This commit was manufactured by cvs2svn to create branch 'vserver'.

[linux-2.6.git] / arch / avr32 / lib / findbit.S

diff --git a/arch/avr32/lib/findbit.S b/arch/avr32/lib/findbit.S
new file mode 100644 (file)
index 0000000..c6b91de
--- /dev/null
+++ b/arch/avr32/lib/findbit.S
@@ -0,0 +1,155 @@
+/*
+ * Copyright (C) 2006 Atmel Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+#include <linux/linkage.h>
+
+       .text
+       /*
+        * unsigned long find_first_zero_bit(const unsigned long *addr,
+        *                                   unsigned long size)
+        */
+ENTRY(find_first_zero_bit)
+       cp.w    r11, 0
+       reteq   r11
+       mov     r9, r11
+1:     ld.w    r8, r12[0]
+       com     r8
+       brne    .L_found
+       sub     r12, -4
+       sub     r9, 32
+       brgt    1b
+       retal   r11
+
+       /*
+        * unsigned long find_next_zero_bit(const unsigned long *addr,
+        *                                  unsigned long size,
+        *                                  unsigned long offset)
+        */
+ENTRY(find_next_zero_bit)
+       lsr     r8, r10, 5
+       sub     r9, r11, r10
+       retle   r11
+
+       lsl     r8, 2
+       add     r12, r8
+       andl    r10, 31, COH
+       breq    1f
+
+       /* offset is not word-aligned. Handle the first (32 - r10) bits */
+       ld.w    r8, r12[0]
+       com     r8
+       sub     r12, -4
+       lsr     r8, r8, r10
+       brne    .L_found
+
+       /* r9 = r9 - (32 - r10) = r9 + r10 - 32 */
+       add     r9, r10
+       sub     r9, 32
+       retle   r11
+
+       /* Main loop. offset must be word-aligned */
+1:     ld.w    r8, r12[0]
+       com     r8
+       brne    .L_found
+       sub     r12, -4
+       sub     r9, 32
+       brgt    1b
+       retal   r11
+
+       /* Common return path for when a bit is actually found. */
+.L_found:
+       brev    r8
+       clz     r10, r8
+       rsub    r9, r11
+       add     r10, r9
+
+       /* XXX: If we don't have to return exactly "size" when the bit
+          is not found, we may drop this "min" thing */
+       min     r12, r11, r10
+       retal   r12
+
+       /*
+        * unsigned long find_first_bit(const unsigned long *addr,
+        *                              unsigned long size)
+        */
+ENTRY(find_first_bit)
+       cp.w    r11, 0
+       reteq   r11
+       mov     r9, r11
+1:     ld.w    r8, r12[0]
+       cp.w    r8, 0
+       brne    .L_found
+       sub     r12, -4
+       sub     r9, 32
+       brgt    1b
+       retal   r11
+
+       /*
+        * unsigned long find_next_bit(const unsigned long *addr,
+        *                             unsigned long size,
+        *                             unsigned long offset)
+        */
+ENTRY(find_next_bit)
+       lsr     r8, r10, 5
+       sub     r9, r11, r10
+       retle   r11
+
+       lsl     r8, 2
+       add     r12, r8
+       andl    r10, 31, COH
+       breq    1f
+
+       /* offset is not word-aligned. Handle the first (32 - r10) bits */
+       ld.w    r8, r12[0]
+       sub     r12, -4
+       lsr     r8, r8, r10
+       brne    .L_found
+
+       /* r9 = r9 - (32 - r10) = r9 + r10 - 32 */
+       add     r9, r10
+       sub     r9, 32
+       retle   r11
+
+       /* Main loop. offset must be word-aligned */
+1:     ld.w    r8, r12[0]
+       cp.w    r8, 0
+       brne    .L_found
+       sub     r12, -4
+       sub     r9, 32
+       brgt    1b
+       retal   r11
+
+ENTRY(generic_find_next_zero_le_bit)
+       lsr     r8, r10, 5
+       sub     r9, r11, r10
+       retle   r11
+
+       lsl     r8, 2
+       add     r12, r8
+       andl    r10, 31, COH
+       breq    1f
+
+       /* offset is not word-aligned. Handle the first (32 - r10) bits */
+       ldswp.w r8, r12[0]
+       sub     r12, -4
+       com     r8
+       lsr     r8, r8, r10
+       brne    .L_found
+
+       /* r9 = r9 - (32 - r10) = r9 + r10 - 32 */
+       add     r9, r10
+       sub     r9, 32
+       retle   r11
+
+       /* Main loop. offset must be word-aligned */
+1:     ldswp.w r8, r12[0]
+       com     r8
+       brne    .L_found
+       sub     r12, -4
+       sub     r9, 32
+       brgt    1b
+       retal   r11