[RFA]: Upstream bfd patch for ia64 vDSO support

[RFA]: Upstream bfd patch for ia64 vDSO support

[Jeff Johnston]

[-- Attachment #1: Type: text/plain, Size: 759 bytes --]

The attached patch has been checked in to binutils by Jim Wilson.  It fixes a 
problem with the ia64 vDSO so that gdb can backtrace from system calls.  I need 
this to check in an ia64/libunwind patch that fixes backtracing through 
syscalls.  I didn't know if one could just check in upstream bfd patches without 
approval so I will ask.  May I commit this?

Without the patch, the bfd_from_remote_memory call sets a bogus address for 
loadbase and symbol_file_add_from_memory fails.

See http://sourceware.org/ml/binutils/2005-05/msg00725.html for the patch and 
references to the discussion

2005-05-31  James E Wilson  <wilson@specifixinc.com>

	* elfcode.h (NAME(bfd_elf,bfd_from_remote_memory)): Check for program
	header PF_R flag on PT_LOAD segments.

[-- Attachment #2: elfcode.h.patch --]
[-- Type: text/plain, Size: 1323 bytes --]

Index: elfcode.h
===================================================================
RCS file: /cvs/src/src/bfd/elfcode.h,v
retrieving revision 1.69
diff -u -p -r1.69 elfcode.h
--- elfcode.h	26 May 2005 07:41:13 -0000	1.69
+++ elfcode.h	1 Jun 2005 19:58:09 -0000
@@ -1656,7 +1656,10 @@ NAME(_bfd_elf,bfd_from_remote_memory)
   for (i = 0; i < i_ehdr.e_phnum; ++i)
     {
       elf_swap_phdr_in (templ, &x_phdrs[i], &i_phdrs[i]);
-      if (i_phdrs[i].p_type == PT_LOAD)
+      /* IA-64 vDSO may have two mappings for one segment, where one mapping
+	 is executable only, and one is read only.  We must not use the
+	 executable one.  */
+      if (i_phdrs[i].p_type == PT_LOAD && (i_phdrs[i].p_flags & PF_R))
 	{
 	  bfd_vma segment_end;
 	  segment_end = (i_phdrs[i].p_offset + i_phdrs[i].p_filesz
@@ -1703,7 +1706,10 @@ NAME(_bfd_elf,bfd_from_remote_memory)
     }
 
   for (i = 0; i < i_ehdr.e_phnum; ++i)
-    if (i_phdrs[i].p_type == PT_LOAD)
+    /* IA-64 vDSO may have two mappings for one segment, where one mapping
+       is executable only, and one is read only.  We must not use the
+       executable one.  */
+    if (i_phdrs[i].p_type == PT_LOAD && (i_phdrs[i].p_flags & PF_R))
       {
 	bfd_vma start = i_phdrs[i].p_offset & -i_phdrs[i].p_align;
 	bfd_vma end = (i_phdrs[i].p_offset + i_phdrs[i].p_filesz

Re: [RFA]: Upstream bfd patch for ia64 vDSO support

[Daniel Jacobowitz]

On Wed, Jun 01, 2005 at 04:13:29PM -0400, Jeff Johnston wrote:
> The attached patch has been checked in to binutils by Jim Wilson.  It fixes 
> a problem with the ia64 vDSO so that gdb can backtrace from system calls.  
> I need this to check in an ia64/libunwind patch that fixes backtracing 
> through syscalls.  I didn't know if one could just check in upstream bfd 
> patches without approval so I will ask.  May I commit this?

GDB and binutils live in the same source repository.  Applying it to
one applies it to the other, except for release branches of course. 
There's nothing more to do.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: [RFA]: Upstream bfd patch for ia64 vDSO support

[Jeff Johnston]

Daniel Jacobowitz wrote:
> On Wed, Jun 01, 2005 at 04:13:29PM -0400, Jeff Johnston wrote:
> 
>>The attached patch has been checked in to binutils by Jim Wilson.  It fixes 
>>a problem with the ia64 vDSO so that gdb can backtrace from system calls.  
>>I need this to check in an ia64/libunwind patch that fixes backtracing 
>>through syscalls.  I didn't know if one could just check in upstream bfd 
>>patches without approval so I will ask.  May I commit this?
> 
> 
> GDB and binutils live in the same source repository.  Applying it to
> one applies it to the other, except for release branches of course. 
> There's nothing more to do.
> 

Ok, thanks for the clarification.

-- Jeff J.