* [PATCH] Support for Xilinx MicroBlaze
@ 2009-10-06 1:35 Michael Eager
2009-10-06 7:14 ` Eli Zaretskii
2009-10-14 1:47 ` Joel Brobecker
0 siblings, 2 replies; 20+ messages in thread
From: Michael Eager @ 2009-10-06 1:35 UTC (permalink / raw)
To: gdb-patches; +Cc: Joel Brobecker, Eli Zaretskii
[-- Attachment #1: Type: text/plain, Size: 1059 bytes --]
Revised patch. I believe that I have addressed all of the comments.
2009-10-05 Michael Eager <eager@eagercon.com>
* config/djgpp/fnchange.lst: Add translations for cpu-microblaze.c,
elf32-microblaze.c, microblaze-rom.c, microblaze-linux-tdep.c,
microblaze-tdep.h, microblaze-tdep.c, microblaze-opc.h, microblaze-opcm.h,
microblaze-dis.c, microblaze-dis.h, sim/microblaze, microblaze.h, and
microblaze.isa.
* configure.tgt: Add targets microblaze*-linux-*, microblaze*-xilinx-*.
* doc/gdb.texinfo: Add MicroBlaze.
* MAINTAINERS: Add self as maintainer for MicroBlaze.
* Makefile.in: Build microblaze-tdep.o, microblaze-linux-tdep.o.
HFILES_NO_SRCDIR: Add microblaze-tdep.h.
* microblaze-linux-tdep.c: New.
* microblaze-rom.c: New.
* microblaze-tdep.c: New.
* microblaze-tdep.h: New.
* NEWS: Announce Xilinx MicroBlaze support.
There are some required changes to opcodes/microblaze* which I'll send
to the Binutils list.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
[-- Attachment #2: mb-gdb.patch --]
[-- Type: text/plain, Size: 55022 bytes --]
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/config/djgpp/fnchange.lst mb-gdb/gdb/config/djgpp/fnchange.lst
--- gdb/gdb/config/djgpp/fnchange.lst 2009-09-19 02:46:44.000000000 -0700
+++ mb-gdb/gdb/config/djgpp/fnchange.lst 2009-10-05 17:44:30.000000000 -0700
@@ -18,6 +18,7 @@
@V@/bfd/cpu-cr16.c @V@/bfd/cpucr16.c
@V@/bfd/cpu-cr16c.c @V@/bfd/cpucr16c.c
@V@/bfd/cpu-ia64-opc.c @V@/bfd/cpuia64-opc.c
+@V@/bfd/cpu-microblaze.c @V@/bfd/cpumb.c
@V@/bfd/cpu-m68hc11.c @V@/bfd/cm68hc11.c
@V@/bfd/cpu-m68hc12.c @V@/bfd/cm68hc12.c
@V@/bfd/efi-app-ia32.c @V@/bfd/efiia32app.c
@@ -34,6 +35,7 @@
@V@/bfd/elf32-frv.c @V@/bfd/elf32fv.c
@V@/bfd/elf32-i370.c @V@/bfd/e32i370.c
@V@/bfd/elf32-i386.c @V@/bfd/e32i86.c
+@V@/bfd/elf32-microblaze.c @V@e32mb.c
@V@/bfd/elf32-m32c.c @V@/bfd/em32c.c
@V@/bfd/elf32-m32r.c @V@/bfd/em32r.c
@V@/bfd/elf32-m68hc11.c @V@/bfd/em68hc11.c
@@ -279,6 +281,10 @@
@V@/gdb/m68kbsd-tdep.c @V@/gdb/m68bsd-tdep.c
@V@/gdb/m68knbsd-nat.c @V@/gdb/m6nbsd-nat.c
@V@/gdb/m68knbsd-tdep.c @V@/gdb/m6nbsd-tdep.c
+@V@/gdb/microblaze-rom.c @V@/gdb/mb-rom.c
+@V@/gdb/microblaze-linux-tdep.c @V@/gdb/mbl-tdep.c
+@V@/gdb/microblaze-tdep.h @V@/gdb/mb-tdep.h
+@V@/gdb/microblaze-tdep.c @V@/gdb/mb-tdep.c
@V@/gdb/mips-linux-nat.c @V@/gdb/mipslnxnat.c
@V@/gdb/mips-linux-tdep.c @V@/gdb/mipslnxtdep.c
@V@/gdb/mipsnbsd-nat.c @V@/gdb/mipsnbnat.c
@@ -511,6 +517,10 @@
@V@/opcodes/ia64-opc.c @V@/opcodes/ia64-opc.c
@V@/opcodes/iq2000-desc.c @V@/opcodes/iq2000desc.c
@V@/opcodes/iq2000-dis.c @V@/opcodes/iq2000dis.c
+@V@/opcodes/microblaze-opc.h @V@/opcodes/mb-opc.h
+@V@/opcodes/microblaze-dis.c @V@/opcodes/mb-dis.c
+@V@/opcodes/microblaze-dis.h @V@/opcodes/mb-dis.h
+@V@/opcodes/microblaze-opcm.h @V@/opcodes/mb-opcm.h
@V@/opcodes/m68hc11-dis.c @V@/opcodes/m68hc11dis.c
@V@/opcodes/m68hc11-opc.c @V@/opcodes/m68hc11opc.c
@V@/opcodes/microblaze-opc.h @V@/opcodes/mbl-opc.h
@@ -540,6 +550,9 @@
@V@/sim/frv/profile-fr500.h @V@/sim/frv/fr500-profile.h
@V@/sim/frv/profile-fr550.c @V@/sim/frv/fr550-profile.c
@V@/sim/frv/profile-fr550.h @V@/sim/frv/fr550-profile.h
+@V@/sim/microblaze @V@/sim/mb
+@V@/sim/microblaze/microblaze.h @V@/sim/mb/mb.h
+@V@/sim/microblaze/microblaze.isa @V@/sim/mb/mb.isa
@V@/sim/m68hc11/dv-m68hc11eepr.c @V@/sim/m68hc11/dv-eepr.c
@V@/sim/m68hc11/dv-m68hc11sio.c @V@/sim/m68hc11/dv-sio.c
@V@/sim/m68hc11/dv-m68hc11spi.c @V@/sim/m68hc11/dv-spi.c
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/configure.tgt mb-gdb/gdb/configure.tgt
--- gdb/gdb/configure.tgt 2009-09-27 13:46:35.000000000 -0700
+++ mb-gdb/gdb/configure.tgt 2009-10-05 10:52:47.000000000 -0700
@@ -313,6 +313,19 @@ mep-*-*)
# No sim needed. Target uses SID.
;;
+microblaze*-linux-*)
+ # Target: Xilinx MicroBlaze running Linux
+ gdb_target_obs="microblaze-tdep.o microblaze-linux-tdep.o microblaze-rom.o \
+ monitor.o dsrec.o solib.o solib-svr4.o corelow.o \
+ symfile-mem.o"
+ gdb_sim=../sim/microblaze/libsim.a
+ ;;
+microblaze*-xilinx-*)
+ # Target: Xilinx MicroBlaze running standalone
+ gdb_target_obs="microblaze-tdep.o microblaze-rom.o monitor.o dsrec.o"
+ gdb_sim=../sim/microblaze/libsim.a
+ ;;
+
mips*-sgi-irix5*)
# Target: MIPS SGI running Irix 5
gdb_target_obs="mips-tdep.o mips-irix-tdep.o solib.o solib-irix.o"
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/doc/gdb.texinfo mb-gdb/gdb/doc/gdb.texinfo
--- gdb/gdb/doc/gdb.texinfo 2009-10-04 10:37:57.000000000 -0700
+++ mb-gdb/gdb/doc/gdb.texinfo 2009-10-05 10:52:56.000000000 -0700
@@ -517,6 +517,9 @@ Tensilica, Inc.@: contributed support fo
who have worked on the Xtensa port of @value{GDBN} in the past include
Steve Tjiang, John Newlin, and Scott Foehner.
+Michael Eager and staff of Xilinx, Inc., contributed support for the
+Xilinx MicroBlaze architecture.
+
@node Sample Session
@chapter A Sample @value{GDBN} Session
@@ -16570,6 +16573,7 @@ acceptable commands.
* ARM:: ARM RDI
* M32R/D:: Renesas M32R/D
* M68K:: Motorola M68K
+* MicroBlaze:: Xilinx MicroBlaze
* MIPS Embedded:: MIPS Embedded
* OpenRISC 1000:: OpenRisc 1000
* PA:: HP PA Embedded
@@ -16821,6 +16825,44 @@ dBUG ROM monitor for Motorola ColdFire.
@end table
+@node MicroBlaze
+@subsection MicroBlaze
+@cindex Xilinx MicroBlaze
+@cindex XMD, Xilinx Microprocessor Debugger
+
+The MicroBlaze is a soft-core processor supported on various Xilinx
+FPGAs, such as Spartan or Virtex series. Boards with these processors
+usually have JTAG ports which connect to a host system running the Xilinx
+Embedded Development Kit (EDK) or Software Development Kit (SDK).
+This host system is used to download the configuration bitstream to
+the target FPGA. The Xilinx Microprocessor Debugger (XMD) program
+communicates with the target board using the JTAG interface and
+presents a @code{gdbserver} interface to the board. By default
+@code{xmd} uses port @code{1234}. (While it is possible to change
+this default port, it requires the use of undocumented @code{xmd}
+commands. Contact Xilinx support if you need to do this.)
+
+Use these GDB commands to connect to the MicroBlaze target processor.
+
+@table @code
+@item target remote :1234
+Use this command to connect to the target if you are running @value{GDBN}
+on the same system as @code{xmd}.
+
+@item target remote @var{xmd-host}:1234
+Use this command to connect to the target if it is connected to @code{xmd}
+running on a different system named @var{xmd-host}.
+
+@item load
+Use this command to download a program to the MicroBlaze target.
+
+@item set debug microblaze @var{n}
+Enable MicroBlaze-specific debugging messages if non-zero.
+
+@item show debug microblaze @var{n}
+Show MicroBlaze-specific debugging level.
+@end table
+
@node MIPS Embedded
@subsection MIPS Embedded
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/MAINTAINERS mb-gdb/gdb/MAINTAINERS
--- gdb/gdb/MAINTAINERS 2009-10-01 16:11:02.000000000 -0700
+++ mb-gdb/gdb/MAINTAINERS 2009-10-05 10:52:46.000000000 -0700
@@ -290,6 +290,10 @@ the native maintainer when resolving ABI
mep --target=mep-elf ,-Werror
Kevin Buettner kevinb@redhat.com
+ microblaze --target=microblaze-xilinx-elf ,-Werror
+ --target=microblaze-linux-gnu ,-Werror
+ Michael Eager eager@eagercon.com
+
mips --target=mips-elf ,-Werror
mn10300 --target=mn10300-elf broken
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/Makefile.in mb-gdb/gdb/Makefile.in
--- gdb/gdb/Makefile.in 2009-09-16 14:55:57.000000000 -0700
+++ mb-gdb/gdb/Makefile.in 2009-10-05 11:59:21.000000000 -0700
@@ -501,6 +501,7 @@ ALL_TARGET_OBS = \
m68kbsd-tdep.o m68klinux-tdep.o m68k-tdep.o \
m88k-tdep.o \
mep-tdep.o \
+ microblaze-tdep.o microblaze-linux-tdep.o \
mips-irix-tdep.o mips-linux-tdep.o \
mipsnbsd-tdep.o mips-tdep.o \
mn10300-linux-tdep.o mn10300-tdep.o \
@@ -750,7 +751,7 @@ config/rs6000/nm-rs6000.h top.h bsd-kvm.
annotate.h sim-regno.h dictionary.h dfp.h main.h frame-unwind.h \
remote-fileio.h i386-linux-tdep.h vax-tdep.h objc-lang.h \
sentinel-frame.h bcache.h symfile.h windows-tdep.h linux-tdep.h \
-gdb_usleep.h jit.h xml-syscall.h ada-operator.inc
+gdb_usleep.h jit.h xml-syscall.h ada-operator.inc microblaze-tdep.h
# Header files that already have srcdir in them, or which are in objdir.
@@ -1434,6 +1435,7 @@ ALLDEPFILES = \
m68kbsd-nat.c m68kbsd-tdep.c \
m68klinux-nat.c m68klinux-tdep.c \
m88k-tdep.c m88kbsd-nat.c \
+ microblaze-tdep.c microblaze-linux-tdep.c \
mingw-hdep.c \
mips-linux-nat.c mips-linux-tdep.c \
mips-irix-tdep.c \
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-linux-tdep.c mb-gdb/gdb/microblaze-linux-tdep.c
--- gdb/gdb/microblaze-linux-tdep.c 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-linux-tdep.c 2009-09-30 11:51:38.000000000 -0700
@@ -0,0 +1,142 @@
+/* Target-dependent code for Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#include "defs.h"
+#include "frame.h"
+#include "inferior.h"
+#include "symtab.h"
+#include "target.h"
+#include "gdbcore.h"
+#include "gdbcmd.h"
+#include "symfile.h"
+#include "objfiles.h"
+#include "regcache.h"
+#include "value.h"
+#include "osabi.h"
+#include "regset.h"
+#include "solib-svr4.h"
+#include "microblaze-tdep.h"
+#include "trad-frame.h"
+#include "frame-unwind.h"
+#include "tramp-frame.h"
+
+
+static int
+microblaze_linux_memory_remove_breakpoint (struct bp_target_info *bp_tgt)
+{
+ CORE_ADDR addr = bp_tgt->placed_address;
+ const gdb_byte *bp;
+ int val;
+ int bplen;
+ gdb_byte old_contents[BREAKPOINT_MAX];
+
+ /* Determine appropriate breakpoint contents and size for this address. */
+ bp = gdbarch_breakpoint_from_pc (gdbarch, &addr, &bplen);
+ if (bp == NULL)
+ error (_("Software breakpoints not implemented for this target."));
+
+ val = target_read_memory (addr, old_contents, bplen);
+
+ /* If our breakpoint is no longer at the address, this means that the
+ program modified the code on us, so it is wrong to put back the
+ old value. */
+ if (val == 0 && memcmp (bp, old_contents, bplen) == 0)
+ val = target_write_memory (addr, bp_tgt->shadow_contents, bplen);
+
+ return val;
+}
+
+static void
+microblaze_linux_sigtramp_cache (struct frame_info *next_frame,
+ struct trad_frame_cache *this_cache,
+ CORE_ADDR func, LONGEST offset,
+ int bias)
+{
+ CORE_ADDR base;
+ CORE_ADDR gpregs;
+ int regnum;
+ struct gdbarch *gdbarch = get_frame_arch (next_frame);
+ struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+
+ base = frame_unwind_register_unsigned (next_frame, SP_REGNUM);
+ if (bias > 0 && frame_pc_unwind (next_frame) != func)
+ /* See below, some signal trampolines increment the stack as their
+ first instruction, need to compensate for that. */
+ base -= bias;
+
+ /* Find the address of the register buffer. */
+ gpregs = base + offset;
+
+ /* Registers saved on stack. */
+ for (regnum = 0; regnum < MICROBLAZE_BTR_REGNUM; regnum++)
+ trad_frame_set_reg_addr (this_cache, regnum,
+ gpregs + regnum * MICROBLAZE_REGISTER_SIZE);
+ trad_frame_set_id (this_cache, frame_id_build (base, func));
+}
+
+
+static void
+microblaze_linux_sighandler_cache_init (const struct tramp_frame *self,
+ struct frame_info *next_frame,
+ struct trad_frame_cache *this_cache,
+ CORE_ADDR func)
+{
+ microblaze_linux_sigtramp_cache (next_frame, this_cache, func,
+ 0 /* Offset to ucontext_t. */
+ + 24 /* Offset to .reg. */,
+ 0);
+}
+
+static struct tramp_frame microblaze_linux_sighandler_tramp_frame =
+{
+ SIGTRAMP_FRAME,
+ 4,
+ {
+ { 0x31800077, -1 }, /* addik R12,R0,119. */
+ { 0xb9cc0008, -1 }, /* brki R14,8. */
+ { TRAMP_SENTINEL_INSN },
+ },
+ microblaze_linux_sighandler_cache_init
+};
+
+
+static void
+microblaze_linux_init_abi (struct gdbarch_info info,
+ struct gdbarch *gdbarch)
+{
+ struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+
+ set_gdbarch_memory_remove_breakpoint (gdbarch,
+ microblaze_linux_memory_remove_breakpoint);
+
+ /* Shared library handling. */
+ set_solib_svr4_fetch_link_map_offsets (gdbarch,
+ svr4_ilp32_fetch_link_map_offsets);
+
+ /* Trampolines. */
+ tramp_frame_prepend_unwinder (gdbarch,
+ µblaze_linux_sighandler_tramp_frame);
+}
+
+void
+_initialize_microblaze_linux_tdep (void)
+{
+ gdbarch_register_osabi (bfd_arch_microblaze, 0, GDB_OSABI_LINUX,
+ microblaze_linux_init_abi);
+}
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-rom.c mb-gdb/gdb/microblaze-rom.c
--- gdb/gdb/microblaze-rom.c 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-rom.c 2009-09-30 10:49:00.000000000 -0700
@@ -0,0 +1,191 @@
+/* Remote debugging interface to Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#include "defs.h"
+#include "gdbcore.h"
+#include "target.h"
+#include "monitor.h"
+#include "gdb_string.h"
+#include "serial.h"
+#include "regcache.h"
+
+static char *picobug_inits[] =
+{"\r", NULL};
+
+static struct target_ops picobug_ops;
+static struct monitor_ops picobug_cmds;
+
+/* Picobug only supports a subset of registers from MCore. In reality,
+ it doesn't support ss1, either. */
+static char *picobug_regnames[] = {
+ "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
+ "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ "psr", "vbr", "epsr", "fpsr", "epc", "fpc", 0, "ss1",
+ "ss2", "ss3", "ss4", 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ "pc" };
+
+
+
+static void
+picobug_open (char *args, int from_tty)
+{
+ monitor_open (args, &picobug_cmds, from_tty);
+}
+/* We choose to write our own dumpregs routine, since the output of
+ the register dumping is rather difficult to encapsulate in a
+ regexp:
+
+picobug> rd
+ pc 2f00031e epc 2f00031e fpc 00000000
+ psr 80000101 epsr 80000101 fpsr 00000000
+ss0-ss4 bad0beef 00000000 00000000 00000000 00000000 vbr 30005c00
+ r0-r7 2f0fff4c 00000090 00000001 00000002 00000003 00000004 00000005 00000006
+ r8-r15 2f0fff64 00000000 00000000 00000000 00000000 00000000 00000000 2f00031e
+*/
+
+static int
+picobug_dumpregs (struct regcache *regcache)
+{
+ struct gdbarch *gdbarch = get_regcache_arch (regcache);
+ char buf[1024];
+ int resp_len;
+ char *p;
+
+ /* Send the dump register command to the monitor and
+ get the reply. */
+ monitor_printf (picobug_cmds.dump_registers);
+ resp_len = monitor_expect_prompt (buf, sizeof (buf));
+
+ p = strtok (buf, " \t\r\n");
+ while (p)
+ {
+ if (strchr (p, '-'))
+ {
+ /* got a range. either r0-r7, r8-r15 or ss0-ss4. */
+ if (strncmp (p, "r0", 2) == 0 || strncmp (p, "r8", 2) == 0)
+ {
+ int rn = (p[1] == '0' ? 0 : 8);
+ int i = 0;
+
+ /* Get the next 8 values and record them. */
+ while (i < 8)
+ {
+ p = strtok (NULL, " \t\r\n");
+ if (p)
+ monitor_supply_register (regcache, rn + i, p);
+ i++;
+ }
+ }
+ else if (strncmp (p, "ss", 2) == 0)
+ {
+ /* get the next five values, ignoring the first. */
+ int rn;
+ p = strtok (NULL, " \t\r\n");
+ for (rn = 39; rn < 43; rn++)
+ {
+ p = strtok (NULL, " \t\r\n");
+ if (p)
+ monitor_supply_register (regcache, rn, p);
+ }
+ }
+ else
+ {
+ break;
+ }
+ }
+ else
+ {
+ /* Simple register type, paired. */
+ char *name = p;
+ int i;
+
+ /* Get and record value. */
+ p = strtok (NULL, " \t\r\n");
+ if (p)
+ {
+ for (i = 0; i < gdbarch_num_regs (gdbarch); i++)
+ {
+ if (picobug_regnames[i]
+ && (strcmp (picobug_regnames[i], name) == 0))
+ break;
+ }
+
+ if (i <= gdbarch_num_regs (gdbarch))
+ monitor_supply_register (regcache, i, p);
+ }
+ }
+ p = strtok (NULL, " \t\r\n");
+ }
+
+ return 0;
+}
+
+static void
+init_picobug_cmds (void)
+{
+ picobug_cmds.flags = MO_GETMEM_NEEDS_RANGE | MO_CLR_BREAK_USES_ADDR
+ | MO_PRINT_PROGRAM_OUTPUT;
+
+ picobug_cmds.init = picobug_inits; /* Init strings */
+ picobug_cmds.cont = "g\n"; /* continue command */
+ picobug_cmds.step = "s\n"; /* single step */
+ picobug_cmds.set_break = "br %x\n"; /* set a breakpoint */
+ picobug_cmds.clr_break = "nobr %x\n"; /* clear a breakpoint */
+ picobug_cmds.clr_all_break = "nobr\n"; /* clear all breakpoints */
+ picobug_cmds.setmem.cmdb = "mm %x %x ;b\n"; /* setmem.cmdb (addr, value) */
+ picobug_cmds.setmem.cmdw = "mm %x %x ;h\n"; /* setmem.cmdw (addr, value) */
+ picobug_cmds.setmem.cmdl = "mm %x %x ;w\n"; /* setmem.cmdl (addr, value) */
+ picobug_cmds.getmem.cmdb = "md %x %x\n"; /* getmem.cmdb (start addr,
+ end addr) */
+ picobug_cmds.getmem.resp_delim = ":"; /* getmem.resp_delim */
+ picobug_cmds.setreg.cmd = "rm %s %x\n"; /* setreg.cmd (name, value) */
+ picobug_cmds.getreg.cmd = "rd %s\n"; /* getreg.cmd (name) */
+ picobug_cmds.getreg.resp_delim = ":"; /* getreg.resp_delim */
+ picobug_cmds.dump_registers = "rd\n"; /* dump_registers */
+ picobug_cmds.dumpregs = picobug_dumpregs; /* dump registers parser */
+ picobug_cmds.load = "lo\n"; /* download command */
+ picobug_cmds.prompt = "picobug> "; /* monitor command prompt */
+ picobug_cmds.line_term = "\n"; /* end-of-line terminator */
+ picobug_cmds.target = &picobug_ops; /* target operations */
+ picobug_cmds.stopbits = SERIAL_1_STOPBITS; /* number of stop bits */
+ picobug_cmds.regnames = picobug_regnames; /* registers names */
+ picobug_cmds.num_breakpoints = 20; /* number of breakpoints */
+ picobug_cmds.magic = MONITOR_OPS_MAGIC; /* magic */
+}
+
+void
+_initialize_picobug_rom ()
+{
+ int i;
+
+ /* Initialize m32r RevC monitor target. */
+ init_picobug_cmds ();
+ init_monitor_ops (&picobug_ops);
+ picobug_ops.to_shortname = "picobug";
+ picobug_ops.to_longname = "picobug monitor";
+ picobug_ops.to_doc = "Debug via the picobug monitor.\n\
+Specify the serial device it is connected to (e.g. /dev/ttya).";
+ picobug_ops.to_open = picobug_open;
+
+ add_target (&picobug_ops);
+}
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-tdep.c mb-gdb/gdb/microblaze-tdep.c
--- gdb/gdb/microblaze-tdep.c 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-tdep.c 2009-10-05 14:09:02.000000000 -0700
@@ -0,0 +1,877 @@
+/* Target-dependent code for Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#include "defs.h"
+#include "arch-utils.h"
+#include "dis-asm.h"
+#include "frame.h"
+#include "trad-frame.h"
+#include "symtab.h"
+#include "value.h"
+#include "gdbcmd.h"
+#include "breakpoint.h"
+#include "inferior.h"
+#include "regcache.h"
+#include "target.h"
+#include "frame.h"
+#include "frame-base.h"
+#include "frame-unwind.h"
+#include "dwarf2-frame.h"
+#include "osabi.h"
+
+#include "gdb_assert.h"
+#include "gdb_string.h"
+#include "target-descriptions.h"
+#include "opcodes/microblaze-opcm.h"
+#include "opcodes/microblaze-dis.h"
+#include "microblaze-tdep.h"
+\f
+/* Instruction macros used for analyzing the prologue. */
+/* This set of instruction macros need to be changed whenever the
+ prologue generated by the compiler could have more instructions or
+ different type of instructions.
+ This set also needs to be verified if it is complete. */
+#define IS_RETURN(op) (op == rtsd || op == rtid)
+#define IS_UPDATE_SP(op, rd, ra) \
+ ((op == addik || op == addi) && rd == REG_SP && ra == REG_SP)
+#define IS_SPILL_SP(op, rd, ra) \
+ ((op == swi || op == sw) && rd == REG_SP && ra == REG_SP)
+#define IS_SPILL_REG(op, rd, ra) \
+ ((op == swi || op == sw) && rd != REG_SP && ra == REG_SP)
+#define IS_ALSO_SPILL_REG(op, rd, ra, rb) \
+ ((op == swi || op == sw) && rd != REG_SP && ra == 0 && rb == REG_SP)
+#define IS_SETUP_FP(op, ra, rb) \
+ ((op == add || op == addik || op == addk) && ra == REG_SP && rb == 0)
+#define IS_SPILL_REG_FP(op, rd, ra, fpregnum) \
+ ((op == swi || op == sw) && rd != REG_SP && ra == fpregnum && ra != 0)
+#define IS_SAVE_HIDDEN_PTR(op, rd, ra, rb) \
+ ((op == add || op == addik) && ra == MICROBLAZE_FIRST_ARGREG && rb == 0)
+
+/* All registers are 32 bits. */
+#define MICROBLAZE_REGISTER_SIZE 4
+
+/* The registers of the Xilinx microblaze processor. */
+
+static const char *microblaze_register_names[] =
+{
+ "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
+ "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
+ "r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
+ "r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31",
+ "rpc", "rmsr", "rear", "resr", "rfsr", "rbtr",
+ "rpvr0", "rpvr1", "rpvr2", "rpvr3", "rpvr4", "rpvr5", "rpvr6",
+ "rpvr7", "rpvr8", "rpvr9", "rpvr10", "rpvr11",
+ "redr", "rpid", "rzpr", "rtlbx", "rtlbsx", "rtlblo", "rtlbhi"
+};
+
+#define MICROBLAZE_NUM_REGS ARRAY_SIZE (microblaze_register_names)
+\f
+static int microblaze_debug = 0;
+
+#define microblaze_insn_debug(args) \
+ { if (microblaze_debug) { printf_filtered args; } }
+\f
+/* Return the name of register REGNUM. */
+
+static const char *
+microblaze_register_name (struct gdbarch *gdbarch, int regnum)
+{
+ if (regnum >= 0 && regnum < MICROBLAZE_NUM_REGS)
+ return microblaze_register_names[regnum];
+ return NULL;
+}
+
+static struct type *
+microblaze_register_type (struct gdbarch *gdbarch, int regnum)
+{
+ if (regnum == MICROBLAZE_SP_REGNUM)
+ return builtin_type (gdbarch)->builtin_data_ptr;
+
+ if (regnum == MICROBLAZE_PC_REGNUM)
+ return builtin_type (gdbarch)->builtin_func_ptr;
+
+ return builtin_type (gdbarch)->builtin_int;
+}
+
+\f
+/* Fetch the instruction at PC. */
+
+unsigned long
+microblaze_fetch_instruction (CORE_ADDR pc)
+{
+ enum bfd_endian byte_order = gdbarch_byte_order (target_gdbarch);
+ gdb_byte buf[4];
+
+ /* If we can't read the instruction at PC, return zero. */
+ if (target_read_memory (pc, buf, sizeof (buf)))
+ return 0;
+
+ return extract_unsigned_integer (buf, 4, byte_order);
+}
+\f
+
+static CORE_ADDR
+microblaze_push_dummy_code (struct gdbarch *gdbarch, CORE_ADDR sp,
+ CORE_ADDR funcaddr,
+ struct value **args, int nargs,
+ struct type *value_type,
+ CORE_ADDR *real_pc, CORE_ADDR *bp_addr,
+ struct regcache *regcache)
+{
+ error (_("push_dummy_code not implemented"));
+ return sp;
+}
+
+
+static CORE_ADDR
+microblaze_push_dummy_call (struct gdbarch *gdbarch, struct value *function,
+ struct regcache *regcache, CORE_ADDR bp_addr,
+ int nargs, struct value **args, CORE_ADDR sp,
+ int struct_return, CORE_ADDR struct_addr)
+{
+ error (_("store_arguments not implemented"));
+ return sp;
+}
+
+static const gdb_byte *
+microblaze_breakpoint_from_pc (struct gdbarch *gdbarch, CORE_ADDR *pc,
+ int *len)
+{
+ static gdb_byte break_insn[] = MICROBLAZE_BREAKPOINT;
+
+ *len = sizeof (break_insn);
+ return break_insn;
+}
+\f
+/* Allocate and initialize a frame cache. */
+
+static struct microblaze_frame_cache *
+microblaze_alloc_frame_cache (void)
+{
+ struct microblaze_frame_cache *cache;
+ int i;
+
+ cache = FRAME_OBSTACK_ZALLOC (struct microblaze_frame_cache);
+
+ /* Base address. */
+ cache->base = 0;
+ cache->pc = 0;
+
+ /* Frameless until proven otherwise. */
+ cache->frameless_p = 1;
+
+ return cache;
+}
+
+/* The base of the current frame is actually in the stack pointer.
+ This happens when there is no frame pointer (microblaze ABI does not
+ require a frame pointer) or when we're stopped in the prologue or
+ epilogue itself. In these cases, microblaze_analyze_prologue will need
+ to update fi->frame before returning or analyzing the register
+ save instructions. */
+#define MICROBLAZE_MY_FRAME_IN_SP 0x1
+
+/* The base of the current frame is in a frame pointer register.
+ This register is noted in frame_extra_info->fp_regnum.
+
+ Note that the existance of an FP might also indicate that the
+ function has called alloca. */
+#define MICROBLAZE_MY_FRAME_IN_FP 0x2
+
+/* Function prologues on the Xilinx microblaze processors consist of:
+
+ - adjustments to the stack pointer (r1) (addi r1, r1, imm)
+ - making a copy of r1 into another register (a "frame" pointer)
+ (add r?, r1, r0)
+ - store word/multiples that use r1 or the frame pointer as the
+ base address (swi r?, r1, imm OR swi r?, fp, imm)
+
+ Note that microblaze really doesn't have a real frame pointer.
+ Instead, the compiler may copy the SP into a register (usually
+ r19) to act as an arg pointer. For our target-dependent purposes,
+ the frame info's "frame" member will be the beginning of the
+ frame. The SP could, in fact, point below this.
+
+ The prologue ends when an instruction fails to meet either of
+ these criteria. */
+
+/* Analyze the prologue to determine where registers are saved,
+ the end of the prologue, etc. Return the address of the first line
+ of "real" code (i.e., the end of the prologue). */
+
+CORE_ADDR
+microblaze_analyze_prologue (CORE_ADDR pc, CORE_ADDR current_pc,
+ struct microblaze_frame_cache *cache)
+{
+ char *name;
+ CORE_ADDR func_addr, func_end, addr, stop, prologue_end_addr = 0;
+ unsigned long insn;
+ int rn, rd, ra, rb, imm;
+ enum microblaze_instr op;
+ int flags = 0;
+ int save_hidden_pointer_found = 0;
+ int non_stack_instruction_found = 0;
+
+ /* Find the start of this function. */
+ find_pc_partial_function (pc, &name, &func_addr, &func_end);
+ if (func_addr < pc)
+ pc = func_addr;
+
+ if (current_pc < pc)
+ return current_pc;
+
+ /* Initialize info about frame. */
+ cache->framesize = 0;
+ cache->fp_regnum = MICROBLAZE_SP_REGNUM;
+ cache->frameless_p = 1;
+
+ /* Start decoding the prologue. We start by checking two special cases:
+
+ 1. We're about to return
+ 2. We're at the first insn of the prologue.
+
+ If we're about to return, our frame has already been deallocated.
+ If we are stopped at the first instruction of a prologue,
+ then our frame has not yet been set up. */
+
+ /* Get the first insn from memory. */
+
+ insn = microblaze_fetch_instruction (pc);
+ op = microblaze_decode_insn (insn, &rd, &ra, &rb, &imm);
+
+ if (IS_RETURN(op))
+ return pc;
+
+ /* Start at beginning of function and analyze until we get to the
+ current pc, or the end of the function, whichever is first. */
+ stop = (current_pc < func_end ? current_pc : func_end);
+
+ microblaze_insn_debug (("MICROBLAZE: Scanning prologue: name=%s, "
+ "func_addr=0x%x, stop=0x%x\n",
+ name, (unsigned int) func_addr,
+ (unsigned int) stop));
+
+ for (addr = func_addr; addr < stop; addr += INST_WORD_SIZE)
+ {
+ insn = microblaze_fetch_instruction (addr);
+ op = microblaze_decode_insn (insn, &rd, &ra, &rb, &imm);
+ if (microblaze_debug)
+ printf_filtered ("MICROBLAZE: %08x %08lx\n", (unsigned int) pc, insn);
+
+ /* This code is very sensitive to what functions are present in the
+ prologue. It assumes that the (addi, addik, swi, sw) can be the
+ only instructions in the prologue. */
+ if (IS_UPDATE_SP(op, rd, ra))
+ {
+ microblaze_insn_debug ((
+ "MICROBLAZE: got addi r1,r1,%d; contnuing\n", imm));
+ if (cache->framesize)
+ break; /* break if framesize already computed. */
+ cache->framesize = -imm; /* stack grows towards low memory. */
+ cache->frameless_p = 0; /* Frame found. */
+ save_hidden_pointer_found = 0;
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SPILL_SP(op, rd, ra))
+ {
+ /* Spill stack pointer. */
+ cache->register_offsets[rd] = imm; /* SP spilled before updating. */
+
+ microblaze_insn_debug (
+ ("MICROBLAZE: swi r1 r1 %d, continuing\n", imm));
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SPILL_REG(op, rd, ra))
+ {
+ /* Spill register. */
+ cache->register_offsets[rd] = imm - cache->framesize;
+
+ microblaze_insn_debug (
+ ("MICROBLAZE: swi %d r1 %d, continuing\n", rd, imm));
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_ALSO_SPILL_REG(op, rd, ra, rb))
+ {
+ /* Spill register. */
+ cache->register_offsets[rd] = 0 - cache->framesize;
+
+ microblaze_insn_debug (
+ ("MICROBLAZE: sw %d r0 r1, continuing\n", rd));
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SETUP_FP(op, ra, rb))
+ {
+ /* We have a frame pointer. Note the register which is
+ acting as the frame pointer. */
+ flags |= MICROBLAZE_MY_FRAME_IN_FP;
+ flags &= ~MICROBLAZE_MY_FRAME_IN_SP;
+ cache->fp_regnum = rd;
+ microblaze_insn_debug (
+ ("MICROBLAZE: Found a frame pointer: r%d\n",
+ cache->fp_regnum));
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SPILL_REG_FP(op, rd, ra, cache->fp_regnum))
+ {
+ cache->register_offsets[rd] = imm - cache->framesize;
+ /* reg spilled after updating. */
+
+ microblaze_insn_debug (
+ ("MICROBLAZE: swi %d %d %d, continuing\n", rd, ra, imm));
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SAVE_HIDDEN_PTR(op, rd, ra, rb))
+ {
+ /* If the first argument is a hidden pointer to the area where the
+ return structure is to be saved, then it is saved as part of the
+ prologue. */
+
+ microblaze_insn_debug (
+ ("MICROBLAZE: add %d %d %d, continuing\n", rd, ra, rb));
+ save_hidden_pointer_found = 1;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+
+ /* As a result of the modification in the next step where we continue
+ to analyze the prologue till we reach a control flow instruction,
+ we need another variable to store when exactly a non-stack
+ instruction was encountered, which is the current definition
+ of a prologue. */
+ if (!non_stack_instruction_found)
+ prologue_end_addr = addr;
+ non_stack_instruction_found = 1;
+
+ /* When optimizations are enabled, it is not guaranteed that prologue
+ instructions are not mixed in with other instructions from the
+ program. Some programs show this behavior at -O2. This can be
+ avoided by adding -fno-schedule-insns2 switch as of now (edk 8.1)
+ In such cases, we scan the function until we see the first control
+ instruction. */
+
+ {
+ unsigned op = (unsigned)insn >> 26;
+
+ if (!(op == 0x26 || op == 0x27 || op == 0x2d || op == 0x2e || op == 0x2f))
+ /* continue if not control flow (branch, return). */
+ continue;
+ else if (op == 0x2c)
+ continue; /* continue if imm. */
+ }
+
+ /* This is not a prologue insn, so stop here. */
+ microblaze_insn_debug (
+ ("microblaze: insn is not a prologue insn -- ending scan\n"));
+ break;
+ }
+
+ microblaze_insn_debug (("microblaze: done analyzing prologue\n"));
+ microblaze_insn_debug (("microblaze: prologue end = 0x%x\n", (int) addr));
+
+ /* If the last instruction was an add rd, r5, r0 then don't count it as
+ part of the prologue. */
+ if (save_hidden_pointer_found)
+ prologue_end_addr -= INST_WORD_SIZE;
+
+ return prologue_end_addr;
+}
+
+static CORE_ADDR
+microblaze_unwind_pc (struct gdbarch *gdbarch, struct frame_info *next_frame)
+{
+ gdb_byte buf[4];
+ CORE_ADDR pc;
+
+ frame_unwind_register (next_frame, MICROBLAZE_PC_REGNUM, buf);
+ pc = extract_typed_address (buf, builtin_type (gdbarch)->builtin_func_ptr);
+ /* For sentinel frame, return address is actual PC. For other frames,
+ return address is pc+8. This is a workaround because gcc does not
+ generate correct return address in CIE. */
+ if (frame_relative_level (next_frame) >= 0)
+ pc += 8;
+ return pc;
+}
+
+/* Return PC of first real instruction of the function starting at
+ START_PC. */
+
+CORE_ADDR
+microblaze_skip_prologue (struct gdbarch *gdbarch, CORE_ADDR start_pc)
+{
+ struct symtab_and_line sal;
+ CORE_ADDR func_start, func_end, ostart_pc;
+ struct microblaze_frame_cache cache;
+
+ /* This is the preferred method, find the end of the prologue by
+ using the debugging information. Debugging info does not always
+ give the right answer since parameters are stored on stack after this.
+ Always analyze the prologue. */
+ if (find_pc_partial_function (start_pc, NULL, &func_start, &func_end))
+ {
+ sal = find_pc_line (func_start, 0);
+
+ if (sal.end < func_end
+ && start_pc <= sal.end)
+ start_pc = sal.end;
+ }
+
+ ostart_pc = microblaze_analyze_prologue (func_start, 0xffffffffUL, &cache);
+
+ if (ostart_pc > start_pc)
+ return ostart_pc;
+ return start_pc;
+}
+
+/* Normal frames. */
+
+struct microblaze_frame_cache *
+microblaze_frame_cache (struct frame_info *next_frame, void **this_cache)
+{
+ struct microblaze_frame_cache *cache;
+ struct gdbarch *gdbarch = get_frame_arch (next_frame);
+ CORE_ADDR func, pc, fp;
+ int rn;
+
+ if (*this_cache)
+ return *this_cache;
+
+ cache = microblaze_alloc_frame_cache ();
+ *this_cache = cache;
+ cache->saved_regs = trad_frame_alloc_saved_regs (next_frame);
+
+ /* Clear offsets to saved regs in frame. */
+ for (rn = 0; rn < gdbarch_num_regs (gdbarch); rn++)
+ cache->register_offsets[rn] = -1;
+
+ func = get_frame_func (next_frame);
+
+ cache->pc = get_frame_address_in_block (next_frame);
+
+ return cache;
+}
+
+static void
+microblaze_frame_this_id (struct frame_info *next_frame, void **this_cache,
+ struct frame_id *this_id)
+{
+ struct microblaze_frame_cache *cache =
+ microblaze_frame_cache (next_frame, this_cache);
+
+ /* This marks the outermost frame. */
+ if (cache->base == 0)
+ return;
+
+ (*this_id) = frame_id_build (cache->base, cache->pc);
+}
+
+static struct value *
+microblaze_frame_prev_register (struct frame_info *this_frame,
+ void **this_cache, int regnum)
+{
+ struct microblaze_frame_cache *cache =
+ microblaze_frame_cache (this_frame, this_cache);
+
+ if (cache->frameless_p)
+ {
+ if (regnum == MICROBLAZE_PC_REGNUM)
+ regnum = 15;
+ if (regnum == MICROBLAZE_SP_REGNUM)
+ regnum = 1;
+ return trad_frame_get_prev_register (this_frame, cache->saved_regs, regnum);
+ }
+ else
+ return trad_frame_get_prev_register (this_frame, cache->saved_regs,
+ regnum);
+
+}
+
+static const struct frame_unwind microblaze_frame_unwind =
+{
+ NORMAL_FRAME,
+ microblaze_frame_this_id,
+ microblaze_frame_prev_register,
+ NULL,
+ default_frame_sniffer
+};
+\f
+static CORE_ADDR
+microblaze_frame_base_address (struct frame_info *next_frame, void **this_cache)
+{
+ struct microblaze_frame_cache *cache =
+ microblaze_frame_cache (next_frame, this_cache);
+
+ return cache->base;
+}
+
+static const struct frame_base microblaze_frame_base =
+{
+ µblaze_frame_unwind,
+ microblaze_frame_base_address,
+ microblaze_frame_base_address,
+ microblaze_frame_base_address
+};
+\f
+/* Extract from an array REGBUF containing the (raw) register state, a
+ function return value of TYPE, and copy that into VALBUF. */
+static void
+microblaze_extract_return_value (struct type *type, struct regcache *regcache,
+ gdb_byte *valbuf)
+{
+ gdb_byte buf[8];
+
+ /* Copy the return value (starting) in RETVAL_REGNUM to VALBUF. */
+ switch (TYPE_LENGTH (type))
+ {
+ case 1: /* return last byte in the register. */
+ regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
+ memcpy(valbuf, buf + MICROBLAZE_REGISTER_SIZE - 1, 1);
+ return;
+ case 2: /* return last 2 bytes in register. */
+ memcpy(valbuf, buf + MICROBLAZE_REGISTER_SIZE - 2, 2);
+ return;
+ case 4: /* for sizes 4 or 8, copy the required length. */
+ case 8:
+ regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
+ regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM+1, buf+4);
+ memcpy (valbuf, buf, TYPE_LENGTH (type));
+ return;
+ default:
+ fatal (_("Unsupported return value size requested"));
+ }
+}
+
+/* Store the return value in VALBUF (of type TYPE) where the caller
+ expects to see it.
+
+ Integers up to four bytes are stored in r3.
+
+ Longs are stored in r3 (most significant word) and r4 (least
+ significant word).
+
+ Small structures are always returned on stack.
+*/
+
+static void
+microblaze_store_return_value (struct type *type, struct regcache *regcache,
+ const gdb_byte *valbuf)
+{
+ int len = TYPE_LENGTH (type);
+ gdb_byte buf[8];
+
+ memset (buf, 0, sizeof(buf));
+
+ /* Integral and pointer return values. */
+
+ if (len > 4)
+ {
+ gdb_assert (len == 8);
+ memcpy (buf, valbuf, 8);
+ regcache_cooked_write (regcache, MICROBLAZE_RETVAL_REGNUM+1, buf + 4);
+ }
+ else
+ /* ??? Do we need to do any sign-extension here? */
+ memcpy (buf + 4 - len, valbuf, len);
+
+ regcache_cooked_write (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
+}
+
+static enum return_value_convention
+microblaze_return_value (struct gdbarch *gdbarch, struct type *func_type,
+ struct type *type, struct regcache *regcache,
+ gdb_byte *readbuf, const gdb_byte *writebuf)
+{
+ if (readbuf)
+ microblaze_extract_return_value (type, regcache, readbuf);
+ if (writebuf)
+ microblaze_store_return_value (type, regcache, writebuf);
+
+ return RETURN_VALUE_REGISTER_CONVENTION;
+}
+
+static int
+microblaze_stabs_argument_has_addr (struct gdbarch *gdbarch, struct type *type)
+{
+ return (TYPE_LENGTH (type) == 16);
+}
+\f
+int
+microblaze_can_use_hardware_watchpoints (enum bptype type, int count, int ot)
+{
+ switch (type)
+ {
+ case bp_hardware_breakpoint:
+ if (count <= 2)
+ return 1;
+ break;
+ case bp_read_watchpoint:
+ case bp_hardware_watchpoint:
+ if (count <= 1)
+ return 1;
+ break;
+ default:
+ return 0;
+ }
+ return -1;
+}
+
+/* Simulate single-step calls, since microblaze does not support
+ single-stepping. */
+
+/* single_step() is called just before we want to resume the inferior,
+ if we want to single-step it but there is no hardware or kernel
+ single-step support. We find all the possible targets of the
+ coming instruction and breakpoint them.
+
+ single_step is also called just after the inferior stops. If we had
+ set up a simulated single-step, we undo our damage. */
+
+int
+microblaze_software_single_step (struct frame_info *frame)
+{
+ struct gdbarch *gdbarch = get_frame_arch (frame);
+ struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+ struct regcache *regcache = get_current_regcache ();
+
+ /* Save the address and the values of the next_pc and the target. */
+ static struct sstep_breaks
+ {
+ CORE_ADDR address;
+ bfd_boolean valid;
+ } stepbreaks[2];
+
+ CORE_ADDR pc;
+ long insn;
+ enum microblaze_instr minstr;
+ int is_unsigned_num;
+ enum microblaze_instr_type insn_type;
+ short delay_slots;
+ int imm;
+ int immfound = 0;
+
+ /* Set a breakpoint at the next instruction. */
+ /* If the current instruction is an imm, set it at the inst after. */
+ /* If the instruction has a delay slot, skip the delay slot. */
+ pc = get_frame_register_unsigned (frame, MICROBLAZE_PC_REGNUM);
+ insn = microblaze_fetch_instruction (pc);
+ minstr = get_insn_microblaze (insn, &is_unsigned_num, &insn_type,
+ &delay_slots);
+ if (insn_type == immediate_inst)
+ {
+ int rd, ra, rb;
+ immfound = 1;
+ minstr = microblaze_decode_insn (insn, &rd, &ra, &rb, &imm);
+ pc = pc + INST_WORD_SIZE;
+ insn = microblaze_fetch_instruction (pc);
+ minstr = get_insn_microblaze (insn, &is_unsigned_num, &insn_type,
+ &delay_slots);
+ }
+ stepbreaks[0].address = pc + (delay_slots * INST_WORD_SIZE) + INST_WORD_SIZE;
+ if (insn_type != return_inst)
+ stepbreaks[0].valid = 1;
+ else
+ stepbreaks[0].valid = 0;
+
+ /* Now check for branch or return instructions. */
+ if (insn_type == branch_inst || insn_type == return_inst)
+ {
+ int limm;
+ int lrd, lra, lrb;
+ ULONGEST ra, rb;
+ bfd_boolean targetvalid;
+ bfd_boolean unconditionalbranch;
+ microblaze_decode_insn(insn, &lrd, &lra, &lrb, &limm);
+ if (lra >= 0 && lra < gdbarch_num_regs (gdbarch))
+ regcache_cooked_read_unsigned (regcache, lra, &ra);
+ else
+ ra = 0;
+ if (lrb >= 0 && lrb < gdbarch_num_regs (gdbarch))
+ regcache_cooked_read_unsigned (regcache, lrb, &rb);
+ else
+ rb = 0;
+ stepbreaks[1].address =
+ microblaze_get_target_address (insn, immfound, imm, pc,
+ ra, rb, &targetvalid,
+ &unconditionalbranch);
+ if (unconditionalbranch)
+ /* This is a unconditional branch: will not come to the next address. */
+ stepbreaks[0].valid = 0;
+ if (targetvalid
+ && (stepbreaks[0].valid == 0
+ || (stepbreaks[0].address != stepbreaks[1].address))
+ && (stepbreaks[1].address != pc))
+ stepbreaks[1].valid = 1;
+ else
+ stepbreaks[1].valid = 0;
+ }
+ else
+ stepbreaks[1].valid = 0;
+
+ /* Insert the breakpoints. */
+ if (stepbreaks[0].valid)
+ insert_single_step_breakpoint (gdbarch, stepbreaks[0].address);
+ if (stepbreaks[1].valid)
+ insert_single_step_breakpoint (gdbarch, stepbreaks[1].address);
+
+ return 1;
+}
+
+static void
+microblaze_write_pc (struct regcache *regcache, CORE_ADDR pc)
+{
+ regcache_cooked_write_unsigned (regcache, MICROBLAZE_PC_REGNUM, pc);
+}
+\f
+static int dwarf2_to_reg_map[78] =
+{ 0 /* r0 */, 1 /* r1 */, 2 /* r2 */, 3 /* r3 */, /* 0- 3 */
+ 4 /* r4 */, 5 /* r5 */, 6 /* r6 */, 7 /* r7 */, /* 4- 7 */
+ 8 /* r8 */, 9 /* r9 */, 10 /* r10 */, 11 /* r11 */, /* 8-11 */
+ 12 /* r12 */, 13 /* r13 */, 14 /* r14 */, 15 /* r15 */, /* 12-15 */
+ 16 /* r16 */, 17 /* r17 */, 18 /* r18 */, 19 /* r19 */, /* 16-19 */
+ 20 /* r20 */, 21 /* r21 */, 22 /* r22 */, 23 /* r23 */, /* 20-23 */
+ 24 /* r24 */, 25 /* r25 */, 26 /* r26 */, 27 /* r27 */, /* 24-25 */
+ 28 /* r28 */, 29 /* r29 */, 30 /* r30 */, 31 /* r31 */, /* 28-31 */
+ -1 /* $f0 */, -1 /* $f1 */, -1 /* $f2 */, -1 /* $f3 */, /* 32-35 */
+ -1 /* $f4 */, -1 /* $f5 */, -1 /* $f6 */, -1 /* $f7 */, /* 36-39 */
+ -1 /* $f8 */, -1 /* $f9 */, -1 /* $f10 */, -1 /* $f11 */, /* 40-43 */
+ -1 /* $f12 */, -1 /* $f13 */, -1 /* $f14 */, -1 /* $f15 */, /* 44-47 */
+ -1 /* $f16 */, -1 /* $f17 */, -1 /* $f18 */, -1 /* $f19 */, /* 48-51 */
+ -1 /* $f20 */, -1 /* $f21 */, -1 /* $f22 */, -1 /* $f23 */, /* 52-55 */
+ -1 /* $f24 */, -1 /* $f25 */, -1 /* $f26 */, -1 /* $f27 */, /* 56-59 */
+ -1 /* $f28 */, -1 /* $f29 */, -1 /* $f30 */, -1 /* $f31 */, /* 60-63 */
+ -1 /* hi */, -1 /* lo */, -1 /* accum*/, 33 /* rmsr */, /* 64-67 */
+ -1 /* $fcc1*/, -1 /* $fcc2*/, -1 /* $fcc3*/, -1 /* $fcc4*/, /* 68-71 */
+ -1 /* $fcc5*/, -1 /* $fcc6*/, -1 /* $fcc7*/, -1 /* $ap */, /* 72-75 */
+ -1 /* $rap */, -1 /* $frp */ /* 76-77 */
+};
+
+static int
+microblaze_dwarf2_reg_to_regnum (struct gdbarch *gdbarch, int reg)
+{
+ gdb_assert (reg < sizeof (dwarf2_to_reg_map));
+ return dwarf2_to_reg_map[reg];
+}
+
+static struct gdbarch *
+microblaze_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
+{
+ struct gdbarch_tdep *tdep;
+ struct gdbarch *gdbarch;
+
+ /* If there is already a candidate, use it. */
+ arches = gdbarch_list_lookup_by_info (arches, &info);
+ if (arches != NULL)
+ return arches->gdbarch;
+
+ /* Allocate space for the new architecture. */
+ tdep = XMALLOC (struct gdbarch_tdep);
+ gdbarch = gdbarch_alloc (&info, tdep);
+
+ set_gdbarch_long_double_bit (gdbarch, 128);
+
+ set_gdbarch_num_regs (gdbarch, MICROBLAZE_NUM_REGS);
+ set_gdbarch_register_name (gdbarch, microblaze_register_name);
+ set_gdbarch_register_type (gdbarch, microblaze_register_type);
+
+ /* Register numbers of various important registers. */
+ set_gdbarch_sp_regnum (gdbarch, MICROBLAZE_SP_REGNUM);
+ set_gdbarch_pc_regnum (gdbarch, MICROBLAZE_PC_REGNUM);
+
+ /* Map Dwarf2 registers to GDB registers. */
+ set_gdbarch_dwarf2_reg_to_regnum (gdbarch, microblaze_dwarf2_reg_to_regnum);
+
+ /* Call dummy code. */
+ set_gdbarch_call_dummy_location (gdbarch, ON_STACK);
+ set_gdbarch_push_dummy_code (gdbarch, microblaze_push_dummy_code);
+ set_gdbarch_push_dummy_call (gdbarch, microblaze_push_dummy_call);
+
+ set_gdbarch_return_value (gdbarch, microblaze_return_value);
+ set_gdbarch_stabs_argument_has_addr
+ (gdbarch, microblaze_stabs_argument_has_addr);
+
+ set_gdbarch_skip_prologue (gdbarch, microblaze_skip_prologue);
+
+ /* Stack grows downward. */
+ set_gdbarch_inner_than (gdbarch, core_addr_lessthan);
+
+ set_gdbarch_breakpoint_from_pc (gdbarch, microblaze_breakpoint_from_pc);
+
+ set_gdbarch_frame_args_skip (gdbarch, 8);
+
+ set_gdbarch_print_insn (gdbarch, print_insn_microblaze);
+
+ set_gdbarch_software_single_step (gdbarch, microblaze_software_single_step);
+ set_gdbarch_write_pc (gdbarch, microblaze_write_pc);
+
+ set_gdbarch_unwind_pc (gdbarch, microblaze_unwind_pc);
+
+ frame_base_set_default (gdbarch, µblaze_frame_base);
+
+ /* Hook in ABI-specific overrides, if they have been registered. */
+ gdbarch_init_osabi (info, gdbarch);
+
+ /* Unwind the frame. */
+ dwarf2_append_unwinders (gdbarch);
+ frame_unwind_append_unwinder (gdbarch, µblaze_frame_unwind);
+ frame_base_append_sniffer (gdbarch, dwarf2_frame_base_sniffer);
+
+ return gdbarch;
+}
+
+int
+microblaze_frame_num_args_p (struct gdbarch *gdbarch)
+{
+ return 0;
+}
+
+/* Provide a prototype to silence -Wmissing-prototypes. */
+void _initialize_microblaze_tdep (void);
+
+void
+_initialize_microblaze_tdep (void)
+{
+ register_gdbarch_init (bfd_arch_microblaze, microblaze_gdbarch_init);
+
+ /* Debug this files internals. */
+ add_setshow_zinteger_cmd ("microblaze", class_maintenance,
+ µblaze_debug, _("\
+Set microblaze debugging."), _("\
+Show microblaze debugging."), _("\
+When non-zero, microblaze specific debugging is enabled."),
+ NULL,
+ NULL,
+ &setdebuglist, &showdebuglist);
+
+}
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-tdep.h mb-gdb/gdb/microblaze-tdep.h
--- gdb/gdb/microblaze-tdep.h 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-tdep.h 2009-10-05 14:34:14.000000000 -0700
@@ -0,0 +1,121 @@
+/* Target-dependent code for Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#ifndef MICROBLAZE_TDEP_H
+#define MICROBLAZE_TDEP_H 1
+
+
+/* Microblaze architecture-specific information. */
+struct gdbarch_tdep
+{
+};
+
+struct microblaze_frame_cache
+{
+ /* Base address. */
+ CORE_ADDR base;
+ CORE_ADDR pc;
+
+ /* Do we have a frame? */
+ int frameless_p;
+
+ /* Frame size. */
+ int framesize;
+
+ /* Frame register. */
+ int fp_regnum;
+
+ /* Offsets to saved registers. */
+ int register_offsets[57]; /* Must match MICROBLAZE_NUM_REGS. */
+
+ /* Table of saved registers. */
+ struct trad_frame_saved_reg *saved_regs;
+};
+
+/* Register numbers. */
+enum microblaze_regnum
+{
+ MICROBLAZE_R0_REGNUM,
+ MICROBLAZE_R1_REGNUM, MICROBLAZE_SP_REGNUM = MICROBLAZE_R1_REGNUM,
+ MICROBLAZE_R2_REGNUM,
+ MICROBLAZE_R3_REGNUM, MICROBLAZE_RETVAL_REGNUM = MICROBLAZE_R3_REGNUM,
+ MICROBLAZE_R4_REGNUM,
+ MICROBLAZE_R5_REGNUM, MICROBLAZE_FIRST_ARGREG = MICROBLAZE_R5_REGNUM,
+ MICROBLAZE_R6_REGNUM,
+ MICROBLAZE_R7_REGNUM,
+ MICROBLAZE_R8_REGNUM,
+ MICROBLAZE_R9_REGNUM,
+ MICROBLAZE_R10_REGNUM, MICROBLAZE_LAST_ARGREG = MICROBLAZE_R10_REGNUM,
+ MICROBLAZE_R11_REGNUM,
+ MICROBLAZE_R12_REGNUM,
+ MICROBLAZE_R13_REGNUM,
+ MICROBLAZE_R14_REGNUM,
+ MICROBLAZE_R15_REGNUM,
+ MICROBLAZE_R16_REGNUM,
+ MICROBLAZE_R17_REGNUM,
+ MICROBLAZE_R18_REGNUM,
+ MICROBLAZE_R19_REGNUM,
+ MICROBLAZE_R20_REGNUM,
+ MICROBLAZE_R21_REGNUM,
+ MICROBLAZE_R22_REGNUM,
+ MICROBLAZE_R23_REGNUM,
+ MICROBLAZE_R24_REGNUM,
+ MICROBLAZE_R25_REGNUM,
+ MICROBLAZE_R26_REGNUM,
+ MICROBLAZE_R27_REGNUM,
+ MICROBLAZE_R28_REGNUM,
+ MICROBLAZE_R29_REGNUM,
+ MICROBLAZE_R30_REGNUM,
+ MICROBLAZE_R31_REGNUM,
+ MICROBLAZE_PC_REGNUM,
+ MICROBLAZE_MSR_REGNUM,
+ MICROBLAZE_EAR_REGNUM,
+ MICROBLAZE_ESR_REGNUM,
+ MICROBLAZE_FSR_REGNUM,
+ MICROBLAZE_BTR_REGNUM,
+ MICROBLAZE_PVR0_REGNUM,
+ MICROBLAZE_PVR1_REGNUM,
+ MICROBLAZE_PVR2_REGNUM,
+ MICROBLAZE_PVR3_REGNUM,
+ MICROBLAZE_PVR4_REGNUM,
+ MICROBLAZE_PVR5_REGNUM,
+ MICROBLAZE_PVR6_REGNUM,
+ MICROBLAZE_PVR7_REGNUM,
+ MICROBLAZE_PVR8_REGNUM,
+ MICROBLAZE_PVR9_REGNUM,
+ MICROBLAZE_PVR10_REGNUM,
+ MICROBLAZE_PVR11_REGNUM,
+ MICROBLAZE_REDR_REGNUM,
+ MICROBLAZE_RPID_REGNUM,
+ MICROBLAZE_RZPR_REGNUM,
+ MICROBLAZE_RTLBX_REGNUM,
+ MICROBLAZE_RTLBSX_REGNUM,
+ MICROBLAZE_RTLBLO_REGNUM,
+ MICROBLAZE_RTLBHI_REGNUM
+};
+
+/* MICROBLAZE_BREAKPOINT defines the breakpoint that should be used.
+ Only used for native debugging. */
+#define MICROBLAZE_BREAKPOINT {0xb9, 0xcc, 0x00, 0x60}
+
+extern CORE_ADDR microblaze_analyze_prologue (CORE_ADDR, CORE_ADDR,
+ struct microblaze_frame_cache *);
+
+
+#endif /* microblaze-tdep.h */
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/NEWS mb-gdb/gdb/NEWS
--- gdb/gdb/NEWS 2009-10-04 10:39:30.000000000 -0700
+++ mb-gdb/gdb/NEWS 2009-10-05 10:52:47.000000000 -0700
@@ -482,6 +482,11 @@ Lattice Mico32 lm32-*
x86 DICOS i[34567]86-*-dicos*
x86_64 DICOS x86_64-*-dicos*
S+core 3 score-*-*
+Xilinx MicroBlaze microblaze-*-*
+
+* New Simulators
+
+Xilinx MicroBlaze microblaze
* The GDB remote stub, gdbserver, now supports x86 Windows CE
(mingw32ce) debugging.
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/opcodes/microblaze-dis.c mb-gdb/opcodes/microblaze-dis.c
--- gdb/opcodes/microblaze-dis.c 2009-09-25 12:59:51.000000000 -0700
+++ mb-gdb/opcodes/microblaze-dis.c 2009-10-05 14:18:58.000000000 -0700
@@ -24,9 +24,10 @@
#define STATIC_TABLE
#define DEFINE_TABLE
-#include "microblaze-opc.h"
#include "dis-asm.h"
#include <strings.h>
+#include "microblaze-opc.h"
+#include "microblaze-dis.h"
#define get_field_rd(instr) get_field (instr, RD_MASK, RD_LOW)
#define get_field_r1(instr) get_field (instr, RA_MASK, RA_LOW)
@@ -35,11 +36,6 @@
#define get_int_field_r1(instr) ((instr & RA_MASK) >> RA_LOW)
-enum microblaze_instr get_insn_microblaze (long, bfd_boolean *,
- enum microblaze_instr_type *, short *);
-unsigned long microblaze_get_target_address (long, bfd_boolean, int, long, long,
- long, bfd_boolean *, bfd_boolean *);
-enum microblaze_instr microblaze_decode_insn (long insn, int *rd, int *ra, int *rb, int *imm);
static char *
get_field (long instr, long mask, unsigned short low)
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/opcodes/microblaze-dis.h mb-gdb/opcodes/microblaze-dis.h
--- gdb/opcodes/microblaze-dis.h 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/opcodes/microblaze-dis.h 2009-10-05 14:11:15.000000000 -0700
@@ -0,0 +1,34 @@
+/* Disassemble Xilinx microblaze instructions.
+
+ Copyright 2009 Free Software Foundation, Inc.
+
+ This file is part of the GNU opcodes library.
+
+ This library 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 3, or (at your option)
+ any later version.
+
+ It 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 file; see the file COPYING. If not, write to the
+ Free Software Foundation, 51 Franklin Street - Fifth Floor, Boston,
+ MA 02110-1301, USA. */
+
+#ifndef MICROBLAZE_DIS_H
+#define MICROBLAZE_DIS_H 1
+
+extern enum microblaze_instr microblaze_decode_insn (long, int *, int *,
+ int *, int *);
+extern unsigned long microblaze_get_target_address (long, bfd_boolean, int,
+ long, long, long, bfd_boolean *, bfd_boolean *);
+
+extern enum microblaze_instr get_insn_microblaze (long, bfd_boolean *,
+ enum microblaze_instr_type *,
+ short *);
+
+#endif /* microblaze-dis.h */
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-06 1:35 [PATCH] Support for Xilinx MicroBlaze Michael Eager
@ 2009-10-06 7:14 ` Eli Zaretskii
2009-10-14 1:47 ` Joel Brobecker
1 sibling, 0 replies; 20+ messages in thread
From: Eli Zaretskii @ 2009-10-06 7:14 UTC (permalink / raw)
To: Michael Eager; +Cc: gdb-patches, brobecker
> Date: Mon, 05 Oct 2009 18:35:36 -0700
> From: Michael Eager <eager@eagercon.com>
> CC: Joel Brobecker <brobecker@adacore.com>, Eli Zaretskii <eliz@gnu.org>
>
> Revised patch. I believe that I have addressed all of the comments.
>
> 2009-10-05 Michael Eager <eager@eagercon.com>
>
> * config/djgpp/fnchange.lst: Add translations for cpu-microblaze.c,
> elf32-microblaze.c, microblaze-rom.c, microblaze-linux-tdep.c,
> microblaze-tdep.h, microblaze-tdep.c, microblaze-opc.h, microblaze-opcm.h,
> microblaze-dis.c, microblaze-dis.h, sim/microblaze, microblaze.h, and
> microblaze.isa.
> * configure.tgt: Add targets microblaze*-linux-*, microblaze*-xilinx-*.
> * doc/gdb.texinfo: Add MicroBlaze.
> * MAINTAINERS: Add self as maintainer for MicroBlaze.
> * Makefile.in: Build microblaze-tdep.o, microblaze-linux-tdep.o.
> HFILES_NO_SRCDIR: Add microblaze-tdep.h.
> * microblaze-linux-tdep.c: New.
> * microblaze-rom.c: New.
> * microblaze-tdep.c: New.
> * microblaze-tdep.h: New.
> * NEWS: Announce Xilinx MicroBlaze support.
Thanks. The changes to fnchange.lst, NEWS and gdb.texinfo are
approved. However, please note that gdb/doc has its own ChangeLog
file, so the entry for gdb.texinfo should go there, not in
gdb/ChangeLog. Also, the entry you put in gdb/doc/ChangeLog should
state the name of the node where you added your text, see the other
entries there for examples.
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-06 1:35 [PATCH] Support for Xilinx MicroBlaze Michael Eager
2009-10-06 7:14 ` Eli Zaretskii
@ 2009-10-14 1:47 ` Joel Brobecker
2009-10-14 17:49 ` Michael Eager
1 sibling, 1 reply; 20+ messages in thread
From: Joel Brobecker @ 2009-10-14 1:47 UTC (permalink / raw)
To: Michael Eager; +Cc: gdb-patches, Eli Zaretskii
> 2009-10-05 Michael Eager <eager@eagercon.com>
>
> * config/djgpp/fnchange.lst: Add translations for cpu-microblaze.c,
> elf32-microblaze.c, microblaze-rom.c, microblaze-linux-tdep.c,
> microblaze-tdep.h, microblaze-tdep.c, microblaze-opc.h, microblaze-opcm.h,
> microblaze-dis.c, microblaze-dis.h, sim/microblaze, microblaze.h, and
> microblaze.isa.
> * configure.tgt: Add targets microblaze*-linux-*, microblaze*-xilinx-*.
> * doc/gdb.texinfo: Add MicroBlaze.
> * MAINTAINERS: Add self as maintainer for MicroBlaze.
> * Makefile.in: Build microblaze-tdep.o, microblaze-linux-tdep.o.
> HFILES_NO_SRCDIR: Add microblaze-tdep.h.
> * microblaze-linux-tdep.c: New.
> * microblaze-rom.c: New.
> * microblaze-tdep.c: New.
> * microblaze-tdep.h: New.
> * NEWS: Announce Xilinx MicroBlaze support.
Overall, this looks like really good work. I just have a few comments,
and I think that the next iteration will be ready to be checked in.
I'll try to review the next revision faster, althought I will be away
for 10 days on a business trip next week.
The indentation for the ChangeLog entry needs to be 8 character, and
that will cause some of the lines to exceed 80 chars (I even prefer
limiting this to something less, such as 78 characters for instance).
As discussed privately, can you hold off on the following change
in MAINTAINERS:
> + Michael Eager eager@eagercon.com
For the record, this is not because we refuse to acknowledge Michael's
contribution of commitment to the port, but simply because this file
is used to list the port maintainers that have been nominated by the
GDB Global Maintainers. See for instance the recent nomination of Jan
Kratochvil and Tristan Gingold.
> + if (picobug_regnames[i]
> + && (strcmp (picobug_regnames[i], name) == 0))
Just a nit, you have some unnecessary parens around strcmp (...) == 0.
> +static CORE_ADDR
> +microblaze_push_dummy_code (struct gdbarch *gdbarch, CORE_ADDR sp,
> + CORE_ADDR funcaddr,
> + struct value **args, int nargs,
> + struct type *value_type,
> + CORE_ADDR *real_pc, CORE_ADDR *bp_addr,
> + struct regcache *regcache)
The formatting of the last line is slightly off...
> + microblaze_insn_debug (("MICROBLAZE: Scanning prologue: name=%s, "
> + "func_addr=0x%x, stop=0x%x\n",
> + name, (unsigned int) func_addr,
> + (unsigned int) stop));
The extra parens are because the macro is defined without the parens
in the call to printf:
+#define microblaze_insn_debug(args) \
+ { if (microblaze_debug) { printf_filtered args; } }
I would add the parens in the macro definition in order to make
the call to the macro more natural. I also recommend, for debugging
traces, that you use printf_unfiltered. Having to hit enter at every
new page when possibly generating tons of debugging traces can drive
someone crazy :).
Speaking of these debug traces, the macro name seems unnecessarily
specialized, as it really is nothing more than a printf. In fact,
you do not use it just a few lines below, but you still output the
same type of trace with the same "MICROBLAZE: ..." format. What
I personally do is define a ..._debug routine with a printf format
such as:
void wtx_opt_objfiles_debug (const char *format, ...)
ATTR_FORMAT (printf, 1, 2);
And with the following implementation:
void
wtx_opt_objfiles_debug (const char *format, ...)
{
if (debug_objfiles)
{
va_list args;
va_start (args, format);
printf_filtered ("DEBUG (objfiles): ");
vprintf_filtered (format, args);
printf_filtered ("\n");
va_end (args);
}
}
(Gaaahhh, what did I just say about using printf_filtered).
I like this approach, even though it's slightly more code, simply
because it allows me to have consistent debug traces without having
to test my debug flag. For instance, the following would become:
> + if (microblaze_debug)
> + printf_filtered ("MICROBLAZE: %08x %08lx\n", (unsigned int) pc, insn);
microblaze_debug ("%08x %08lx\n", (unsigned int) pc, insn);
Actually, there is a routine to print CORE_ADDR values: paddress.
So it should be:
microblaze_debug ("%s %08lx\n", paddress (pc), insn);
> + cache->register_offsets[rd] = imm - cache->framesize;
> + /* reg spilled after updating. */
Comments in GDB are placed before the associated code (unfortunately,
if you ask me).
> + microblaze_insn_debug (
> + ("MICROBLAZE: swi %d %d %d, continuing\n", rd, ra, imm));
It would look better if the parens on the first line did not terminate
the line. This is just a general remark, as you won't need 2 parens
if you fix the macro or adopt the ..._debug function or macro as I
suggest above.
> + if (!(op == 0x26 || op == 0x27 || op == 0x2d || op == 0x2e || op == 0x2f))
You may prefer to use
if (op != 0x26 && op != 0x27 && ..)
instead of "if (!(op == 0x26 || op == 0x27 || ...))" which I find slightly
confusing because you have to remember to keep reading until you find the
end of the expression to which the "!" applies. But this is a matter of
personal taste, so feel free to chose the version you prefer.
> + /* For sentinel frame, return address is actual PC. For other frames,
> + return address is pc+8. This is a workaround because gcc does not
> + generate correct return address in CIE. */
You do what you have to do, but this is *BAD* idea (IMO). Unless you can
detect the case when GCC generates bad return addresses or not in CIE,
you'll end up having a broken debugger as soon as the compiler gets
fixed. Introducing work arounds for compiler deficiencies is often fine,
but I don't think that this should be done at the cost of proper operation.
> + case 1: /* return last byte in the register. */
> + regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
> + memcpy(valbuf, buf + MICROBLAZE_REGISTER_SIZE - 1, 1);
> + return;
> + case 2: /* return last 2 bytes in register. */
> + memcpy(valbuf, buf + MICROBLAZE_REGISTER_SIZE - 2, 2);
> + return;
> + case 4: /* for sizes 4 or 8, copy the required length. */
> + case 8:
> + regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
> + regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM+1, buf+4);
> + memcpy (valbuf, buf, TYPE_LENGTH (type));
> + return;
> + default:
> + fatal (_("Unsupported return value size requested"));
fatal is verbotten except in a few specific cases. Can you use
internal_error instead?
> +int
> +microblaze_can_use_hardware_watchpoints (enum bptype type, int count, int ot)
This function appears to be unused? Let's delete it if that is the case.
> +int
> +microblaze_software_single_step (struct frame_info *frame)
Can this function be made static?
> + struct regcache *regcache = get_current_regcache ();
This one raised a red flag, as we try to avoid depending on global
variables. But I'm not sure what the kosher way of getting the regcache
would be. I thought there would be method to get the regcache from
a frame, but apparently not. Perhaps the right way is to use
get_thread_arch_regcache (inferior_ptid, gdbarch), but I'm not sure.
I'll ask Ulrich, who knows this area a lot better.
> +int
> +microblaze_frame_num_args_p (struct gdbarch *gdbarch)
Looks unused...
> + /* Debug this files internals. */
Tiny nit: Missing second space after the period.
> +extern CORE_ADDR microblaze_analyze_prologue (CORE_ADDR, CORE_ADDR,
> + struct microblaze_frame_cache *);
Let's make that function static and remove that declaration.
--
Joel
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-14 1:47 ` Joel Brobecker
@ 2009-10-14 17:49 ` Michael Eager
2009-10-14 17:56 ` Daniel Jacobowitz
0 siblings, 1 reply; 20+ messages in thread
From: Michael Eager @ 2009-10-14 17:49 UTC (permalink / raw)
To: Joel Brobecker; +Cc: gdb-patches, Eli Zaretskii
Joel Brobecker wrote:
>> 2009-10-05 Michael Eager <eager@eagercon.com>
>> + /* For sentinel frame, return address is actual PC. For other frames,
>> + return address is pc+8. This is a workaround because gcc does not
>> + generate correct return address in CIE. */
>
> You do what you have to do, but this is *BAD* idea (IMO). Unless you can
> detect the case when GCC generates bad return addresses or not in CIE,
> you'll end up having a broken debugger as soon as the compiler gets
> fixed. Introducing work arounds for compiler deficiencies is often fine,
> but I don't think that this should be done at the cost of proper operation.
Yes, I agree. But ...
I looked at fixing this in gcc a while ago. The translation from rtx to CIE
entry doesn't allow expressing PC+8. If I recall, there was another architecture
(sparc?) which also has a kludge to overcome this deficiency.
Perhaps it is possible to determine whether the CIE contains pc or pc+8
and make the right choice. I'll take a look at this, but I think that
this info is lost long before the return address is used. I really don't
want to put MicroBlaze-specific checks in the target-independent code which
parses the CIE.
If it's not possible to tell whether the CIE is generated correctly or not,
then the situation is that gdb will be unable to backtrace on a code generated
by a fixed compiler, or alternately, if gdb is changed, it will be unable to
backtrace on existing object files. Neither is an acceptable choice.
>> + struct regcache *regcache = get_current_regcache ();
>
> This one raised a red flag, as we try to avoid depending on global
> variables. But I'm not sure what the kosher way of getting the regcache
> would be. I thought there would be method to get the regcache from
> a frame, but apparently not. Perhaps the right way is to use
> get_thread_arch_regcache (inferior_ptid, gdbarch), but I'm not sure.
> I'll ask Ulrich, who knows this area a lot better.
I wasn't able to find a better way to get the regcache either. Perhaps
it's known farther up the call tree, but it isn't passed to the
software single step routine.
On the other hand, I'm not sure that this is functionality is
necessary in the real world. In an embedded environment, I believe
that the target manager (XDM) handles single stepping; on Linux, ptrace
does this. I'll check on this and see if I can eliminate the code.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-14 17:49 ` Michael Eager
@ 2009-10-14 17:56 ` Daniel Jacobowitz
2009-10-14 18:01 ` Michael Eager
0 siblings, 1 reply; 20+ messages in thread
From: Daniel Jacobowitz @ 2009-10-14 17:56 UTC (permalink / raw)
To: Michael Eager; +Cc: Joel Brobecker, gdb-patches, Eli Zaretskii
On Wed, Oct 14, 2009 at 10:48:57AM -0700, Michael Eager wrote:
> >>+ struct regcache *regcache = get_current_regcache ();
> >
> >This one raised a red flag, as we try to avoid depending on global
> >variables. But I'm not sure what the kosher way of getting the regcache
> >would be. I thought there would be method to get the regcache from
> >a frame, but apparently not. Perhaps the right way is to use
> >get_thread_arch_regcache (inferior_ptid, gdbarch), but I'm not sure.
> >I'll ask Ulrich, who knows this area a lot better.
>
> I wasn't able to find a better way to get the regcache either. Perhaps
> it's known farther up the call tree, but it isn't passed to the
> software single step routine.
Why do you need the regcache, rather than getting registers from the
frame?
--
Daniel Jacobowitz
CodeSourcery
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-14 17:56 ` Daniel Jacobowitz
@ 2009-10-14 18:01 ` Michael Eager
0 siblings, 0 replies; 20+ messages in thread
From: Michael Eager @ 2009-10-14 18:01 UTC (permalink / raw)
To: Michael Eager, Joel Brobecker, gdb-patches, Eli Zaretskii
Daniel Jacobowitz wrote:
> On Wed, Oct 14, 2009 at 10:48:57AM -0700, Michael Eager wrote:
>>>> + struct regcache *regcache = get_current_regcache ();
>>> This one raised a red flag, as we try to avoid depending on global
>>> variables. But I'm not sure what the kosher way of getting the regcache
>>> would be. I thought there would be method to get the regcache from
>>> a frame, but apparently not. Perhaps the right way is to use
>>> get_thread_arch_regcache (inferior_ptid, gdbarch), but I'm not sure.
>>> I'll ask Ulrich, who knows this area a lot better.
>> I wasn't able to find a better way to get the regcache either. Perhaps
>> it's known farther up the call tree, but it isn't passed to the
>> software single step routine.
>
> Why do you need the regcache, rather than getting registers from the
> frame?
Not sure, offhand. It's the way the routine was originally written.
But this is moot -- the code isn't needed and will be removed.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-16 12:53 ` Daniel Jacobowitz
@ 2009-10-16 16:14 ` Michael Eager
0 siblings, 0 replies; 20+ messages in thread
From: Michael Eager @ 2009-10-16 16:14 UTC (permalink / raw)
To: Tom Tromey, Joel Brobecker, Michael Eager, gdb-patches
Daniel Jacobowitz wrote:
> On Thu, Oct 15, 2009 at 12:03:45PM -0600, Tom Tromey wrote:
>> In fact I'd be interested to hear how gdb has dealt with this problem in
>> the past. I'm also sure this won't be the last time we're in this
>> situation.
>
> In general, I'm not reluctant to make folks use an older GDB with an
> older compiler. Users who get a prebuilt toolchain from Xilinx -
> probably the majority - won't be affected by the switch.
It may be possible to make this change as part of Xilinx's
next toolchain release. I'll discuss this with them. But I
would be reluctant to break backward compatibility unless
there was a more compelling reason.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 18:04 ` Tom Tromey
2009-10-15 18:31 ` Joel Brobecker
@ 2009-10-16 12:53 ` Daniel Jacobowitz
2009-10-16 16:14 ` Michael Eager
1 sibling, 1 reply; 20+ messages in thread
From: Daniel Jacobowitz @ 2009-10-16 12:53 UTC (permalink / raw)
To: Tom Tromey; +Cc: Joel Brobecker, Michael Eager, gdb-patches
On Thu, Oct 15, 2009 at 12:03:45PM -0600, Tom Tromey wrote:
> In fact I'd be interested to hear how gdb has dealt with this problem in
> the past. I'm also sure this won't be the last time we're in this
> situation.
In general, I'm not reluctant to make folks use an older GDB with an
older compiler. Users who get a prebuilt toolchain from Xilinx -
probably the majority - won't be affected by the switch.
--
Daniel Jacobowitz
CodeSourcery
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 16:46 ` Joel Brobecker
2009-10-15 17:14 ` Michael Eager
@ 2009-10-15 19:31 ` Michael Eager
1 sibling, 0 replies; 20+ messages in thread
From: Michael Eager @ 2009-10-15 19:31 UTC (permalink / raw)
To: Joel Brobecker; +Cc: gdb-patches
Joel Brobecker wrote:
>> 2009-10-15 Michael Eager <eager@eagercon.com>
>>
>> * config/djgpp/fnchange.lst: Add translations for cpu-microblaze.c,
>> elf32-microblaze.c, microblaze-rom.c, microblaze-linux-tdep.c,
>> microblaze-tdep.h, microblaze-tdep.c, microblaze-opc.h,
>> microblaze-opcm.h, microblaze-dis.c, microblaze-dis.h, sim/microblaze,
>> microblaze.h, and microblaze.isa.
>> * configure.tgt: Add targets microblaze*-linux-*, microblaze*-xilinx-*.
>> * doc/gdb.texinfo: Add MicroBlaze.
>> * Makefile.in: Build microblaze-tdep.o, microblaze-linux-tdep.o.
>> HFILES_NO_SRCDIR: Add microblaze-tdep.h.
>> * microblaze-linux-tdep.c: New.
>> * microblaze-tdep.c: New.
>> * microblaze-tdep.h: New.
>> * NEWS: Announce Xilinx MicroBlaze support.
>
> Approved. I believe that Eli already approved the doc and NEWS bit,
> and I didn't see any change to this part, so you should be all set.
Checked in.
Changelog corrected -- the doc changes were checked in when Eli
approved them.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 18:04 ` Tom Tromey
@ 2009-10-15 18:31 ` Joel Brobecker
2009-10-16 12:53 ` Daniel Jacobowitz
1 sibling, 0 replies; 20+ messages in thread
From: Joel Brobecker @ 2009-10-15 18:31 UTC (permalink / raw)
To: Tom Tromey; +Cc: Michael Eager, gdb-patches
> In fact I'd be interested to hear how gdb has dealt with this problem in
> the past. I'm also sure this won't be the last time we're in this
> situation.
As far as I know, we have been trying to be accomodating, but not at
the cost of doing the wrong thing. This should have been fixed in
the compiler since the beginning, but now users are stuck between
a rock and a hard place. I don't know much how CIEs are processed
in GDB, so I don't know offhand how we can provide any kind of info
back to the CIE consumer in mb-tdep.
--
Joel
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 17:18 ` Joel Brobecker
@ 2009-10-15 18:06 ` Michael Eager
0 siblings, 0 replies; 20+ messages in thread
From: Michael Eager @ 2009-10-15 18:06 UTC (permalink / raw)
To: Joel Brobecker; +Cc: tromey, gdb-patches
Joel Brobecker wrote:
>> I'm just as happy to leave this problem unfixed. Gcc generates an
>> incorrect CIE entry, which is philosophically not a good thing.
>> But fixing this causes far more problems than it solves.
>
> One of the things that we perhaps didn't consider is debugging
> code compiled with a compiler other than GCC... Not sure if one
> even exists, though, but that compiler might be generating correct
> debugging info, which would trigger the "bug" in GDB.
There's no other compiler for MicroBlaze. (I was going to mention
this hypothetical situation, but this has very low probability
of ever being a real concern.)
> Anyway, if "GCC" is unwilling to fix the problem, then GDB has to do
> its best. It's disappointing when that happens, but it's not the first
> time we introduce work-arounds. This one is a little harder to accept
> because it causes GDB to malfunction in case of correct debugging info.
> But it's still better this way, in practice.
It's not so much that GCC is unwilling to fix the problem. If
I wanted to, I could extend the CIE generation code and modify
the MicroBlaze-specific code in both GCC and GDB. This would
break debugging existing programs and libraries with the new gdb
as well as break debugging newly compiled code with an old gdb.
(There are ways that this might be synchronized with Xilinx's
release of tool chains to their customers, but I'd be very
reluctant to recommend that a new tool chain be incompatible
with older tool chains unless there was a very compelling reason.)
If anyone can suggest a reasonable way to identify whether the
CIE entry for return address is lr or lr+offset, I'll pursue making
gdb accept both. This is certainly clear while parsing the CIE,
but it is particularly ugly to put a target-dependent test in
this code.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 18:00 ` Tom Tromey
@ 2009-10-15 18:04 ` Tom Tromey
2009-10-15 18:31 ` Joel Brobecker
2009-10-16 12:53 ` Daniel Jacobowitz
0 siblings, 2 replies; 20+ messages in thread
From: Tom Tromey @ 2009-10-15 18:04 UTC (permalink / raw)
To: Joel Brobecker; +Cc: Michael Eager, gdb-patches
Tom> How would this be done?
Tom> Users might be using an old gcc or a new one.
Joel> The only solution I see to this problem would be that users of a recent
Joel> GCC would have to use a recent GDB.
Tom> The problem is that this also implies that users stuck on an old compiler
Tom> will be stuck with an old gdb.
Just to be clear, I don't want to be a spoiler about this.
It is fine by me to move forward like this.
In fact I'd be interested to hear how gdb has dealt with this problem in
the past. I'm also sure this won't be the last time we're in this
situation.
Tom
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 16:47 ` Joel Brobecker
@ 2009-10-15 18:00 ` Tom Tromey
2009-10-15 18:04 ` Tom Tromey
0 siblings, 1 reply; 20+ messages in thread
From: Tom Tromey @ 2009-10-15 18:00 UTC (permalink / raw)
To: Joel Brobecker; +Cc: Michael Eager, gdb-patches
>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:
Tom> How would this be done?
Tom> Users might be using an old gcc or a new one.
Joel> The only solution I see to this problem would be that users of a recent
Joel> GCC would have to use a recent GDB.
The problem is that this also implies that users stuck on an old compiler
will be stuck with an old gdb.
Joel> Worse case scenario, a microblase-specific setting.
Yeah. Or parsing the dwarf producer string.
Tom
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 17:12 ` Michael Eager
@ 2009-10-15 17:18 ` Joel Brobecker
2009-10-15 18:06 ` Michael Eager
0 siblings, 1 reply; 20+ messages in thread
From: Joel Brobecker @ 2009-10-15 17:18 UTC (permalink / raw)
To: Michael Eager; +Cc: tromey, gdb-patches
> I'm just as happy to leave this problem unfixed. Gcc generates an
> incorrect CIE entry, which is philosophically not a good thing.
> But fixing this causes far more problems than it solves.
One of the things that we perhaps didn't consider is debugging
code compiled with a compiler other than GCC... Not sure if one
even exists, though, but that compiler might be generating correct
debugging info, which would trigger the "bug" in GDB.
Anyway, if "GCC" is unwilling to fix the problem, then GDB has to do
its best. It's disappointing when that happens, but it's not the first
time we introduce work-arounds. This one is a little harder to accept
because it causes GDB to malfunction in case of correct debugging info.
But it's still better this way, in practice.
--
Joel
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 16:46 ` Joel Brobecker
@ 2009-10-15 17:14 ` Michael Eager
2009-10-15 19:31 ` Michael Eager
1 sibling, 0 replies; 20+ messages in thread
From: Michael Eager @ 2009-10-15 17:14 UTC (permalink / raw)
To: Joel Brobecker; +Cc: gdb-patches
Joel Brobecker wrote:
>> 2009-10-15 Michael Eager <eager@eagercon.com>
>>
>> * config/djgpp/fnchange.lst: Add translations for cpu-microblaze.c,
>> elf32-microblaze.c, microblaze-rom.c, microblaze-linux-tdep.c,
>> microblaze-tdep.h, microblaze-tdep.c, microblaze-opc.h,
>> microblaze-opcm.h, microblaze-dis.c, microblaze-dis.h, sim/microblaze,
>> microblaze.h, and microblaze.isa.
>> * configure.tgt: Add targets microblaze*-linux-*, microblaze*-xilinx-*.
>> * doc/gdb.texinfo: Add MicroBlaze.
>> * Makefile.in: Build microblaze-tdep.o, microblaze-linux-tdep.o.
>> HFILES_NO_SRCDIR: Add microblaze-tdep.h.
>> * microblaze-linux-tdep.c: New.
>> * microblaze-tdep.c: New.
>> * microblaze-tdep.h: New.
>> * NEWS: Announce Xilinx MicroBlaze support.
>
> Approved. I believe that Eli already approved the doc and NEWS bit,
> and I didn't see any change to this part, so you should be all set.
Thanks.
>
> Just one tiny little thing:
>
>> + microblaze --target=microblaze-xilinx-elf ,-Werror
>> + --target=microblaze-linux-gnu ,-Werror
>> + Michael Eager eager@eagercon.com
>
> As discussed previously, please do not add yourself from the official
> maintainer for that target just yet.
I had pulled this from the patch but it seems to have crept back in.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 16:44 ` Tom Tromey
2009-10-15 16:47 ` Joel Brobecker
@ 2009-10-15 17:12 ` Michael Eager
2009-10-15 17:18 ` Joel Brobecker
1 sibling, 1 reply; 20+ messages in thread
From: Michael Eager @ 2009-10-15 17:12 UTC (permalink / raw)
To: tromey; +Cc: gdb-patches, Joel Brobecker
Tom Tromey wrote:
>>>>>> "Michael" == Michael Eager <eager@eagercon.com> writes:
>
> Michael> When (if) this is fixed in gcc there will need to coordinate
> Michael> this with a corresponding fix in gdb.
>
> How would this be done?
> Users might be using an old gcc or a new one.
Not easy. It would require coordination with users. Lots.
Perhaps a test to see if code was generated by a particular version
of gcc or something equally ugly.
Even if gcc was fixed to accept reg+8 for the return address,
there would need to be a change to the MicroBlaze-specific
INCOMING_RETURN_ADDR_RTX for it to take any effect. (There's
a warning note about this in gcc, which I might make more emphatic.)
So this is not going to happen by accident.
I'm just as happy to leave this problem unfixed. Gcc generates an
incorrect CIE entry, which is philosophically not a good thing.
But fixing this causes far more problems than it solves.
> Michael> diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/NEWS mb-gdb/gdb/NEWS
> Michael> --- gdb/gdb/NEWS 2009-10-14 11:15:51.000000000 -0700
> Michael> +++ mb-gdb/gdb/NEWS 2009-10-14 19:15:57.000000000 -0700
> Michael> @@ -484,6 +484,11 @@ Lattice Mico32 lm32-*
> Michael> x86 DICOS i[34567]86-*-dicos*
> Michael> x86_64 DICOS x86_64-*-dicos*
> Michael> S+core 3 score-*-*
> Michael> +Xilinx MicroBlaze microblaze-*-*
> Michael> +
> Michael> +* New Simulators
> Michael> +
> Michael> +Xilinx MicroBlaze microblaze
>
> I think this changes the "7.0" section, not the "since 7.0" section.
Oops. The "Changes in GDB 7.0" was added after I updated the file
and I didn't notice it. I'll move the entry.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 16:44 ` Tom Tromey
@ 2009-10-15 16:47 ` Joel Brobecker
2009-10-15 18:00 ` Tom Tromey
2009-10-15 17:12 ` Michael Eager
1 sibling, 1 reply; 20+ messages in thread
From: Joel Brobecker @ 2009-10-15 16:47 UTC (permalink / raw)
To: Tom Tromey; +Cc: Michael Eager, gdb-patches
> Michael> When (if) this is fixed in gcc there will need to coordinate
> Michael> this with a corresponding fix in gdb.
>
> How would this be done?
> Users might be using an old gcc or a new one.
The only solution I see to this problem would be that users of a recent
GCC would have to use a recent GDB. Very unfortunate, but I don't see
any other solution. Worse case scenario, a microblase-specific setting.
--
Joel
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 16:32 Michael Eager
2009-10-15 16:44 ` Tom Tromey
@ 2009-10-15 16:46 ` Joel Brobecker
2009-10-15 17:14 ` Michael Eager
2009-10-15 19:31 ` Michael Eager
1 sibling, 2 replies; 20+ messages in thread
From: Joel Brobecker @ 2009-10-15 16:46 UTC (permalink / raw)
To: Michael Eager; +Cc: gdb-patches
> 2009-10-15 Michael Eager <eager@eagercon.com>
>
> * config/djgpp/fnchange.lst: Add translations for cpu-microblaze.c,
> elf32-microblaze.c, microblaze-rom.c, microblaze-linux-tdep.c,
> microblaze-tdep.h, microblaze-tdep.c, microblaze-opc.h,
> microblaze-opcm.h, microblaze-dis.c, microblaze-dis.h, sim/microblaze,
> microblaze.h, and microblaze.isa.
> * configure.tgt: Add targets microblaze*-linux-*, microblaze*-xilinx-*.
> * doc/gdb.texinfo: Add MicroBlaze.
> * Makefile.in: Build microblaze-tdep.o, microblaze-linux-tdep.o.
> HFILES_NO_SRCDIR: Add microblaze-tdep.h.
> * microblaze-linux-tdep.c: New.
> * microblaze-tdep.c: New.
> * microblaze-tdep.h: New.
> * NEWS: Announce Xilinx MicroBlaze support.
Approved. I believe that Eli already approved the doc and NEWS bit,
and I didn't see any change to this part, so you should be all set.
Just one tiny little thing:
> + microblaze --target=microblaze-xilinx-elf ,-Werror
> + --target=microblaze-linux-gnu ,-Werror
> + Michael Eager eager@eagercon.com
As discussed previously, please do not add yourself from the official
maintainer for that target just yet.
Thanks,
--
Joel
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH] Support for Xilinx MicroBlaze
2009-10-15 16:32 Michael Eager
@ 2009-10-15 16:44 ` Tom Tromey
2009-10-15 16:47 ` Joel Brobecker
2009-10-15 17:12 ` Michael Eager
2009-10-15 16:46 ` Joel Brobecker
1 sibling, 2 replies; 20+ messages in thread
From: Tom Tromey @ 2009-10-15 16:44 UTC (permalink / raw)
To: Michael Eager; +Cc: gdb-patches, Joel Brobecker
>>>>> "Michael" == Michael Eager <eager@eagercon.com> writes:
Michael> When (if) this is fixed in gcc there will need to coordinate
Michael> this with a corresponding fix in gdb.
How would this be done?
Users might be using an old gcc or a new one.
Michael> diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/NEWS mb-gdb/gdb/NEWS
Michael> --- gdb/gdb/NEWS 2009-10-14 11:15:51.000000000 -0700
Michael> +++ mb-gdb/gdb/NEWS 2009-10-14 19:15:57.000000000 -0700
Michael> @@ -484,6 +484,11 @@ Lattice Mico32 lm32-*
Michael> x86 DICOS i[34567]86-*-dicos*
Michael> x86_64 DICOS x86_64-*-dicos*
Michael> S+core 3 score-*-*
Michael> +Xilinx MicroBlaze microblaze-*-*
Michael> +
Michael> +* New Simulators
Michael> +
Michael> +Xilinx MicroBlaze microblaze
I think this changes the "7.0" section, not the "since 7.0" section.
Tom
^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH] Support for Xilinx MicroBlaze
@ 2009-10-15 16:32 Michael Eager
2009-10-15 16:44 ` Tom Tromey
2009-10-15 16:46 ` Joel Brobecker
0 siblings, 2 replies; 20+ messages in thread
From: Michael Eager @ 2009-10-15 16:32 UTC (permalink / raw)
To: gdb-patches; +Cc: Joel Brobecker
[-- Attachment #1: Type: text/plain, Size: 1216 bytes --]
I've addressed most of the issues raised about the previous patch.
I was not able to figure out a way to identify whether the return address
in the CIE was generated correctly (pc+8) or not, so the work-around in the
microblaze_unwind_pc is unchanged. If anyone has an idea, I'll be
happy to look into this further. When (if) this is fixed in gcc there
will need to coordinate this with a corresponding fix in gdb.
2009-10-15 Michael Eager <eager@eagercon.com>
* config/djgpp/fnchange.lst: Add translations for cpu-microblaze.c,
elf32-microblaze.c, microblaze-rom.c, microblaze-linux-tdep.c,
microblaze-tdep.h, microblaze-tdep.c, microblaze-opc.h,
microblaze-opcm.h, microblaze-dis.c, microblaze-dis.h, sim/microblaze,
microblaze.h, and microblaze.isa.
* configure.tgt: Add targets microblaze*-linux-*, microblaze*-xilinx-*.
* doc/gdb.texinfo: Add MicroBlaze.
* Makefile.in: Build microblaze-tdep.o, microblaze-linux-tdep.o.
HFILES_NO_SRCDIR: Add microblaze-tdep.h.
* microblaze-linux-tdep.c: New.
* microblaze-tdep.c: New.
* microblaze-tdep.h: New.
* NEWS: Announce Xilinx MicroBlaze support.
--
Michael Eager eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306 650-325-8077
[-- Attachment #2: mb-gdb.patch --]
[-- Type: text/plain, Size: 45580 bytes --]
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/config/djgpp/fnchange.lst mb-gdb/gdb/config/djgpp/fnchange.lst
--- gdb/gdb/config/djgpp/fnchange.lst 2009-09-19 02:46:44.000000000 -0700
+++ mb-gdb/gdb/config/djgpp/fnchange.lst 2009-10-05 17:44:30.000000000 -0700
@@ -18,6 +18,7 @@
@V@/bfd/cpu-cr16.c @V@/bfd/cpucr16.c
@V@/bfd/cpu-cr16c.c @V@/bfd/cpucr16c.c
@V@/bfd/cpu-ia64-opc.c @V@/bfd/cpuia64-opc.c
+@V@/bfd/cpu-microblaze.c @V@/bfd/cpumb.c
@V@/bfd/cpu-m68hc11.c @V@/bfd/cm68hc11.c
@V@/bfd/cpu-m68hc12.c @V@/bfd/cm68hc12.c
@V@/bfd/efi-app-ia32.c @V@/bfd/efiia32app.c
@@ -34,6 +35,7 @@
@V@/bfd/elf32-frv.c @V@/bfd/elf32fv.c
@V@/bfd/elf32-i370.c @V@/bfd/e32i370.c
@V@/bfd/elf32-i386.c @V@/bfd/e32i86.c
+@V@/bfd/elf32-microblaze.c @V@e32mb.c
@V@/bfd/elf32-m32c.c @V@/bfd/em32c.c
@V@/bfd/elf32-m32r.c @V@/bfd/em32r.c
@V@/bfd/elf32-m68hc11.c @V@/bfd/em68hc11.c
@@ -279,6 +281,10 @@
@V@/gdb/m68kbsd-tdep.c @V@/gdb/m68bsd-tdep.c
@V@/gdb/m68knbsd-nat.c @V@/gdb/m6nbsd-nat.c
@V@/gdb/m68knbsd-tdep.c @V@/gdb/m6nbsd-tdep.c
+@V@/gdb/microblaze-rom.c @V@/gdb/mb-rom.c
+@V@/gdb/microblaze-linux-tdep.c @V@/gdb/mbl-tdep.c
+@V@/gdb/microblaze-tdep.h @V@/gdb/mb-tdep.h
+@V@/gdb/microblaze-tdep.c @V@/gdb/mb-tdep.c
@V@/gdb/mips-linux-nat.c @V@/gdb/mipslnxnat.c
@V@/gdb/mips-linux-tdep.c @V@/gdb/mipslnxtdep.c
@V@/gdb/mipsnbsd-nat.c @V@/gdb/mipsnbnat.c
@@ -511,6 +517,10 @@
@V@/opcodes/ia64-opc.c @V@/opcodes/ia64-opc.c
@V@/opcodes/iq2000-desc.c @V@/opcodes/iq2000desc.c
@V@/opcodes/iq2000-dis.c @V@/opcodes/iq2000dis.c
+@V@/opcodes/microblaze-opc.h @V@/opcodes/mb-opc.h
+@V@/opcodes/microblaze-dis.c @V@/opcodes/mb-dis.c
+@V@/opcodes/microblaze-dis.h @V@/opcodes/mb-dis.h
+@V@/opcodes/microblaze-opcm.h @V@/opcodes/mb-opcm.h
@V@/opcodes/m68hc11-dis.c @V@/opcodes/m68hc11dis.c
@V@/opcodes/m68hc11-opc.c @V@/opcodes/m68hc11opc.c
@V@/opcodes/microblaze-opc.h @V@/opcodes/mbl-opc.h
@@ -540,6 +550,9 @@
@V@/sim/frv/profile-fr500.h @V@/sim/frv/fr500-profile.h
@V@/sim/frv/profile-fr550.c @V@/sim/frv/fr550-profile.c
@V@/sim/frv/profile-fr550.h @V@/sim/frv/fr550-profile.h
+@V@/sim/microblaze @V@/sim/mb
+@V@/sim/microblaze/microblaze.h @V@/sim/mb/mb.h
+@V@/sim/microblaze/microblaze.isa @V@/sim/mb/mb.isa
@V@/sim/m68hc11/dv-m68hc11eepr.c @V@/sim/m68hc11/dv-eepr.c
@V@/sim/m68hc11/dv-m68hc11sio.c @V@/sim/m68hc11/dv-sio.c
@V@/sim/m68hc11/dv-m68hc11spi.c @V@/sim/m68hc11/dv-spi.c
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/configure.tgt mb-gdb/gdb/configure.tgt
--- gdb/gdb/configure.tgt 2009-09-27 13:46:35.000000000 -0700
+++ mb-gdb/gdb/configure.tgt 2009-10-05 10:52:47.000000000 -0700
@@ -313,6 +313,19 @@ mep-*-*)
# No sim needed. Target uses SID.
;;
+microblaze*-linux-*)
+ # Target: Xilinx MicroBlaze running Linux
+ gdb_target_obs="microblaze-tdep.o microblaze-linux-tdep.o microblaze-rom.o \
+ monitor.o dsrec.o solib.o solib-svr4.o corelow.o \
+ symfile-mem.o"
+ gdb_sim=../sim/microblaze/libsim.a
+ ;;
+microblaze*-xilinx-*)
+ # Target: Xilinx MicroBlaze running standalone
+ gdb_target_obs="microblaze-tdep.o microblaze-rom.o monitor.o dsrec.o"
+ gdb_sim=../sim/microblaze/libsim.a
+ ;;
+
mips*-sgi-irix5*)
# Target: MIPS SGI running Irix 5
gdb_target_obs="mips-tdep.o mips-irix-tdep.o solib.o solib-irix.o"
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/MAINTAINERS mb-gdb/gdb/MAINTAINERS
--- gdb/gdb/MAINTAINERS 2009-10-14 11:15:51.000000000 -0700
+++ mb-gdb/gdb/MAINTAINERS 2009-10-14 19:15:56.000000000 -0700
@@ -291,6 +291,10 @@ the native maintainer when resolving ABI
mep --target=mep-elf ,-Werror
Kevin Buettner kevinb@redhat.com
+ microblaze --target=microblaze-xilinx-elf ,-Werror
+ --target=microblaze-linux-gnu ,-Werror
+ Michael Eager eager@eagercon.com
+
mips --target=mips-elf ,-Werror
mn10300 --target=mn10300-elf broken
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/Makefile.in mb-gdb/gdb/Makefile.in
--- gdb/gdb/Makefile.in 2009-09-16 14:55:57.000000000 -0700
+++ mb-gdb/gdb/Makefile.in 2009-10-05 11:59:21.000000000 -0700
@@ -501,6 +501,7 @@ ALL_TARGET_OBS = \
m68kbsd-tdep.o m68klinux-tdep.o m68k-tdep.o \
m88k-tdep.o \
mep-tdep.o \
+ microblaze-tdep.o microblaze-linux-tdep.o \
mips-irix-tdep.o mips-linux-tdep.o \
mipsnbsd-tdep.o mips-tdep.o \
mn10300-linux-tdep.o mn10300-tdep.o \
@@ -750,7 +751,7 @@ config/rs6000/nm-rs6000.h top.h bsd-kvm.
annotate.h sim-regno.h dictionary.h dfp.h main.h frame-unwind.h \
remote-fileio.h i386-linux-tdep.h vax-tdep.h objc-lang.h \
sentinel-frame.h bcache.h symfile.h windows-tdep.h linux-tdep.h \
-gdb_usleep.h jit.h xml-syscall.h ada-operator.inc
+gdb_usleep.h jit.h xml-syscall.h ada-operator.inc microblaze-tdep.h
# Header files that already have srcdir in them, or which are in objdir.
@@ -1434,6 +1435,7 @@ ALLDEPFILES = \
m68kbsd-nat.c m68kbsd-tdep.c \
m68klinux-nat.c m68klinux-tdep.c \
m88k-tdep.c m88kbsd-nat.c \
+ microblaze-tdep.c microblaze-linux-tdep.c \
mingw-hdep.c \
mips-linux-nat.c mips-linux-tdep.c \
mips-irix-tdep.c \
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-linux-tdep.c mb-gdb/gdb/microblaze-linux-tdep.c
--- gdb/gdb/microblaze-linux-tdep.c 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-linux-tdep.c 2009-09-30 11:51:38.000000000 -0700
@@ -0,0 +1,142 @@
+/* Target-dependent code for Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#include "defs.h"
+#include "frame.h"
+#include "inferior.h"
+#include "symtab.h"
+#include "target.h"
+#include "gdbcore.h"
+#include "gdbcmd.h"
+#include "symfile.h"
+#include "objfiles.h"
+#include "regcache.h"
+#include "value.h"
+#include "osabi.h"
+#include "regset.h"
+#include "solib-svr4.h"
+#include "microblaze-tdep.h"
+#include "trad-frame.h"
+#include "frame-unwind.h"
+#include "tramp-frame.h"
+
+
+static int
+microblaze_linux_memory_remove_breakpoint (struct bp_target_info *bp_tgt)
+{
+ CORE_ADDR addr = bp_tgt->placed_address;
+ const gdb_byte *bp;
+ int val;
+ int bplen;
+ gdb_byte old_contents[BREAKPOINT_MAX];
+
+ /* Determine appropriate breakpoint contents and size for this address. */
+ bp = gdbarch_breakpoint_from_pc (gdbarch, &addr, &bplen);
+ if (bp == NULL)
+ error (_("Software breakpoints not implemented for this target."));
+
+ val = target_read_memory (addr, old_contents, bplen);
+
+ /* If our breakpoint is no longer at the address, this means that the
+ program modified the code on us, so it is wrong to put back the
+ old value. */
+ if (val == 0 && memcmp (bp, old_contents, bplen) == 0)
+ val = target_write_memory (addr, bp_tgt->shadow_contents, bplen);
+
+ return val;
+}
+
+static void
+microblaze_linux_sigtramp_cache (struct frame_info *next_frame,
+ struct trad_frame_cache *this_cache,
+ CORE_ADDR func, LONGEST offset,
+ int bias)
+{
+ CORE_ADDR base;
+ CORE_ADDR gpregs;
+ int regnum;
+ struct gdbarch *gdbarch = get_frame_arch (next_frame);
+ struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+
+ base = frame_unwind_register_unsigned (next_frame, SP_REGNUM);
+ if (bias > 0 && frame_pc_unwind (next_frame) != func)
+ /* See below, some signal trampolines increment the stack as their
+ first instruction, need to compensate for that. */
+ base -= bias;
+
+ /* Find the address of the register buffer. */
+ gpregs = base + offset;
+
+ /* Registers saved on stack. */
+ for (regnum = 0; regnum < MICROBLAZE_BTR_REGNUM; regnum++)
+ trad_frame_set_reg_addr (this_cache, regnum,
+ gpregs + regnum * MICROBLAZE_REGISTER_SIZE);
+ trad_frame_set_id (this_cache, frame_id_build (base, func));
+}
+
+
+static void
+microblaze_linux_sighandler_cache_init (const struct tramp_frame *self,
+ struct frame_info *next_frame,
+ struct trad_frame_cache *this_cache,
+ CORE_ADDR func)
+{
+ microblaze_linux_sigtramp_cache (next_frame, this_cache, func,
+ 0 /* Offset to ucontext_t. */
+ + 24 /* Offset to .reg. */,
+ 0);
+}
+
+static struct tramp_frame microblaze_linux_sighandler_tramp_frame =
+{
+ SIGTRAMP_FRAME,
+ 4,
+ {
+ { 0x31800077, -1 }, /* addik R12,R0,119. */
+ { 0xb9cc0008, -1 }, /* brki R14,8. */
+ { TRAMP_SENTINEL_INSN },
+ },
+ microblaze_linux_sighandler_cache_init
+};
+
+
+static void
+microblaze_linux_init_abi (struct gdbarch_info info,
+ struct gdbarch *gdbarch)
+{
+ struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
+
+ set_gdbarch_memory_remove_breakpoint (gdbarch,
+ microblaze_linux_memory_remove_breakpoint);
+
+ /* Shared library handling. */
+ set_solib_svr4_fetch_link_map_offsets (gdbarch,
+ svr4_ilp32_fetch_link_map_offsets);
+
+ /* Trampolines. */
+ tramp_frame_prepend_unwinder (gdbarch,
+ µblaze_linux_sighandler_tramp_frame);
+}
+
+void
+_initialize_microblaze_linux_tdep (void)
+{
+ gdbarch_register_osabi (bfd_arch_microblaze, 0, GDB_OSABI_LINUX,
+ microblaze_linux_init_abi);
+}
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-rom.c mb-gdb/gdb/microblaze-rom.c
--- gdb/gdb/microblaze-rom.c 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-rom.c 2009-10-14 14:41:11.000000000 -0700
@@ -0,0 +1,191 @@
+/* Remote debugging interface to Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#include "defs.h"
+#include "gdbcore.h"
+#include "target.h"
+#include "monitor.h"
+#include "gdb_string.h"
+#include "serial.h"
+#include "regcache.h"
+
+static char *picobug_inits[] =
+{"\r", NULL};
+
+static struct target_ops picobug_ops;
+static struct monitor_ops picobug_cmds;
+
+/* Picobug only supports a subset of registers from MCore. In reality,
+ it doesn't support ss1, either. */
+static char *picobug_regnames[] = {
+ "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
+ "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ "psr", "vbr", "epsr", "fpsr", "epc", "fpc", 0, "ss1",
+ "ss2", "ss3", "ss4", 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0, 0, 0, 0, 0, 0,
+ "pc" };
+
+
+
+static void
+picobug_open (char *args, int from_tty)
+{
+ monitor_open (args, &picobug_cmds, from_tty);
+}
+/* We choose to write our own dumpregs routine, since the output of
+ the register dumping is rather difficult to encapsulate in a
+ regexp:
+
+picobug> rd
+ pc 2f00031e epc 2f00031e fpc 00000000
+ psr 80000101 epsr 80000101 fpsr 00000000
+ss0-ss4 bad0beef 00000000 00000000 00000000 00000000 vbr 30005c00
+ r0-r7 2f0fff4c 00000090 00000001 00000002 00000003 00000004 00000005 00000006
+ r8-r15 2f0fff64 00000000 00000000 00000000 00000000 00000000 00000000 2f00031e
+*/
+
+static int
+picobug_dumpregs (struct regcache *regcache)
+{
+ struct gdbarch *gdbarch = get_regcache_arch (regcache);
+ char buf[1024];
+ int resp_len;
+ char *p;
+
+ /* Send the dump register command to the monitor and
+ get the reply. */
+ monitor_printf (picobug_cmds.dump_registers);
+ resp_len = monitor_expect_prompt (buf, sizeof (buf));
+
+ p = strtok (buf, " \t\r\n");
+ while (p)
+ {
+ if (strchr (p, '-'))
+ {
+ /* got a range. either r0-r7, r8-r15 or ss0-ss4. */
+ if (strncmp (p, "r0", 2) == 0 || strncmp (p, "r8", 2) == 0)
+ {
+ int rn = (p[1] == '0' ? 0 : 8);
+ int i = 0;
+
+ /* Get the next 8 values and record them. */
+ while (i < 8)
+ {
+ p = strtok (NULL, " \t\r\n");
+ if (p)
+ monitor_supply_register (regcache, rn + i, p);
+ i++;
+ }
+ }
+ else if (strncmp (p, "ss", 2) == 0)
+ {
+ /* get the next five values, ignoring the first. */
+ int rn;
+ p = strtok (NULL, " \t\r\n");
+ for (rn = 39; rn < 43; rn++)
+ {
+ p = strtok (NULL, " \t\r\n");
+ if (p)
+ monitor_supply_register (regcache, rn, p);
+ }
+ }
+ else
+ {
+ break;
+ }
+ }
+ else
+ {
+ /* Simple register type, paired. */
+ char *name = p;
+ int i;
+
+ /* Get and record value. */
+ p = strtok (NULL, " \t\r\n");
+ if (p)
+ {
+ for (i = 0; i < gdbarch_num_regs (gdbarch); i++)
+ {
+ if (picobug_regnames[i]
+ && strcmp (picobug_regnames[i], name) == 0)
+ break;
+ }
+
+ if (i <= gdbarch_num_regs (gdbarch))
+ monitor_supply_register (regcache, i, p);
+ }
+ }
+ p = strtok (NULL, " \t\r\n");
+ }
+
+ return 0;
+}
+
+static void
+init_picobug_cmds (void)
+{
+ picobug_cmds.flags = MO_GETMEM_NEEDS_RANGE | MO_CLR_BREAK_USES_ADDR
+ | MO_PRINT_PROGRAM_OUTPUT;
+
+ picobug_cmds.init = picobug_inits; /* Init strings */
+ picobug_cmds.cont = "g\n"; /* continue command */
+ picobug_cmds.step = "s\n"; /* single step */
+ picobug_cmds.set_break = "br %x\n"; /* set a breakpoint */
+ picobug_cmds.clr_break = "nobr %x\n"; /* clear a breakpoint */
+ picobug_cmds.clr_all_break = "nobr\n"; /* clear all breakpoints */
+ picobug_cmds.setmem.cmdb = "mm %x %x ;b\n"; /* setmem.cmdb (addr, value) */
+ picobug_cmds.setmem.cmdw = "mm %x %x ;h\n"; /* setmem.cmdw (addr, value) */
+ picobug_cmds.setmem.cmdl = "mm %x %x ;w\n"; /* setmem.cmdl (addr, value) */
+ picobug_cmds.getmem.cmdb = "md %x %x\n"; /* getmem.cmdb (start addr,
+ end addr) */
+ picobug_cmds.getmem.resp_delim = ":"; /* getmem.resp_delim */
+ picobug_cmds.setreg.cmd = "rm %s %x\n"; /* setreg.cmd (name, value) */
+ picobug_cmds.getreg.cmd = "rd %s\n"; /* getreg.cmd (name) */
+ picobug_cmds.getreg.resp_delim = ":"; /* getreg.resp_delim */
+ picobug_cmds.dump_registers = "rd\n"; /* dump_registers */
+ picobug_cmds.dumpregs = picobug_dumpregs; /* dump registers parser */
+ picobug_cmds.load = "lo\n"; /* download command */
+ picobug_cmds.prompt = "picobug> "; /* monitor command prompt */
+ picobug_cmds.line_term = "\n"; /* end-of-line terminator */
+ picobug_cmds.target = &picobug_ops; /* target operations */
+ picobug_cmds.stopbits = SERIAL_1_STOPBITS; /* number of stop bits */
+ picobug_cmds.regnames = picobug_regnames; /* registers names */
+ picobug_cmds.num_breakpoints = 20; /* number of breakpoints */
+ picobug_cmds.magic = MONITOR_OPS_MAGIC; /* magic */
+}
+
+void
+_initialize_picobug_rom ()
+{
+ int i;
+
+ /* Initialize m32r RevC monitor target. */
+ init_picobug_cmds ();
+ init_monitor_ops (&picobug_ops);
+ picobug_ops.to_shortname = "picobug";
+ picobug_ops.to_longname = "picobug monitor";
+ picobug_ops.to_doc = "Debug via the picobug monitor.\n\
+Specify the serial device it is connected to (e.g. /dev/ttya).";
+ picobug_ops.to_open = picobug_open;
+
+ add_target (&picobug_ops);
+}
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-tdep.c mb-gdb/gdb/microblaze-tdep.c
--- gdb/gdb/microblaze-tdep.c 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-tdep.c 2009-10-15 09:14:37.000000000 -0700
@@ -0,0 +1,752 @@
+/* Target-dependent code for Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#include "defs.h"
+#include "arch-utils.h"
+#include "dis-asm.h"
+#include "frame.h"
+#include "trad-frame.h"
+#include "symtab.h"
+#include "value.h"
+#include "gdbcmd.h"
+#include "breakpoint.h"
+#include "inferior.h"
+#include "regcache.h"
+#include "target.h"
+#include "frame.h"
+#include "frame-base.h"
+#include "frame-unwind.h"
+#include "dwarf2-frame.h"
+#include "osabi.h"
+
+#include "gdb_assert.h"
+#include "gdb_string.h"
+#include "target-descriptions.h"
+#include "opcodes/microblaze-opcm.h"
+#include "opcodes/microblaze-dis.h"
+#include "microblaze-tdep.h"
+\f
+/* Instruction macros used for analyzing the prologue. */
+/* This set of instruction macros need to be changed whenever the
+ prologue generated by the compiler could have more instructions or
+ different type of instructions.
+ This set also needs to be verified if it is complete. */
+#define IS_RETURN(op) (op == rtsd || op == rtid)
+#define IS_UPDATE_SP(op, rd, ra) \
+ ((op == addik || op == addi) && rd == REG_SP && ra == REG_SP)
+#define IS_SPILL_SP(op, rd, ra) \
+ ((op == swi || op == sw) && rd == REG_SP && ra == REG_SP)
+#define IS_SPILL_REG(op, rd, ra) \
+ ((op == swi || op == sw) && rd != REG_SP && ra == REG_SP)
+#define IS_ALSO_SPILL_REG(op, rd, ra, rb) \
+ ((op == swi || op == sw) && rd != REG_SP && ra == 0 && rb == REG_SP)
+#define IS_SETUP_FP(op, ra, rb) \
+ ((op == add || op == addik || op == addk) && ra == REG_SP && rb == 0)
+#define IS_SPILL_REG_FP(op, rd, ra, fpregnum) \
+ ((op == swi || op == sw) && rd != REG_SP && ra == fpregnum && ra != 0)
+#define IS_SAVE_HIDDEN_PTR(op, rd, ra, rb) \
+ ((op == add || op == addik) && ra == MICROBLAZE_FIRST_ARGREG && rb == 0)
+
+/* All registers are 32 bits. */
+#define MICROBLAZE_REGISTER_SIZE 4
+
+/* The registers of the Xilinx microblaze processor. */
+
+static const char *microblaze_register_names[] =
+{
+ "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
+ "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
+ "r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
+ "r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31",
+ "rpc", "rmsr", "rear", "resr", "rfsr", "rbtr",
+ "rpvr0", "rpvr1", "rpvr2", "rpvr3", "rpvr4", "rpvr5", "rpvr6",
+ "rpvr7", "rpvr8", "rpvr9", "rpvr10", "rpvr11",
+ "redr", "rpid", "rzpr", "rtlbx", "rtlbsx", "rtlblo", "rtlbhi"
+};
+
+#define MICROBLAZE_NUM_REGS ARRAY_SIZE (microblaze_register_names)
+\f
+static int microblaze_debug_flag = 0;
+
+void
+microblaze_debug (const char *fmt, ...)
+{
+ if (microblaze_debug_flag)
+ {
+ va_list args;
+
+ va_start (args, fmt);
+ printf_unfiltered ("MICROBLAZE: ");
+ vprintf_unfiltered (fmt, args);
+ va_end (args);
+ }
+}
+\f
+/* Return the name of register REGNUM. */
+
+static const char *
+microblaze_register_name (struct gdbarch *gdbarch, int regnum)
+{
+ if (regnum >= 0 && regnum < MICROBLAZE_NUM_REGS)
+ return microblaze_register_names[regnum];
+ return NULL;
+}
+
+static struct type *
+microblaze_register_type (struct gdbarch *gdbarch, int regnum)
+{
+ if (regnum == MICROBLAZE_SP_REGNUM)
+ return builtin_type (gdbarch)->builtin_data_ptr;
+
+ if (regnum == MICROBLAZE_PC_REGNUM)
+ return builtin_type (gdbarch)->builtin_func_ptr;
+
+ return builtin_type (gdbarch)->builtin_int;
+}
+
+\f
+/* Fetch the instruction at PC. */
+
+unsigned long
+microblaze_fetch_instruction (CORE_ADDR pc)
+{
+ enum bfd_endian byte_order = gdbarch_byte_order (target_gdbarch);
+ gdb_byte buf[4];
+
+ /* If we can't read the instruction at PC, return zero. */
+ if (target_read_memory (pc, buf, sizeof (buf)))
+ return 0;
+
+ return extract_unsigned_integer (buf, 4, byte_order);
+}
+\f
+
+static CORE_ADDR
+microblaze_push_dummy_code (struct gdbarch *gdbarch, CORE_ADDR sp,
+ CORE_ADDR funcaddr,
+ struct value **args, int nargs,
+ struct type *value_type,
+ CORE_ADDR *real_pc, CORE_ADDR *bp_addr,
+ struct regcache *regcache)
+{
+ error (_("push_dummy_code not implemented"));
+ return sp;
+}
+
+
+static CORE_ADDR
+microblaze_push_dummy_call (struct gdbarch *gdbarch, struct value *function,
+ struct regcache *regcache, CORE_ADDR bp_addr,
+ int nargs, struct value **args, CORE_ADDR sp,
+ int struct_return, CORE_ADDR struct_addr)
+{
+ error (_("store_arguments not implemented"));
+ return sp;
+}
+
+static const gdb_byte *
+microblaze_breakpoint_from_pc (struct gdbarch *gdbarch, CORE_ADDR *pc,
+ int *len)
+{
+ static gdb_byte break_insn[] = MICROBLAZE_BREAKPOINT;
+
+ *len = sizeof (break_insn);
+ return break_insn;
+}
+\f
+/* Allocate and initialize a frame cache. */
+
+static struct microblaze_frame_cache *
+microblaze_alloc_frame_cache (void)
+{
+ struct microblaze_frame_cache *cache;
+ int i;
+
+ cache = FRAME_OBSTACK_ZALLOC (struct microblaze_frame_cache);
+
+ /* Base address. */
+ cache->base = 0;
+ cache->pc = 0;
+
+ /* Frameless until proven otherwise. */
+ cache->frameless_p = 1;
+
+ return cache;
+}
+
+/* The base of the current frame is actually in the stack pointer.
+ This happens when there is no frame pointer (microblaze ABI does not
+ require a frame pointer) or when we're stopped in the prologue or
+ epilogue itself. In these cases, microblaze_analyze_prologue will need
+ to update fi->frame before returning or analyzing the register
+ save instructions. */
+#define MICROBLAZE_MY_FRAME_IN_SP 0x1
+
+/* The base of the current frame is in a frame pointer register.
+ This register is noted in frame_extra_info->fp_regnum.
+
+ Note that the existance of an FP might also indicate that the
+ function has called alloca. */
+#define MICROBLAZE_MY_FRAME_IN_FP 0x2
+
+/* Function prologues on the Xilinx microblaze processors consist of:
+
+ - adjustments to the stack pointer (r1) (addi r1, r1, imm)
+ - making a copy of r1 into another register (a "frame" pointer)
+ (add r?, r1, r0)
+ - store word/multiples that use r1 or the frame pointer as the
+ base address (swi r?, r1, imm OR swi r?, fp, imm)
+
+ Note that microblaze really doesn't have a real frame pointer.
+ Instead, the compiler may copy the SP into a register (usually
+ r19) to act as an arg pointer. For our target-dependent purposes,
+ the frame info's "frame" member will be the beginning of the
+ frame. The SP could, in fact, point below this.
+
+ The prologue ends when an instruction fails to meet either of
+ these criteria. */
+
+/* Analyze the prologue to determine where registers are saved,
+ the end of the prologue, etc. Return the address of the first line
+ of "real" code (i.e., the end of the prologue). */
+
+static CORE_ADDR
+microblaze_analyze_prologue (struct gdbarch *gdbarch, CORE_ADDR pc,
+ CORE_ADDR current_pc,
+ struct microblaze_frame_cache *cache)
+{
+ char *name;
+ CORE_ADDR func_addr, func_end, addr, stop, prologue_end_addr = 0;
+ unsigned long insn;
+ int rn, rd, ra, rb, imm;
+ enum microblaze_instr op;
+ int flags = 0;
+ int save_hidden_pointer_found = 0;
+ int non_stack_instruction_found = 0;
+
+ /* Find the start of this function. */
+ find_pc_partial_function (pc, &name, &func_addr, &func_end);
+ if (func_addr < pc)
+ pc = func_addr;
+
+ if (current_pc < pc)
+ return current_pc;
+
+ /* Initialize info about frame. */
+ cache->framesize = 0;
+ cache->fp_regnum = MICROBLAZE_SP_REGNUM;
+ cache->frameless_p = 1;
+
+ /* Start decoding the prologue. We start by checking two special cases:
+
+ 1. We're about to return
+ 2. We're at the first insn of the prologue.
+
+ If we're about to return, our frame has already been deallocated.
+ If we are stopped at the first instruction of a prologue,
+ then our frame has not yet been set up. */
+
+ /* Get the first insn from memory. */
+
+ insn = microblaze_fetch_instruction (pc);
+ op = microblaze_decode_insn (insn, &rd, &ra, &rb, &imm);
+
+ if (IS_RETURN(op))
+ return pc;
+
+ /* Start at beginning of function and analyze until we get to the
+ current pc, or the end of the function, whichever is first. */
+ stop = (current_pc < func_end ? current_pc : func_end);
+
+ microblaze_debug ("Scanning prologue: name=%s, func_addr=%s, stop=%s\n",
+ name, paddress (gdbarch, func_addr),
+ paddress (gdbarch, stop));
+
+ for (addr = func_addr; addr < stop; addr += INST_WORD_SIZE)
+ {
+ insn = microblaze_fetch_instruction (addr);
+ op = microblaze_decode_insn (insn, &rd, &ra, &rb, &imm);
+ microblaze_debug ("%s %08lx\n", paddress (gdbarch, pc), insn);
+
+ /* This code is very sensitive to what functions are present in the
+ prologue. It assumes that the (addi, addik, swi, sw) can be the
+ only instructions in the prologue. */
+ if (IS_UPDATE_SP(op, rd, ra))
+ {
+ microblaze_debug ("got addi r1,r1,%d; contnuing\n", imm);
+ if (cache->framesize)
+ break; /* break if framesize already computed. */
+ cache->framesize = -imm; /* stack grows towards low memory. */
+ cache->frameless_p = 0; /* Frame found. */
+ save_hidden_pointer_found = 0;
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SPILL_SP(op, rd, ra))
+ {
+ /* Spill stack pointer. */
+ cache->register_offsets[rd] = imm; /* SP spilled before updating. */
+
+ microblaze_debug ("swi r1 r1 %d, continuing\n", imm);
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SPILL_REG(op, rd, ra))
+ {
+ /* Spill register. */
+ cache->register_offsets[rd] = imm - cache->framesize;
+
+ microblaze_debug ("swi %d r1 %d, continuing\n", rd, imm);
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_ALSO_SPILL_REG(op, rd, ra, rb))
+ {
+ /* Spill register. */
+ cache->register_offsets[rd] = 0 - cache->framesize;
+
+ microblaze_debug ("sw %d r0 r1, continuing\n", rd);
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SETUP_FP(op, ra, rb))
+ {
+ /* We have a frame pointer. Note the register which is
+ acting as the frame pointer. */
+ flags |= MICROBLAZE_MY_FRAME_IN_FP;
+ flags &= ~MICROBLAZE_MY_FRAME_IN_SP;
+ cache->fp_regnum = rd;
+ microblaze_debug ("Found a frame pointer: r%d\n", cache->fp_regnum);
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SPILL_REG_FP(op, rd, ra, cache->fp_regnum))
+ {
+ /* reg spilled after updating. */
+ cache->register_offsets[rd] = imm - cache->framesize;
+
+ microblaze_debug ("swi %d %d %d, continuing\n", rd, ra, imm);
+ save_hidden_pointer_found = 0;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+ else if (IS_SAVE_HIDDEN_PTR(op, rd, ra, rb))
+ {
+ /* If the first argument is a hidden pointer to the area where the
+ return structure is to be saved, then it is saved as part of the
+ prologue. */
+
+ microblaze_debug ("add %d %d %d, continuing\n", rd, ra, rb);
+ save_hidden_pointer_found = 1;
+ if (!cache->framesize)
+ non_stack_instruction_found = 0;
+ continue;
+ }
+
+ /* As a result of the modification in the next step where we continue
+ to analyze the prologue till we reach a control flow instruction,
+ we need another variable to store when exactly a non-stack
+ instruction was encountered, which is the current definition
+ of a prologue. */
+ if (!non_stack_instruction_found)
+ prologue_end_addr = addr;
+ non_stack_instruction_found = 1;
+
+ /* When optimizations are enabled, it is not guaranteed that prologue
+ instructions are not mixed in with other instructions from the
+ program. Some programs show this behavior at -O2. This can be
+ avoided by adding -fno-schedule-insns2 switch as of now (edk 8.1)
+ In such cases, we scan the function until we see the first control
+ instruction. */
+
+ {
+ unsigned op = (unsigned)insn >> 26;
+
+ /* continue if not control flow (branch, return). */
+ if (op != 0x26 && op != 0x27 && op != 0x2d && op != 0x2e && op != 0x2f)
+ continue;
+ else if (op == 0x2c)
+ continue; /* continue if imm. */
+ }
+
+ /* This is not a prologue insn, so stop here. */
+ microblaze_debug ("insn is not a prologue insn -- ending scan\n");
+ break;
+ }
+
+ microblaze_debug ("done analyzing prologue\n");
+ microblaze_debug ("prologue end = 0x%x\n", (int) addr);
+
+ /* If the last instruction was an add rd, r5, r0 then don't count it as
+ part of the prologue. */
+ if (save_hidden_pointer_found)
+ prologue_end_addr -= INST_WORD_SIZE;
+
+ return prologue_end_addr;
+}
+
+static CORE_ADDR
+microblaze_unwind_pc (struct gdbarch *gdbarch, struct frame_info *next_frame)
+{
+ gdb_byte buf[4];
+ CORE_ADDR pc;
+
+ frame_unwind_register (next_frame, MICROBLAZE_PC_REGNUM, buf);
+ pc = extract_typed_address (buf, builtin_type (gdbarch)->builtin_func_ptr);
+ /* For sentinel frame, return address is actual PC. For other frames,
+ return address is pc+8. This is a workaround because gcc does not
+ generate correct return address in CIE. */
+ if (frame_relative_level (next_frame) >= 0)
+ pc += 8;
+ return pc;
+}
+
+/* Return PC of first real instruction of the function starting at
+ START_PC. */
+
+CORE_ADDR
+microblaze_skip_prologue (struct gdbarch *gdbarch, CORE_ADDR start_pc)
+{
+ struct symtab_and_line sal;
+ CORE_ADDR func_start, func_end, ostart_pc;
+ struct microblaze_frame_cache cache;
+
+ /* This is the preferred method, find the end of the prologue by
+ using the debugging information. Debugging info does not always
+ give the right answer since parameters are stored on stack after this.
+ Always analyze the prologue. */
+ if (find_pc_partial_function (start_pc, NULL, &func_start, &func_end))
+ {
+ sal = find_pc_line (func_start, 0);
+
+ if (sal.end < func_end
+ && start_pc <= sal.end)
+ start_pc = sal.end;
+ }
+
+ ostart_pc = microblaze_analyze_prologue (gdbarch, func_start, 0xffffffffUL,
+ &cache);
+
+ if (ostart_pc > start_pc)
+ return ostart_pc;
+ return start_pc;
+}
+
+/* Normal frames. */
+
+struct microblaze_frame_cache *
+microblaze_frame_cache (struct frame_info *next_frame, void **this_cache)
+{
+ struct microblaze_frame_cache *cache;
+ struct gdbarch *gdbarch = get_frame_arch (next_frame);
+ CORE_ADDR func, pc, fp;
+ int rn;
+
+ if (*this_cache)
+ return *this_cache;
+
+ cache = microblaze_alloc_frame_cache ();
+ *this_cache = cache;
+ cache->saved_regs = trad_frame_alloc_saved_regs (next_frame);
+
+ /* Clear offsets to saved regs in frame. */
+ for (rn = 0; rn < gdbarch_num_regs (gdbarch); rn++)
+ cache->register_offsets[rn] = -1;
+
+ func = get_frame_func (next_frame);
+
+ cache->pc = get_frame_address_in_block (next_frame);
+
+ return cache;
+}
+
+static void
+microblaze_frame_this_id (struct frame_info *next_frame, void **this_cache,
+ struct frame_id *this_id)
+{
+ struct microblaze_frame_cache *cache =
+ microblaze_frame_cache (next_frame, this_cache);
+
+ /* This marks the outermost frame. */
+ if (cache->base == 0)
+ return;
+
+ (*this_id) = frame_id_build (cache->base, cache->pc);
+}
+
+static struct value *
+microblaze_frame_prev_register (struct frame_info *this_frame,
+ void **this_cache, int regnum)
+{
+ struct microblaze_frame_cache *cache =
+ microblaze_frame_cache (this_frame, this_cache);
+
+ if (cache->frameless_p)
+ {
+ if (regnum == MICROBLAZE_PC_REGNUM)
+ regnum = 15;
+ if (regnum == MICROBLAZE_SP_REGNUM)
+ regnum = 1;
+ return trad_frame_get_prev_register (this_frame, cache->saved_regs, regnum);
+ }
+ else
+ return trad_frame_get_prev_register (this_frame, cache->saved_regs,
+ regnum);
+
+}
+
+static const struct frame_unwind microblaze_frame_unwind =
+{
+ NORMAL_FRAME,
+ microblaze_frame_this_id,
+ microblaze_frame_prev_register,
+ NULL,
+ default_frame_sniffer
+};
+\f
+static CORE_ADDR
+microblaze_frame_base_address (struct frame_info *next_frame, void **this_cache)
+{
+ struct microblaze_frame_cache *cache =
+ microblaze_frame_cache (next_frame, this_cache);
+
+ return cache->base;
+}
+
+static const struct frame_base microblaze_frame_base =
+{
+ µblaze_frame_unwind,
+ microblaze_frame_base_address,
+ microblaze_frame_base_address,
+ microblaze_frame_base_address
+};
+\f
+/* Extract from an array REGBUF containing the (raw) register state, a
+ function return value of TYPE, and copy that into VALBUF. */
+static void
+microblaze_extract_return_value (struct type *type, struct regcache *regcache,
+ gdb_byte *valbuf)
+{
+ gdb_byte buf[8];
+
+ /* Copy the return value (starting) in RETVAL_REGNUM to VALBUF. */
+ switch (TYPE_LENGTH (type))
+ {
+ case 1: /* return last byte in the register. */
+ regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
+ memcpy(valbuf, buf + MICROBLAZE_REGISTER_SIZE - 1, 1);
+ return;
+ case 2: /* return last 2 bytes in register. */
+ memcpy(valbuf, buf + MICROBLAZE_REGISTER_SIZE - 2, 2);
+ return;
+ case 4: /* for sizes 4 or 8, copy the required length. */
+ case 8:
+ regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
+ regcache_cooked_read (regcache, MICROBLAZE_RETVAL_REGNUM+1, buf+4);
+ memcpy (valbuf, buf, TYPE_LENGTH (type));
+ return;
+ default:
+ internal_error (__FILE__, __LINE__,
+ _("Unsupported return value size requested"));
+ }
+}
+
+/* Store the return value in VALBUF (of type TYPE) where the caller
+ expects to see it.
+
+ Integers up to four bytes are stored in r3.
+
+ Longs are stored in r3 (most significant word) and r4 (least
+ significant word).
+
+ Small structures are always returned on stack.
+*/
+
+static void
+microblaze_store_return_value (struct type *type, struct regcache *regcache,
+ const gdb_byte *valbuf)
+{
+ int len = TYPE_LENGTH (type);
+ gdb_byte buf[8];
+
+ memset (buf, 0, sizeof(buf));
+
+ /* Integral and pointer return values. */
+
+ if (len > 4)
+ {
+ gdb_assert (len == 8);
+ memcpy (buf, valbuf, 8);
+ regcache_cooked_write (regcache, MICROBLAZE_RETVAL_REGNUM+1, buf + 4);
+ }
+ else
+ /* ??? Do we need to do any sign-extension here? */
+ memcpy (buf + 4 - len, valbuf, len);
+
+ regcache_cooked_write (regcache, MICROBLAZE_RETVAL_REGNUM, buf);
+}
+
+static enum return_value_convention
+microblaze_return_value (struct gdbarch *gdbarch, struct type *func_type,
+ struct type *type, struct regcache *regcache,
+ gdb_byte *readbuf, const gdb_byte *writebuf)
+{
+ if (readbuf)
+ microblaze_extract_return_value (type, regcache, readbuf);
+ if (writebuf)
+ microblaze_store_return_value (type, regcache, writebuf);
+
+ return RETURN_VALUE_REGISTER_CONVENTION;
+}
+
+static int
+microblaze_stabs_argument_has_addr (struct gdbarch *gdbarch, struct type *type)
+{
+ return (TYPE_LENGTH (type) == 16);
+}
+
+static void
+microblaze_write_pc (struct regcache *regcache, CORE_ADDR pc)
+{
+ regcache_cooked_write_unsigned (regcache, MICROBLAZE_PC_REGNUM, pc);
+}
+\f
+static int dwarf2_to_reg_map[78] =
+{ 0 /* r0 */, 1 /* r1 */, 2 /* r2 */, 3 /* r3 */, /* 0- 3 */
+ 4 /* r4 */, 5 /* r5 */, 6 /* r6 */, 7 /* r7 */, /* 4- 7 */
+ 8 /* r8 */, 9 /* r9 */, 10 /* r10 */, 11 /* r11 */, /* 8-11 */
+ 12 /* r12 */, 13 /* r13 */, 14 /* r14 */, 15 /* r15 */, /* 12-15 */
+ 16 /* r16 */, 17 /* r17 */, 18 /* r18 */, 19 /* r19 */, /* 16-19 */
+ 20 /* r20 */, 21 /* r21 */, 22 /* r22 */, 23 /* r23 */, /* 20-23 */
+ 24 /* r24 */, 25 /* r25 */, 26 /* r26 */, 27 /* r27 */, /* 24-25 */
+ 28 /* r28 */, 29 /* r29 */, 30 /* r30 */, 31 /* r31 */, /* 28-31 */
+ -1 /* $f0 */, -1 /* $f1 */, -1 /* $f2 */, -1 /* $f3 */, /* 32-35 */
+ -1 /* $f4 */, -1 /* $f5 */, -1 /* $f6 */, -1 /* $f7 */, /* 36-39 */
+ -1 /* $f8 */, -1 /* $f9 */, -1 /* $f10 */, -1 /* $f11 */, /* 40-43 */
+ -1 /* $f12 */, -1 /* $f13 */, -1 /* $f14 */, -1 /* $f15 */, /* 44-47 */
+ -1 /* $f16 */, -1 /* $f17 */, -1 /* $f18 */, -1 /* $f19 */, /* 48-51 */
+ -1 /* $f20 */, -1 /* $f21 */, -1 /* $f22 */, -1 /* $f23 */, /* 52-55 */
+ -1 /* $f24 */, -1 /* $f25 */, -1 /* $f26 */, -1 /* $f27 */, /* 56-59 */
+ -1 /* $f28 */, -1 /* $f29 */, -1 /* $f30 */, -1 /* $f31 */, /* 60-63 */
+ -1 /* hi */, -1 /* lo */, -1 /* accum*/, 33 /* rmsr */, /* 64-67 */
+ -1 /* $fcc1*/, -1 /* $fcc2*/, -1 /* $fcc3*/, -1 /* $fcc4*/, /* 68-71 */
+ -1 /* $fcc5*/, -1 /* $fcc6*/, -1 /* $fcc7*/, -1 /* $ap */, /* 72-75 */
+ -1 /* $rap */, -1 /* $frp */ /* 76-77 */
+};
+
+static int
+microblaze_dwarf2_reg_to_regnum (struct gdbarch *gdbarch, int reg)
+{
+ gdb_assert (reg < sizeof (dwarf2_to_reg_map));
+ return dwarf2_to_reg_map[reg];
+}
+
+static struct gdbarch *
+microblaze_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
+{
+ struct gdbarch_tdep *tdep;
+ struct gdbarch *gdbarch;
+
+ /* If there is already a candidate, use it. */
+ arches = gdbarch_list_lookup_by_info (arches, &info);
+ if (arches != NULL)
+ return arches->gdbarch;
+
+ /* Allocate space for the new architecture. */
+ tdep = XMALLOC (struct gdbarch_tdep);
+ gdbarch = gdbarch_alloc (&info, tdep);
+
+ set_gdbarch_long_double_bit (gdbarch, 128);
+
+ set_gdbarch_num_regs (gdbarch, MICROBLAZE_NUM_REGS);
+ set_gdbarch_register_name (gdbarch, microblaze_register_name);
+ set_gdbarch_register_type (gdbarch, microblaze_register_type);
+
+ /* Register numbers of various important registers. */
+ set_gdbarch_sp_regnum (gdbarch, MICROBLAZE_SP_REGNUM);
+ set_gdbarch_pc_regnum (gdbarch, MICROBLAZE_PC_REGNUM);
+
+ /* Map Dwarf2 registers to GDB registers. */
+ set_gdbarch_dwarf2_reg_to_regnum (gdbarch, microblaze_dwarf2_reg_to_regnum);
+
+ /* Call dummy code. */
+ set_gdbarch_call_dummy_location (gdbarch, ON_STACK);
+ set_gdbarch_push_dummy_code (gdbarch, microblaze_push_dummy_code);
+ set_gdbarch_push_dummy_call (gdbarch, microblaze_push_dummy_call);
+
+ set_gdbarch_return_value (gdbarch, microblaze_return_value);
+ set_gdbarch_stabs_argument_has_addr
+ (gdbarch, microblaze_stabs_argument_has_addr);
+
+ set_gdbarch_skip_prologue (gdbarch, microblaze_skip_prologue);
+
+ /* Stack grows downward. */
+ set_gdbarch_inner_than (gdbarch, core_addr_lessthan);
+
+ set_gdbarch_breakpoint_from_pc (gdbarch, microblaze_breakpoint_from_pc);
+
+ set_gdbarch_frame_args_skip (gdbarch, 8);
+
+ set_gdbarch_print_insn (gdbarch, print_insn_microblaze);
+
+ set_gdbarch_write_pc (gdbarch, microblaze_write_pc);
+
+ set_gdbarch_unwind_pc (gdbarch, microblaze_unwind_pc);
+
+ frame_base_set_default (gdbarch, µblaze_frame_base);
+
+ /* Hook in ABI-specific overrides, if they have been registered. */
+ gdbarch_init_osabi (info, gdbarch);
+
+ /* Unwind the frame. */
+ dwarf2_append_unwinders (gdbarch);
+ frame_unwind_append_unwinder (gdbarch, µblaze_frame_unwind);
+ frame_base_append_sniffer (gdbarch, dwarf2_frame_base_sniffer);
+
+ return gdbarch;
+}
+
+/* Provide a prototype to silence -Wmissing-prototypes. */
+void _initialize_microblaze_tdep (void);
+
+void
+_initialize_microblaze_tdep (void)
+{
+ register_gdbarch_init (bfd_arch_microblaze, microblaze_gdbarch_init);
+
+ /* Debug this files internals. */
+ add_setshow_zinteger_cmd ("microblaze", class_maintenance,
+ µblaze_debug_flag, _("\
+Set microblaze debugging."), _("\
+Show microblaze debugging."), _("\
+When non-zero, microblaze specific debugging is enabled."),
+ NULL,
+ NULL,
+ &setdebuglist, &showdebuglist);
+
+}
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/microblaze-tdep.h mb-gdb/gdb/microblaze-tdep.h
--- gdb/gdb/microblaze-tdep.h 1969-12-31 16:00:00.000000000 -0800
+++ mb-gdb/gdb/microblaze-tdep.h 2009-10-14 14:41:06.000000000 -0700
@@ -0,0 +1,117 @@
+/* Target-dependent code for Xilinx MicroBlaze.
+
+ Copyright 2009 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 3 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, see <http://www.gnu.org/licenses/>. */
+
+#ifndef MICROBLAZE_TDEP_H
+#define MICROBLAZE_TDEP_H 1
+
+
+/* Microblaze architecture-specific information. */
+struct gdbarch_tdep
+{
+};
+
+struct microblaze_frame_cache
+{
+ /* Base address. */
+ CORE_ADDR base;
+ CORE_ADDR pc;
+
+ /* Do we have a frame? */
+ int frameless_p;
+
+ /* Frame size. */
+ int framesize;
+
+ /* Frame register. */
+ int fp_regnum;
+
+ /* Offsets to saved registers. */
+ int register_offsets[57]; /* Must match MICROBLAZE_NUM_REGS. */
+
+ /* Table of saved registers. */
+ struct trad_frame_saved_reg *saved_regs;
+};
+
+/* Register numbers. */
+enum microblaze_regnum
+{
+ MICROBLAZE_R0_REGNUM,
+ MICROBLAZE_R1_REGNUM, MICROBLAZE_SP_REGNUM = MICROBLAZE_R1_REGNUM,
+ MICROBLAZE_R2_REGNUM,
+ MICROBLAZE_R3_REGNUM, MICROBLAZE_RETVAL_REGNUM = MICROBLAZE_R3_REGNUM,
+ MICROBLAZE_R4_REGNUM,
+ MICROBLAZE_R5_REGNUM, MICROBLAZE_FIRST_ARGREG = MICROBLAZE_R5_REGNUM,
+ MICROBLAZE_R6_REGNUM,
+ MICROBLAZE_R7_REGNUM,
+ MICROBLAZE_R8_REGNUM,
+ MICROBLAZE_R9_REGNUM,
+ MICROBLAZE_R10_REGNUM, MICROBLAZE_LAST_ARGREG = MICROBLAZE_R10_REGNUM,
+ MICROBLAZE_R11_REGNUM,
+ MICROBLAZE_R12_REGNUM,
+ MICROBLAZE_R13_REGNUM,
+ MICROBLAZE_R14_REGNUM,
+ MICROBLAZE_R15_REGNUM,
+ MICROBLAZE_R16_REGNUM,
+ MICROBLAZE_R17_REGNUM,
+ MICROBLAZE_R18_REGNUM,
+ MICROBLAZE_R19_REGNUM,
+ MICROBLAZE_R20_REGNUM,
+ MICROBLAZE_R21_REGNUM,
+ MICROBLAZE_R22_REGNUM,
+ MICROBLAZE_R23_REGNUM,
+ MICROBLAZE_R24_REGNUM,
+ MICROBLAZE_R25_REGNUM,
+ MICROBLAZE_R26_REGNUM,
+ MICROBLAZE_R27_REGNUM,
+ MICROBLAZE_R28_REGNUM,
+ MICROBLAZE_R29_REGNUM,
+ MICROBLAZE_R30_REGNUM,
+ MICROBLAZE_R31_REGNUM,
+ MICROBLAZE_PC_REGNUM,
+ MICROBLAZE_MSR_REGNUM,
+ MICROBLAZE_EAR_REGNUM,
+ MICROBLAZE_ESR_REGNUM,
+ MICROBLAZE_FSR_REGNUM,
+ MICROBLAZE_BTR_REGNUM,
+ MICROBLAZE_PVR0_REGNUM,
+ MICROBLAZE_PVR1_REGNUM,
+ MICROBLAZE_PVR2_REGNUM,
+ MICROBLAZE_PVR3_REGNUM,
+ MICROBLAZE_PVR4_REGNUM,
+ MICROBLAZE_PVR5_REGNUM,
+ MICROBLAZE_PVR6_REGNUM,
+ MICROBLAZE_PVR7_REGNUM,
+ MICROBLAZE_PVR8_REGNUM,
+ MICROBLAZE_PVR9_REGNUM,
+ MICROBLAZE_PVR10_REGNUM,
+ MICROBLAZE_PVR11_REGNUM,
+ MICROBLAZE_REDR_REGNUM,
+ MICROBLAZE_RPID_REGNUM,
+ MICROBLAZE_RZPR_REGNUM,
+ MICROBLAZE_RTLBX_REGNUM,
+ MICROBLAZE_RTLBSX_REGNUM,
+ MICROBLAZE_RTLBLO_REGNUM,
+ MICROBLAZE_RTLBHI_REGNUM
+};
+
+/* MICROBLAZE_BREAKPOINT defines the breakpoint that should be used.
+ Only used for native debugging. */
+#define MICROBLAZE_BREAKPOINT {0xb9, 0xcc, 0x00, 0x60}
+
+#endif /* microblaze-tdep.h */
diff -urNp --exclude '*.swp' --exclude DEV-PHASE --exclude .svn --exclude CVS gdb/gdb/NEWS mb-gdb/gdb/NEWS
--- gdb/gdb/NEWS 2009-10-14 11:15:51.000000000 -0700
+++ mb-gdb/gdb/NEWS 2009-10-14 19:15:57.000000000 -0700
@@ -484,6 +484,11 @@ Lattice Mico32 lm32-*
x86 DICOS i[34567]86-*-dicos*
x86_64 DICOS x86_64-*-dicos*
S+core 3 score-*-*
+Xilinx MicroBlaze microblaze-*-*
+
+* New Simulators
+
+Xilinx MicroBlaze microblaze
* The GDB remote stub, gdbserver, now supports x86 Windows CE
(mingw32ce) debugging.
^ permalink raw reply [flat|nested] 20+ messages in thread
end of thread, other threads:[~2009-10-16 16:14 UTC | newest]
Thread overview: 20+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-10-06 1:35 [PATCH] Support for Xilinx MicroBlaze Michael Eager
2009-10-06 7:14 ` Eli Zaretskii
2009-10-14 1:47 ` Joel Brobecker
2009-10-14 17:49 ` Michael Eager
2009-10-14 17:56 ` Daniel Jacobowitz
2009-10-14 18:01 ` Michael Eager
2009-10-15 16:32 Michael Eager
2009-10-15 16:44 ` Tom Tromey
2009-10-15 16:47 ` Joel Brobecker
2009-10-15 18:00 ` Tom Tromey
2009-10-15 18:04 ` Tom Tromey
2009-10-15 18:31 ` Joel Brobecker
2009-10-16 12:53 ` Daniel Jacobowitz
2009-10-16 16:14 ` Michael Eager
2009-10-15 17:12 ` Michael Eager
2009-10-15 17:18 ` Joel Brobecker
2009-10-15 18:06 ` Michael Eager
2009-10-15 16:46 ` Joel Brobecker
2009-10-15 17:14 ` Michael Eager
2009-10-15 19:31 ` Michael Eager
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox