summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorMichael Snyder <msnyder@vmware.com>1998-06-23 18:22:43 +0000
committerMichael Snyder <msnyder@vmware.com>1998-06-23 18:22:43 +0000
commita00dfa360a4115f6f1c698b501018c8a21c020ec (patch)
tree0ac96080889dcde058518d316e1c8f131382fe1a
parentebb6416960c0011b7032da235ed0d5986d1d94b8 (diff)
downloadppe42-binutils-a00dfa360a4115f6f1c698b501018c8a21c020ec.tar.gz
ppe42-binutils-a00dfa360a4115f6f1c698b501018c8a21c020ec.zip
Tue Jun 23 11:14:04 1998 Michael Snyder <msnyder@cleaver.cygnus.com>
* config/mips/tm-irix5.h: Modify to work better on irix 6, by making FP registers 8 bytes instead of 4. REGISTER_BYTES: redefine. REGISTER_BYTE(): redefine. REGISTER_VIRTUAL_TYPE: redefine. MIPS_LAST_ARG_REGNUM: redefine. * irix5-nat.c (fetch_core_registers): read 8 bytes per FP register. * mips-tdep.c (FP_REGISTER_DOUBLE): new macro to distinguish targets with 8-byte FP registers (don't use TARGET_MIPS64). (STACK_ARGSIZE): new macro, how much space is taken up on the stack for each function argument (don't use TARGET_MIPS64). (mips_push_arguments): modify logic to work better on Irix 6 (n32 ABI).
-rw-r--r--gdb/ChangeLog14
-rw-r--r--gdb/config/mips/tm-irix5.h31
-rw-r--r--gdb/irix5-nat.c18
-rw-r--r--gdb/mips-tdep.c78
4 files changed, 123 insertions, 18 deletions
diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index 6be91502c3..e9eaf2c5b4 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,17 @@
+Tue Jun 23 11:14:04 1998 Michael Snyder <msnyder@cleaver.cygnus.com>
+
+ * config/mips/tm-irix5.h: Modify to work better on irix 6, by
+ making FP registers 8 bytes instead of 4.
+ REGISTER_BYTES: redefine. REGISTER_BYTE(): redefine.
+ REGISTER_VIRTUAL_TYPE: redefine. MIPS_LAST_ARG_REGNUM: redefine.
+ * irix5-nat.c (fetch_core_registers): read 8 bytes per FP register.
+ * mips-tdep.c (FP_REGISTER_DOUBLE): new macro to distinguish
+ targets with 8-byte FP registers (don't use TARGET_MIPS64).
+ (STACK_ARGSIZE): new macro, how much space is taken up on the
+ stack for each function argument (don't use TARGET_MIPS64).
+ (mips_push_arguments): modify logic to work better on Irix 6
+ (n32 ABI).
+
Tue Jun 23 12:29:53 1998 Jillian Ye <jillian@cygnus.com>
* configure.in: Add -lXext to mips_extra_libs
diff --git a/gdb/config/mips/tm-irix5.h b/gdb/config/mips/tm-irix5.h
index d70afd57c0..ad98e88185 100644
--- a/gdb/config/mips/tm-irix5.h
+++ b/gdb/config/mips/tm-irix5.h
@@ -19,6 +19,37 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
#include "mips/tm-irix3.h"
+#if defined (_MIPS_SIM_NABI32) && _MIPS_SIM == _MIPS_SIM_NABI32
+/*
+ * Irix 6 (n32 ABI) has 32-bit GP regs and 64-bit FP regs
+ */
+
+#undef REGISTER_BYTES
+#define REGISTER_BYTES (MIPS_NUMREGS * 8 + (NUM_REGS - MIPS_NUMREGS) * MIPS_REGSIZE)
+
+#undef REGISTER_BYTE
+#define REGISTER_BYTE(N) \
+ (((N) < FP0_REGNUM) ? (N) * MIPS_REGSIZE : \
+ ((N) < FP0_REGNUM + 32) ? \
+ FP0_REGNUM * MIPS_REGSIZE + \
+ ((N) - FP0_REGNUM) * sizeof(double) : \
+ 32 * sizeof(double) + ((N) - 32) * MIPS_REGSIZE)
+
+#undef REGISTER_VIRTUAL_TYPE
+#define REGISTER_VIRTUAL_TYPE(N) \
+ (((N) >= FP0_REGNUM && (N) < FP0_REGNUM+32) ? builtin_type_double \
+ : ((N) == 32 /*SR*/) ? builtin_type_uint32 \
+ : ((N) >= 70 && (N) <= 89) ? builtin_type_uint32 \
+ : builtin_type_int)
+
+#undef MIPS_LAST_ARG_REGNUM
+#define MIPS_LAST_ARG_REGNUM 11 /* N32 uses R4 through R11 for args */
+
+#undef MIPS_NUM_ARG_REGS
+#define MIPS_NUM_ARG_REGS 8
+
+#endif /* N32 */
+
/* When calling functions on Irix 5 (or any MIPS SVR4 ABI compliant
platform) $25 must hold the function address. Dest_Reg is a macro
used in CALL_DUMMY in tm-mips.h. */
diff --git a/gdb/irix5-nat.c b/gdb/irix5-nat.c
index 6ea5deafed..88bbf96e10 100644
--- a/gdb/irix5-nat.c
+++ b/gdb/irix5-nat.c
@@ -192,7 +192,8 @@ fetch_core_registers (core_reg_sect, core_reg_size, which, reg_addr)
{
memcpy ((char *)registers, core_reg_sect, core_reg_size);
}
- else if (core_reg_size == (2 * REGISTER_BYTES) && MIPS_REGSIZE == 4)
+ else if (MIPS_REGSIZE == 4 &&
+ core_reg_size == (2 * MIPS_REGSIZE) * NUM_REGS)
{
/* This is a core file from a N32 executable, 64 bits are saved
for all registers. */
@@ -210,7 +211,20 @@ fetch_core_registers (core_reg_sect, core_reg_size, which, reg_addr)
*dstp++ = *srcp++;
*dstp++ = *srcp++;
*dstp++ = *srcp++;
- srcp += 4;
+ if (REGISTER_RAW_SIZE(regno) == 4)
+ {
+ /* copying 4 bytes from eight bytes?
+ I don't see how this can be right... */
+ srcp += 4;
+ }
+ else
+ {
+ /* copy all 8 bytes (sizeof(double)) */
+ *dstp++ = *srcp++;
+ *dstp++ = *srcp++;
+ *dstp++ = *srcp++;
+ *dstp++ = *srcp++;
+ }
}
else
{
diff --git a/gdb/mips-tdep.c b/gdb/mips-tdep.c
index 41dc0cad01..ec0fa969e7 100644
--- a/gdb/mips-tdep.c
+++ b/gdb/mips-tdep.c
@@ -1,5 +1,5 @@
/* Target-dependent code for the MIPS architecture, for GDB, the GNU Debugger.
- Copyright 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996
+ Copyright 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998
Free Software Foundation, Inc.
Contributed by Alessandro Forin(af@cs.cmu.edu) at CMU
and by Per Bothner(bothner@cs.wisc.edu) at U.Wisconsin.
@@ -38,6 +38,9 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
#define VM_MIN_ADDRESS (CORE_ADDR)0x400000
+/* Do not use "TARGET_IS_MIPS64" to test the size of floating point registers */
+#define FP_REGISTER_DOUBLE (REGISTER_VIRTUAL_SIZE(FP0_REGNUM) == 8)
+
/* FIXME: Put this declaration in frame.h. */
extern struct obstack frame_cache_obstack;
@@ -282,6 +285,9 @@ mips32_decode_reg_save (inst, gen_mask, float_mask)
if ((inst & 0xffe00000) == 0xafa00000 /* sw reg,n($sp) */
|| (inst & 0xffe00000) == 0xafc00000 /* sw reg,n($r30) */
+ /* start-sanitize-r5900 */
+ || (inst & 0xffe00000) == 0x7fa00000 /* sq reg,n($sp) */
+ /* end-sanitize-r5900 */
|| (inst & 0xffe00000) == 0xffa00000) /* sd reg,n($sp) */
{
/* It might be possible to use the instruction to
@@ -916,6 +922,15 @@ mips_find_saved_regs (fci)
{
fci->saved_regs->regs[ireg] = reg_position;
reg_position -= MIPS_REGSIZE;
+ /* start-sanitize-r5900 */
+#ifdef R5900_128BIT_GPR_HACK
+ /* Gross. The r5900 has 128bit wide registers, but MIPS_REGSIZE is
+ still 64bits. See the comments in tm.h for a discussion of the
+ various problems this causes. */
+ if (ireg <= RA_REGNUM)
+ reg_position -= MIPS_REGSIZE;
+#endif
+ /* end-sanitize-r5900 */
}
/* The MIPS16 entry instruction saves $s0 and $s1 in the reverse order
@@ -1395,6 +1410,15 @@ restart:
PROC_REG_MASK(&temp_proc_desc) |= 1 << reg;
set_reg_offset (reg, sp + low_word + 8 - MIPS_REGSIZE);
}
+ /* start-sanitize-r5900 */
+ else if ((high_word & 0xFFE0) == 0x7fa0) /* sq reg,offset($sp) */
+ {
+ /* I don't think we have to worry about the Irix 6.2 N32 ABI
+ issue noted int he sd reg, offset($sp) case above. */
+ PROC_REG_MASK(&temp_proc_desc) |= 1 << reg;
+ set_reg_offset (reg, sp + low_word);
+ }
+ /* end-sanitize-r5900 */
else if (high_word == 0x27be) /* addiu $30,$sp,size */
{
/* Old gcc frame, r30 is virtual frame pointer. */
@@ -1708,6 +1732,20 @@ setup_arbitrary_frame (argc, argv)
return create_new_frame (argv[0], argv[1]);
}
+/*
+ * STACK_ARGSIZE -- how many bytes does a pushed function arg take up on the stack?
+ *
+ * For n32 ABI, eight.
+ * For all others, he same as the size of a general register.
+ */
+#if defined (_MIPS_SIM_NABI32) && _MIPS_SIM == _MIPS_SIM_NABI32
+#define MIPS_NABI32 1
+#define STACK_ARGSIZE 8
+#else
+#define MIPS_NABI32 0
+#define STACK_ARGSIZE MIPS_REGSIZE
+#endif
+
CORE_ADDR
mips_push_arguments(nargs, args, sp, struct_return, struct_addr)
int nargs;
@@ -1775,7 +1813,7 @@ mips_push_arguments(nargs, args, sp, struct_return, struct_addr)
/* 32-bit ABIs always start floating point arguments in an
even-numbered floating point register. */
- if (!GDB_TARGET_IS_MIPS64 && typecode == TYPE_CODE_FLT
+ if (!FP_REGISTER_DOUBLE && typecode == TYPE_CODE_FLT
&& (float_argreg & 1))
float_argreg++;
@@ -1792,7 +1830,7 @@ mips_push_arguments(nargs, args, sp, struct_return, struct_addr)
&& float_argreg <= MIPS_LAST_FP_ARG_REGNUM
&& mips_fpu != MIPS_FPU_NONE)
{
- if (!GDB_TARGET_IS_MIPS64 && len == 8)
+ if (!FP_REGISTER_DOUBLE && len == 8)
{
int low_offset = TARGET_BYTE_ORDER == BIG_ENDIAN ? 4 : 0;
unsigned long regval;
@@ -1822,7 +1860,7 @@ mips_push_arguments(nargs, args, sp, struct_return, struct_addr)
if (!MIPS_EABI)
{
write_register (argreg, regval);
- argreg += GDB_TARGET_IS_MIPS64 ? 1 : 2;
+ argreg += FP_REGISTER_DOUBLE ? 1 : 2;
}
}
}
@@ -1850,15 +1888,15 @@ mips_push_arguments(nargs, args, sp, struct_return, struct_addr)
int longword_offset = 0;
if (TARGET_BYTE_ORDER == BIG_ENDIAN)
- if (MIPS_REGSIZE == 8 &&
+ if (STACK_ARGSIZE == 8 &&
(typecode == TYPE_CODE_INT ||
typecode == TYPE_CODE_PTR ||
typecode == TYPE_CODE_FLT) && len <= 4)
- longword_offset = MIPS_REGSIZE - len;
+ longword_offset = STACK_ARGSIZE - len;
else if ((typecode == TYPE_CODE_STRUCT ||
typecode == TYPE_CODE_UNION) &&
- TYPE_LENGTH (arg_type) < MIPS_REGSIZE)
- longword_offset = MIPS_REGSIZE - len;
+ TYPE_LENGTH (arg_type) < STACK_ARGSIZE)
+ longword_offset = STACK_ARGSIZE - len;
write_memory (sp + stack_offset + longword_offset,
val, partial_len);
@@ -1907,12 +1945,14 @@ mips_push_arguments(nargs, args, sp, struct_return, struct_addr)
begins at (4 * MIPS_REGSIZE) in the old ABI. This
leaves room for the "home" area for register parameters.
- In the new EABI, the 8 register parameters do not
- have "home" stack space reserved for them, so the
+ In the new EABI (and the NABI32), the 8 register parameters
+ do not have "home" stack space reserved for them, so the
stack offset does not get incremented until after
we have used up the 8 parameter registers. */
- if (!(MIPS_EABI && argnum < 8))
- stack_offset += ROUND_UP (partial_len, MIPS_REGSIZE);
+
+ if (!(MIPS_EABI || MIPS_NABI32) ||
+ argnum >= 8)
+ stack_offset += ROUND_UP (partial_len, STACK_ARGSIZE);
}
}
}
@@ -1990,7 +2030,8 @@ mips_push_dummy_frame()
/* Save general CPU registers */
PROC_REG_MASK(proc_desc) = GEN_REG_SAVE_MASK;
- PROC_REG_OFFSET(proc_desc) = sp - old_sp; /* offset of (Saved R31) from FP */
+ /* PROC_REG_OFFSET is the offset of the first saved register from FP. */
+ PROC_REG_OFFSET(proc_desc) = sp - old_sp - MIPS_REGSIZE;
for (ireg = 32; --ireg >= 0; )
if (PROC_REG_MASK(proc_desc) & (1 << ireg))
mips_push_register (&sp, ireg);
@@ -1999,7 +2040,9 @@ mips_push_dummy_frame()
PROC_FREG_MASK(proc_desc) =
mips_fpu == MIPS_FPU_DOUBLE ? FLOAT_REG_SAVE_MASK
: mips_fpu == MIPS_FPU_SINGLE ? FLOAT_SINGLE_REG_SAVE_MASK : 0;
- PROC_FREG_OFFSET(proc_desc) = sp - old_sp; /* offset of (Saved D18) from FP */
+ /* PROC_FREG_OFFSET is the offset of the first saved *double* register
+ from FP. */
+ PROC_FREG_OFFSET(proc_desc) = sp - old_sp - 8;
for (ireg = 32; --ireg >= 0; )
if (PROC_FREG_MASK(proc_desc) & (1 << ireg))
mips_push_register (&sp, ireg + FP0_REGNUM);
@@ -2086,7 +2129,7 @@ mips_print_register (regnum, all)
/* If an even floating point register, also print as double. */
if (TYPE_CODE (REGISTER_VIRTUAL_TYPE (regnum)) == TYPE_CODE_FLT
&& !((regnum-FP0_REGNUM) & 1))
- if (REGISTER_RAW_SIZE(regnum) == 4) /* this would be silly on MIPS64 */
+ if (REGISTER_RAW_SIZE(regnum) == 4) /* this would be silly on MIPS64 or N32 (Irix 6) */
{
char dbuffer[2 * MAX_REGISTER_RAW_SIZE];
@@ -2112,7 +2155,7 @@ mips_print_register (regnum, all)
/* If virtual format is floating, print it that way. */
if (TYPE_CODE (REGISTER_VIRTUAL_TYPE (regnum)) == TYPE_CODE_FLT)
- if (REGISTER_RAW_SIZE(regnum) == 8)
+ if (FP_REGISTER_DOUBLE)
{ /* show 8-byte floats as float AND double: */
int offset = 4 * (TARGET_BYTE_ORDER == BIG_ENDIAN);
@@ -2387,6 +2430,9 @@ mips32_skip_prologue (pc, lenient)
inst == 0x03a8e823) /* subu $sp,$sp,$t0 */
seen_sp_adjust = 1;
else if (((inst & 0xFFE00000) == 0xAFA00000 /* sw reg,n($sp) */
+ /* start-sanitize-r5900 */
+ || (inst & 0xFFE00000) == 0x7FA00000 /* sq reg,n($sp) */
+ /* end-sanitize-r5900 */
|| (inst & 0xFFE00000) == 0xFFA00000) /* sd reg,n($sp) */
&& (inst & 0x001F0000)) /* reg != $zero */
continue;
OpenPOWER on IntegriCloud