gdb cvs vs x86_64-apple-darwin10

gdb cvs vs x86_64-apple-darwin10

[Jack Howarth]

   Has anyone tried building the current gdb cvs on
Snow Leopard. I was able to build gdb with...

../src/configure --disable-werror --build=x86_64-apple-darwin10 --target=x86_64-apple-darwin10 --target=x86_64-apple-darwin10
make

however the resulting gdb doesn't run properly. I get an error...

[Macintosh-2:~/gdb_cvs/dist/bin] howarth% ./gdb /Users/howarth/a.out
GNU gdb (GDB) 6.8.50.20090830-cvs
Copyright (C) 2009 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-apple-darwin10".
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>...
unable to read unknown load command 0x22
unable to read unknown load command 0x22
Reading symbols from /Users/howarth/a.out...(no debugging symbols found)...done.
(gdb) 

when I try to load an x86_64 executable created in gcc 4.4.1 with -g. Does
the x86_64-apple-darwin9 target work?
               Jack

Re: gdb cvs vs x86_64-apple-darwin10

[Tristan Gingold]

On Aug 30, 2009, at 10:20 PM, Jack Howarth wrote:

>   Has anyone tried building the current gdb cvs on
> Snow Leopard. I was able to build gdb with...
>
> ../src/configure --disable-werror --build=x86_64-apple-darwin10 -- 
> target=x86_64-apple-darwin10 --target=x86_64-apple-darwin10
> make
>
> however the resulting gdb doesn't run properly. I get an error...

I haven't yet installed snow leopard and I expect some adjustments as  
Apple frequently changes/updates/improves the Mach-O format.

Tristan.

Re: gdb cvs vs x86_64-apple-darwin10

[Jack Howarth]

On Mon, Aug 31, 2009 at 09:57:52AM +0200, Tristan Gingold wrote:
>
> On Aug 30, 2009, at 10:20 PM, Jack Howarth wrote:
>
>>   Has anyone tried building the current gdb cvs on
>> Snow Leopard. I was able to build gdb with...
>>
>> ../src/configure --disable-werror --build=x86_64-apple-darwin10 -- 
>> target=x86_64-apple-darwin10 --target=x86_64-apple-darwin10
>> make
>>
>> however the resulting gdb doesn't run properly. I get an error...
>
> I haven't yet installed snow leopard and I expect some adjustments as  
> Apple frequently changes/updates/improves the Mach-O format.
>
> Tristan.

Tristan,
   Getting x86_64-apple-darwin10 working should be a higher priority
for gdb 7.0 since it sounds like after the VTA branch merge into gcc
4.5, we will need to have access to gdb 7.0 to debug code generated
with gcc 4.5 on darwin.
                       Jack

Re: gdb cvs vs x86_64-apple-darwin10

[Jonas Maebe]

On 31 Aug 2009, at 15:22, Jack Howarth wrote:

>   Getting x86_64-apple-darwin10 working should be a higher priority
> for gdb 7.0 since it sounds like after the VTA branch merge into gcc
> 4.5, we will need to have access to gdb 7.0 to debug code generated
> with gcc 4.5 on darwin

Do you think Apple will ever ship anything >gcc 4.2 on Mac OS X? As  
far as I can tell, they are trying to move completely to Clang+LLVM.

Of course, maybe you build your own gcc (and AdaCore most certainly  
does build its own gcc).


Jonas

Re: gdb cvs vs x86_64-apple-darwin10

[Tristan Gingold]

On Aug 31, 2009, at 3:22 PM, Jack Howarth wrote:
> Tristan,
>   Getting x86_64-apple-darwin10 working should be a higher priority
> for gdb 7.0 since it sounds like after the VTA branch merge into gcc
> 4.5, we will need to have access to gdb 7.0 to debug code generated
> with gcc 4.5 on darwin.

Hi,

can you clarify your issue: is it Snow Leopard vs Leopard (as I  
thought first)
or is it fsf-gdb vs Apple gdb ?

I might be wrong but VTA concerns only optimized code, so we will be  
able to debug code generated
without optimizations.

Tristan.

Re: gdb cvs vs x86_64-apple-darwin10

[Jack Howarth]

On Mon, Aug 31, 2009 at 03:29:11PM +0200, Jonas Maebe wrote:
>
> On 31 Aug 2009, at 15:22, Jack Howarth wrote:
>
>>   Getting x86_64-apple-darwin10 working should be a higher priority
>> for gdb 7.0 since it sounds like after the VTA branch merge into gcc
>> 4.5, we will need to have access to gdb 7.0 to debug code generated
>> with gcc 4.5 on darwin
>
> Do you think Apple will ever ship anything >gcc 4.2 on Mac OS X? As far 
> as I can tell, they are trying to move completely to Clang+LLVM.
>
> Of course, maybe you build your own gcc (and AdaCore most certainly does 
> build its own gcc).
>
>
> Jonas

Jonas,
   No. I strongly doubt Apple will ever ship a gcc > 4.2.1 due to a combination
of their objections to GPLv3 and their preception that gcc has been treading
water in terms of improvements. As the FSF gcc developers are fond of saying,
it's hard to make the elephant dance so Apple wanted to start with a more
modern code base and design with clang/llvm. FYI, I maintain the gcc packages for
fink and so far we have had pretty good luck with FSF gcc maintaining quality...

http://gcc.gnu.org/ml/gcc-testresults/2009-07/msg01866.html

although there will be forks coming like with LTO. Apple wants to use a
different approach for implementing LTO than FSF gcc does. I was told by
the llvm folks...

> The FSF is designing their LTO to not need linker integration.  This
> will present a whole host of problems for them, but they are insistent
> on it.  :)

I am unclear if the FSF design could be adapted for darwin without Apple's
participation on the linker side.
                  Jack

Re: gdb cvs vs x86_64-apple-darwin10

[Jack Howarth]

On Mon, Aug 31, 2009 at 03:33:08PM +0200, Tristan Gingold wrote:
>
> On Aug 31, 2009, at 3:22 PM, Jack Howarth wrote:
>> Tristan,
>>   Getting x86_64-apple-darwin10 working should be a higher priority
>> for gdb 7.0 since it sounds like after the VTA branch merge into gcc
>> 4.5, we will need to have access to gdb 7.0 to debug code generated
>> with gcc 4.5 on darwin.
>
> Hi,
>
> can you clarify your issue: is it Snow Leopard vs Leopard (as I thought 
> first)
> or is it fsf-gdb vs Apple gdb ?
>
> I might be wrong but VTA concerns only optimized code, so we will be  
> able to debug code generated
> without optimizations.
>
> Tristan.

Tristan,
   If you follow the thread starting at...

http://gcc.gnu.org/ml/gcc-patches/2009-08/msg01629.html

you will see other platforms like AIX are concerned as well.
The best answer I have read so far...

http://gcc.gnu.org/ml/gcc-patches/2009-08/msg01656.html

...suggests that FSF gcc will 'try' to maintain compatibility
in the generated debug code but may not go out of their way
to keep hacks to work around bugs in the older version of
gdb.
                Jack