[patch] Fix gdb.cp/gdb2495.exp regression with gcc-4.7

[Jan Kratochvil <jan dot kratochvil at redhat dot com>]

Hi,

(gdb) p exceptions.throw_function()
Program received signal SIGSEGV, Segmentation fault.
x86_64_fallback_frame_state (fs=0x7fffffffdc80, context=0x7fffffffdef0) at ../../../gcc/config/i386/linux-unwind.h:47
47       if (*(unsigned char *)(pc+0) == 0x48
The program being debugged was signaled while in a function called from GDB.
[...]
(gdb) FAIL: gdb.cp/gdb2495.exp: Call a function that raises an exception without a handler.

This happens with
	gcc (GCC) 4.7.0 20111222 (experimental)
on Fedora Rawhide (pre-17) x86_64 as the function before _start is PLT and
PLTs have no proper .eh_frame entries.  Still such .eh_frame PLT entry sure
does not apply for the <function called from gdb> frame.  Therefore it unwinds
into garbage and then crashes on it (PC is 0x1 abo-ve).

No regressions on {x86_64,x86_64-m32}-fedora16-linux-gnu.

Thanks,
Jan


gdb/
2011-12-22  Jan Kratochvil  <jan.kratochvil@redhat.com>

	Fix regression for gdb.cp/gdb2495.exp with gcc-4.7.
	* gdbarch.sh (max_insn_length): Set the default length to 31.
	* gdbarch.c: Regenerate.
	* gdbarch.h: Regenerate.
	* infcall.c: Include disasm.h.
	(call_function_by_hand) <AT_ENTRY_POINT>: New variables insn and
	insn_len.  Adjust DUMMY_ADDR with them if possible.

--- a/gdb/gdbarch.sh
+++ b/gdb/gdbarch.sh
@@ -655,7 +655,8 @@ v:int:vbit_in_delta:::0:0::0
 F:void:skip_permanent_breakpoint:struct regcache *regcache:regcache
 
 # The maximum length of an instruction on this architecture.
-V:ULONGEST:max_insn_length:::0:0
+# It should be at least as maximum of all the supported architectures.
+V:ULONGEST:max_insn_length:::31:31
 
 # Copy the instruction at FROM to TO, and make any adjustments
 # necessary to single-step it at that address.
--- a/gdb/infcall.c
+++ b/gdb/infcall.c
@@ -38,6 +38,7 @@
 #include "ada-lang.h"
 #include "gdbthread.h"
 #include "exceptions.h"
+#include "disasm.h"
 
 /* If we can't find a function's name from its address,
    we print this instead.  */
@@ -634,9 +635,32 @@ call_function_by_hand (struct value *function, int nargs, struct value **args)
     case AT_ENTRY_POINT:
       {
 	CORE_ADDR dummy_addr;
+	gdb_byte *insn;
+	CORE_ADDR insn_len;
 
 	real_pc = funaddr;
 	dummy_addr = entry_point_address ();
+
+	/* If the inferior call throws an uncaught C++ exception the inferior
+	   unwinder will try to unwind all the frames incl. the dummy frame.
+	   Using the entry_point_address directly will try to find FDE at the
+	   function right before the entry_point_address address as the
+	   unwinder subtracts 1 to get at the call instruction.  FDE of the
+	   preceding function, if found, would be invalid for the dummy frame
+	   and it would crash the inferior's unwinder.  Therefore attempt to
+	   skip the very first instruction of entry_point_address.  */
+
+	insn_len = gdbarch_max_insn_length (gdbarch);
+	insn = alloca (insn_len);
+	if (target_read_memory (dummy_addr, insn, insn_len) == 0)
+	  dummy_addr += gdb_buffered_insn_length (gdbarch, insn, insn_len,
+						  dummy_addr);
+	else
+	  {
+	    /* No problem probably occurs without this adjustment.  INSN_LEN
+	       may be for example larger than the entry_point_address code.  */
+	  }
+
 	/* A call dummy always consists of just a single breakpoint, so
 	   its address is the same as the address of the dummy.  */
 	bp_addr = dummy_addr;