* sh_link not set for section `.ARM.exidx'
@ 2007-08-07 13:41 Richard Purdie
2007-08-09 0:29 ` Alan Modra
0 siblings, 1 reply; 7+ messages in thread
From: Richard Purdie @ 2007-08-07 13:41 UTC (permalink / raw)
To: binutils
Hi,
I've been making ARM EABI builds with a toolchain with binutils
2.17.50.0.5. The build system splits debug symbols into separate
binaries by doing:
objcopy --only-keep-debug x .debug/x
strip x
objcopy --add-gnu-debuglink=.debug/x x
as per the objcopy manpage and which worked find under the previous ABI.
Under EABI each time it does this (and each time libbfd touches a debug
binary on device) I see:
BFD: x: warning: sh_link not set for section `.ARM.exidx'
The results from objdump look interesting:
objdump -x striped-binary:
12 .text 000358e0 0000a1b8 0000a1b8 000021b8 2**2
CONTENTS, ALLOC, LOAD, READONLY, CODE
13 .fini 00000010 0003fa98 0003fa98 00037a98 2**2
CONTENTS, ALLOC, LOAD, READONLY, CODE
14 .rodata 0000a328 0003faa8 0003faa8 00037aa8 2**3
CONTENTS, ALLOC, LOAD, READONLY, DATA
15 .ARM.exidx 00000008 00049dd0 00049dd0 00041dd0 2**2
CONTENTS, ALLOC, LOAD, READONLY, DATA
16 .eh_frame 00000004 00049dd8 00049dd8 00041dd8 2**2
CONTENTS, ALLOC, LOAD, READONLY, DATA
17 .init_array 00000004 00051f04 00051f04 00041f04 2**2
CONTENTS, ALLOC, LOAD, DATA
18 .fini_array 00000004 00051f08 00051f08 00041f08 2**2
CONTENTS, ALLOC, LOAD, DATA
objdump -x debug-binary:
12 .text 000358e0 0000a1b8 0000a1b8 00000134 2**2
ALLOC, READONLY, CODE
13 .fini 00000010 0003fa98 0003fa98 00000134 2**2
ALLOC, READONLY, CODE
14 .rodata 0000a328 0003faa8 0003faa8 00000134 2**3
ALLOC, READONLY
15 .ARM.exidx 00000008 00049dd0 00049dd0 00000134 2**2
CONTENTS, ALLOC, LOAD, READONLY, DATA
16 .eh_frame 00000004 00049dd8 00049dd8 00000134 2**2
ALLOC, READONLY
17 .init_array 00000004 00051f04 00051f04 00001f04 2**2
ALLOC
18 .fini_array 00000004 00051f08 00051f08 00001f04 2**2
since .ARM.exidx still has CONTENTS/DARA/LOAD flags yet its file offset
has changed to 00000134 which I'm guessing means it was emptied. I'm
unable to remove those flags with objcopy.
Is there anything I can do to clear this error up? I wondered about just
removing that section from the debug libraries but I don't know what
implications that might have and whether there are any references that
could cause problems. I would prefer to solve this properly...
Does anyone have any ideas?
Cheers,
Richard
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: sh_link not set for section `.ARM.exidx'
2007-08-07 13:41 sh_link not set for section `.ARM.exidx' Richard Purdie
@ 2007-08-09 0:29 ` Alan Modra
2007-08-09 8:35 ` Richard Purdie
0 siblings, 1 reply; 7+ messages in thread
From: Alan Modra @ 2007-08-09 0:29 UTC (permalink / raw)
To: Richard Purdie; +Cc: binutils
Please send me an arm-eabi binary that shows this problem and I'll
take a look.
--
Alan Modra
Australia Development Lab, IBM
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: sh_link not set for section `.ARM.exidx'
2007-08-09 0:29 ` Alan Modra
@ 2007-08-09 8:35 ` Richard Purdie
2007-08-09 12:09 ` Alan Modra
0 siblings, 1 reply; 7+ messages in thread
From: Richard Purdie @ 2007-08-09 8:35 UTC (permalink / raw)
To: Alan Modra; +Cc: binutils
On Thu, 2007-08-09 at 09:59 +0930, Alan Modra wrote:
> Please send me an arm-eabi binary that shows this problem and I'll
> take a look.
I've put an example binary online in case anyone else wants to look too:
http://www.rpsys.net/openzaurus/temp/binutils/apmd
and stripped/debug versions of this showing the problem:
http://www.rpsys.net/openzaurus/temp/binutils/apmd-debug
http://www.rpsys.net/openzaurus/temp/binutils/apmd-stripped
Thanks,
Richard
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: sh_link not set for section `.ARM.exidx'
2007-08-09 8:35 ` Richard Purdie
@ 2007-08-09 12:09 ` Alan Modra
2007-08-10 11:33 ` Richard Purdie
0 siblings, 1 reply; 7+ messages in thread
From: Alan Modra @ 2007-08-09 12:09 UTC (permalink / raw)
To: Richard Purdie; +Cc: binutils
On Thu, Aug 09, 2007 at 09:35:25AM +0100, Richard Purdie wrote:
> On Thu, 2007-08-09 at 09:59 +0930, Alan Modra wrote:
> > Please send me an arm-eabi binary that shows this problem and I'll
> > take a look.
>
> I've put an example binary online in case anyone else wants to look too:
>
> http://www.rpsys.net/openzaurus/temp/binutils/apmd
Looks to be no problem with latest CVS binutils.
--
Alan Modra
Australia Development Lab, IBM
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: sh_link not set for section `.ARM.exidx'
2007-08-09 12:09 ` Alan Modra
@ 2007-08-10 11:33 ` Richard Purdie
2007-08-10 12:51 ` Daniel Jacobowitz
0 siblings, 1 reply; 7+ messages in thread
From: Richard Purdie @ 2007-08-10 11:33 UTC (permalink / raw)
To: Alan Modra; +Cc: binutils
On Thu, 2007-08-09 at 21:38 +0930, Alan Modra wrote:
> On Thu, Aug 09, 2007 at 09:35:25AM +0100, Richard Purdie wrote:
> > On Thu, 2007-08-09 at 09:59 +0930, Alan Modra wrote:
> > > Please send me an arm-eabi binary that shows this problem and I'll
> > > take a look.
> >
> > I've put an example binary online in case anyone else wants to look too:
> >
> > http://www.rpsys.net/openzaurus/temp/binutils/apmd
>
> Looks to be no problem with latest CVS binutils.
Thanks for testing that, after quite a bit of effort I managed to build
a suitable cross binutils and can confirm it doesn't show the problem.
To make binutils cvs build I had to disable libgloss and rda. Are those
failures interesting or a known issue?
I can't see how rda could ever cross compile given its configure
script...
Cheers,
Richard
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: sh_link not set for section `.ARM.exidx'
2007-08-10 11:33 ` Richard Purdie
@ 2007-08-10 12:51 ` Daniel Jacobowitz
2007-08-10 15:24 ` Richard Purdie
0 siblings, 1 reply; 7+ messages in thread
From: Daniel Jacobowitz @ 2007-08-10 12:51 UTC (permalink / raw)
To: Richard Purdie; +Cc: Alan Modra, binutils
On Fri, Aug 10, 2007 at 12:32:41PM +0100, Richard Purdie wrote:
> Thanks for testing that, after quite a bit of effort I managed to build
> a suitable cross binutils and can confirm it doesn't show the problem.
>
> To make binutils cvs build I had to disable libgloss and rda. Are those
> failures interesting or a known issue?
>
> I can't see how rda could ever cross compile given its configure
> script...
You can check out just binutils, rather than the whole "src"
repository, or get a binutils snapshot from the FTP site - then those
(and many other directories) will not be included. They're part of
different programs.
--
Daniel Jacobowitz
CodeSourcery
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: sh_link not set for section `.ARM.exidx'
2007-08-10 12:51 ` Daniel Jacobowitz
@ 2007-08-10 15:24 ` Richard Purdie
0 siblings, 0 replies; 7+ messages in thread
From: Richard Purdie @ 2007-08-10 15:24 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: Alan Modra, binutils
On Fri, 2007-08-10 at 08:51 -0400, Daniel Jacobowitz wrote:
> You can check out just binutils, rather than the whole "src"
> repository, or get a binutils snapshot from the FTP site - then those
> (and many other directories) will not be included. They're part of
> different programs.
Thanks for the tip, I need bfd and some of the other bits but I've now
realised I can just remove the directories I don't want which is neater
than what I was trying before!
Cheers,
Richard
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2007-08-10 15:24 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-08-07 13:41 sh_link not set for section `.ARM.exidx' Richard Purdie
2007-08-09 0:29 ` Alan Modra
2007-08-09 8:35 ` Richard Purdie
2007-08-09 12:09 ` Alan Modra
2007-08-10 11:33 ` Richard Purdie
2007-08-10 12:51 ` Daniel Jacobowitz
2007-08-10 15:24 ` Richard Purdie
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).