsr #24
+		orr	r7, r7, r8, lsl #8
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		mov	r4, r8, lsr #24
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+2:		ands	ip, r2, #12
+		beq	4f
+		tst	ip, #8
+		beq	3f
+USER_LDR(	ldrt	r5, [r0], #4)
+USER_LDR(	ldrt	r6, [r0], #4)
+		orr	r4, r4, r5, lsl #8
+		mov	r5, r5, lsr #24
+		orr	r5, r5, r6, lsl #8
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		mov	r4, r6, lsr #24
+		tst	ip, #4
+		beq	4f
+3:
+USER_LDR(	ldrt	r5, [r0], #4)
+		orr	r4, r4, r5, lsl #8
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+		mov	r4, r5, lsr #24
+4:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		tst	r2, #2
+		beq	.exit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+USER_LDR(	ldrt	r4, [r0], #4)
+		strb	r4, [r1], #1
+		adcs	r3, r3, r4, lsl #24
+		mov	r4, r4, lsr #8
+		b	.exit
+
+		.section .fixup,"ax"
+		.align	4
+6001:		mov	r4, #-EFAULT
+		ldr	r5, [sp, #4*8]
+		str	r4, [r5]
+		LOADREGS(ea,fp,{r4 - r8, fp, sp, pc})
+
+/* Function: __u32 csum_partial_copy (const char *src, char *dst, int len, __u32 sum)
+ * Params  : r0 = src, r1 = dst, r2 = len, r3 = checksum
+ * Returns : r0 = new checksum
+ */
+ENTRY(csum_partial_copy)
+		mov	ip, sp
+		stmfd	sp!, {r4 - r8, fp, ip, lr, pc}
+		sub	fp, ip, #4
+		cmp	r2, #4
+		blt	Ltoo_small
+		tst	r1, #2			@ Test destination alignment
+		beq	Ldst_aligned
+		subs	r2, r2, #2		@ We dont know if SRC is aligned...
+		ldrb	ip, [r0], #1
+		ldrb	r8, [r0], #1
+		orr	ip, ip, r8, lsl #8
+		adds	r3, r3, ip
+		adcs	r3, r3, #0
+		strb	ip, [r1], #1
+		mov	ip, ip, lsr #8
+		strb	ip, [r1], #1		@ Destination now aligned
+Ldst_aligned:	tst	r0, #3
+		bne	Lsrc_not_aligned
+		adds	r3, r3, #0
+		bics	ip, r2, #15		@ Routine for src & dst aligned
+		beq	3f
+1:		ldmia	r0!, {r4, r5, r6, r7}
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+3:		ands	ip, r2, #12
+		beq	5f
+		tst	ip, #8
+		beq	4f
+		ldmia	r0!, {r4, r5}
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		tst	ip, #4
+		beq	5f
+4:		ldr	r4, [r0], #4
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+5:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		ldr	r4, [r0], #4
+		tst	r2, #2
+		beq	Lexit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		b	Lexit
+
+Ltoo_small:	teq	r2, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		cmp	r2, #2
+		blt	Ltoo_small1
+		ldrb	ip, [r0], #1
+		ldrb	r8, [r0], #1
+		orr	ip, ip, r8, lsl #8
+		adds	r3, r3, ip
+		strb	ip, [r1], #1
+		strb	r8, [r1], #1
+Lexit:		tst	r2, #1
+Ltoo_small1:	ldrneb	ip, [r0], #1
+		strneb	ip, [r1], #1
+		adcnes	r3, r3, ip
+		adcs	r0, r3, #0
+		LOADREGS(ea,fp,{r4 - r8, fp, sp, pc})
+
+Lsrc_not_aligned:
+		cmp	r2, #4
+		blt	Ltoo_small
+		and	ip, r0, #3
+		bic	r0, r0, #3
+		ldr	r4, [r0], #4
+		cmp	ip, #2
+		beq	Lsrc2_aligned
+		bhi	Lsrc3_aligned
+		mov	r4, r4, lsr #8
+		adds	r3, r3, #0
+		bics	ip, r2, #15
+		beq	2f
+1:		ldmia	r0!, {r5, r6, r7, r8}
+		orr	r4, r4, r5, lsl #24
+		mov	r5, r5, lsr #8
+		orr	r5, r5, r6, lsl #24
+		mov	r6, r6, lsr #8
+		orr	r6, r6, r7, lsl #24
+		mov	r7, r7, lsr #8
+		orr	r7, r7, r8, lsl #24
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		mov	r4, r8, lsr #8
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+2:		ands	ip, r2, #12
+		beq	4f
+		tst	ip, #8
+		beq	3f
+		ldmia	r0!, {r5, r6}
+		orr	r4, r4, r5, lsl #24
+		mov	r5, r5, lsr #8
+		orr	r5, r5, r6, lsl #24
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		mov	r4, r6, lsr #8
+		tst	ip, #4
+		beq	4f
+3:		ldr	r5, [r0], #4
+		orr	r4, r4, r5, lsl #24
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+		mov	r4, r5, lsr #8
+4:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		tst	r2, #2
+		beq	Lexit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		b	Lexit
+
+Lsrc2_aligned:	mov	r4, r4, lsr #16
+		adds	r3, r3, #0
+		bics	ip, r2, #15
+		beq	2f
+1:		ldmia	r0!, {r5, r6, r7, r8}
+		orr	r4, r4, r5, lsl #16
+		mov	r5, r5, lsr #16
+		orr	r5, r5, r6, lsl #16
+		mov	r6, r6, lsr #16
+		orr	r6, r6, r7, lsl #16
+		mov	r7, r7, lsr #16
+		orr	r7, r7, r8, lsl #16
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		mov	r4, r8, lsr #16
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+2:		ands	ip, r2, #12
+		beq	4f
+		tst	ip, #8
+		beq	3f
+		ldmia	r0!, {r5, r6}
+		orr	r4, r4, r5, lsl #16
+		mov	r5, r5, lsr #16
+		orr	r5, r5, r6, lsl #16
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		mov	r4, r6, lsr #16
+		tst	ip, #4
+		beq	4f
+3:		ldr	r5, [r0], #4
+		orr	r4, r4, r5, lsl #16
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+		mov	r4, r5, lsr #16
+4:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		tst	r2, #2
+		beq	Lexit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+		mov	r4, r4, lsr #8
+		strb	r4, [r1], #1
+		ldrb	r4, [r0], #1
+		b	Lexit
+
+Lsrc3_aligned:	mov	r4, r4, lsr #24
+		adds	r3, r3, #0
+		bics	ip, r2, #15
+		beq	2f
+1:		ldmia	r0!, {r5, r6, r7, r8}
+		orr	r4, r4, r5, lsl #8
+		mov	r5, r5, lsr #24
+		orr	r5, r5, r6, lsl #8
+		mov	r6, r6, lsr #24
+		orr	r6, r6, r7, lsl #8
+		mov	r7, r7, lsr #24
+		orr	r7, r7, r8, lsl #8
+		stmia	r1!, {r4, r5, r6, r7}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		adcs	r3, r3, r6
+		adcs	r3, r3, r7
+		mov	r4, r8, lsr #24
+		sub	ip, ip, #16
+		teq	ip, #0
+		bne	1b
+2:		ands	ip, r2, #12
+		beq	4f
+		tst	ip, #8
+		beq	3f
+		ldmia	r0!, {r5, r6}
+		orr	r4, r4, r5, lsl #8
+		mov	r5, r5, lsr #24
+		orr	r5, r5, r6, lsl #8
+		stmia	r1!, {r4, r5}
+		adcs	r3, r3, r4
+		adcs	r3, r3, r5
+		mov	r4, r6, lsr #24
+		tst	ip, #4
+		beq	4f
+3:		ldr	r5, [r0], #4
+		orr	r4, r4, r5, lsl #8
+		str	r4, [r1], #4
+		adcs	r3, r3, r4
+		mov	r4, r5, lsr #24
+4:		ands	r2, r2, #3
+		adceq	r0, r3, #0
+		LOADREGS(eqea,fp,{r4 - r8, fp, sp, pc})
+		tst	r2, #2
+		beq	Lexit
+		adcs	r3, r3, r4, lsl #16
+		strb	r4, [r1], #1
+		ldr	r4, [r0], #4
+		strb	r4, [r1], #1
+		adcs	r3, r3, r4, lsl #24
+		mov	r4, r4, lsr #8
+		b	Lexit

FUNET's LINUX-ADM group, linux-adm@nic.funet.fi
TCL-scripts by Sam Shen, slshen@lbl.gov