diff options
author | Jason Molenda <jsm@bugshack.cygnus.com> | 1999-11-02 04:44:27 +0000 |
---|---|---|
committer | Jason Molenda <jsm@bugshack.cygnus.com> | 1999-11-02 04:44:27 +0000 |
commit | 1f80975aceb82ecd4fc9bf4e988e6d1b7c7bbca1 (patch) | |
tree | 89b91ca4bdc6d7fa4fc721e8bb68d90484cc5358 /gdb/rdi-share | |
parent | 2c74239c50b6b7f9e08083a8692621d4d314d011 (diff) | |
download | gdb-1f80975aceb82ecd4fc9bf4e988e6d1b7c7bbca1.tar.gz |
Initial revision
Diffstat (limited to 'gdb/rdi-share')
-rw-r--r-- | gdb/rdi-share/angel_bytesex.c | 57 | ||||
-rw-r--r-- | gdb/rdi-share/angel_bytesex.h | 42 | ||||
-rw-r--r-- | gdb/rdi-share/angel_endian.h | 125 |
3 files changed, 224 insertions, 0 deletions
diff --git a/gdb/rdi-share/angel_bytesex.c b/gdb/rdi-share/angel_bytesex.c new file mode 100644 index 00000000000..b98c971708e --- /dev/null +++ b/gdb/rdi-share/angel_bytesex.c @@ -0,0 +1,57 @@ +/* + * Copyright (C) 1995 Advanced RISC Machines Limited. All rights reserved. + * + * This software may be freely used, copied, modified, and distributed + * provided that the above copyright notice is preserved in all copies of the + * software. + */ + +/* + * angel_bytesex.c - Code to support byte-sex independence + * Copyright: (C) 1991, Advanced RISC Machines Ltd., Cambridge, England. + */ + +/* + * RCS $Revision: 1.1 $ + * Checkin $Date: 1999/11/01 15:29:57 $ + */ + +#include "angel_bytesex.h" + +static int reversing_bytes = 0; + +void bytesex_reverse(yes_or_no) +int yes_or_no; +{ reversing_bytes = yes_or_no; +} + +int bytesex_reversing() +{ + return reversing_bytes; +} + +int32 bytesex_hostval(v) +int32 v; +{ /* Return v with the same endian-ness as the host */ + /* This mess generates better ARM code than the more obvious mess */ + /* and may eventually peephole to optimal code... */ + if (reversing_bytes) + { unsigned32 t; + /* t = v ^ (v ror 16) */ + t = v ^ ((v << 16) | (((unsigned32)v) >> 16)); + t &= ~0xff0000; + /* v = v ror 8 */ + v = (v << 24) | (((unsigned32)v) >> 8); + v = v ^ (t >> 8); + } + return v; +} + +int32 bytesex_hostval_16(v) +int32 v; +{ + if (reversing_bytes) { + v = ((v >> 8) & 0xff) | ((v << 8) & 0xff00); + } + return v; +} diff --git a/gdb/rdi-share/angel_bytesex.h b/gdb/rdi-share/angel_bytesex.h new file mode 100644 index 00000000000..40786ed2ba9 --- /dev/null +++ b/gdb/rdi-share/angel_bytesex.h @@ -0,0 +1,42 @@ +/* + * Copyright (C) 1995 Advanced RISC Machines Limited. All rights reserved. + * + * This software may be freely used, copied, modified, and distributed + * provided that the above copyright notice is preserved in all copies of the + * software. + */ + +/* + Title: Code to support byte-sex independence + Copyright: (C) 1991, Advanced RISC Machines Ltd., Cambridge, England. +*/ +/* + * RCS $Revision: 1.1 $ + * Checkin $Date: 1999/11/01 15:29:57 $ + */ + +#ifndef angel_bytesex_h +#define angel_bytesex_h + +#include "host.h" + +void bytesex_reverse(int yes_or_no); +/* + * Turn sex-reversal on or off - 0 means off, non-0 means on. + */ + +int bytesex_reversing(void); +/* + * Return non-0 if reversing the byte sex, else 0. + */ + +int32 bytesex_hostval(int32 v); +/* + * Return v or byte-reversed v, according to whether sex-reversval + * is on or off. + */ + +int32 bytesex_hostval_16(int32 v); +/* Return v or byte-reversed v for a 16 bit value */ + +#endif diff --git a/gdb/rdi-share/angel_endian.h b/gdb/rdi-share/angel_endian.h new file mode 100644 index 00000000000..6e871397041 --- /dev/null +++ b/gdb/rdi-share/angel_endian.h @@ -0,0 +1,125 @@ +/* + * Copyright (C) 1995 Advanced RISC Machines Limited. All rights reserved. + * + * This software may be freely used, copied, modified, and distributed + * provided that the above copyright notice is preserved in all copies of the + * software. + */ + +/* -*-C-*- + * + * $Revision: 1.1 $ + * $Date: 1999/11/01 13:31:32 $ + * + * + * angel_endian.h - target endianness independent read/write primitives. + */ + +#ifndef angel_endian_h +#define angel_endian_h + +/* + * The endianness of the data being processed needs to be known, but + * the host endianness is not required (since the data is constructed + * using bytes). At the moment these are provided as macros. This + * gives the compiler freedom in optimising individual calls. However, + * if space is at a premium then functions should be provided. + * + * NOTE: These macros assume that the data has been packed in the same format + * as the packing on the build host. If this is not the case then + * the wrong addresses could be used when dealing with structures. + * + */ + +/* + * For all the following routines the target endianness is defined by the + * following boolean definitions. + */ +#define BE (1 == 1) /* TRUE : big-endian */ +#define LE (1 == 0) /* FALSE : little-endian */ + +/* + * The following type definitions are used by the endianness converting + * macros. + */ +typedef unsigned char U8; +typedef U8 *P_U8; +typedef const U8 *CP_U8; + +typedef unsigned short U16; +typedef U16 *P_U16; + +typedef unsigned int U32; +typedef U32 *P_U32; + +/* + * If the endianness of the host and target are known (fixed) and the same + * then the following macro definitions can be used. These just directly copy + * the data. + * + * #define READ(e,a) (a) + * #define WRITE(e,a,v) ((a) = (v)) + * #define PREAD(e,a) (a) + * #define PWRITE(e,a,v) (*(a) = (v)) + */ + +/* + * These macros assume that a byte (char) is 8bits in size, and that the + * endianness is not important when reading or writing bytes. + */ +#define PUT8(a,v) (*((P_U8)(a)) = (U8)(v)) +#define PUT16LE(a,v) (PUT8(a,((v) & 0xFF)), \ + PUT8((((P_U8)(a)) + sizeof(char)),((v) >> 8))) +#define PUT16BE(a,v) (PUT8(a,((v) >> 8)), \ + PUT8((((P_U8)(a)) + sizeof(char)),((v) & 0xFF))) +#define PUT32LE(a,v) (PUT16LE(a,v), \ + PUT16LE((((P_U8)(a)) + sizeof(short)),((v) >> 16))) +#define PUT32BE(a,v) (PUT16BE(a,((v) >> 16)), \ + PUT16BE((((P_U8)(a)) + sizeof(short)),v)) + +#define GET8(a) (*((CP_U8)(a))) +#define GET16LE(a) (GET8(a) | (((U16)GET8(((CP_U8)(a)) + sizeof(char))) << 8)) +#define GET16BE(a) ((((U16)GET8(a)) << 8) | GET8(((CP_U8)(a)) + sizeof(char))) +#define GET32LE(a) (GET16LE(a) | \ + (((U32)GET16LE(((CP_U8)(a)) + sizeof(short))) << 16)) +#define GET32BE(a) ((((U32)GET16BE(a)) << 16) | \ + GET16BE(((CP_U8)(a)) + sizeof(short))) + +/* + * These macros simplify the code in respect to reading and writing the + * correct size data when dealing with endianness. "e" is TRUE if we are + * dealing with big-endian data, FALSE if we are dealing with little-endian. + */ + +/* void WRITE(int endianness, void *address, unsigned value); */ + +#define WRITE16(e,a,v) ((e) ? PUT16BE(&(a),v) : PUT16LE(&(a),v)) +#define WRITE32(e,a,v) ((e) ? PUT32BE(&(a),v) : PUT32LE(&(a),v)) +#define WRITE(e,a,v) ((sizeof(v) == sizeof(char)) ? \ + PUT8(&(a),v) : ((sizeof(v) == sizeof(short)) ? \ + WRITE16(e,a,v) : WRITE32(e,a,v))) + +/* unsigned READ(int endianness, void *address) */ +#define READ16(e,a) ((e) ? GET16BE(&(a)) : GET16LE(&(a))) +#define READ32(e,a) ((e) ? GET32BE(&(a)) : GET32LE(&(a))) +#define READ(e,a) ((sizeof(a) == sizeof(char)) ? \ + GET8((CP_U8)&(a)) : ((sizeof(a) == sizeof(short)) ? \ + READ16(e,a) : READ32(e,a))) + +/* void PWRITE(int endianness, void *address, unsigned value); */ +#define PWRITE16(e,a,v) ((e) ? PUT16BE(a,v) : PUT16LE(a,v)) +#define PWRITE32(e,a,v) ((e) ? PUT32BE(a,v) : PUT32LE(a,v)) +#define PWRITE(e,a,v) ((sizeof(v) == sizeof(char)) ? \ + PUT8(a,v) : ((sizeof(v) == sizeof(short)) ? \ + PWRITE16(e,a,v) : PWRITE32(e,a,v))) + +/* unsigned PREAD(int endianness, void *address) */ +#define PREAD16(e,a) ((e) ? GET16BE(a) : GET16LE(a)) +#define PREAD32(e,a) ((e) ? GET32BE(a) : GET32LE(a)) +#define PREAD(e,a) ((sizeof(*(a)) == sizeof(char)) ? \ + GET8((CP_U8)a) : ((sizeof(*(a)) == sizeof(short)) ? \ + PREAD16(e,a) : PREAD32(e,a))) + +#endif /* !defined(angel_endian_h) */ + +/* EOF angel_endian.h */ |