This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] h8300 "info registers" broken


On 02/08/2014 06:36 PM, Yoshinori Sato wrote:


> #10 0x00000000005b954e in frame_unwind_register (frame=<optimized out>, 
>     regnum=13, buf=<optimized out>) at ../../gdb/frame.c:1064

Hard to reason about an optimized build...  Please try with "-g3 -O0".

I don't have a h8300-elf toolchain handy, and the h8300-linux
toolchain I found doesn't seem to want to link executables,
but I managed to try something by building an .o file, and debugging
that.  I don't see a crash, but instead GDB complains CCR
is unavailable.

(gdb) info target
Symbols from "/home/pedro/h8300-main.o".
simulator:
        Attached to sim running program /home/pedro/h8300-main.o

#instructions executed           0
#cycles (v approximate)          0
#real time taken            0.0000
#virtual time taken         0.0000
#compiles                        0
#cache size                   1024
        While running this, GDB does not access memory from...
Local exec file:
        `/home/pedro/h8300-main.o', file type elf32-h8300.
        Entry point: 0x0
        0x00000000 - 0x00000024 is .text
        0x00000024 - 0x00000024 is .data
        0x00000024 - 0x00000024 is .bss
(gdb) b *0
Breakpoint 1 at 0x0: file main.c, line 4.
(gdb) r
Starting program: /home/pedro/h8300-main.o

Breakpoint 1, foo (i=0x0 <foo>) at main.c:4
4       {
(gdb) info registers
r0             0x0000  0
r1             0x0000  0
r2             0x0000  0
r3             0x0000  0
r4             0x0000  0
r5             0x0000  0
r6             0x0000  0
sp             0x0000  0
Register 13 is not available
(gdb) info registers ccr
Register 13 is not available

The problem seems to me that the h8300 port does not define
a register_sim_regno gdbarch hook, and thus when fetching
registers off of the sim, we end up in legacy_register_sim_regno
trying to figure out the sim register number for the raw CCR register:

int
legacy_register_sim_regno (struct gdbarch *gdbarch, int regnum)
{
  /* Only makes sense to supply raw registers.  */
  gdb_assert (regnum >= 0 && regnum < gdbarch_num_regs (gdbarch));
  /* NOTE: cagney/2002-05-13: The old code did it this way and it is
     suspected that some GDB/SIM combinations may rely on this
     behavour.  The default should be one2one_register_sim_regno
     (below).  */
  if (gdbarch_register_name (gdbarch, regnum) != NULL
      && gdbarch_register_name (gdbarch, regnum)[0] != '\0')
    return regnum;
  else
    return LEGACY_SIM_REGNO_IGNORE;
}

And because the raw ccr register does not have a name, that returns
LEGACY_SIM_REGNO_IGNORE.  Which means that we never actually
read the ccr raw value.  Before the <unavailable> support, this
must have meant that ccr was _always_ read as 0...  At least, I'm
not seeing how this ever worked.

Looking at sim/h8300/sim-main.h, it seems like the sim's register
numbers are compatible with gdb's.

This patch below "works" for me, as in, I can now print CCR,
but that's about all I tested (and am willing to test) myself.

Look me know how this looks to you.

---
 gdb/h8300-tdep.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/gdb/h8300-tdep.c b/gdb/h8300-tdep.c
index ffffbc9..ac34a9b 100644
--- a/gdb/h8300-tdep.c
+++ b/gdb/h8300-tdep.c
@@ -939,6 +939,20 @@ h8300h_return_value (struct gdbarch *gdbarch, struct value *function,

 static struct cmd_list_element *setmachinelist;

+static int
+h8300_register_sim_regno (struct gdbarch *gdbarch, int regnum)
+{
+  /* Only makes sense to supply raw registers.  */
+  gdb_assert (regnum >= 0 && regnum < gdbarch_num_regs (gdbarch));
+
+  /* We hide the raw ccr from the user by making it nameless.  Because
+     the default register_sim_regno hook returns
+     LEGACY_SIM_REGNO_IGNORE for unnamed registers, we need to
+     override it.  The sim register numbering is compatible with
+     gdb's, so there isn't anything to do.  */
+  return regnum;
+}
+
 static const char *
 h8300_register_name (struct gdbarch *gdbarch, int regno)
 {
@@ -1230,6 +1244,8 @@ h8300_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)

   gdbarch = gdbarch_alloc (&info, 0);

+  set_gdbarch_register_sim_regno (gdbarch, h8300_register_sim_regno);
+
   switch (info.bfd_arch_info->mach)
     {
     case bfd_mach_h8300:
-- 
1.7.11.7



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]