Testing: I tested on native Red Hat Linux 7 using gcc 3.0.
The "info float" command exercises print_i387_value.
2001-07-06 Michael Chastain <chastain@redhat.com>
* i387-tdep.c (print_i387_value): Fix pointer glitch.
===
Index: i387-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/i387-tdep.c,v
retrieving revision 1.11
diff -c -1 -0 -p -r1.11 i387-tdep.c
*** gdb/i387-tdep.c 2001/07/04 21:14:05 1.11
--- gdb/i387-tdep.c 2001/07/06 12:47:00
*************** print_i387_value (char *raw)
*** 163,184 ****
DOUBLEST value;
int len = TARGET_LONG_DOUBLE_BIT / TARGET_CHAR_BIT;
char *tmp = alloca (len);
/* This code only works on targets where ... */
gdb_assert (TARGET_LONG_DOUBLE_FORMAT == &floatformat_i387_ext);
/* Take care of the padding. FP reg is 80 bits. The same value in
memory is 96 bits. */
gdb_assert (FPU_REG_RAW_SIZE < len);
! memcpy (&tmp, raw, FPU_REG_RAW_SIZE);
! memset (&tmp + FPU_REG_RAW_SIZE, 0, len - FPU_REG_RAW_SIZE);
/* Extract the value as a DOUBLEST. */
/* Use extract_floating() rather than floatformat_to_doublest().
The latter is lossy in nature. Once GDB gets a host/target
independent and non-lossy FP it will become possible to bypass
extract_floating() and call floatformat*() directly. Note also
the assumptions about TARGET_LONG_DOUBLE above. */
value = extract_floating (tmp, len);
/* We try to print 19 digits. The last digit may or may not contain
--- 163,184 ----
DOUBLEST value;
int len = TARGET_LONG_DOUBLE_BIT / TARGET_CHAR_BIT;
char *tmp = alloca (len);
/* This code only works on targets where ... */
gdb_assert (TARGET_LONG_DOUBLE_FORMAT == &floatformat_i387_ext);
/* Take care of the padding. FP reg is 80 bits. The same value in
memory is 96 bits. */
gdb_assert (FPU_REG_RAW_SIZE < len);
! memcpy (tmp, raw, FPU_REG_RAW_SIZE);
! memset (tmp + FPU_REG_RAW_SIZE, 0, len - FPU_REG_RAW_SIZE);
/* Extract the value as a DOUBLEST. */
/* Use extract_floating() rather than floatformat_to_doublest().
The latter is lossy in nature. Once GDB gets a host/target
independent and non-lossy FP it will become possible to bypass
extract_floating() and call floatformat*() directly. Note also
the assumptions about TARGET_LONG_DOUBLE above. */
value = extract_floating (tmp, len);
/* We try to print 19 digits. The last digit may or may not contain
commented with "paranoia" into gdb_assert.
(select_event_lwp_callback): Likewise.
(cancel_breakpoints_callback): Bail out early if LP is the event
LWP. Add comment about backup up breakpoints. Fix formatting and
debug message.
(select_event_lwp): Make solely repsonsible for switching event
LWP. Fix formatting and remove bogus "ERROR" debug message.
Don't backup breakpoints from here.
(lin_lwp_wait): Don't touch LP->status, let select_event_lwp
handle that. Only call select_event_lwp if we're not waiting for
a specific LWP, i.e. when PID == -1. Backup breakpoints from here.
* lin-lwp.c: Prevent thread starvation by using a monte carlo
method to choose which of several event threads to handle next.
(stop_wait_callback): Defer pushback of breakpoint events until
later; add SIGTRAP events to the queue of unhandled events.
Keep calling waitpid until SIGSTOP retrieved. If more than one
non-SIGSTOP event is retrieved, push them back onto the process
queue using kill.
(count_events_callback, select_singlestep_lwp_callback,
select_event_lwp_callback, cancel_breakpoints_callback,
select_event_lwp): New functions. Implement monte carlo method
for selecting which of several SIGTRAP threads to handle next.
Push back the breakpoint event for all threads other than the
selected one.
(lin_lwp_wait): Call select_event_lwp to decide which of several
sigtrapped lwps to handle next.
(resume_callback): Disable code that attempts to handle
step_resume breakpoints. Let core gdb handle this.
show_regs, d10v_read_pc, d10v_write_pc, d10v_read_sp,
d10v_write_sp, d10v_write_fp, d10v_read_fp,
d10v_push_return_address): Call the functions d10v_make_daddr,
d10v_make_iaddr, d10v_convert_iaddr_to_raw, and
d10v_convert_daddr_to_raw, not the global macros D10V_MAKE_DADDR,
D10V_MAKE_IADDR, D10V_CONVERT_IADDR_TO_RAW, and
D10V_CONVERT_DADDR_TO_RAW.
structure and not macros.
(Host Definition): Document that much of this chapter is obsolete.
(Target Architecture Definition): Update list of files that make
up a target architecture.
(Coding): Update.
(read_partial_die): Delete fourth argument; we return this info in
the struct partial_die_info object itself now.
(dwarf2_build_psymtabs_hard, scan_partial_symbols): Use the
has_pc_info field of the partial die struct, rather than passing a
variable by reference to read_partial_die.
code in loop condition. This seemed to be trying to round
info_ptr up to the next four-byte boundary, but that's not what it
actually did. If we discover the problem the old code was really
trying to address, we can fix it properly.
* thread-db.c (attach_thread): Check for TD_THR_ZOMBIE in addition
to TD_THR_UNKNOWN when looking for defunct zombie threads.
(thread_db_thread_alive): Ditto.
(find_new_threads_callback): Ditto.