All of lore.kernel.org
 help / color / mirror / Atom feed
* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
@ 2017-01-30 11:21 John Paul Adrian Glaubitz
  2017-01-31 20:41 ` Rob Landley
                   ` (6 more replies)
  0 siblings, 7 replies; 8+ messages in thread
From: John Paul Adrian Glaubitz @ 2017-01-30 11:21 UTC (permalink / raw)
  To: linux-sh

Hi!

We're currently working on improving SuperH support in Debian and Linux
in general since the architecture is currently being redeveloped as an
open source architecture called J-Core [1]. J-Core supports the SH-2
instruction set only, but it's expected to extend to SH-3 and SH-4
in the future. Hence, having working native SH support for gdb on Linux
would be very desirable.

>  * Tom Tromey <tromey@redhat.com> pointed out that any significant
>    contributors would need to sign copyright assignment papers, since
>    the patch is far from trivial.  Otherwise, GNU won't merge it, and it
>    will bitrot constantly.
>    <http://www.cygwin.com/ml/gdb-patches/2011-11/msg00499.html>

This would actually the first and most important thing to do. I have
tried contacting Takashi YOSHII but so far I had no success. One that
could help would be Nobuhiro Iwamatsu, so I put him in CC. There is
also Yutaka Niibe whom we can ask.

Adrian

> [1] http://j-core.org/

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
  2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
@ 2017-01-31 20:41 ` Rob Landley
  2017-01-31 20:47 ` John Paul Adrian Glaubitz
                   ` (5 subsequent siblings)
  6 siblings, 0 replies; 8+ messages in thread
From: Rob Landley @ 2017-01-31 20:41 UTC (permalink / raw)
  To: linux-sh

On 01/30/2017 05:21 AM, John Paul Adrian Glaubitz wrote:
> Hi!
> 
> We're currently working on improving SuperH support in Debian and Linux
> in general since the architecture is currently being redeveloped as an
> open source architecture called J-Core [1]. J-Core supports the SH-2
> instruction set only, but it's expected to extend to SH-3 and SH-4
> in the future. Hence, having working native SH support for gdb on Linux
> would be very desirable.
> 
>>  * Tom Tromey <tromey@redhat.com> pointed out that any significant
>>    contributors would need to sign copyright assignment papers, since
>>    the patch is far from trivial.  Otherwise, GNU won't merge it, and it
>>    will bitrot constantly.
>>    <http://www.cygwin.com/ml/gdb-patches/2011-11/msg00499.html>
> 
> This would actually the first and most important thing to do. I have
> tried contacting Takashi YOSHII but so far I had no success.

I asked the j-core engineers in Japan if they knew him, but the name
didn't ring any bells. (Jeff Dionne might know, but he's busy at
Distributech this week.)

Rob

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
  2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
  2017-01-31 20:41 ` Rob Landley
@ 2017-01-31 20:47 ` John Paul Adrian Glaubitz
  2017-02-07 20:51 ` John Paul Adrian Glaubitz
                   ` (4 subsequent siblings)
  6 siblings, 0 replies; 8+ messages in thread
From: John Paul Adrian Glaubitz @ 2017-01-31 20:47 UTC (permalink / raw)
  To: linux-sh

On 01/31/2017 09:41 PM, Rob Landley wrote:
> I asked the j-core engineers in Japan if they knew him, but the name
> didn't ring any bells. (Jeff Dionne might know, but he's busy at
> Distributech this week.)

I asked Yutaka Niibe and he knows him back from Hitachi and he said, he
still has a channel to contact him. He promised to try to reach out to
him and ask for the copyright assignment.

Yutaka was involved hiring all the developers for Hitachi/Renesas back
then when they started porting Linux to the architecture, that's why
he knows most of the people involved. He also got me in touch with
Renesas in Tokyo who sent me the last SH-7724 Evoboard they had in
stock ;).

As for the copyright assignment, Yutaka told me that as far as he
remembers, assigning copyrights to an external entity wasn't possible
back then which is why it didn't happen in this case. But since Hitachi
is no longer involved, it will hopefully no problem for Yoshii-san
to do the copyright assignment.

Adrian

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
  2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
  2017-01-31 20:41 ` Rob Landley
  2017-01-31 20:47 ` John Paul Adrian Glaubitz
@ 2017-02-07 20:51 ` John Paul Adrian Glaubitz
  2017-03-10  1:22 ` Nobuhiro Iwamatsu
                   ` (3 subsequent siblings)
  6 siblings, 0 replies; 8+ messages in thread
From: John Paul Adrian Glaubitz @ 2017-02-07 20:51 UTC (permalink / raw)
  To: linux-sh

[-- Attachment #1: Type: text/plain, Size: 1119 bytes --]

Hi!

I have ported Yoshii-san's code to GDB 7.12, attaching an updated
patch. A git repository with my linux-sh branch can be found in [1].

Yutaka Niibe has already reached out to Renesas in Japan and is trying to
get into touch with Yoshii-san, this might take a while though. However,
I have talked with someone at the FSFE knowledgeable with open source
licenses and it looks like we actually don't need the copyright assignment
at all.

If we needed the copyright assignment, we'd probably need it from Renesas
anyway and not from Yoshii-san directly as he wrote the patch while working
for Renesas who consequently are the copyright owners. And since Renesas
already submitted their SH code to gcc, I assume they also already assigned
the FSF copyright assignment (if that's actually needed).

More updates coming soon.

Thanks,
Adrian

> [1] https://github.com/glaubitz/binutils-gdb/tree/linux-sh

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

[-- Attachment #2: sh-linux-native-support.patch --]
[-- Type: text/x-patch, Size: 39364 bytes --]

Description: Add native support for Linux SH
Author: Takashi Yoshii <takasi-y@ops.dti.ne.jp>
        John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Last-Update: 2017-02-07

--- gdb-7.12.orig/gdb/Makefile.in
+++ gdb-7.12/gdb/Makefile.in
@@ -1750,6 +1750,7 @@ ALLDEPFILES = \
 	score-tdep.c \
 	ser-go32.c ser-pipe.c ser-tcp.c ser-mingw.c \
 	sh-tdep.c sh64-tdep.c shnbsd-tdep.c shnbsd-nat.c \
+	sh-linux-nat.c sh-linux-tdep.c \
 	sol2-tdep.c \
 	solib-svr4.c \
 	sparc-linux-nat.c sparc-linux-tdep.c \
--- /dev/null
+++ gdb-7.12/gdb/config/sh/linux.mh
@@ -0,0 +1,14 @@
+# Host: Renesas Super-H running GNU/Linux
+
+NAT_FILE= nm-linux.h
+NATDEPFILES= inf-ptrace.o fork-child.o \
+	sh-linux-nat.o \
+	proc-service.o linux-thread-db.o \
+	linux-nat.o linux-osdata.o linux-fork.o linux-procfs.o linux-ptrace.o \
+	linux-personality.o \
+	linux-waitpid.o linux-namespaces.o
+NAT_CDEPS = $(srcdir)/proc-service.list
+
+# The dynamically loaded libthread_db needs access to symbols in the
+# gdb executable.
+LOADLIBES = -ldl $(RDYNAMIC)
--- /dev/null
+++ gdb-7.12/gdb/config/sh/nm-linux.h
@@ -0,0 +1,54 @@
+/* Native-dependent definitions for SuperH running Linux, for GDB.
+   Copyright 2004 Free Software Foundation, Inc.
+
+   This file is part of GDB.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 2 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program; if not, write to the Free Software
+   Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+#ifndef NM_LINUX_H
+#define NM_LINUX_H
+
+/* Get generic Linux native definitions.  */
+#include "config/nm-linux.h"
+/* Support for the user area.  */
+
+/* Return the size of the user struct.  */
+extern int kernel_u_size (void);
+#define KERNEL_U_SIZE kernel_u_size()
+
+/* This is the amount to substract from u.u_ar0 to get the offset in
+   the core file of the register values.  */
+#define KERNEL_U_ADDR 0
+
+#define U_REGS_OFFSET 0
+
+extern CORE_ADDR register_u_addr (CORE_ADDR blockend, int regnum);
+#define REGISTER_U_ADDR(addr, blockend, regnum) \
+  (addr) = register_u_addr (blockend, regnum)
+
+/* Override copies of {fetch,store}_inferior_registers in `infptrace.c'.  */
+#define FETCH_INFERIOR_REGISTERS
+
+/* Nevertheless, define CANNOT_{FETCH,STORE}_REGISTER, because we
+   might fall back on the code `infptrace.c' (well a copy of that code
+   in `sh-linux-nat.c' for now) and we can access only the
+   general-purpose registers in that way.  */
+extern int cannot_fetch_register (int regno);
+extern int cannot_store_register (int regno);
+#define CANNOT_FETCH_REGISTER(regno) cannot_fetch_register (regno)
+#define CANNOT_STORE_REGISTER(regno) cannot_store_register (regno)
+
+#endif /* NM_LINUX_H */
--- /dev/null
+++ gdb-7.12/gdb/config/sh/xm-linux.h
@@ -0,0 +1,32 @@
+/* Native support for GNU/Linux, for GDB, the GNU debugger.
+   Copyright (C) 2000 Free Software Foundation, Inc.
+
+This file is part of GDB.
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation; either version 2 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program; if not, write to the Free Software
+Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
+
+#ifndef XM_LINUX_H
+#define XM_LINUX_H
+
+#define HOST_BYTE_ORDER LITTLE_ENDIAN
+
+#define HAVE_TERMIOS
+
+#define NEED_POSIX_SETPGID
+
+/* Need R_OK etc, but USG isn't defined.  */
+#include <unistd.h>
+
+#endif	/* #ifndef XM_LINUX_H */
--- gdb-7.12.orig/gdb/configure.host
+++ gdb-7.12/gdb/configure.host
@@ -149,6 +149,7 @@ powerpc*-*-linux*)	gdb_host=linux ;;
 
 s390*-*-linux*)		gdb_host=linux ;;
 
+sh*-*-linux*)		gdb_host=linux ;;
 sh*-*-netbsdelf* | sh*-*-knetbsd*-gnu)
 			gdb_host=nbsd ;;
 sh*-*-openbsd*)		gdb_host=nbsd ;;
--- /dev/null
+++ gdb-7.12/gdb/sh-linux-nat.c
@@ -0,0 +1,268 @@
+/* Low level SH interface to ptrace, for GDB when running native.
+   Copyright (C) 2002, 2004 Free Software Foundation, Inc.
+
+This file is part of GDB.
+
+This program is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation; either version 2 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program; if not, write to the Free Software
+Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */
+
+#include "defs.h"
+#include "inferior.h"
+#include "gdbcore.h"
+#include "regcache.h"
+#include "linux-nat.h"
+#include "target.h"
+#include "arch-utils.h"
+
+#include "gdb_assert.h"
+#include <sys/ptrace.h>
+#include <sys/user.h>
+#include <sys/procfs.h>
+#include <asm/ptrace.h>
+
+/* Prototypes for supply_gregset etc. */
+#include "gregset.h"
+#include "sh-tdep.h"
+
+/* Defines ps_err_e, struct ps_prochandle.  */
+#include "gdb_proc_service.h"
+
+//#include <asm/elf.h>
+
+#define SH_LINUX_NUM_REGS	40
+/* This table must line up with REGISTER_NAME in "sh-tdep.c".  */
+static const int regmap[] =
+{
+  /* general registers 0-15 */
+  REG_REG0   , REG_REG0+1 , REG_REG0+2 , REG_REG0+3,
+  REG_REG0+4 , REG_REG0+5 , REG_REG0+6 , REG_REG0+7,
+  REG_REG0+8 , REG_REG0+9 , REG_REG0+10, REG_REG0+11,
+  REG_REG0+12, REG_REG0+13, REG_REG0+14, REG_REG0+15,
+  /* 16 - 22 */
+  REG_PC, REG_PR, REG_GBR, -1, REG_MACH, REG_MACL, REG_SR,
+  /* 23, 24 */
+  REG_FPUL, REG_FPSCR,
+  /* floating point registers 25 - 40 */
+  REG_FPREG0   , REG_FPREG0+1 , REG_FPREG0+2 , REG_FPREG0+3 ,
+  REG_FPREG0+4 , REG_FPREG0+5 , REG_FPREG0+6 , REG_FPREG0+7 ,
+  REG_FPREG0+8 , REG_FPREG0+9 , REG_FPREG0+10, REG_FPREG0+11,
+  REG_FPREG0+12, REG_FPREG0+13, REG_FPREG0+14, REG_FPREG0+15,
+};
+
+CORE_ADDR
+register_u_addr (CORE_ADDR blockend, int regnum)
+{
+  if (regnum < 0 || regnum >= sizeof regmap/sizeof regmap[0])
+    return (CORE_ADDR)-1;
+  return (blockend + 4 * regmap[regnum]);
+}
+
+\f
+/* Return the address in the core dump or inferior of register REGNO.
+   BLOCKEND is the address of the end of the user structure.  */
+
+CORE_ADDR
+register_addr (int regno, CORE_ADDR blockend)
+{
+  CORE_ADDR addr;
+
+  if (regno < 0 || regno >= SH_LINUX_NUM_REGS) {
+    internal_error (__FILE__, __LINE__,
+		  _("Got request for bad register number %d."), regno);
+  }
+
+  REGISTER_U_ADDR (addr, blockend, regno);
+
+  return addr;
+}
+
+/* Fetch one register.  */
+
+static void
+fetch_register (struct regcache *regcache, int tid, int regno)
+{
+  int val;
+
+  if (cannot_fetch_register (regno))
+    {
+      regcache_raw_supply (regcache, regno, NULL);
+      return;
+    }
+
+  errno = 0;
+  val = ptrace (PTRACE_PEEKUSER, tid, register_addr (regno, 0), 0);
+  if (errno != 0)
+    perror_with_name (_("Couldn't get registers"));
+
+  regcache_raw_supply (regcache, regno, &val);
+}
+
+/* Store one register. */
+
+static void
+store_register (struct regcache *regcache, int tid, int regno)
+{
+  int val;
+
+  if (cannot_store_register (regno))
+    return;
+
+  errno = 0;
+  regcache_raw_collect (regcache, regno, &val);
+  ptrace (PTRACE_POKEUSER, tid, register_addr (regno, 0), val);
+  if (errno != 0)
+    perror_with_name (_("Couldn't write registers"));
+}
+
+/* Transfering the general-purpose registers between GDB, inferiors
+   and core files.  */
+
+/* Fill GDB's register array with the general-purpose register values
+   in *GREGSETP.  */
+
+void
+supply_gregset (struct regcache *regcache, const elf_gregset_t *gregsetp)
+{
+  elf_greg_t *regp = (elf_greg_t *) gregsetp;
+  int i;
+
+  for (i = 0; i < 23; i++)
+    if (regmap[i] == -1)
+      regcache_raw_supply (regcache, i, NULL);
+    else
+      regcache_raw_supply (regcache, i, (char *) (regp + regmap[i]));
+}
+
+/* Fill register REGNO (if it is a general-purpose register) in
+   *GREGSETPS with the value in GDB's register array.  If REGNO is -1,
+   do this for all registers.  */
+
+void
+fill_gregset (const struct regcache *regcache, elf_gregset_t *gregsetp, int regno)
+{
+  elf_greg_t *regp = (elf_greg_t *) gregsetp;
+  int i;
+
+  for (i = 0; i < 23; i++)
+    if (regmap[i] != -1 && (regno == -1 || regno == i))
+      regcache_raw_collect (regcache, i, (char *) (regp + regmap[i]));
+}
+
+/* Transfering floating-point registers between GDB, inferiors and cores.  */
+
+/* Fill GDB's register array with the floating-point register values in
+   *FPREGSETP.  */
+
+void
+supply_fpregset (struct regcache *regcache, const elf_fpregset_t *fpregsetp)
+{
+  int i;
+  long *regp = (long *)fpregsetp;
+
+  for (i = 0; i < 16; i++)
+    regcache_raw_supply (regcache, 25 + i, (char *) (regp + i));
+  regcache_raw_supply (regcache, FPUL_REGNUM, (char *) (regp + REG_FPUL - REG_FPREG0));
+  regcache_raw_supply (regcache, FPSCR_REGNUM, (char *) (regp + REG_FPSCR - REG_FPREG0));
+}
+
+/* Fill register REGNO (if it is a floating-point register) in
+   *FPREGSETP with the value in GDB's register array.  If REGNO is -1,
+   do this for all registers.  */
+
+void
+fill_fpregset (const struct regcache *regcache, elf_fpregset_t *fpregsetp, int regno)
+{
+  int i;
+  long *regp = (long *)fpregsetp;
+
+  for (i = 0; i < 16; i++)
+    if ((regno == -1) || (regno == i))
+      regcache_raw_collect (regcache, 25 + i, (char *) (regp + i));
+  if ((regno == -1) || regno == FPSCR_REGNUM)
+    regcache_raw_collect (regcache, FPSCR_REGNUM, (char *) (regp + REG_FPSCR - REG_FPREG0));
+  if ((regno == -1) || regno == FPUL_REGNUM)
+    regcache_raw_collect (regcache, FPUL_REGNUM, (char *) (regp + REG_FPUL - REG_FPREG0));
+}
+
+/* Transferring arbitrary registers between GDB and inferior.  */
+
+/* Check if register REGNO in the child process is accessible.
+   If we are accessing registers directly via the U area, only the
+   general-purpose registers are available.
+   All registers should be accessible if we have GETREGS support.  */
+   
+int
+cannot_fetch_register (int regno)
+{
+  return (regno < 0 || regno >= sizeof regmap / sizeof regmap[0] || regmap[regno] == -1);
+}
+
+int
+cannot_store_register (int regno)
+{
+  return (regno < 0 || regno >= sizeof regmap / sizeof regmap[0] || regmap[regno] == -1);
+}
+
+/* Fetch register values from the inferior.
+   If REGNO is negative, do this for all registers.
+   Otherwise, REGNO specifies which register (so we can save time). */
+
+static void
+sh_linux_fetch_inferior_registers (struct target_ops *ops, struct regcache *regcache, int regno)
+{
+  int i;
+  int tid;
+
+  /* GNU/Linux LWP ID's are process ID's.  */
+  if ((tid = ptid_get_lwp (inferior_ptid)) == 0)
+    tid = ptid_get_pid (inferior_ptid);	/* Not a threaded program.  */
+
+  for (i = 0; i < SH_LINUX_NUM_REGS; i++)
+    if (regno == -1 || regno == i)
+      fetch_register (regcache, tid, i);
+}
+/* Store our register values back into the inferior.
+   If REGNO is negative, do this for all registers.
+   Otherwise, REGNO specifies which register (so we can save time).  */
+
+static void
+sh_linux_store_inferior_registers (struct target_ops *ops, struct regcache *regcache, int regno)
+{
+  int i;
+  int tid;
+
+  /* GNU/Linux LWP ID's are process ID's.  */
+  if ((tid = ptid_get_lwp (inferior_ptid)) == 0)
+    tid = ptid_get_pid (inferior_ptid);	/* Not a threaded program.  */
+
+  for (i = 0; i < SH_LINUX_NUM_REGS; i++)
+    if (regno == -1 || regno == i)
+      store_register (regcache, tid, i);
+}
+
+void
+_initialize_sh_linux_nat (void)
+{
+  struct target_ops *t;
+
+  /* Fill in the generic GNU/Linux methods.  */
+  t = linux_target ();
+
+  /* Add our register access methods.  */
+  t->to_fetch_registers = sh_linux_fetch_inferior_registers;
+  t->to_store_registers = sh_linux_store_inferior_registers;
+
+  /* Register the target.  */
+  linux_nat_add_target (t);
+}
--- gdb-7.12.orig/gdb/sh-linux-tdep.c
+++ gdb-7.12/gdb/sh-linux-tdep.c
@@ -18,14 +18,36 @@
    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
 
 #include "defs.h"
+#include "gdbcore.h"
+#include "frame.h"
+#include "frame-base.h"
+#include "frame-unwind.h"
+#include "dwarf2-frame.h"
+#include "value.h"
+#include "regcache.h"
+#include "inferior.h"
 #include "osabi.h"
 
+#include "reggroups.h"
+#include "arch-utils.h"
+#include "floatformat.h"
 #include "solib-svr4.h"
 #include "symtab.h"
+#include "command.h"
+#include "gdb_assert.h"
 
 #include "trad-frame.h"
 #include "tramp-frame.h"
 
+#include <sys/ptrace.h>
+#include <sys/types.h>
+#include <sys/param.h>
+#include <sys/user.h>
+#include <sys/syscall.h>
+
+#include <asm/ptrace.h>
+
+#include "regset.h"
 #include "glibc-tdep.h"
 #include "sh-tdep.h"
 #include "linux-tdep.h"
@@ -180,11 +202,505 @@ static struct tramp_frame sh_linux_rt_si
   sh_linux_rt_sigreturn_init
 };
 
+/* Recognizing signal handler frames.  */
+
+/* GNU/Linux has two flavors of signals.  Normal signal handlers, and
+   "realtime" (RT) signals.  The RT signals can provide additional
+   information to the signal handler if the SA_SIGINFO flag is set
+   when establishing a signal handler using `sigaction'.  It is not
+   unlikely that future versions of GNU/Linux will support SA_SIGINFO
+   for normal signals too.  */
+
+/* When the SH Linux kernel calls a signal handler and the
+   SA_RESTORER flag isn't set, the return address points to a bit of
+   code on the stack.  This function returns whether the PC appears to
+   be within this bit of code.
+
+   The instruction sequence for normal signals is
+       mov.w  1f,r3
+       trapa  #16
+       or     r0, r0
+       or     r0, r0
+       or     r0, r0
+       or     r0, r0
+       or     r0, r0
+    1: .word  __NR_sigreturn
+   or 0x9305 0xc310 0x200b 0x200b 0x200b 0x200b 0x200b 0x0077.
+
+   Checking for the code sequence should be somewhat reliable, because
+   the effect is to call the system call sigreturn.  This is unlikely
+   to occur anywhere other than a signal trampoline.
+
+   It kind of sucks that we have to read memory from the process in
+   order to identify a signal trampoline, but there doesn't seem to be
+   any other way.  The PC_IN_SIGTRAMP macro in tm-linux.h arranges to
+   only call us if no function name could be identified, which should
+   be the case since the code is on the stack.
+
+   Detection of signal trampolines for handlers that set the
+   SA_RESTORER flag is in general not possible.  Unfortunately this is
+   what the GNU C Library has been doing for quite some time now.
+   However, as of version 2.1.2, the GNU C Library uses signal
+   trampolines (named __restore and __restore_rt) that are identical
+   to the ones used by the kernel.  Therefore, these trampolines are
+   supported too.  */
+
+#define MOVW(n)	 (0x9300|((n)-2))	/* Move mem word at PC+n to R3 */
+#define TRAP16	 0xc310			/* Syscall w/no args (NR in R3) */
+#define OR_R0_R0 0x200b			/* or r0,r0 (insert to avoid hardware bug) */
+
+#define LINUX_SIGTRAMP_INSN0	MOVW(7)		/* Move mem word at PC+7 to R3 */
+#define LINUX_SIGTRAMP_INSN1	TRAP16		/* Syscall w/no args (NR in R3) */
+#define LINUX_SIGTRAMP_INSN2	OR_R0_R0	/* or r0,r0 (insert to avoid hardware bug) */
+
+static const unsigned short linux_sigtramp_code[] =
+{
+  LINUX_SIGTRAMP_INSN0,
+  LINUX_SIGTRAMP_INSN1,
+  LINUX_SIGTRAMP_INSN2,
+  LINUX_SIGTRAMP_INSN2,
+  LINUX_SIGTRAMP_INSN2,
+  LINUX_SIGTRAMP_INSN2,
+  LINUX_SIGTRAMP_INSN2,
+  __NR_sigreturn
+};
+
+#define LINUX_SIGTRAMP_LEN (sizeof linux_sigtramp_code)
+
+/* If PC is in a sigtramp routine, return the address of the start of
+   the routine.  Otherwise, return 0.  */
+
+static CORE_ADDR
+sh_linux_sigtramp_start (struct frame_info *next_frame)
+{
+  CORE_ADDR pc = get_frame_pc (next_frame);
+  gdb_byte buf[LINUX_SIGTRAMP_LEN];
+
+  /* We only recognize a signal trampoline if PC is at the start of
+     one of the three instructions.  We optimize for finding the PC at
+     the start, as will be the case when the trampoline is not the
+     first frame on the stack.  We assume that in the case where the
+     PC is not at the start of the instruction sequence, there will be
+     a few trailing readable bytes on the stack.  */
+
+  if (!safe_frame_unwind_memory (next_frame, pc, buf, LINUX_SIGTRAMP_LEN))
+    return 0;
+
+  if (buf[0] != LINUX_SIGTRAMP_INSN0)
+    {
+      if (buf[0] != LINUX_SIGTRAMP_INSN1)
+        return 0;
+
+      pc -= 2;
+
+      if (!safe_frame_unwind_memory (next_frame, pc, buf, LINUX_SIGTRAMP_LEN))
+	return 0;
+    }
+
+  if (memcmp (buf, linux_sigtramp_code, LINUX_SIGTRAMP_LEN) != 0)
+    return 0;
+
+  return pc;
+}
+
+/* This function does the same for RT signals.  Here the instruction
+   sequence is
+       mov.w  1f,r3
+       trapa  #16
+       or     r0, r0
+       or     r0, r0
+       or     r0, r0
+       or     r0, r0
+       or     r0, r0
+    1: .word  __NR_rt_sigreturn
+   or 0x9305 0xc310 0x200b 0x200b 0x200b 0x200b 0x200b 0x00ad.
+
+   The effect is to call the system call rt_sigreturn.  */
+
+#define LINUX_RT_SIGTRAMP_INSN0		MOVW(7)		/* Move mem word at PC+7 to R3 */
+#define LINUX_RT_SIGTRAMP_INSN1		TRAP16		/* Syscall w/no args (NR in R3) */
+#define LINUX_RT_SIGTRAMP_INSN2		OR_R0_R0	/* or r0,r0 (insert to avoid hardware bug) */
+
+static const unsigned short linux_rt_sigtramp_code[] =
+{
+  LINUX_RT_SIGTRAMP_INSN0,
+  LINUX_RT_SIGTRAMP_INSN1,
+  LINUX_RT_SIGTRAMP_INSN2,
+  LINUX_RT_SIGTRAMP_INSN2,
+  LINUX_RT_SIGTRAMP_INSN2,
+  LINUX_RT_SIGTRAMP_INSN2,
+  LINUX_RT_SIGTRAMP_INSN2,
+  __NR_rt_sigreturn
+};
+
+#define LINUX_RT_SIGTRAMP_LEN (sizeof linux_rt_sigtramp_code)
+
+/* If PC is in a RT sigtramp routine, return the address of the start
+   of the routine.  Otherwise, return 0.  */
+
+static CORE_ADDR
+sh_linux_rt_sigtramp_start (struct frame_info *next_frame)
+{
+  CORE_ADDR pc = get_frame_pc (next_frame);
+  gdb_byte buf[LINUX_RT_SIGTRAMP_LEN];
+
+  /* We only recognize a signal trampoline if PC is at the start of
+     one of the two instructions.  We optimize for finding the PC at
+     the start, as will be the case when the trampoline is not the
+     first frame on the stack.  We assume that in the case where the
+     PC is not at the start of the instruction sequence, there will be
+     a few trailing readable bytes on the stack.  */
+
+  if (!safe_frame_unwind_memory (next_frame, pc, buf, LINUX_RT_SIGTRAMP_LEN))
+    return 0;
+
+  if (buf[0] != LINUX_RT_SIGTRAMP_INSN0)
+    {
+      if (buf[0] != LINUX_RT_SIGTRAMP_INSN1)
+	return 0;
+
+      pc -= 2;
+
+      if (!safe_frame_unwind_memory (next_frame, pc, buf,
+				     LINUX_RT_SIGTRAMP_LEN))
+	return 0;
+    }
+
+  if (memcmp (buf, linux_rt_sigtramp_code, LINUX_RT_SIGTRAMP_LEN) != 0)
+    return 0;
+
+  return pc;
+}
+
+/* Return whether PC is in a GNU/Linux sigtramp routine.  */
+
+static int
+sh_linux_sigtramp_p (struct frame_info *this_frame)
+{
+  CORE_ADDR pc = get_frame_pc (this_frame);
+  const char *name;
+
+  find_pc_partial_function (pc, &name, NULL, NULL);
+
+  /* If we have NAME, we can optimize the search.  The trampolines are
+     named __restore and __restore_rt.  However, they aren't dynamically
+     exported from the shared C library, so the trampoline may appear to
+     be part of the preceding function.  This should always be sigaction,
+     __sigaction, or __libc_sigaction (all aliases to the same function).  */
+  if (name == NULL || strstr (name, "sigaction") != NULL)
+    return (sh_linux_sigtramp_start (this_frame) != 0
+	    || sh_linux_rt_sigtramp_start (this_frame) != 0);
+
+  return (strcmp ("__restore", name) == 0
+	  || strcmp ("__restore_rt", name) == 0);
+}
+
+/* Offset to struct sigcontext in ucontext, from <asm/ucontext.h>.  */
+#define SH_LINUX_UCONTEXT_SIGCONTEXT_OFFSET 12
+
+
+/* Assuming NEXT_FRAME is a frame following a GNU/Linux sigtramp
+   routine, return the address of the associated sigcontext structure.  */
+
+static CORE_ADDR
+sh_linux_sigcontext_addr (struct frame_info *this_frame)
+{
+  CORE_ADDR pc;
+  CORE_ADDR sp;
+
+  sp = get_frame_register_unsigned (this_frame, SP_REGNUM);
+
+  pc = sh_linux_sigtramp_start (this_frame);
+  if (pc)
+    {
+      return sp;
+    }
+
+  pc = sh_linux_rt_sigtramp_start (this_frame);
+  if (pc)
+    {
+      CORE_ADDR ucontext_addr;
+
+      /* The sigcontext structure is part of the user context.  A
+	 pointer to the user context is passed as the third argument
+	 to the signal handler.  */
+      ucontext_addr = get_frame_register_unsigned (this_frame, ARG0_REGNUM+2);
+      return ucontext_addr + SH_LINUX_UCONTEXT_SIGCONTEXT_OFFSET;
+    }
+
+  error ("Couldn't recognize signal trampoline.");
+  return 0;
+}
+
+/* Signal trampolines.  */
+extern struct sh_frame_cache *sh_alloc_frame_cache (void);
+
+static struct sh_frame_cache *
+sh_linux_sigtramp_frame_cache (struct frame_info *this_frame, void **this_cache)
+{
+  struct sh_frame_cache *cache;
+  struct gdbarch_tdep *tdep = gdbarch_tdep (get_current_arch ());
+  CORE_ADDR sigcontext_addr;
+
+  if (*this_cache)
+    return (struct sh_frame_cache *) *this_cache;
+
+  cache = sh_alloc_frame_cache ();
+
+  cache->base = get_frame_register_unsigned (this_frame, SP_REGNUM);
+  sigcontext_addr = tdep->sigcontext_addr (this_frame);
+  if (tdep->sc_reg_offset)
+    {
+      int i;
+
+      gdb_assert (tdep->sc_num_regs <= SH_NUM_REGS);
+
+      for (i = 0; i < tdep->sc_num_regs; i++)
+	if (tdep->sc_reg_offset[i] != -1)
+	  cache->saved_regs[i] = sigcontext_addr + tdep->sc_reg_offset[i];
+    }
+
+  *this_cache = cache;
+  return cache;
+}
+
+static void
+sh_linux_sigtramp_frame_this_id (struct frame_info *this_frame, void **this_cache,
+			     struct frame_id *this_id)
+{
+  struct sh_frame_cache *cache =
+    sh_linux_sigtramp_frame_cache (this_frame, this_cache);
+
+  (*this_id) = frame_id_build (cache->base + 64, cache->pc);
+}
+
+extern struct value * sh_frame_prev_register (struct frame_info *this_frame, void **this_cache, int regnum);
+static struct value *
+sh_linux_sigtramp_frame_prev_register (struct frame_info *this_frame,
+				   void **this_cache, int regnum)
+{
+  sh_linux_sigtramp_frame_cache (this_frame, this_cache);
+
+  return sh_frame_prev_register (this_frame, this_cache, regnum);
+}
+
+static int
+sh_linux_sigtramp_frame_sniffer (const struct frame_unwind *self,
+				 struct frame_info *this_frame,
+				 void **this_prologue_cache)
+{
+  struct gdbarch_tdep *tdep = gdbarch_tdep (get_frame_arch (this_frame));
+
+  /* We shouldn't even bother if we don't have a sigcontext_addr
+     handler.  */
+  if (tdep->sigcontext_addr == NULL)
+    return 0;
+
+  if (tdep->sigtramp_p != NULL)
+    {
+      if (tdep->sigtramp_p (this_frame))
+	return 1;
+    }
+
+  return 0;
+}
+
+static const struct frame_unwind sh_linux_sigtramp_frame_unwind =
+{
+  SIGTRAMP_FRAME,
+  default_frame_unwind_stop_reason,
+  sh_linux_sigtramp_frame_this_id,
+  sh_linux_sigtramp_frame_prev_register,
+  NULL,
+  sh_linux_sigtramp_frame_sniffer
+};
+
+/* Supply register REGNUM from the buffer specified by GREGS and LEN
+   in the general-purpose register set REGSET to register cache
+   REGCACHE.  If REGNUM is -1, do this for all registers in REGSET.  */
+
+void
+sh_linux_supply_gregset (const struct regset *regset, struct regcache *regcache,
+		     int regnum, const void *gregs, size_t len)
+{
+  struct gdbarch *gdbarch = get_regcache_arch (regcache);
+  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+  const gdb_byte *regs = (const gdb_byte *) gregs;
+  int i;
+
+  gdb_assert (len == tdep->sizeof_gregset);
+
+  for (i = 0; i < tdep->gregset_num_regs; i++)
+    {
+      if ((regnum == i || regnum == -1)
+	  && tdep->gregset_reg_offset[i] != -1)
+	regcache_raw_supply (regcache, i, regs + tdep->gregset_reg_offset[i]);
+    }
+}
+
+/* Collect register REGNUM from the register cache REGCACHE and store
+   it in the buffer specified by GREGS and LEN as described by the
+   general-purpose register set REGSET.  If REGNUM is -1, do this for
+   all registers in REGSET.  */
+
+void
+sh_linux_collect_gregset (const struct regset *regset,
+		      const struct regcache *regcache,
+		      int regnum, void *gregs, size_t len)
+{
+  struct gdbarch *gdbarch = get_regcache_arch (regcache);
+  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+  gdb_byte *regs = (gdb_byte *) gregs;
+  int i;
+
+  gdb_assert (len == tdep->sizeof_gregset);
+
+  for (i = 0; i < tdep->gregset_num_regs; i++)
+    {
+      if ((regnum == i || regnum == -1)
+	  && tdep->gregset_reg_offset[i] != -1)
+	regcache_raw_collect (regcache, i, regs + tdep->gregset_reg_offset[i]);
+    }
+}
+
+/* Supply register REGNUM from the buffer specified by FPREGS and LEN
+   in the floating-point register set REGSET to register cache
+   REGCACHE.  If REGNUM is -1, do this for all registers in REGSET.  */
+
+static void
+sh_linux_supply_fpregset (const struct regset *regset, struct regcache *regcache,
+		      int regnum, const void *fpregs, size_t len)
+{
+  struct gdbarch *gdbarch = get_regcache_arch (regcache);
+  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+  const gdb_byte *regs = (const gdb_byte *) fpregs;
+  int i;
+
+  gdb_assert (len == tdep->sizeof_fpregset);
+  for (i = 0; i < 16; i++)
+    {
+      if (regnum == i+25 || regnum == -1)
+	regcache_raw_supply (regcache, i+25, regs + i*4);
+    }
+  if (regnum == FPSCR_REGNUM || regnum == -1)
+    regcache_raw_supply (regcache, FPSCR_REGNUM, regs + 32*4);
+  if (regnum == FPUL_REGNUM || regnum == -1)
+    regcache_raw_supply (regcache, FPUL_REGNUM, regs + 33*4);
+}
+
+/* Collect register REGNUM from the register cache REGCACHE and store
+   it in the buffer specified by FPREGS and LEN as described by the
+   floating-point register set REGSET.  If REGNUM is -1, do this for
+   all registers in REGSET.  */
+
+static void
+sh_linux_collect_fpregset (const struct regset *regset,
+		       const struct regcache *regcache,
+		       int regnum, void *fpregs, size_t len)
+{
+  struct gdbarch *gdbarch = get_regcache_arch (regcache);
+  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+  gdb_byte *regs = (gdb_byte *) fpregs;
+  int i;
+
+  gdb_assert (len == tdep->sizeof_fpregset);
+  for (i = 0; i < 16; i++)
+    {
+      if (regnum == i+25 || regnum == -1)
+	regcache_raw_collect (regcache, i+25, regs + i*4);
+    }
+  if (regnum == FPSCR_REGNUM || regnum == -1)
+    regcache_raw_collect (regcache, FPSCR_REGNUM, regs + 32*4);
+  if (regnum == FPUL_REGNUM || regnum == -1)
+    regcache_raw_collect (regcache, FPUL_REGNUM, regs + 33*4);
+}
+
+static const struct regset sh_linux_gregset =
+  {
+    NULL, sh_linux_supply_gregset, sh_linux_collect_gregset
+  };
+
+static const struct regset sh_linux_fpregset =
+  {
+    NULL, sh_linux_supply_fpregset, sh_linux_collect_fpregset
+  };
+
+/* Iterate over core file register note sections.  */
+
+static void
+sh_linux_iterate_over_regset_sections (struct gdbarch *gdbarch,
+					 iterate_over_regset_sections_cb *cb,
+					 void *cb_data,
+					 const struct regcache *regcache)
+{
+  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+  cb (".reg", tdep->sizeof_gregset, &sh_linux_gregset, NULL, cb_data);
+  cb (".reg2", tdep->sizeof_fpregset, &sh_linux_fpregset, NULL, cb_data);
+}
+
+/* The register sets used in GNU/Linux ELF core-dumps are identical to
+   the register sets in `struct user' that are used for a.out
+   core-dumps.  These are also used by ptrace(2).  The corresponding
+   types are `elf_gregset_t' for the general-purpose registers (with
+   `elf_greg_t' the type of a single GP register) and `elf_fpregset_t'
+   for the floating-point registers.
+
+   Those types used to be available under the names `gregset_t' and
+   `fpregset_t' too, and GDB used those names in the past.  But those
+   names are now used for the register sets used in the `mcontext_t'
+   type, which have a different size and layout.  */
+
+/* Mapping between the general-purpose registers in `struct user'
+   format and GDB's register cache layout.  */
+
+/* From <sys/reg.h>.  */
+static int sh_linux_gregset_reg_offset[] =
+{
+ 0,	4,	8,	12,	16,	20,	24,	28,
+ 32,	36,	40,	44,	48,	52,	56,	60,
+
+ REG_PC*4,   REG_PR*4,   REG_GBR*4,  -1,
+ REG_MACH*4, REG_MACL*4, REG_SR*4,
+};
+
+/* Mapping between the general-purpose registers in `struct
+   sigcontext' format and GDB's register cache layout.  */
+
+/* From <asm/sigcontext.h>.  */
+static int sh_linux_sc_reg_offset[] =
+{
+ 4,	8,	12,	16,	20,	24,	28,	32,
+ 36,	40,	44,	48,	52,	56,	60,	64,
+ 68,	72,	80,	-1,
+ 84,	88,	76
+};
+
+
 static void
 sh_linux_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch)
 {
+  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+  bfd abfd;
+
   linux_init_abi (info, gdbarch);
 
+  tdep->gregset_reg_offset = sh_linux_gregset_reg_offset;
+  tdep->gregset_num_regs = ARRAY_SIZE (sh_linux_gregset_reg_offset);
+  tdep->sizeof_gregset = 23 * 4;
+
+  tdep->jb_pc_offset = 32;     /* From <bits/setjmp.h>.  */
+
+  tdep->sigtramp_p = sh_linux_sigtramp_p;
+  tdep->sigcontext_addr = sh_linux_sigcontext_addr;
+  tdep->sc_reg_offset = sh_linux_sc_reg_offset;
+  tdep->sc_num_regs = ARRAY_SIZE (sh_linux_sc_reg_offset);
+
+  frame_unwind_append_unwinder(gdbarch, &sh_linux_sigtramp_frame_unwind);
+
+  /* Core file support. */
+  set_gdbarch_iterate_over_regset_sections
+    (gdbarch, sh_linux_iterate_over_regset_sections);
+
   /* GNU/Linux uses SVR4-style shared libraries.  */
   set_gdbarch_skip_trampoline_code (gdbarch, find_solib_trampoline_target);
   set_solib_svr4_fetch_link_map_offsets
--- gdb-7.12.orig/gdb/sh-tdep.c
+++ gdb-7.12/gdb/sh-tdep.c
@@ -21,6 +21,9 @@
    sac@cygnus.com.  */
 
 #include "defs.h"
+#include "arch-utils.h"
+#include "command.h"
+#include "dummy-frame.h"
 #include "frame.h"
 #include "frame-base.h"
 #include "frame-unwind.h"
@@ -35,6 +38,7 @@
 #include "arch-utils.h"
 #include "floatformat.h"
 #include "regcache.h"
+#include "regset.h"
 #include "doublest.h"
 #include "osabi.h"
 #include "reggroups.h"
@@ -67,23 +71,6 @@ static const char *const sh_cc_enum[] =
 
 static const char *sh_active_calling_convention = sh_cc_gcc;
 
-#define SH_NUM_REGS 67
-
-struct sh_frame_cache
-{
-  /* Base address.  */
-  CORE_ADDR base;
-  LONGEST sp_offset;
-  CORE_ADDR pc;
-
-  /* Flag showing that a frame has been created in the prologue code.  */
-  int uses_fp;
-
-  /* Saved registers.  */
-  CORE_ADDR saved_regs[SH_NUM_REGS];
-  CORE_ADDR saved_sp;
-};
-
 static int
 sh_is_renesas_calling_convention (struct type *func_type)
 {
@@ -1043,7 +1030,7 @@ sh_treat_as_flt_p (struct type *type)
     return 0;
   /* Otherwise if the type of that member is float, the whole type is
      treated as float.  */
-  if (TYPE_CODE (TYPE_FIELD_TYPE (type, 0)) == TYPE_CODE_FLT)
+  if (TYPE_CODE (check_typedef (TYPE_FIELD_TYPE (type, 0))) == TYPE_CODE_FLT)
     return 1;
   /* Otherwise it's not treated as float.  */
   return 0;
@@ -1093,7 +1080,7 @@ sh_push_dummy_call_fpu (struct gdbarch *
      in four registers available.  Loop thru args from first to last.  */
   for (argnum = 0; argnum < nargs; argnum++)
     {
-      type = value_type (args[argnum]);
+      type = check_typedef (value_type (args[argnum]));
       len = TYPE_LENGTH (type);
       val = sh_justify_value_in_reg (gdbarch, args[argnum], len);
 
@@ -1819,7 +1806,7 @@ sh_dwarf2_frame_init_reg (struct gdbarch
     reg->how = DWARF2_FRAME_REG_UNDEFINED;
 }
 
-static struct sh_frame_cache *
+struct sh_frame_cache *
 sh_alloc_frame_cache (void)
 {
   struct sh_frame_cache *cache;
@@ -1846,7 +1833,7 @@ sh_alloc_frame_cache (void)
   return cache;
 }
 
-static struct sh_frame_cache *
+struct sh_frame_cache *
 sh_frame_cache (struct frame_info *this_frame, void **this_cache)
 {
   struct gdbarch *gdbarch = get_frame_arch (this_frame);
@@ -1913,9 +1900,9 @@ sh_frame_cache (struct frame_info *this_
   return cache;
 }
 
-static struct value *
-sh_frame_prev_register (struct frame_info *this_frame,
-			void **this_cache, int regnum)
+struct value *
+sh_frame_prev_register (struct frame_info *this_frame, void **this_cache,
+			int regnum)
 {
   struct gdbarch *gdbarch = get_frame_arch (this_frame);
   struct sh_frame_cache *cache = sh_frame_cache (this_frame, this_cache);
@@ -1929,7 +1916,7 @@ sh_frame_prev_register (struct frame_inf
      the current frame.  Frob regnum so that we pull the value from
      the correct place.  */
   if (regnum == gdbarch_pc_regnum (gdbarch))
-    regnum = PR_REGNUM;
+    regnum = PR_REGNUM; /* XXX: really? */
 
   if (regnum < SH_NUM_REGS && cache->saved_regs[regnum] != -1)
     return frame_unwind_got_memory (this_frame, regnum,
@@ -2238,8 +2225,8 @@ sh_return_in_first_hidden_param_p (struc
 static struct gdbarch *
 sh_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
 {
-  struct gdbarch *gdbarch;
   struct gdbarch_tdep *tdep;
+  struct gdbarch *gdbarch;
 
   /* SH5 is handled entirely in sh64-tdep.c.  */
   if (info.bfd_arch_info->mach == bfd_mach_sh5)
@@ -2255,6 +2242,18 @@ sh_gdbarch_init (struct gdbarch_info inf
   tdep = XCNEW (struct gdbarch_tdep);
   gdbarch = gdbarch_alloc (&info, tdep);
 
+  /* General-purpose registers.  */
+  tdep->gregset = NULL;
+  tdep->gregset_reg_offset = NULL;
+  tdep->gregset_num_regs = 23;
+  tdep->sizeof_gregset = 0;
+
+  /* Floating-point registers.  */
+  tdep->fpregset = NULL;
+  tdep->sizeof_fpregset = 34*4;
+
+  tdep->jb_pc_offset = -1;
+
   set_gdbarch_short_bit (gdbarch, 2 * TARGET_CHAR_BIT);
   set_gdbarch_int_bit (gdbarch, 4 * TARGET_CHAR_BIT);
   set_gdbarch_long_bit (gdbarch, 4 * TARGET_CHAR_BIT);
@@ -2405,10 +2404,11 @@ sh_gdbarch_init (struct gdbarch_info inf
       break;
     }
 
+  dwarf2_append_unwinders (gdbarch);
+
   /* Hook in ABI-specific overrides, if they have been registered.  */
   gdbarch_init_osabi (info, gdbarch);
 
-  dwarf2_append_unwinders (gdbarch);
   frame_unwind_append_unwinder (gdbarch, &sh_stub_unwind);
   frame_unwind_append_unwinder (gdbarch, &sh_frame_unwind);
 
--- gdb-7.12.orig/gdb/sh-tdep.h
+++ gdb-7.12/gdb/sh-tdep.h
@@ -21,6 +21,12 @@
 
 /* Contributed by Steve Chamberlain sac@cygnus.com.  */
 
+struct frame_info;
+struct gdbarch;
+struct reggroup;
+struct regset;
+struct regcache;
+
 /* Registers for all SH variants.  Used also by sh3-rom.c.  */
 enum
   {
@@ -29,6 +35,7 @@ enum
     ARG0_REGNUM = 4,
     ARGLAST_REGNUM = 7,
     FP_REGNUM = 14,
+    SP_REGNUM = 15,
     PC_REGNUM = 16,
     PR_REGNUM = 17,
     GBR_REGNUM = 18,
@@ -82,6 +89,25 @@ enum
     FV_LAST_REGNUM = 79
   };
 
+#define SH_NUM_REGS 67
+
+struct sh_frame_cache
+{
+  /* Base address.  */
+  CORE_ADDR base;
+  LONGEST sp_offset;
+  CORE_ADDR pc;
+
+  /* Flag showing that a frame has been created in the prologue code. */
+  int uses_fp;
+
+  /* Saved registers.  */
+  CORE_ADDR saved_regs[SH_NUM_REGS];
+  CORE_ADDR saved_sp;
+};
+
+extern struct sh_frame_cache *sh_frame_cache (struct frame_info *next_frame, void **this_cache);
+
 /* This structure describes a register in a core-file.  */
 struct sh_corefile_regmap
 {
@@ -89,18 +115,41 @@ struct sh_corefile_regmap
   unsigned int offset;
 };
 
+/* sh architecture specific information.  */
 struct gdbarch_tdep
 {
+  /* General-purpose registers.  */
+  struct regset *gregset;
+  int *gregset_reg_offset;
+  int gregset_num_regs;
+  size_t sizeof_gregset;
+
+  /* Floating-point registers.  */
+  struct regset *fpregset;
+  size_t sizeof_fpregset;
+
+  /* Offset of saved PC in jmp_buf.  */
+  int jb_pc_offset;
+
+  /* Detect sigtramp.  */
+  int (*sigtramp_p) (struct frame_info *);
+
+  /* Get address of sigcontext for sigtramp.  */
+  CORE_ADDR (*sigcontext_addr) (struct frame_info *);
+
+  /* Offset of registers in `struct sigcontext'.  */
+  int *sc_reg_offset;
+  int sc_num_regs;
+
   /* Non-NULL when debugging from a core file.  Provides the offset
      where each general-purpose register is stored inside the associated
      core file section.  */
   struct sh_corefile_regmap *core_gregmap;
-  int sizeof_gregset;
+
   /* Non-NULL when debugging from a core file and when FP registers are
      available.  Provides the offset where each FP register is stored
      inside the associated core file section.  */
   struct sh_corefile_regmap *core_fpregmap;
-  int sizeof_fpregset;
 };
 
 extern const struct regset sh_corefile_gregset;
--- gdb-7.12.orig/gdb/testsuite/gdb.asm/asm-source.exp
+++ gdb-7.12/gdb/testsuite/gdb.asm/asm-source.exp
@@ -116,6 +116,11 @@ switch -glob -- [istarget] {
             append link-flags " -m elf32ppc"
         }
     }
+    "sh*-linux*" {
+        set asm-arch sh-linux
+        set asm-flags "-I${srcdir}/${subdir} -I${objdir}/${subdir}"
+	set debug-flags "-gdwarf-2"
+    }
     "sh*-*-*" {
         set asm-arch sh
 	set debug-flags "-gdwarf-2"
--- /dev/null
+++ gdb-7.12/gdb/testsuite/gdb.asm/sh-linux.inc
@@ -0,0 +1,78 @@
+# You'll find a bunch of nop opcodes in the below macros.  They are
+# there to keep the code correctly aligned.  Be careful to maintain
+# them when changing the code.
+
+	comment "subroutine declare"
+	.purgem gdbasm_declare
+	.macro gdbasm_declare name
+	.align	1
+	.global	\name
+\name:
+	.endm
+
+	comment "subroutine prologue"
+	.macro gdbasm_enter
+	mov.l   r14,@-r15
+	sts.l   pr,@-r15
+	mov     r15,r14
+	nop
+	.endm
+
+	comment "subroutine epilogue"
+	.macro gdbasm_leave
+	mov     r14,r15
+	lds.l   @r15+,pr
+	mov.l   @r15+,r14
+	rts
+	nop
+	nop
+	.endm
+
+	comment "subroutine end"
+	.purgem gdbasm_end
+	.macro gdbasm_end name
+	.size   \name, .-_foo1
+	.align	1
+	.endm
+
+	comment "subroutine call"
+	.macro gdbasm_call subr
+	mov.l   .Lconst\@,r1
+	bra	.Lafterconst\@
+	nop
+	.align	2
+.Lconst\@:
+	.long	\subr
+.Lafterconst\@:
+	jsr	@r1
+	nop
+	.endm
+
+	.macro gdbasm_several_nops
+	nop
+	nop
+	nop
+	nop
+	.endm
+
+	comment "exit (0)"
+	.macro gdbasm_exit0
+	sleep
+	nop
+	.endm
+
+	comment "crt0 startup"
+	.macro gdbasm_startup
+	mov	#0,r14
+	.endm
+
+	comment "Declare a data variable"
+	.purgem gdbasm_datavar
+	.macro gdbasm_datavar name value
+	.data
+	.align 2
+	.type	\name, @object
+	.size	\name, 4
+\name:
+	.long	\value
+	.endm
--- gdb-7.12.orig/gdb/testsuite/gdb.asm/sh.inc
+++ gdb-7.12/gdb/testsuite/gdb.asm/sh.inc
@@ -40,9 +40,8 @@
 	mov.l   .Lconst\@,r1
 	bra	.Lafterconst\@
 	nop
-	nop
-.Lconst\@:
 	.align	2
+.Lconst\@:
 	.long	\subr
 	.align	1
 .Lafterconst\@:

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
  2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
                   ` (2 preceding siblings ...)
  2017-02-07 20:51 ` John Paul Adrian Glaubitz
@ 2017-03-10  1:22 ` Nobuhiro Iwamatsu
  2017-03-10  3:16 ` Nobuhiro Iwamatsu
                   ` (2 subsequent siblings)
  6 siblings, 0 replies; 8+ messages in thread
From: Nobuhiro Iwamatsu @ 2017-03-10  1:22 UTC (permalink / raw)
  To: linux-sh

Hi,

Sorry, replay was late.

Takashi already leaved from Renesas, he is working other company.
I am contacting him now, I am planning to ask him about this matter.

Best regards,
  Nobuhiro

2017-02-08 5:51 GMT+09:00 John Paul Adrian Glaubitz
<glaubitz@physik.fu-berlin.de>:
> Hi!
>
> I have ported Yoshii-san's code to GDB 7.12, attaching an updated
> patch. A git repository with my linux-sh branch can be found in [1].
>
> Yutaka Niibe has already reached out to Renesas in Japan and is trying to
> get into touch with Yoshii-san, this might take a while though. However,
> I have talked with someone at the FSFE knowledgeable with open source
> licenses and it looks like we actually don't need the copyright assignment
> at all.
>
> If we needed the copyright assignment, we'd probably need it from Renesas
> anyway and not from Yoshii-san directly as he wrote the patch while working
> for Renesas who consequently are the copyright owners. And since Renesas
> already submitted their SH code to gcc, I assume they also already assigned
> the FSF copyright assignment (if that's actually needed).
>
> More updates coming soon.
>
> Thanks,
> Adrian
>
>> [1] https://github.com/glaubitz/binutils-gdb/tree/linux-sh
>
> --
>  .''`.  John Paul Adrian Glaubitz
> : :' :  Debian Developer - glaubitz@debian.org
> `. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
>   `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913



-- 
Nobuhiro Iwamatsu
   iwamatsu at {nigauri.org / debian.org}
   GPG ID: 40AD1FA6

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
  2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
                   ` (3 preceding siblings ...)
  2017-03-10  1:22 ` Nobuhiro Iwamatsu
@ 2017-03-10  3:16 ` Nobuhiro Iwamatsu
  2017-03-11  8:34 ` John Paul Adrian Glaubitz
  2017-05-17  9:04 ` John Paul Adrian Glaubitz
  6 siblings, 0 replies; 8+ messages in thread
From: Nobuhiro Iwamatsu @ 2017-03-10  3:16 UTC (permalink / raw)
  To: linux-sh

Hi, again.

I add Takashi's  to this thread.
I contacted to Takashi, and talked about this problem.
He said I would reply on this matter.
Would you please wait for his reply?

Best regards,
  Nobuhiro

2017-03-10 10:22 GMT+09:00 Nobuhiro Iwamatsu <iwamatsu@nigauri.org>:
> Hi,
>
> Sorry, replay was late.
>
> Takashi already leaved from Renesas, he is working other company.
> I am contacting him now, I am planning to ask him about this matter.
>
> Best regards,
>   Nobuhiro
>
> 2017-02-08 5:51 GMT+09:00 John Paul Adrian Glaubitz
> <glaubitz@physik.fu-berlin.de>:
>> Hi!
>>
>> I have ported Yoshii-san's code to GDB 7.12, attaching an updated
>> patch. A git repository with my linux-sh branch can be found in [1].
>>
>> Yutaka Niibe has already reached out to Renesas in Japan and is trying to
>> get into touch with Yoshii-san, this might take a while though. However,
>> I have talked with someone at the FSFE knowledgeable with open source
>> licenses and it looks like we actually don't need the copyright assignment
>> at all.
>>
>> If we needed the copyright assignment, we'd probably need it from Renesas
>> anyway and not from Yoshii-san directly as he wrote the patch while working
>> for Renesas who consequently are the copyright owners. And since Renesas
>> already submitted their SH code to gcc, I assume they also already assigned
>> the FSF copyright assignment (if that's actually needed).
>>
>> More updates coming soon.
>>
>> Thanks,
>> Adrian
>>
>>> [1] https://github.com/glaubitz/binutils-gdb/tree/linux-sh
>>
>> --
>>  .''`.  John Paul Adrian Glaubitz
>> : :' :  Debian Developer - glaubitz@debian.org
>> `. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
>>   `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913
>
>
>
> --
> Nobuhiro Iwamatsu
>    iwamatsu at {nigauri.org / debian.org}
>    GPG ID: 40AD1FA6



-- 
Nobuhiro Iwamatsu
   iwamatsu at {nigauri.org / debian.org}
   GPG ID: 40AD1FA6

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
  2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
                   ` (4 preceding siblings ...)
  2017-03-10  3:16 ` Nobuhiro Iwamatsu
@ 2017-03-11  8:34 ` John Paul Adrian Glaubitz
  2017-05-17  9:04 ` John Paul Adrian Glaubitz
  6 siblings, 0 replies; 8+ messages in thread
From: John Paul Adrian Glaubitz @ 2017-03-11  8:34 UTC (permalink / raw)
  To: linux-sh

Hi Nobuhiro!

On 03/10/2017 04:16 AM, Nobuhiro Iwamatsu wrote:
> I add Takashi's  to this thread.
> I contacted to Takashi, and talked about this problem.
> He said I would reply on this matter.
> Would you please wait for his reply?

Thank you very much for your initiative. Having native support for
Linux SH in GDB will be very useful for the J-Core project.

Now, I have already talked to the GDB people regarding this matter
and they told me that Yoshii-san needs to sign the copyright assignment
himself [1]. Yao Qi has provided a link which provide the necessary
forms which need to be filled out by Yoshii-san. If he has any problems
choosing the right form or needs additional advise, he can ask on the
gdb-patches mailing list himself [2] or send an email to assign@gnu.org
who will get back to him very quickly.

On a sidenote: Yao Qi just recently happened to talk to Masami Hiramatsu
who happened to have contacted to Yoshii-san at Linaro Connect [3], so
with everyone involved, I am confident to get this resolved soon!

I have one last important question regarding SuperH in general:

Can you make a statement regarding the remaining SH7785LCR boards you
have? Would you still be willing to set these up as buildds? If not,
would it be possible that you hand these boards over to me? I could
receive them in Japan, so they don't have to go abroad.

We desperately need additional buildd hardware for the sh4 port since
we currently have my one SH7785LCR board and two qemu-sh4 instances
running only. I have more hardware (NextVoD boxes) but these currently
lack support for a modern kernel (they are stuck at 2.6.32). It also
doesn't matter if the boards are broken, those are definitely fixable.

Please let's get the buildd issue resolved as well.

Thanks,

Adrian

> [1] https://sourceware.org/ml/gdb-patches/2017-02/msg00507.html
> [2] https://www.gnu.org/software/gdb/mailing-lists/
> [3] https://sourceware.org/ml/gdb-patches/2017-03/msg00127.html

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: Bug#576242: gdb: Please support Renesas SH(sh4)
  2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
                   ` (5 preceding siblings ...)
  2017-03-11  8:34 ` John Paul Adrian Glaubitz
@ 2017-05-17  9:04 ` John Paul Adrian Glaubitz
  6 siblings, 0 replies; 8+ messages in thread
From: John Paul Adrian Glaubitz @ 2017-05-17  9:04 UTC (permalink / raw)
  To: linux-sh

Hello!

On 03/10/2017 04:16 AM, Nobuhiro Iwamatsu wrote:
> I add Takashi's  to this thread.
> I contacted to Takashi, and talked about this problem.
> He said I would reply on this matter.
> Would you please wait for his reply?

Has there been any progress on this issue?

Please note that if we wait too long, merging the patch becomes
more difficult over the time as the upstream source progress.

I'm currently rebasing the patch from time to time to make
sure it remains mergable [1].

Please don't let this take forever. Signing the FSF copyright
assignment can be done quickly and Takashi just needs to do
it to clear the path.

Adrian

> [1] https://github.com/glaubitz/binutils-gdb/tree/linux-sh

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer - glaubitz@debian.org
`. `'   Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

^ permalink raw reply	[flat|nested] 8+ messages in thread

end of thread, other threads:[~2017-05-17  9:04 UTC | newest]

Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-01-30 11:21 Bug#576242: gdb: Please support Renesas SH(sh4) John Paul Adrian Glaubitz
2017-01-31 20:41 ` Rob Landley
2017-01-31 20:47 ` John Paul Adrian Glaubitz
2017-02-07 20:51 ` John Paul Adrian Glaubitz
2017-03-10  1:22 ` Nobuhiro Iwamatsu
2017-03-10  3:16 ` Nobuhiro Iwamatsu
2017-03-11  8:34 ` John Paul Adrian Glaubitz
2017-05-17  9:04 ` John Paul Adrian Glaubitz

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.