[patch] Fix PIE for 64bit gdb -> 32bit inferior

[patch] Fix PIE for 64bit gdb -> 32bit inferior

[Jan Kratochvil]

Hi,

according to the suggestion by Mark Kettenis the fix for biarch PIE is in fact
simple.

The displacement for cross-arch inferiors can use 64bit math operations if its
bits 32..63 are properly set to either 0x00000000 << 32 or 0xffffffff << 32 as
appropriate.

No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.  It fixes some
FAILs of gdb.base/break-interp.exp on {x86_64-m32}-fedora12-linux-gnu
(therefore host=x86_64-fedora12-linux-gnu target=i686-fedora12-linux-gnu).


Thanks,
Jan


2010-02-11  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* solib-svr4.c (enable_break <target_auxv_search>): New variable
	addr_bit.  Adjust LOAD_ADDR sign for cross-arch inferiors.

--- a/gdb/solib-svr4.c
+++ b/gdb/solib-svr4.c
@@ -1428,7 +1428,32 @@ enable_break (struct svr4_info *info, int from_tty)
          from our so_list, then try using the AT_BASE auxilliary entry.  */
       if (!load_addr_found)
         if (target_auxv_search (&current_target, AT_BASE, &load_addr) > 0)
-          load_addr_found = 1;
+	  {
+	    int addr_bit = gdbarch_addr_bit (target_gdbarch);
+
+	    /* Ensure LOAD_ADDR has proper sign in its possible upper bits so
+	       that `+ load_addr' will overflow CORE_ADDR width not creating
+	       invalid addresses like 0x101234567 for 32bit inferiors on 64bit
+	       GDB.  */
+
+	    if (addr_bit < (sizeof (ULONGEST) * HOST_CHAR_BIT))
+	      {
+		CORE_ADDR space_size = (ULONGEST) 1 << addr_bit;
+		CORE_ADDR tmp_entry_point = exec_entry_point (tmp_bfd,
+							      tmp_bfd_target);
+
+		gdb_assert (load_addr < space_size);
+
+		/* TMP_ENTRY_POINT exceeding SPACE_SIZE would be for prelinked
+		   64bit ld.so with 32bit executable, it should not happen.  */
+
+		if (tmp_entry_point < space_size
+		    && tmp_entry_point + load_addr >= space_size)
+		  load_addr -= space_size;
+	      }
+
+	    load_addr_found = 1;
+	  }
 
       /* Otherwise we find the dynamic linker's base address by examining
 	 the current pc (which should point at the entry point for the

Re: [patch] Fix PIE for 64bit gdb -> 32bit inferior

[Tom Tromey]

>>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:

Jan> according to the suggestion by Mark Kettenis the fix for biarch PIE
Jan> is in fact simple.

Thanks.

Jan> 2010-02-11  Jan Kratochvil  <jan.kratochvil@redhat.com>
Jan> 	* solib-svr4.c (enable_break <target_auxv_search>): New variable
Jan> 	addr_bit.  Adjust LOAD_ADDR sign for cross-arch inferiors.

This is ok.

Tom

Re: [patch] Fix PIE for 64bit gdb -> 32bit inferior

[Jan Kratochvil]

On Wed, 17 Feb 2010 00:34:29 +0100, Tom Tromey wrote:
> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
> Jan> 2010-02-11  Jan Kratochvil  <jan.kratochvil@redhat.com>
> Jan> 	* solib-svr4.c (enable_break <target_auxv_search>): New variable
> Jan> 	addr_bit.  Adjust LOAD_ADDR sign for cross-arch inferiors.
> 
> This is ok.

Checked-in:
	http://sourceware.org/ml/gdb-cvs/2010-02/msg00135.html


Thanks,
Jan

Re: [patch] Fix PIE for 64bit gdb -> 32bit inferior

[H.J. Lu]

On Wed, Feb 17, 2010 at 3:38 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> On Wed, 17 Feb 2010 00:34:29 +0100, Tom Tromey wrote:
>> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
>> Jan> 2010-02-11  Jan Kratochvil  <jan.kratochvil@redhat.com>
>> Jan>  * solib-svr4.c (enable_break <target_auxv_search>): New variable
>> Jan>  addr_bit.  Adjust LOAD_ADDR sign for cross-arch inferiors.
>>
>> This is ok.
>
> Checked-in:
>        http://sourceware.org/ml/gdb-cvs/2010-02/msg00135.html
>

I think it caused:

http://www.sourceware.org/bugzilla/show_bug.cgi?id=11293


-- 
H.J.