Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Joel Brobecker]

> Unfortunately I can't get it to work at all; I always get this error:
> 
>   $ ./x86_64-darwin/bin/gdb -c core.54937 myprog
>   GNU gdb (GDB) 7.6.2
>    ...
>   Reading symbols from myprog...Reading symbols from
> myprog.dSYM/Contents/Resources/DWARF/myprog...done.
>   done.
>   "core.54937": no core file handler recognizes format

I don't think we support core files on this platform.

-- 
Joel

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Klee Dienes]

I wrote the core dump support for GDB while I was working at NeXT/Apple, and would be happy to help you if you want to fold it back into stock GDB.  There should still be a copyright assignment on file for the time period it was written (but a lot could have changed since then, so it’s worth checking).  You can also build the Apple GDB from the last sources released at http://opensource.apple.com/source/gdb/gdb-1344/ — the core dump support seems to be present but not quite right. 

On Jan 30, 2014, at 9:09 PM, Joel Brobecker <brobecker@adacore.com> wrote:

>> Unfortunately I can't get it to work at all; I always get this error:
>> 
>>  $ ./x86_64-darwin/bin/gdb -c core.54937 myprog
>>  GNU gdb (GDB) 7.6.2
>>   ...
>>  Reading symbols from myprog...Reading symbols from
>> myprog.dSYM/Contents/Resources/DWARF/myprog...done.
>>  done.
>>  "core.54937": no core file handler recognizes format
> 
> I don't think we support core files on this platform.
> 
> -- 
> Joel

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Matt Rice]

On Thu, Jan 30, 2014 at 7:20 PM, Klee Dienes <klee@mit.edu> wrote:
> There should still be a copyright assignment on file for the time period it was written

I think we need to be wary that copyright on file for the time period
may not mean that the copyright for that specific piece of code was
assigned, and note apples unwillingness to do so in the past.

I don't suppose it exists in any branch, or was at least submitted for
inclusion where it can be had from the archives?

http://gcc.gnu.org/ml/gcc/2010-09/msg00136.html

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Klee Dienes]

I don’t work at Apple and don’t speak for them.  I can tell you that I was the person responsible for the initial copyright assignment from Apple for GDB, and that until I left Apple in 2005 it was my explicit job to try to merge code back to the FSF codebase under copyright assignment (I was not very successful, but that’s a separate issue).

I don’t have any of my records from before 2007, unfortunately, so it would take a bit of archaeology to say any more than that.  I don’t think we ever succeeded in pushing a branch to sourceware.  Since we were releasing via opensource.apple.com fairly often (and had live CVS available if I remember right), there wasn’t much need to push elsewhere.

On Jan 31, 2014, at 2:58 AM, Matt Rice <ratmice@gmail.com> wrote:

> On Thu, Jan 30, 2014 at 7:20 PM, Klee Dienes <klee@mit.edu> wrote:
>> There should still be a copyright assignment on file for the time period it was written
> 
> I think we need to be wary that copyright on file for the time period
> may not mean that the copyright for that specific piece of code was
> assigned, and note apples unwillingness to do so in the past.
> 
> I don't suppose it exists in any branch, or was at least submitted for
> inclusion where it can be had from the archives?
> 
> http://gcc.gnu.org/ml/gcc/2010-09/msg00136.html

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Stan Shebs]

On 1/30/14 11:58 PM, Matt Rice wrote:
> On Thu, Jan 30, 2014 at 7:20 PM, Klee Dienes <klee@mit.edu> wrote:

(Hi Klee!_

>> There should still be a copyright assignment on file for the time period it was written
> 
> I think we need to be wary that copyright on file for the time period
> may not mean that the copyright for that specific piece of code was
> assigned, and note apples unwillingness to do so in the past.

No, Apple had a blanket assignment to the FSF for all of their GNU work.
Presumably it's still in effect, but it's not something that could be
rescinded retroactively, so old code unaffected in any case.

(Nostalgia time - we actually had regular internal meetings on
open-source activities, and an in-house lawyer specifically assigned
to be the expert on legal aspects.)

Stan
stan@codesourcery.com

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Paul Smith]

On Fri, 2014-01-31 at 10:18 -0800, Stan Shebs wrote:
> On 1/30/14 11:58 PM, Matt Rice wrote:
> > On Thu, Jan 30, 2014 at 7:20 PM, Klee Dienes <klee@mit.edu> wrote:
> 
> (Hi Klee!_
> 
> >> There should still be a copyright assignment on file for the time
> period it was written
> > 
> > I think we need to be wary that copyright on file for the time period
> > may not mean that the copyright for that specific piece of code was
> > assigned, and note apples unwillingness to do so in the past.
> 
> No, Apple had a blanket assignment to the FSF for all of their GNU work.
> Presumably it's still in effect, but it's not something that could be
> rescinded retroactively, so old code unaffected in any case.

So the idea is that it should be possible to get the last GDB version
provided by Apple (this was GDB 6.3.x with Xcode 4.x I think), bring the
core dump handling code under FSF copyright due to the blanket
assignment, and port it to the newer GDB?

The latest GDB binary I have from Apple is GNU gdb 6.3.50-20050815
(Apple version gdb-1824) (from Xcode 4.6.3).  It looks like the source
for this version is available here:

http://www.opensource.apple.com/tarballs/gdb/gdb-1824.tar.gz

For now maybe I'll just build this version myself; it doesn't have all
the latest GDB goodness but it's got to work better than LLDB.

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Joseph S. Myers]

On Fri, 31 Jan 2014, Stan Shebs wrote:

> No, Apple had a blanket assignment to the FSF for all of their GNU work.

But did the code only get assigned by that assignment when it was actually 
submitted to the FSF by Apple, rather than simply by being created and 
published?

My understanding for GCC was that anything on the 4.0-based branch in the 
FSF repository was assigned by virtue of being committed to that 
repository, but the 4.2-based sources weren't (hence, in particular, 
implementation of blocks for GCC would need either something to be done 
about assignment of the 4.2-based code from Apple, or a complete 
reimplementation; cf <http://gcc.gnu.org/ml/gcc/2013-11/msg00039.html>) - 
and that it was generically the case for such assignments that changes 
became assigned at the point where they were submitted by or with the 
consent of the previous owner.

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Matt Rice]

On Fri, Jan 31, 2014 at 12:55 PM, Joseph S. Myers
<joseph@codesourcery.com> wrote:
> On Fri, 31 Jan 2014, Stan Shebs wrote:
>
>> No, Apple had a blanket assignment to the FSF for all of their GNU work.
>
> But did the code only get assigned by that assignment when it was actually
> submitted to the FSF by Apple, rather than simply by being created and
> published?

also wondering if copyright notices might play any role in this case e.g.
add a new file, add FSF copyright notices publish not commit, not sure what
copyright notices the code in question contains, but it may be pertinent.

Re: GDB w/ Mac OS cores: "no core file handler recognizes format"

[Tristan Gingold]

On 31 Jan 2014, at 03:09, Joel Brobecker <brobecker@adacore.com> wrote:

>> Unfortunately I can't get it to work at all; I always get this error:
>> 
>>  $ ./x86_64-darwin/bin/gdb -c core.54937 myprog
>>  GNU gdb (GDB) 7.6.2
>>   ...
>>  Reading symbols from myprog...Reading symbols from
>> myprog.dSYM/Contents/Resources/DWARF/myprog...done.
>>  done.
>>  "core.54937": no core file handler recognizes format
> 
> I don't think we support core files on this platform.

Indeed.  We initially wanted to implement it, but we realized that there is no way
to link a register context in the dump core with a thread.  So dump cores weren't
that useful in the context of Ada...

Now adding the support shouldn't be difficult.

Tristan.