Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Kevin Buettner <kevinb@redhat.com>
To: gdb-patches@sourceware.org
Subject: [RFC] FRV: Skip call to __main in main()
Date: Sat, 15 Nov 2008 18:40:00 -0000	[thread overview]
Message-ID: <20081114161353.3495cc1e@mesquite.lan> (raw)

For frv-elf (and perhaps other frv targets too), GCC now emits a call
to __main() in the prologue of main().  The patch below adds code to
test for this call and to skip over it so that placing a breakpoint in
main() will correctly cause the breakpoint to be placed on the first
statement instead of in the prologue.

I submitted a similar patch for PPC a while back and was asked to fix
the hard-coded constants.  However, the convention in frv-tdep.c seems
to be to use hard-coded constants preceded by a comment block showing
how the constant was derived.  I chose to follow that convention in
this patch.

Comments?

	* frv-tdep.c (frv_skip_main_prologue): New function.
	(frv_gdbarch_init): Register frv_skip_main_prologue.

Index: frv-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/frv-tdep.c,v
retrieving revision 1.122
diff -u -p -r1.122 frv-tdep.c
--- frv-tdep.c	11 Sep 2008 14:23:15 -0000	1.122
+++ frv-tdep.c	14 Nov 2008 22:40:25 -0000
@@ -999,6 +999,85 @@ frv_skip_prologue (struct gdbarch *gdbar
 }
 
 
+/* Examine the instruction pointed to by PC.  If it corresponds to
+   a call to __main, return the address of the next instruction.
+   Otherwise, return PC.  */
+
+static CORE_ADDR
+frv_skip_main_prologue (struct gdbarch *gdbarch, CORE_ADDR pc)
+{
+  gdb_byte buf[4];
+  unsigned long op;
+  CORE_ADDR orig_pc = pc;
+
+  if (target_read_memory (pc, buf, 4))
+    return pc;
+  op = extract_unsigned_integer (buf, 4);
+
+  /* In PIC code, GR15 may be loaded from some offset off of FP prior
+     to the call instruction.
+     
+     Skip over this instruction if present.  It won't be present in
+     non-PIC code, and even in PIC code, it might not be present. 
+     (This is due to the fact that GR15, the FDPIC register, already
+     contains the correct value.)
+
+     The general form of the LDI is given first, followed by the
+     specific instruction with the GRi and GRk filled in as FP and
+     GR15.
+
+     ldi @(GRi, d12), GRk
+     P KKKKKK 0110010 IIIIII SSSSSSSSSSSS = 0x00c80000
+     0 000000 1111111 000000 000000000000 = 0x01fc0000
+	 .    .   .    .   .    .   .   .
+     ldi @(FP, d12), GR15
+     P KKKKKK 0110010 IIIIII SSSSSSSSSSSS = 0x1ec82000
+     0 001111 1111111 000010 000000000000 = 0x7ffff000
+	 .    .   .    .   .    .   .   .               */
+
+  if ((op & 0x7ffff000) == 0x1ec82000)
+    {
+      pc += 4;
+      if (target_read_memory (pc, buf, 4))
+	return orig_pc;
+      op = extract_unsigned_integer (buf, 4);
+    }
+
+  /* The format of an FRV CALL instruction is as follows:
+
+     call label24
+     P HHHHHH 0001111 LLLLLLLLLLLLLLLLLL = 0x003c0000
+     0 000000 1111111 000000000000000000 = 0x01fc0000
+         .    .   .    .   .   .   .   .
+
+     where label24 is constructed by concatenating the H bits with the
+     L bits.  The call target is PC + (4 * sign_ext(label24)).  */
+
+  if ((op & 0x01fc0000) == 0x003c0000)
+    {
+      LONGEST displ;
+      CORE_ADDR call_dest;
+      struct minimal_symbol *s;
+
+      displ = ((op & 0xfe000000) >> 7) | (op & 0x0003ffff);
+      if ((displ & 0x00800000) != 0)
+	displ |= ~((LONGEST) 0x00ffffff);
+
+      call_dest = pc + 4 * displ;
+      s = lookup_minimal_symbol_by_pc (call_dest);
+
+      if (s != NULL
+          && SYMBOL_LINKAGE_NAME (s) != NULL
+	  && strcmp (SYMBOL_LINKAGE_NAME (s), "__main") == 0)
+	{
+	  pc += 4;
+	  return pc;
+	}
+    }
+  return orig_pc;
+}
+
+
 static struct frv_unwind_cache *
 frv_frame_unwind_cache (struct frame_info *this_frame,
 			 void **this_prologue_cache)
@@ -1501,6 +1580,7 @@ frv_gdbarch_init (struct gdbarch_info in
   set_gdbarch_pseudo_register_write (gdbarch, frv_pseudo_register_write);
 
   set_gdbarch_skip_prologue (gdbarch, frv_skip_prologue);
+  set_gdbarch_skip_main_prologue (gdbarch, frv_skip_main_prologue);
   set_gdbarch_breakpoint_from_pc (gdbarch, frv_breakpoint_from_pc);
   set_gdbarch_adjust_breakpoint_address
     (gdbarch, frv_adjust_breakpoint_address);


             reply	other threads:[~2008-11-14 23:14 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-11-15 18:40 Kevin Buettner [this message]
2008-11-15 21:13 ` Daniel Jacobowitz
2008-11-20  1:58 ` Kevin Buettner

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20081114161353.3495cc1e@mesquite.lan \
    --to=kevinb@redhat.com \
    --cc=gdb-patches@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox