Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Nick Clifton]

Hi Michael,

: 2000-06-07  Michael Sokolov  <msokolov@ivan.Harhan.ORG>
: 
: 	* emultempl/mipsecoff.em (gld{EMULATION_NAME}_after_open): Require all
: 	input objects to be ECOFF.
: 	(check_sections): einfo takes %B, not %P, to print a BFD name.

Approved,

Cheers
	Nick

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Jeffrey A Law]

  In message < 0006081756.AA18259@ivan.Harhan.ORG >you write:
  > the maintainer, and he told me I needed to file the copyright assignment. I
  > snailed the form to the FSF on 2000-05-02, but heard nothing back since the
  > Do I need to get a confirmation from them that my assignment is accepted
  > before I can get commit access? If so, whom and how do I bug? I think over
  > a month is long enough.
Yes, we need to be absolutely sure you've got a valid assignment before
allowing you to check in changes.

You can contact assign@gnu.org to find out if it's been recorded.
jeff

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[David O'Brien]

On Thu, Jun 08, 2000 at 12:56:17PM -0500, Michael Sokolov wrote:
> Nick Clifton <nickc@cygnus.com> wrote:
> > Approved,
> 
> Can you please apply it, I don't have commit access.

While I was in "commit mode", I took care of this.
 
-- 
-- David    (obrien@NUXI.com)

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Michael Sokolov]

Nick Clifton <nickc@cygnus.com> wrote:

> Approved,

Can you please apply it, I don't have commit access.

Speaking of commit access, I asked Ian for it at the end of April, when he was
the maintainer, and he told me I needed to file the copyright assignment. I
snailed the form to the FSF on 2000-05-02, but heard nothing back since then.
Do I need to get a confirmation from them that my assignment is accepted before
I can get commit access? If so, whom and how do I bug? I think over a month is
long enough.

--
Michael Sokolov		Harhan Engineering Laboratory
Public Service Agent	International Free Computing Task Force
			International Engineering and Science Task Force
			615 N GOOD LATIMER EXPY STE #4
			DALLAS TX 75204-5852 USA

Phone: +1-214-824-7693 (Harhan Eng Lab office)
E-mail: msokolov@ivan.Harhan.ORG (ARPA TCP/SMTP) (UUCP coming soon)

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Michael Sokolov]

Philip Blundell <pb@labs.futuretv.com> wrote:

> Not for the 2.10 release; I don't think this is critical enough.  It can go 
> into 2.10.1.

OK, then I only need it on the trunk for now. Nick?

--
Michael Sokolov		Harhan Engineering Laboratory
Public Service Agent	International Free Computing Task Force
			International Engineering and Science Task Force
			615 N GOOD LATIMER EXPY STE #4
			DALLAS TX 75204-5852 USA

Phone: +1-214-824-7693 (Harhan Eng Lab office)
E-mail: msokolov@ivan.Harhan.ORG (ARPA TCP/SMTP) (UUCP coming soon)

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Philip Blundell]

In message < 0006080140.AA17104@ivan.Harhan.ORG >, Michael Sokolov writes:
>I just realized that since my patch fixes a bug (of core dump type), it should
>go into binutils-2.10. Is there still any chance of getting it in?

Not for the 2.10 release; I don't think this is critical enough.  It can go 
into 2.10.1.

p.

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Michael Sokolov]

I just realized that since my patch fixes a bug (of core dump type), it should
go into binutils-2.10. Is there still any chance of getting it in?

--
Michael Sokolov		Harhan Engineering Laboratory
Public Service Agent	International Free Computing Task Force
			International Engineering and Science Task Force
			615 N GOOD LATIMER EXPY STE #4
			DALLAS TX 75204-5852 USA

Phone: +1-214-824-7693 (Harhan Eng Lab office)
E-mail: msokolov@ivan.Harhan.ORG (ARPA TCP/SMTP) (UUCP coming soon)

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Michael Sokolov]

Ian Lance Taylor <ian@zembu.com> wrote:

> Just check for bfd_target_ecoff_flavour here.  Don't make assumptions
> about the target name.

Alan Modra <alan@linuxcare.com.au> wrote:

> if (bfd_get_flavour (abfd) == bfd_target_ecoff_flavour)

Alan, you reversed the sense of the condition! :-)

The revised patch is below.

--
Michael Sokolov		Harhan Engineering Laboratory
Public Service Agent	International Free Computing Task Force
			International Engineering and Science Task Force
			615 N GOOD LATIMER EXPY STE #4
			DALLAS TX 75204-5852 USA

Phone: +1-214-824-7693 (Harhan Eng Lab office)
E-mail: msokolov@ivan.Harhan.ORG (ARPA TCP/SMTP) (UUCP coming soon)

2000-06-07  Michael Sokolov  <msokolov@ivan.Harhan.ORG>

	* emultempl/mipsecoff.em (gld{EMULATION_NAME}_after_open): Require all
	input objects to be ECOFF.
	(check_sections): einfo takes %B, not %P, to print a BFD name.

Index: emultempl/mipsecoff.em
===================================================================
RCS file: /cvs/src/src/ld/emultempl/mipsecoff.em,v
retrieving revision 1.4
diff -c -3 -p -r1.4 mipsecoff.em
*** mipsecoff.em	2000/02/29 05:53:53	1.4
--- mipsecoff.em	2000/06/08 01:06:32
*************** gld${EMULATION_NAME}_after_open ()
*** 69,74 ****
--- 69,79 ----
      {
        asection *datasec;
  
+       /* As first-order business, make sure that each input BFD is ECOFF. It
+          better be, as we are directly calling an ECOFF backend function.  */
+       if (bfd_get_flavour (abfd) != bfd_target_ecoff_flavour)
+ 	einfo ("%F%B: all input objects must be ECOFF for --embedded-relocs\n");
+ 
        datasec = bfd_get_section_by_name (abfd, ".sdata");
  
        /* Note that we assume that the reloc_count field has already
*************** check_sections (abfd, sec, sdatasec)
*** 113,119 ****
    if ((bfd_get_section_flags (abfd, sec) & SEC_CODE) == 0
        && sec != (asection *) sdatasec
        && sec->reloc_count != 0)
!     einfo ("%P%X: section %s has relocs; can not use --embedded-relocs\n",
  	   abfd, bfd_get_section_name (abfd, sec));
  }
  
--- 118,124 ----
    if ((bfd_get_section_flags (abfd, sec) & SEC_CODE) == 0
        && sec != (asection *) sdatasec
        && sec->reloc_count != 0)
!     einfo ("%B%X: section %s has relocs; can not use --embedded-relocs\n",
  	   abfd, bfd_get_section_name (abfd, sec));
  }
  

Re: PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Ian Lance Taylor]

   Date: Wed, 7 Jun 00 19:45:17 CDT
   From: msokolov@ivan.Harhan.ORG (Michael Sokolov)

   +       /* As first-order business, make sure that each input BFD is ECOFF. It
   +          better be, as we are directly calling an ECOFF backend function.  */
   +       if (strncmp (bfd_get_target (abfd), "ecoff", 5))
   + 	einfo ("%F%B: all input objects must be ECOFF for --embedded-relocs\n");
   + 

Just check for bfd_target_ecoff_flavour here.  Don't make assumptions
about the target name.

Ian

PATCH to ld fixes the MIPS ECOFF --embedded-relocs BFD check bug

[Michael Sokolov]

The patch below fixes the bug just discussed with Ian. While fixing it I also
fixed an obvious bug nearby under the obviously-correct rule.

Does this patch need a copyright assignment? I snailed my form to the FSF on
2000-05-02, but still no response. Does it always take this long? I'm in the
USA.

--
Michael Sokolov		Harhan Engineering Laboratory
Public Service Agent	International Free Computing Task Force
			International Engineering and Science Task Force
			615 N GOOD LATIMER EXPY STE #4
			DALLAS TX 75204-5852 USA

Phone: +1-214-824-7693 (Harhan Eng Lab office)
E-mail: msokolov@ivan.Harhan.ORG (ARPA TCP/SMTP) (UUCP coming soon)

2000-06-07  Michael Sokolov  <msokolov@ivan.Harhan.ORG>

	* emultempl/mipsecoff.em (gld{EMULATION_NAME}_after_open): Require all
	input objects to be ECOFF.
	(check_sections): einfo takes %B, not %P, to print a BFD name.

Index: emultempl/mipsecoff.em
===================================================================
RCS file: /cvs/src/src/ld/emultempl/mipsecoff.em,v
retrieving revision 1.4
diff -c -3 -p -r1.4 mipsecoff.em
*** mipsecoff.em	2000/02/29 05:53:53	1.4
--- mipsecoff.em	2000/06/08 00:29:31
*************** gld${EMULATION_NAME}_after_open ()
*** 69,74 ****
--- 69,79 ----
      {
        asection *datasec;
  
+       /* As first-order business, make sure that each input BFD is ECOFF. It
+          better be, as we are directly calling an ECOFF backend function.  */
+       if (strncmp (bfd_get_target (abfd), "ecoff", 5))
+ 	einfo ("%F%B: all input objects must be ECOFF for --embedded-relocs\n");
+ 
        datasec = bfd_get_section_by_name (abfd, ".sdata");
  
        /* Note that we assume that the reloc_count field has already
*************** check_sections (abfd, sec, sdatasec)
*** 113,119 ****
    if ((bfd_get_section_flags (abfd, sec) & SEC_CODE) == 0
        && sec != (asection *) sdatasec
        && sec->reloc_count != 0)
!     einfo ("%P%X: section %s has relocs; can not use --embedded-relocs\n",
  	   abfd, bfd_get_section_name (abfd, sec));
  }
  
--- 118,124 ----
    if ((bfd_get_section_flags (abfd, sec) & SEC_CODE) == 0
        && sec != (asection *) sdatasec
        && sec->reloc_count != 0)
!     einfo ("%B%X: section %s has relocs; can not use --embedded-relocs\n",
  	   abfd, bfd_get_section_name (abfd, sec));
  }