[PATCH] debugedit: add MIPS support

[PATCH] debugedit: add MIPS support

[Dmitry V. Levin]

From: "Ivan A. Melnikov" <iv@altlinux.org>
Date: Fri, 13 Nov 2020 15:51:08 +0400

According to the specification[1], all MIPS .debug_* sections are
marked with ELF type SHT_MIPS_DWARF.  The format of the section
data stays the same, so we have can handle e.g. .debug_info
section as we used to.

As SHT_MIPS_DWARF is from processor-specific range, we have to check
that we're actually dealing with MIPS ELF file before handling such
sections.

[1] MIPS Extensions to DWARF Version 2.0. -- Silicon Graphics
    Computer Systems, rev 1.17, 29 Aug 2001

Refs: https://bugzilla.altlinux.org/39284
Signed-off-by: Ivan A. Melnikov <iv@altlinux.org>
Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
---
 tools/debugedit.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/tools/debugedit.c b/tools/debugedit.c
index 7c57e16..917c59b 100644
--- a/tools/debugedit.c
+++ b/tools/debugedit.c
@@ -3440,6 +3440,16 @@ main (int argc, char *argv[])
 
       switch (dso->shdr[i].sh_type)
 	{
+	case SHT_MIPS_DWARF:
+	  /* According to the specification, all MIPS .debug_* sections are
+	     marked with ELF type SHT_MIPS_DWARF. As SHT_MIPS_DWARF is from
+	     processor-specific range, we have to check that we're actually
+	     dealing with MIPS ELF file before handling such sections.  */
+	  if (dso->ehdr.e_machine != EM_MIPS
+	      && dso->ehdr.e_machine != EM_MIPS_RS3_LE) {
+	    break;
+	  }
+	  /*@fallthrough@*/
 	case SHT_PROGBITS:
 	  name = strptr (dso, dso->ehdr.e_shstrndx, dso->shdr[i].sh_name);
 	  /* TODO: Handle stabs */
-- 
ldv

Re: [PATCH] debugedit: add MIPS support

[Mark Wielaard]

Hi Dmitry, Hi Ivan,

On Fri, Apr 02, 2021 at 08:00:00AM +0000, Dmitry V. Levin wrote:
> From: "Ivan A. Melnikov" <iv@altlinux.org>
> Date: Fri, 13 Nov 2020 15:51:08 +0400
> 
> According to the specification[1], all MIPS .debug_* sections are
> marked with ELF type SHT_MIPS_DWARF.  The format of the section
> data stays the same, so we have can handle e.g. .debug_info
> section as we used to.
> 
> As SHT_MIPS_DWARF is from processor-specific range, we have to check
> that we're actually dealing with MIPS ELF file before handling such
> sections.
> 
> [1] MIPS Extensions to DWARF Version 2.0. -- Silicon Graphics
>     Computer Systems, rev 1.17, 29 Aug 2001
> 
> Refs: https://bugzilla.altlinux.org/39284
> Signed-off-by: Ivan A. Melnikov <iv@altlinux.org>
> Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>

This looks correct. I have pushed it to main.  I don't have a MIPS
setup to check, but don't you also need some R_MIPS value for
setup_relbuf?

Thanks,

Mark

Re: [PATCH] debugedit: add MIPS support

[Ivan A. Melnikov]

On Sat, Apr 03, 2021 at 01:03:09AM +0200, Mark Wielaard wrote:
> Hi Dmitry, Hi Ivan,

Hi Mark,

Sorry for slow reply.

> On Fri, Apr 02, 2021 at 08:00:00AM +0000, Dmitry V. Levin wrote:
> > From: "Ivan A. Melnikov" <iv@altlinux.org>
> > Date: Fri, 13 Nov 2020 15:51:08 +0400
> > 
> > According to the specification[1], all MIPS .debug_* sections are
> > marked with ELF type SHT_MIPS_DWARF.  The format of the section
> > data stays the same, so we have can handle e.g. .debug_info
> > section as we used to.
> > 
> > As SHT_MIPS_DWARF is from processor-specific range, we have to check
> > that we're actually dealing with MIPS ELF file before handling such
> > sections.
> > 
> > [1] MIPS Extensions to DWARF Version 2.0. -- Silicon Graphics
> >     Computer Systems, rev 1.17, 29 Aug 2001
> > 
> > Refs: https://bugzilla.altlinux.org/39284
> > Signed-off-by: Ivan A. Melnikov <iv@altlinux.org>
> > Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
> 
> This looks correct. I have pushed it to main. 

Thank you.

> I don't have a MIPS setup to check, but don't you
> also need some R_MIPS value for setup_relbuf?

The thing is that there are several MIPS ABIs. Our
primary target is MIPS32 with o32 (traditional) ABI.
With this ABI, RELA relocations are not used,
even for kernel modules, so all the relocations
in .debug_info sections are "uniteresting", as the
comment in setup_relbuf calls them.

So, my patch is enough to support MIPS32 systems
with o32 ABI.

On the other hand, on MIPS64 systems with n64 ABI
(the most whidespread ABI on 64-bit MIPS) at least
4 types of RELA relocations are used. They,
indeed, should be added to setup_relbuf, and some
of them may even require additional handling.
I'm looking into that, but I haven't figured out
all the details yet. I'll send a patch here when
I'll have something.

-- 
  wbr,
    iv m.