/* Target-dependent code for the MIPS architecture, for GDB, the GNU Debugger.
- Copyright 1988, 1989, 1990, 1991, 1992 Free Software Foundation, Inc.
+ Copyright 1988, 1989, 1990, 1991, 1992, 1993, 1994
+ Free Software Foundation, Inc.
#include "value.h"
#include "gdbcmd.h"
#include "language.h"
-
-#ifdef USG
-#include <sys/types.h>
-#endif
-
-#include <sys/param.h>
-#include <sys/dir.h>
-#include <signal.h>
-#include <sys/ioctl.h>
-
-#ifdef sgi
-/* Must do it this way only for SGIs, as other mips platforms get their
- JB_ symbols from machine/pcb.h (included via sys/user.h). */
-#include <setjmp.h>
-#endif
-
#include "gdbcore.h"
#include "symfile.h"
#include "objfiles.h"
+#include "gdbtypes.h"
-#ifndef MIPSMAGIC
-#ifdef MIPSEL
-#define MIPSMAGIC MIPSELMAGIC
-#else
-#define MIPSMAGIC MIPSEBMAGIC
-#endif
-#endif
+#include "opcode/mips.h"
#define VM_MIN_ADDRESS (unsigned)0x400000
-#include <sys/user.h> /* After a.out.h */
-#include <sys/file.h>
-#include <sys/stat.h>
-
+/* FIXME: Put this declaration in frame.h. */
+extern struct obstack frame_cache_obstack;
\f
+#if 0
+static int mips_in_lenient_prologue PARAMS ((CORE_ADDR, CORE_ADDR));
+#endif
+
+/* Some MIPS boards don't support floating point, so we permit the
+ user to turn it off. */
+int mips_fpu = 1;
+
+/* Heuristic_proc_start may hunt through the text section for a long
+ time across a 2400 baud serial line. Allows the user to limit this
+ search. */
+static unsigned int heuristic_fence_post = 0;
+
#define PROC_LOW_ADDR(proc) ((proc)->pdr.adr) /* least address */
#define PROC_HIGH_ADDR(proc) ((proc)->pdr.iline) /* upper address bound */
#define PROC_FRAME_OFFSET(proc) ((proc)->pdr.frameoffset)
{
struct mips_extra_func_info info;
struct linked_proc_info *next;
-} * linked_proc_desc_table = NULL;
+} *linked_proc_desc_table = NULL;
\f
-#define READ_FRAME_REG(fi, regno) read_next_frame_reg((fi)->next, regno)
+/* Guaranteed to set fci->saved_regs to some values (it never leaves it
+ NULL). */
+
+void
+mips_find_saved_regs (fci)
+ FRAME fci;
+{
+ int ireg;
+ CORE_ADDR reg_position;
+ /* r0 bit means kernel trap */
+ int kernel_trap;
+ /* What registers have been saved? Bitmasks. */
+ unsigned long gen_mask, float_mask;
+ mips_extra_func_info_t proc_desc;
+
+ fci->saved_regs = (struct frame_saved_regs *)
+ obstack_alloc (&frame_cache_obstack, sizeof(struct frame_saved_regs));
+ memset (fci->saved_regs, 0, sizeof (struct frame_saved_regs));
+
+ proc_desc = fci->proc_desc;
+ if (proc_desc == NULL)
+ /* I'm not sure how/whether this can happen. Normally when we can't
+ find a proc_desc, we "synthesize" one using heuristic_proc_desc
+ and set the saved_regs right away. */
+ return;
+
+ kernel_trap = PROC_REG_MASK(proc_desc) & 1;
+ gen_mask = kernel_trap ? 0xFFFFFFFF : PROC_REG_MASK(proc_desc);
+ float_mask = kernel_trap ? 0xFFFFFFFF : PROC_FREG_MASK(proc_desc);
+
+ if (/* In any frame other than the innermost, we assume that all
+ registers have been saved. This assumes that all register
+ saves in a function happen before the first function
+ call. */
+ fci->next == NULL
+
+ /* In a dummy frame we know exactly where things are saved. */
+ && !PROC_DESC_IS_DUMMY (proc_desc)
+
+ /* Not sure exactly what kernel_trap means, but if it means
+ the kernel saves the registers without a prologue doing it,
+ we better not examine the prologue to see whether registers
+ have been saved yet. */
+ && !kernel_trap)
+ {
+ /* We need to figure out whether the registers that the proc_desc
+ claims are saved have been saved yet. */
+
+ CORE_ADDR addr;
+ int status;
+ char buf[4];
+ unsigned long inst;
+
+ /* Bitmasks; set if we have found a save for the register. */
+ unsigned long gen_save_found = 0;
+ unsigned long float_save_found = 0;
+
+ for (addr = PROC_LOW_ADDR (proc_desc);
+ addr < fci->pc /*&& (gen_mask != gen_save_found
+ || float_mask != float_save_found)*/;
+ addr += 4)
+ {
+ status = read_memory_nobpt (addr, buf, 4);
+ if (status)
+ memory_error (status, addr);
+ inst = extract_unsigned_integer (buf, 4);
+ if (/* sw reg,n($sp) */
+ (inst & 0xffe00000) == 0xafa00000
+
+ /* sw reg,n($r30) */
+ || (inst & 0xffe00000) == 0xafc00000
+
+ /* sd reg,n($sp) */
+ || (inst & 0xffe00000) == 0xffa00000)
+ {
+ /* It might be possible to use the instruction to
+ find the offset, rather than the code below which
+ is based on things being in a certain order in the
+ frame, but figuring out what the instruction's offset
+ is relative to might be a little tricky. */
+ int reg = (inst & 0x001f0000) >> 16;
+ gen_save_found |= (1 << reg);
+ }
+ else if (/* swc1 freg,n($sp) */
+ (inst & 0xffe00000) == 0xe7a00000
+
+ /* swc1 freg,n($r30) */
+ || (inst & 0xffe00000) == 0xe7c00000
+
+ /* sdc1 freg,n($sp) */
+ || (inst & 0xffe00000) == 0xf7a00000)
+
+ {
+ int reg = ((inst & 0x001f0000) >> 16);
+ float_save_found |= (1 << reg);
+ }
+ }
+ gen_mask = gen_save_found;
+ float_mask = float_save_found;
+ }
+
+ /* Fill in the offsets for the registers which gen_mask says
+ were saved. */
+ reg_position = fci->frame + PROC_REG_OFFSET (proc_desc);
+ for (ireg= 31; gen_mask; --ireg, gen_mask <<= 1)
+ if (gen_mask & 0x80000000)
+ {
+ fci->saved_regs->regs[ireg] = reg_position;
+ reg_position -= MIPS_REGSIZE;
+ }
+ /* Fill in the offsets for the registers which float_mask says
+ were saved. */
+ reg_position = fci->frame + PROC_FREG_OFFSET (proc_desc);
+
+ /* The freg_offset points to where the first *double* register
+ is saved. So skip to the high-order word. */
+ reg_position += 4;
+ for (ireg = 31; float_mask; --ireg, float_mask <<= 1)
+ if (float_mask & 0x80000000)
+ {
+ fci->saved_regs->regs[FP0_REGNUM+ireg] = reg_position;
+ reg_position -= MIPS_REGSIZE;
+ }
+
+ fci->saved_regs->regs[PC_REGNUM] = fci->saved_regs->regs[RA_REGNUM];
+}
static int
read_next_frame_reg(fi, regno)
FRAME fi;
int regno;
{
-#define SIGFRAME_BASE sizeof(struct sigcontext)
-#define SIGFRAME_PC_OFF (-SIGFRAME_BASE+ 2*sizeof(int))
-#define SIGFRAME_SP_OFF (-SIGFRAME_BASE+32*sizeof(int))
-#define SIGFRAME_RA_OFF (-SIGFRAME_BASE+34*sizeof(int))
+ /* If it is the frame for sigtramp we have a complete sigcontext
+ somewhere above the frame and we get the saved registers from there.
+ If the stack layout for sigtramp changes we might have to change these
+ constants and the companion fixup_sigtramp in mdebugread.c */
+#ifndef SIGFRAME_BASE
+/* To satisfy alignment restrictions the sigcontext is located 4 bytes
+ above the sigtramp frame. */
+#define SIGFRAME_BASE 4
+#define SIGFRAME_PC_OFF (SIGFRAME_BASE + 2 * 4)
+#define SIGFRAME_REGSAVE_OFF (SIGFRAME_BASE + 3 * 4)
+#endif
+#ifndef SIGFRAME_REG_SIZE
+#define SIGFRAME_REG_SIZE 4
+#endif
for (; fi; fi = fi->next)
- if (in_sigtramp(fi->pc, 0)) {
- /* No idea if this code works. --PB. */
+ {
+ if (fi->signal_handler_caller)
+ {
int offset;
if (regno == PC_REGNUM) offset = SIGFRAME_PC_OFF;
- else if (regno == RA_REGNUM) offset = SIGFRAME_RA_OFF;
- else if (regno == SP_REGNUM) offset = SIGFRAME_SP_OFF;
+ else if (regno < 32) offset = (SIGFRAME_REGSAVE_OFF
+ + regno * SIGFRAME_REG_SIZE);
else return 0;
- return read_memory_integer(fi->frame + offset, 4);
- }
+ return read_memory_integer(fi->frame + offset, MIPS_REGSIZE);
+ }
else if (regno == SP_REGNUM) return fi->frame;
- else if (fi->saved_regs->regs[regno])
- return read_memory_integer(fi->saved_regs->regs[regno], 4);
- return read_register(regno);
+ else
+ {
+ if (fi->saved_regs == NULL)
+ mips_find_saved_regs (fi);
+ if (fi->saved_regs->regs[regno])
+ return read_memory_integer(fi->saved_regs->regs[regno], MIPS_REGSIZE);
+ }
+ }
+ return read_register (regno);
}
int
FRAME frame;
{
mips_extra_func_info_t proc_desc = frame->proc_desc;
- int pcreg = proc_desc ? PROC_PC_REG(proc_desc) : RA_REGNUM;
+ /* We have to get the saved pc from the sigcontext
+ if it is a signal handler frame. */
+ int pcreg = frame->signal_handler_caller ? PC_REGNUM
+ : (proc_desc ? PROC_PC_REG(proc_desc) : RA_REGNUM);
if (proc_desc && PROC_DESC_IS_DUMMY(proc_desc))
return read_memory_integer(frame->frame - 4, 4);
static struct mips_extra_func_info temp_proc_desc;
static struct frame_saved_regs temp_saved_regs;
+/* This fencepost looks highly suspicious to me. Removing it also
+ seems suspicious as it could affect remote debugging across serial
+ lines. */
+
static CORE_ADDR
heuristic_proc_start(pc)
CORE_ADDR pc;
{
CORE_ADDR start_pc = pc;
- CORE_ADDR fence = start_pc - 200;
+ CORE_ADDR fence = start_pc - heuristic_fence_post;
if (start_pc == 0) return 0;
- if (fence < VM_MIN_ADDRESS) fence = VM_MIN_ADDRESS;
+
+ if (heuristic_fence_post == UINT_MAX
+ || fence < VM_MIN_ADDRESS)
+ fence = VM_MIN_ADDRESS;
/* search back for previous return */
for (start_pc -= 4; ; start_pc -= 4)
- if (start_pc < fence) return 0;
+ if (start_pc < fence)
+ {
+ /* It's not clear to me why we reach this point when
+ stop_soon_quietly, but with this test, at least we
+ don't print out warnings for every child forked (eg, on
+ if (!stop_soon_quietly)
+ {
+ static int blurb_printed = 0;
+
+ if (fence == VM_MIN_ADDRESS)
+ warning("Hit beginning of text section without finding");
+ else
+ warning("Hit heuristic-fence-post without finding");
+
+ warning("enclosing function for address 0x%x", pc);
+ if (!blurb_printed)
+ {
+ printf_filtered ("\
+This warning occurs if you are debugging a function without any symbols\n\
+(for example, in a stripped executable). In that case, you may wish to\n\
+increase the size of the search with the `set heuristic-fence-post' command.\n\
+\n\
+Otherwise, you told GDB there was a function where there isn't one, or\n\
+(more likely) you have encountered a bug in GDB.\n");
+ blurb_printed = 1;
+ }
+ }
+
+ return 0;
+ }
else if (ABOUT_TO_RETURN(start_pc))
break;
CORE_ADDR cur_pc;
int frame_size;
int has_frame_reg = 0;
- int reg30; /* Value of $r30. Used by gcc for frame-pointer */
+ int reg30 = 0; /* Value of $r30. Used by gcc for frame-pointer */
unsigned long reg_mask = 0;
if (start_pc == 0) return NULL;
- bzero(&temp_proc_desc, sizeof(temp_proc_desc));
- bzero(&temp_saved_regs, sizeof(struct frame_saved_regs));
+ memset(&temp_proc_desc, '\0', sizeof(temp_proc_desc));
+ memset(&temp_saved_regs, '\0', sizeof(struct frame_saved_regs));
PROC_LOW_ADDR(&temp_proc_desc) = start_pc;
if (start_pc + 200 < limit_pc) limit_pc = start_pc + 200;
restart:
frame_size = 0;
for (cur_pc = start_pc; cur_pc < limit_pc; cur_pc += 4) {
+ char buf[4];
unsigned long word;
int status;
- status = read_memory_nobpt (cur_pc, (char *)&word, 4);
- if (status) memory_error (status, cur_pc);
- SWAP_TARGET_AND_HOST (&word, sizeof (word));
+ status = read_memory_nobpt (cur_pc, buf, 4);
+ if (status) memory_error (status, cur_pc);
+ word = extract_unsigned_integer (buf, 4);
+
if ((word & 0xFFFF0000) == 0x27bd0000) /* addiu $sp,$sp,-i */
frame_size += (-word) & 0xFFFF;
else if ((word & 0xFFFF0000) == 0x23bd0000) /* addu $sp,$sp,-i */
else if ((word & 0xFFE00000) == 0xafa00000) { /* sw reg,offset($sp) */
int reg = (word & 0x001F0000) >> 16;
reg_mask |= 1 << reg;
- temp_saved_regs.regs[reg] = sp + (short)word;
+ temp_saved_regs.regs[reg] = sp + (word & 0xffff);
}
else if ((word & 0xFFFF0000) == 0x27be0000) { /* addiu $30,$sp,size */
- if ((unsigned short)word != frame_size)
- reg30 = sp + (unsigned short)word;
+ if ((word & 0xffff) != frame_size)
+ reg30 = sp + (word & 0xffff);
else if (!has_frame_reg) {
int alloca_adjust;
has_frame_reg = 1;
reg30 = read_next_frame_reg(next_frame, 30);
- alloca_adjust = reg30 - (sp + (unsigned short)word);
+ alloca_adjust = reg30 - (sp + (word & 0xffff));
if (alloca_adjust > 0) {
/* FP > SP + frame_size. This may be because
- /* of an alloca or somethings similar.
+ * of an alloca or somethings similar.
* Fix sp to "pre-alloca" value, and try again.
*/
sp += alloca_adjust;
else if ((word & 0xFFE00000) == 0xafc00000) { /* sw reg,offset($30) */
int reg = (word & 0x001F0000) >> 16;
reg_mask |= 1 << reg;
- temp_saved_regs.regs[reg] = reg30 + (short)word;
+ temp_saved_regs.regs[reg] = reg30 + (word & 0xffff);
}
}
if (has_frame_reg) {
{
mips_extra_func_info_t proc_desc;
struct block *b = block_for_pc(pc);
- struct symbol *sym =
- b ? lookup_symbol(".gdbinfo.", b, LABEL_NAMESPACE, 0, NULL) : NULL;
+ struct symbol *sym;
+ CORE_ADDR startaddr;
+
+ find_pc_partial_function (pc, NULL, &startaddr, NULL);
+ if (b == NULL)
+ sym = NULL;
+ else
+ {
+ if (startaddr > BLOCK_START (b))
+ /* This is the "pathological" case referred to in a comment in
+ print_frame_info. It might be better to move this check into
+ symbol reading. */
+ sym = NULL;
+ else
+ sym = lookup_symbol (MIPS_EFI_SYMBOL_NAME, b, LABEL_NAMESPACE,
+ 0, NULL);
+ }
if (sym)
{
if (PROC_LOW_ADDR(&link->info) <= pc
&& PROC_HIGH_ADDR(&link->info) > pc)
return &link->info;
+
+ if (startaddr == 0)
+ startaddr = heuristic_proc_start (pc);
+
proc_desc =
- heuristic_proc_desc(heuristic_proc_start(pc), pc, next_frame);
+ heuristic_proc_desc (startaddr, pc, next_frame);
}
return proc_desc;
}
return 0;
cached_proc_desc = proc_desc;
- return read_next_frame_reg(frame, PROC_FRAME_REG(proc_desc))
- + PROC_FRAME_OFFSET(proc_desc);
+
+ /* If no frame pointer and frame size is zero, we must be at end
+ of stack (or otherwise hosed). If we don't check frame size,
+ we loop forever if we see a zero size frame. */
+ if (PROC_FRAME_REG (proc_desc) == SP_REGNUM
+ && PROC_FRAME_OFFSET (proc_desc) == 0
+ /* The previous frame from a sigtramp frame might be frameless
+ and have frame size zero. */
+ && !frame->signal_handler_caller)
+ return 0;
+ else
+ return read_next_frame_reg(frame, PROC_FRAME_REG(proc_desc))
+ + PROC_FRAME_OFFSET(proc_desc);
}
void
init_extra_frame_info(fci)
struct frame_info *fci;
{
- extern struct obstack frame_cache_obstack;
/* Use proc_desc calculated in frame_chain */
- mips_extra_func_info_t proc_desc = fci->next ? cached_proc_desc :
- find_proc_desc(fci->pc, fci->next);
+ mips_extra_func_info_t proc_desc =
+ fci->next ? cached_proc_desc : find_proc_desc(fci->pc, fci->next);
- fci->saved_regs = (struct frame_saved_regs*)
- obstack_alloc (&frame_cache_obstack, sizeof(struct frame_saved_regs));
- bzero(fci->saved_regs, sizeof(struct frame_saved_regs));
+ fci->saved_regs = NULL;
fci->proc_desc =
- proc_desc == &temp_proc_desc ? 0 : proc_desc;
+ proc_desc == &temp_proc_desc ? 0 : proc_desc;
if (proc_desc)
{
- int ireg;
- CORE_ADDR reg_position;
- unsigned long mask;
- /* r0 bit means kernel trap */
- int kernel_trap = PROC_REG_MASK(proc_desc) & 1;
-
- if (fci->frame == 0)
- {
- /* Fixup frame-pointer - only needed for top frame */
- /* This may not be quite right, if proc has a real frame register */
- if (fci->pc == PROC_LOW_ADDR(proc_desc))
- fci->frame = read_register (SP_REGNUM);
- else
- fci->frame = READ_FRAME_REG(fci, PROC_FRAME_REG(proc_desc))
- + PROC_FRAME_OFFSET(proc_desc);
- }
+ /* Fixup frame-pointer - only needed for top frame */
+ /* This may not be quite right, if proc has a real frame register.
+ Get the value of the frame relative sp, procedure might have been
+ interrupted by a signal at it's very start. */
+ if (fci->pc == PROC_LOW_ADDR (proc_desc)
+ && !PROC_DESC_IS_DUMMY (proc_desc))
+ fci->frame = read_next_frame_reg (fci->next, SP_REGNUM);
+ else
+ fci->frame =
+ read_next_frame_reg (fci->next, PROC_FRAME_REG (proc_desc))
+ + PROC_FRAME_OFFSET (proc_desc);
if (proc_desc == &temp_proc_desc)
+ {
+ fci->saved_regs = (struct frame_saved_regs*)
+ obstack_alloc (&frame_cache_obstack,
+ sizeof (struct frame_saved_regs));
*fci->saved_regs = temp_saved_regs;
- else
- {
- /* find which general-purpose registers were saved */
- reg_position = fci->frame + PROC_REG_OFFSET(proc_desc);
- mask = kernel_trap ? 0xFFFFFFFF : PROC_REG_MASK(proc_desc);
- for (ireg= 31; mask; --ireg, mask <<= 1)
- if (mask & 0x80000000)
- {
- fci->saved_regs->regs[ireg] = reg_position;
- reg_position -= 4;
- }
- /* find which floating-point registers were saved */
- reg_position = fci->frame + PROC_FREG_OFFSET(proc_desc);
- /* The freg_offset points to where the first *double* register is saved.
- * So skip to the high-order word. */
- reg_position += 4;
- mask = kernel_trap ? 0xFFFFFFFF : PROC_FREG_MASK(proc_desc);
- for (ireg = 31; mask; --ireg, mask <<= 1)
- if (mask & 0x80000000)
- {
- fci->saved_regs->regs[FP0_REGNUM+ireg] = reg_position;
- reg_position -= 4;
- }
- }
+ fci->saved_regs->regs[PC_REGNUM] = fci->saved_regs->regs[RA_REGNUM];
+ }
/* hack: if argument regs are saved, guess these contain args */
if ((PROC_REG_MASK(proc_desc) & 0xF0) == 0) fci->num_args = -1;
else if ((PROC_REG_MASK(proc_desc) & 0x40) == 0) fci->num_args = 3;
else if ((PROC_REG_MASK(proc_desc) & 0x20) == 0) fci->num_args = 2;
else if ((PROC_REG_MASK(proc_desc) & 0x10) == 0) fci->num_args = 1;
-
- fci->saved_regs->regs[PC_REGNUM] = fci->saved_regs->regs[RA_REGNUM];
}
}
arguments without difficulty. */
FRAME
-setup_arbitrary_frame (stack, pc)
- FRAME_ADDR stack;
- CORE_ADDR pc;
+setup_arbitrary_frame (argc, argv)
+ int argc;
+ FRAME_ADDR *argv;
{
- return create_new_frame (stack, pc);
+ if (argc != 2)
+ error ("MIPS frame specifications require two arguments: sp and pc");
+
+ return create_new_frame (argv[0], argv[1]);
}
int struct_return;
CORE_ADDR struct_addr;
{
- CORE_ADDR buf;
register i;
- int accumulate_size = struct_return ? 4 : 0;
+ int accumulate_size = struct_return ? MIPS_REGSIZE : 0;
struct mips_arg { char *contents; int len; int offset; };
struct mips_arg *mips_args =
- (struct mips_arg*)alloca(nargs * sizeof(struct mips_arg));
+ (struct mips_arg*)alloca((nargs + 4) * sizeof(struct mips_arg));
register struct mips_arg *m_arg;
+ int fake_args = 0;
+
for (i = 0, m_arg = mips_args; i < nargs; i++, m_arg++) {
extern value value_arg_coerce();
value arg = value_arg_coerce (args[i]);
* breaks their varargs implementation...). A correct solution
* requires an simulation of gcc's 'alignof' (and use of 'alignof'
* in stdarg.h/varargs.h).
+ * On the 64 bit r4000 we always pass the first four arguments
+ * using eight bytes each, so that we can load them up correctly
+ * in CALL_DUMMY.
*/
- if (m_arg->len > 4) accumulate_size = (accumulate_size + 7) & -8;
+ if (m_arg->len > 4)
+ accumulate_size = (accumulate_size + 7) & -8;
m_arg->offset = accumulate_size;
- accumulate_size = (accumulate_size + m_arg->len + 3) & -4;
m_arg->contents = VALUE_CONTENTS(arg);
+#ifndef GDB_TARGET_IS_MIPS64
+ accumulate_size = (accumulate_size + m_arg->len + 3) & -4;
+#else
+ if (accumulate_size >= 4 * MIPS_REGSIZE)
+ accumulate_size = (accumulate_size + m_arg->len + 3) &~ 4;
+ else
+ {
+ static char zeroes[8] = { 0 };
+ int len = m_arg->len;
+
+ if (len < 8)
+ {
+#if TARGET_BYTE_ORDER == BIG_ENDIAN
+ m_arg->offset += 8 - len;
+#endif
+ ++m_arg;
+ m_arg->len = 8 - len;
+ m_arg->contents = zeroes;
+#if TARGET_BYTE_ORDER == BIG_ENDIAN
+ m_arg->offset = accumulate_size;
+#else
+ m_arg->offset = accumulate_size + len;
+#endif
+ ++fake_args;
+ }
+ accumulate_size = (accumulate_size + len + 7) & ~8;
+ }
+#endif
}
accumulate_size = (accumulate_size + 7) & (-8);
- if (accumulate_size < 16) accumulate_size = 16;
+ if (accumulate_size < 4 * MIPS_REGSIZE)
+ accumulate_size = 4 * MIPS_REGSIZE;
sp -= accumulate_size;
- for (i = nargs; m_arg--, --i >= 0; )
+ for (i = nargs + fake_args; m_arg--, --i >= 0; )
write_memory(sp + m_arg->offset, m_arg->contents, m_arg->len);
- if (struct_return) {
- buf = struct_addr;
- write_memory(sp, (char *)&buf, sizeof(CORE_ADDR));
- }
+ if (struct_return)
+ {
+ char buf[TARGET_PTR_BIT / HOST_CHAR_BIT];
+
+ store_address (buf, sizeof buf, struct_addr);
+ write_memory (sp, buf, sizeof buf);
+ }
return sp;
}
/* MASK(i,j) == (1<<i) + (1<<(i+1)) + ... + (1<<j)). Assume i<=j<31. */
-#define MASK(i,j) ((1 << (j)+1)-1 ^ (1 << (i))-1)
+#define MASK(i,j) (((1 << ((j)+1))-1) ^ ((1 << (i))-1))
void
mips_push_dummy_frame()
{
+ char buffer[MAX_REGISTER_RAW_SIZE];
int ireg;
struct linked_proc_info *link = (struct linked_proc_info*)
xmalloc(sizeof(struct linked_proc_info));
mips_extra_func_info_t proc_desc = &link->info;
CORE_ADDR sp = read_register (SP_REGNUM);
CORE_ADDR save_address;
- REGISTER_TYPE buffer;
link->next = linked_proc_desc_table;
linked_proc_desc_table = link;
#define PUSH_FP_REGNUM 16 /* must be a register preserved across calls */
* Saved D18 (i.e. F19, F18)
* ...
* Saved D0 (i.e. F1, F0)
- * CALL_DUMMY (subroutine stub; see m-mips.h)
+ * CALL_DUMMY (subroutine stub; see tm-mips.h)
* Parameter build area (not yet implemented)
* (low memory)
*/
PROC_REG_MASK(proc_desc) = GEN_REG_SAVE_MASK;
- PROC_FREG_MASK(proc_desc) = FLOAT_REG_SAVE_MASK;
+ PROC_FREG_MASK(proc_desc) = mips_fpu ? FLOAT_REG_SAVE_MASK : 0;
PROC_REG_OFFSET(proc_desc) = /* offset of (Saved R31) from FP */
-sizeof(long) - 4 * SPECIAL_REG_SAVE_COUNT;
PROC_FREG_OFFSET(proc_desc) = /* offset of (Saved D18) from FP */
for (ireg = 32; --ireg >= 0; )
if (PROC_REG_MASK(proc_desc) & (1 << ireg))
{
- buffer = read_register (ireg);
- write_memory (save_address, (char *)&buffer, sizeof(REGISTER_TYPE));
+ read_register_gen (ireg, buffer);
+
+ /* Need to fix the save_address decrement below, and also make sure
+ that we don't run into problems with the size of the dummy frame
+ or any of the offsets within it. */
+ if (REGISTER_RAW_SIZE (ireg) > 4)
+ error ("Cannot call functions on mips64");
+
+ write_memory (save_address, buffer, REGISTER_RAW_SIZE (ireg));
save_address -= 4;
}
- /* save floating-points registers */
- save_address = sp + PROC_FREG_OFFSET(proc_desc);
+ /* save floating-points registers starting with high order word */
+ save_address = sp + PROC_FREG_OFFSET(proc_desc) + 4;
for (ireg = 32; --ireg >= 0; )
if (PROC_FREG_MASK(proc_desc) & (1 << ireg))
{
- buffer = read_register (ireg + FP0_REGNUM);
- write_memory (save_address, (char *)&buffer, 4);
+ read_register_gen (ireg + FP0_REGNUM, buffer);
+
+ if (REGISTER_RAW_SIZE (ireg + FP0_REGNUM) > 4)
+ error ("Cannot call functions on mips64");
+
+ write_memory (save_address, buffer,
+ REGISTER_RAW_SIZE (ireg + FP0_REGNUM));
save_address -= 4;
}
write_register (PUSH_FP_REGNUM, sp);
PROC_FRAME_REG(proc_desc) = PUSH_FP_REGNUM;
PROC_FRAME_OFFSET(proc_desc) = 0;
- buffer = read_register (PC_REGNUM);
- write_memory (sp - 4, (char *)&buffer, sizeof(REGISTER_TYPE));
- buffer = read_register (HI_REGNUM);
- write_memory (sp - 8, (char *)&buffer, sizeof(REGISTER_TYPE));
- buffer = read_register (LO_REGNUM);
- write_memory (sp - 12, (char *)&buffer, sizeof(REGISTER_TYPE));
- buffer = read_register (FCRCS_REGNUM);
- write_memory (sp - 16, (char *)&buffer, sizeof(REGISTER_TYPE));
- sp -= 4 * (GEN_REG_SAVE_COUNT+FLOAT_REG_SAVE_COUNT+SPECIAL_REG_SAVE_COUNT);
+ read_register_gen (PC_REGNUM, buffer);
+ write_memory (sp - 4, buffer, REGISTER_RAW_SIZE (PC_REGNUM));
+ read_register_gen (HI_REGNUM, buffer);
+ write_memory (sp - 8, buffer, REGISTER_RAW_SIZE (HI_REGNUM));
+ read_register_gen (LO_REGNUM, buffer);
+ write_memory (sp - 12, buffer, REGISTER_RAW_SIZE (LO_REGNUM));
+ if (mips_fpu)
+ read_register_gen (FCRCS_REGNUM, buffer);
+ else
+ memset (buffer, 0, REGISTER_RAW_SIZE (FCRCS_REGNUM));
+ write_memory (sp - 16, buffer, REGISTER_RAW_SIZE (FCRCS_REGNUM));
+ sp -= 4 * (GEN_REG_SAVE_COUNT
+ + (mips_fpu ? FLOAT_REG_SAVE_COUNT : 0)
+ + SPECIAL_REG_SAVE_COUNT);
write_register (SP_REGNUM, sp);
PROC_LOW_ADDR(proc_desc) = sp - CALL_DUMMY_SIZE + CALL_DUMMY_START_OFFSET;
PROC_HIGH_ADDR(proc_desc) = sp;
void
mips_pop_frame()
-{ register int regnum;
+{
+ register int regnum;
FRAME frame = get_current_frame ();
CORE_ADDR new_sp = frame->frame;
+
mips_extra_func_info_t proc_desc = frame->proc_desc;
- if (PROC_DESC_IS_DUMMY(proc_desc))
+
+ write_register (PC_REGNUM, FRAME_SAVED_PC(frame));
+ if (frame->saved_regs == NULL)
+ mips_find_saved_regs (frame);
+ if (proc_desc)
{
- struct linked_proc_info **ptr = &linked_proc_desc_table;;
- for (; &ptr[0]->info != proc_desc; ptr = &ptr[0]->next )
- if (ptr[0] == NULL) abort();
- *ptr = ptr[0]->next;
- free (ptr[0]);
- write_register (HI_REGNUM, read_memory_integer(new_sp - 8, 4));
- write_register (LO_REGNUM, read_memory_integer(new_sp - 12, 4));
- write_register (FCRCS_REGNUM, read_memory_integer(new_sp - 16, 4));
+ for (regnum = 32; --regnum >= 0; )
+ if (PROC_REG_MASK(proc_desc) & (1 << regnum))
+ write_register (regnum,
+ read_memory_integer (frame->saved_regs->regs[regnum],
+ 4));
+ for (regnum = 32; --regnum >= 0; )
+ if (PROC_FREG_MASK(proc_desc) & (1 << regnum))
+ write_register (regnum + FP0_REGNUM,
+ read_memory_integer (frame->saved_regs->regs[regnum + FP0_REGNUM], 4));
}
- write_register (PC_REGNUM, FRAME_SAVED_PC(frame));
- if (frame->proc_desc) {
- for (regnum = 32; --regnum >= 0; )
- if (PROC_REG_MASK(proc_desc) & (1 << regnum))
- write_register (regnum,
- read_memory_integer (frame->saved_regs->regs[regnum], 4));
- for (regnum = 32; --regnum >= 0; )
- if (PROC_FREG_MASK(proc_desc) & (1 << regnum))
- write_register (regnum + FP0_REGNUM,
- read_memory_integer (frame->saved_regs->regs[regnum + FP0_REGNUM], 4));
- }
write_register (SP_REGNUM, new_sp);
flush_cached_frames ();
- set_current_frame (create_new_frame (new_sp, read_pc ()));
+ /* We let mips_init_extra_frame_info figure out the frame pointer */
+ set_current_frame (create_new_frame (0, read_pc ()));
+
+ if (proc_desc && PROC_DESC_IS_DUMMY(proc_desc))
+ {
+ struct linked_proc_info *pi_ptr, *prev_ptr;
+
+ for (pi_ptr = linked_proc_desc_table, prev_ptr = NULL;
+ pi_ptr != NULL;
+ prev_ptr = pi_ptr, pi_ptr = pi_ptr->next)
+ {
+ if (&pi_ptr->info == proc_desc)
+ break;
+ }
+
+ if (pi_ptr == NULL)
+ error ("Can't locate dummy extra frame info\n");
+
+ if (prev_ptr != NULL)
+ prev_ptr->next = pi_ptr->next;
+ else
+ linked_proc_desc_table = pi_ptr->next;
+
+ free (pi_ptr);
+
+ write_register (HI_REGNUM, read_memory_integer(new_sp - 8, 4));
+ write_register (LO_REGNUM, read_memory_integer(new_sp - 12, 4));
+ if (mips_fpu)
+ write_register (FCRCS_REGNUM, read_memory_integer(new_sp - 16, 4));
+ }
}
static void
mips_print_register (regnum, all)
int regnum, all;
{
- unsigned char raw_buffer[MAX_REGISTER_RAW_SIZE * 2]; /* *2 for doubles */
- REGISTER_TYPE val;
+ unsigned char raw_buffer[MAX_REGISTER_RAW_SIZE];
+ struct type *our_type =
+ init_type (TYPE_CODE_INT,
+ /* We will fill in the length for each register. */
+ 0,
+ TYPE_FLAG_UNSIGNED,
+ NULL,
+ NULL);
+
+ /* Get the data in raw format. */
+ if (read_relative_register_raw_bytes (regnum, raw_buffer))
+ {
+ printf_filtered ("%s: [Invalid]", reg_names[regnum]);
+ return;
+ }
- /* Get the data in raw format. */
- if (read_relative_register_raw_bytes (regnum, raw_buffer))
- {
- printf_filtered ("%s: [Invalid]", reg_names[regnum]);
- return;
- }
-
- /* If an even floating pointer register, also print as double. */
- if (regnum >= FP0_REGNUM && regnum < FP0_REGNUM+32
- && !((regnum-FP0_REGNUM) & 1)) {
- read_relative_register_raw_bytes (regnum+1, raw_buffer+4);
- printf_filtered ("(d%d: ", regnum-FP0_REGNUM);
- val_print (builtin_type_double, raw_buffer, 0,
- stdout, 0, 1, 0, Val_pretty_default);
- printf_filtered ("); ");
- }
- fputs_filtered (reg_names[regnum], stdout);
-#ifndef NUMERIC_REG_NAMES
- if (regnum < 32)
- printf_filtered ("(r%d): ", regnum);
- else
+ /* If an even floating pointer register, also print as double. */
+ if (regnum >= FP0_REGNUM && regnum < FP0_REGNUM+32
+ && !((regnum-FP0_REGNUM) & 1)) {
+ char dbuffer[MAX_REGISTER_RAW_SIZE];
+
+ read_relative_register_raw_bytes (regnum, dbuffer);
+ read_relative_register_raw_bytes (regnum+1, dbuffer+4);
+#ifdef REGISTER_CONVERT_TO_TYPE
+ REGISTER_CONVERT_TO_TYPE(regnum, builtin_type_double, dbuffer);
#endif
- printf_filtered (": ");
+ printf_filtered ("(d%d: ", regnum-FP0_REGNUM);
+ val_print (builtin_type_double, dbuffer, 0,
+ gdb_stdout, 0, 1, 0, Val_pretty_default);
+ printf_filtered ("); ");
+ }
+ fputs_filtered (reg_names[regnum], gdb_stdout);
+
+ /* The problem with printing numeric register names (r26, etc.) is that
+ the user can't use them on input. Probably the best solution is to
+ fix it so that either the numeric or the funky (a2, etc.) names
+ are accepted on input. */
+ if (regnum < 32)
+ printf_filtered ("(r%d): ", regnum);
+ else
+ printf_filtered (": ");
- /* If virtual format is floating, print it that way. */
- if (TYPE_CODE (REGISTER_VIRTUAL_TYPE (regnum)) == TYPE_CODE_FLT
- && ! INVALID_FLOAT (raw_buffer, REGISTER_VIRTUAL_SIZE(regnum))) {
- val_print (REGISTER_VIRTUAL_TYPE (regnum), raw_buffer, 0,
- stdout, 0, 1, 0, Val_pretty_default);
- }
- /* Else print as integer in hex. */
- else
- {
- long val;
-
- bcopy (raw_buffer, &val, sizeof (long));
- SWAP_TARGET_AND_HOST ((char *)&val, sizeof (long));
- if (val == 0)
- printf_filtered ("0");
- else if (all)
- printf_filtered (local_hex_format(), val);
- else
- printf_filtered ("%s=%d", local_hex_string(val), val);
- }
+ /* If virtual format is floating, print it that way. */
+ if (TYPE_CODE (REGISTER_VIRTUAL_TYPE (regnum)) == TYPE_CODE_FLT
+ && ! INVALID_FLOAT (raw_buffer, REGISTER_VIRTUAL_SIZE(regnum))) {
+ val_print (REGISTER_VIRTUAL_TYPE (regnum), raw_buffer, 0,
+ gdb_stdout, 0, 1, 0, Val_pretty_default);
+ }
+ /* Else print as integer in hex. */
+ else
+ {
+ print_scalar_formatted (raw_buffer,
+ REGISTER_VIRTUAL_TYPE (regnum),
+ 'x',
+ 0,
+ gdb_stdout);
+ }
}
/* Replacement for generic do_registers_info. */
#endif
return -1;
}
-
\f
-/* Bad floats: Returns 0 if P points to a valid IEEE floating point number,
- 1 if P points to a denormalized number or a NaN. LEN says whether this is
- a single-precision or double-precision float */
-#define SINGLE_EXP_BITS 8
-#define DOUBLE_EXP_BITS 11
+/* Is this a branch with a delay slot? */
+static int is_delayed PARAMS ((unsigned long));
+
+static int
+is_delayed (insn)
+ unsigned long insn;
+{
+ int i;
+ for (i = 0; i < NUMOPCODES; ++i)
+ if (mips_opcodes[i].pinfo != INSN_MACRO
+ && (insn & mips_opcodes[i].mask) == mips_opcodes[i].match)
+ break;
+ return (i < NUMOPCODES
+ && (mips_opcodes[i].pinfo & (INSN_UNCOND_BRANCH_DELAY
+ | INSN_COND_BRANCH_DELAY
+ | INSN_COND_BRANCH_LIKELY)));
+}
+
int
-isa_NAN(p, len)
- int *p, len;
+mips_step_skips_delay (pc)
+ CORE_ADDR pc;
{
- int exponent;
- if (len == 4)
- {
- exponent = *p;
- exponent = exponent << 1 >> (32 - SINGLE_EXP_BITS - 1);
- return ((exponent == -1) || (! exponent && *p));
- }
- else if (len == 8)
- {
- exponent = *(p+1);
- exponent = exponent << 1 >> (32 - DOUBLE_EXP_BITS - 1);
- return ((exponent == -1) || (! exponent && *p * *(p+1)));
- }
- else return 1;
+ char buf[4];
+
+ if (target_read_memory (pc, buf, 4) != 0)
+ /* If error reading memory, guess that it is not a delayed branch. */
+ return 0;
+ return is_delayed (extract_unsigned_integer (buf, 4));
}
-\f
-/* To skip prologues, I use this predicate. Returns either PC
- itself if the code at PC does not look like a function prologue,
- PC+4 if it does (our caller does not need anything more fancy). */
+
+/* To skip prologues, I use this predicate. Returns either PC itself
+ if the code at PC does not look like a function prologue; otherwise
+ returns an address that (if we're lucky) follows the prologue. If
+ LENIENT, then we must skip everything which is involved in setting
+ up the frame (it's OK to skip more, just so long as we don't skip
+ anything which might clobber the registers which are being saved.
+ We must skip more in the case where part of the prologue is in the
+ delay slot of a non-prologue instruction). */
CORE_ADDR
-mips_skip_prologue(pc)
+mips_skip_prologue (pc, lenient)
CORE_ADDR pc;
+ int lenient;
{
- struct symbol *f;
- struct block *b;
unsigned long inst;
int offset;
+ int seen_sp_adjust = 0;
+
+ /* Skip the typical prologue instructions. These are the stack adjustment
+ instruction and the instructions that save registers on the stack
+ or in the gcc frame. */
+ for (offset = 0; offset < 100; offset += 4)
+ {
+ char buf[4];
+ int status;
- /* For -g modules and most functions anyways the
- first instruction adjusts the stack.
- But we allow some number of stores before the stack adjustment.
- (These are emitted by varags functions compiled by gcc-2.0. */
- for (offset = 0; offset < 100; offset += 4) {
- inst = read_memory_integer(pc + offset, 4);
- if ((inst & 0xffff0000) == 0x27bd0000) /* addiu $sp,$sp,offset */
- return pc + offset + 4;
- if ((inst & 0xFFE00000) != 0xAFA00000) /* sw reg,n($sp) */
+ status = read_memory_nobpt (pc + offset, buf, 4);
+ if (status)
+ memory_error (status, pc + offset);
+ inst = extract_unsigned_integer (buf, 4);
+
+#if 0
+ if (lenient && is_delayed (inst))
+ continue;
+#endif
+
+ if ((inst & 0xffff0000) == 0x27bd0000) /* addiu $sp,$sp,offset */
+ seen_sp_adjust = 1;
+ else if ((inst & 0xFFE00000) == 0xAFA00000 && (inst & 0x001F0000))
+ continue; /* sw reg,n($sp) */
+ /* reg != $zero */
+ else if ((inst & 0xFFE00000) == 0xE7A00000) /* swc1 freg,n($sp) */
+ continue;
+ else if ((inst & 0xF3E00000) == 0xA3C00000 && (inst & 0x001F0000))
+ /* sx reg,n($s8) */
+ continue; /* reg != $zero */
+ else if (inst == 0x03A0F021) /* move $s8,$sp */
+ continue;
+ else if ((inst & 0xFF9F07FF) == 0x00800021) /* move reg,$a0-$a3 */
+ continue;
+ else if ((inst & 0xffff0000) == 0x3c1c0000) /* lui $gp,n */
+ continue;
+ else if ((inst & 0xffff0000) == 0x279c0000) /* addiu $gp,$gp,n */
+ continue;
+ else if (inst == 0x0399e021 /* addu $gp,$gp,$t9 */
+ || inst == 0x033ce021) /* addu $gp,$t9,$gp */
+ continue;
+ else
break;
}
+ return pc + offset;
+
+/* FIXME schauer. The following code seems no longer necessary if we
+ always skip the typical prologue instructions. */
+
+#if 0
+ if (seen_sp_adjust)
+ return pc + offset;
/* Well, it looks like a frameless. Let's make sure.
Note that we are not called on the current PC,
b = block_for_pc(pc);
if (!b) return pc;
- f = lookup_symbol(".gdbinfo.", b, LABEL_NAMESPACE, 0, NULL);
+ f = lookup_symbol(MIPS_EFI_SYMBOL_NAME, b, LABEL_NAMESPACE, 0, NULL);
if (!f) return pc;
/* Ideally, I would like to use the adjusted info
from mips_frame_info(), but for all practical
return pc + 4;
return pc;
+#endif
}
-/* Figure out where the longjmp will land.
- We expect the first arg to be a pointer to the jmp_buf structure from which
- we extract the pc (JB_PC) that we will land at. The pc is copied into PC.
- This routine returns true on success. */
+#if 0
+/* The lenient prologue stuff should be superceded by the code in
+ init_extra_frame_info which looks to see whether the stores mentioned
+ in the proc_desc have actually taken place. */
-int
-get_longjmp_target(pc)
- CORE_ADDR *pc;
+/* Is address PC in the prologue (loosely defined) for function at
+ STARTADDR? */
+
+static int
+mips_in_lenient_prologue (startaddr, pc)
+ CORE_ADDR startaddr;
+ CORE_ADDR pc;
{
- CORE_ADDR jb_addr;
+ CORE_ADDR end_prologue = mips_skip_prologue (startaddr, 1);
+ return pc >= startaddr && pc < end_prologue;
+}
+#endif
- jb_addr = read_register(A0_REGNUM);
+/* Given a return value in `regbuf' with a type `valtype',
+ extract and copy its value into `valbuf'. */
+void
+mips_extract_return_value (valtype, regbuf, valbuf)
+ struct type *valtype;
+ char regbuf[REGISTER_BYTES];
+ char *valbuf;
+{
+ int regnum;
+
+ regnum = TYPE_CODE (valtype) == TYPE_CODE_FLT && mips_fpu ? FP0_REGNUM : 2;
- if (target_read_memory(jb_addr + JB_PC * JB_ELEMENT_SIZE, pc,
- sizeof(CORE_ADDR)))
- return 0;
+ memcpy (valbuf, regbuf + REGISTER_BYTE (regnum), TYPE_LENGTH (valtype));
+#ifdef REGISTER_CONVERT_TO_TYPE
+ REGISTER_CONVERT_TO_TYPE(regnum, valtype, valbuf);
+#endif
+}
+
+/* Given a return value in `regbuf' with a type `valtype',
+ write it's value into the appropriate register. */
+void
+mips_store_return_value (valtype, valbuf)
+ struct type *valtype;
+ char *valbuf;
+{
+ int regnum;
+ char raw_buffer[MAX_REGISTER_RAW_SIZE];
+
+ regnum = TYPE_CODE (valtype) == TYPE_CODE_FLT && mips_fpu ? FP0_REGNUM : 2;
+ memcpy(raw_buffer, valbuf, TYPE_LENGTH (valtype));
+
+#ifdef REGISTER_CONVERT_FROM_TYPE
+ REGISTER_CONVERT_FROM_TYPE(regnum, valtype, raw_buffer);
+#endif
+
+ write_register_bytes(REGISTER_BYTE (regnum), raw_buffer, TYPE_LENGTH (valtype));
+}
+
+/* These exist in mdebugread.c. */
+extern CORE_ADDR sigtramp_address, sigtramp_end;
+extern void fixup_sigtramp PARAMS ((void));
- SWAP_TARGET_AND_HOST(pc, sizeof(CORE_ADDR));
+/* Exported procedure: Is PC in the signal trampoline code */
- return 1;
+int
+in_sigtramp (pc, ignore)
+ CORE_ADDR pc;
+ char *ignore; /* function name */
+{
+ if (sigtramp_address == 0)
+ fixup_sigtramp ();
+ return (pc >= sigtramp_address && pc < sigtramp_end);
+}
+
+static void reinit_frame_cache_sfunc PARAMS ((char *, int,
+ struct cmd_list_element *));
+
+/* Just like reinit_frame_cache, but with the right arguments to be
+ callable as an sfunc. */
+static void
+reinit_frame_cache_sfunc (args, from_tty, c)
+ char *args;
+ int from_tty;
+ struct cmd_list_element *c;
+{
+ reinit_frame_cache ();
+}
+
+void
+_initialize_mips_tdep ()
+{
+ struct cmd_list_element *c;
+
+ /* Let the user turn off floating point and set the fence post for
+ heuristic_proc_start. */
+
+ add_show_from_set
+ (add_set_cmd ("mipsfpu", class_support, var_boolean,
+ (char *) &mips_fpu,
+ "Set use of floating point coprocessor.\n\
+Turn off to avoid using floating point instructions when calling functions\n\
+or dealing with return values.", &setlist),
+ &showlist);
+
+ /* We really would like to have both "0" and "unlimited" work, but
+ command.c doesn't deal with that. So make it a var_zinteger
+ because the user can always use "999999" or some such for unlimited. */
+ c = add_set_cmd ("heuristic-fence-post", class_support, var_zinteger,
+ (char *) &heuristic_fence_post,
+ "\
+Set the distance searched for the start of a function.\n\
+If you are debugging a stripped executable, GDB needs to search through the\n\
+program for the start of a function. This command sets the distance of the\n\
+search. The only need to set it is when debugging a stripped executable.",
+ &setlist);
+ /* We need to throw away the frame cache when we set this, since it
+ might change our ability to get backtraces. */
+ c->function.sfunc = reinit_frame_cache_sfunc;
+ add_show_from_set (c, &showlist);
}
projects / binutils.git / blobdiff