Re: why no gcc3_compiled symbol

Re: why no gcc3_compiled symbol

[Robert Dewar]

> Not to mention the ironic part of all this, which is that other 
> compilers started adding  gcc2_compiled symbols too, defeating the 
> whole purpose of it in the first place.

it's like intel compatible processors pretending they are made  by intel :-)

Re: why no gcc3_compiled symbol

[John David Anglin]

> ... except, of course, that the file date inclusion breaks GCC bootstrap
> comparison (as you know) :)

I am sure there are ways to fix that if it was deemed useful to include
such information.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: why no gcc3_compiled symbol

[Nix]

On Thu, 1 Aug 2002, John David Anglin spake:
> HP appears to include file dates and compiler version as well.  For
> people seriously involved in software maintenance, this is obviously
> very useful information to have.

... except, of course, that the file date inclusion breaks GCC bootstrap
comparison (as you know) :)

-- 
`There's something satisfying about killing JWZ over and over again.'
                                        -- 1i, personal communication

Re: why no gcc3_compiled symbol

[John David Anglin]

> There's a wishlist PR, c/2676, asking for the command-line options to
> be recorded in the object files.  I implemented most of it, but there's
> no useful way of getting the information back out, because nobody ever
> finished writing the BFD/libbfd.a documentation.  So the patch rotted.

HP appears to include file dates and compiler version as well.  For
people seriously involved in software maintenance, this is obviously
very useful information to have.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: why no gcc3_compiled symbol

[Phil Edwards]

On Wed, Jul 31, 2002 at 08:21:32PM +2000, John David Anglin wrote:
> > Debuggers and related tools use it to identify the object.
> 
> I had a question today from a developer working on purify for
> hppa64-hpux.  He wanted to know how to identify a gcc compiled
> object.  As far as I know, there isn't any direct way.  The HP
> compiler/ assembler stick a bunch of stuff into .note.  However,
> gcc/binutils doesn't seem to do this.

There's a wishlist PR, c/2676, asking for the command-line options to
be recorded in the object files.  I implemented most of it, but there's
no useful way of getting the information back out, because nobody ever
finished writing the BFD/libbfd.a documentation.  So the patch rotted.


Phil

-- 
If ye love wealth greater than liberty, the tranquility of servitude greater
than the animating contest for freedom, go home and leave us in peace.  We seek
not your counsel, nor your arms.  Crouch down and lick the hand that feeds you;
and may posterity forget that ye were our countrymen.            - Samuel Adams

Re: why no gcc3_compiled symbol

[John David Anglin]

> Debuggers and related tools use it to identify the object.

I had a question today from a developer working on purify for
hppa64-hpux.  He wanted to know how to identify a gcc compiled
object.  As far as I know, there isn't any direct way.  The HP
compiler/ assembler stick a bunch of stuff into .note.  However,
gcc/binutils doesn't seem to do this.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: why no gcc3_compiled symbol

[Daniel Berlin]

On Wednesday, July 31, 2002, at 03:44  PM, Zack Weinberg wrote:

> On Wed, Jul 31, 2002 at 03:23:07PM -0400, Jack Howarth wrote:
>> ...what I was actually wondering was if anyone knew the original
>> intent of the gcc2_compiled symbol?  Was its presence intended to
>> keep object files from being empty of symbols or was there some
>> other reason for it?
>
> Its purpose was to indicate to GDB that an object file had been
> compiled by GCC 2 as opposed to some other compiler.  It turned
> out that GDB did not need this information, and the presence of
> the symbol caused other problems, so we removed it.

Not to mention the ironic part of all this, which is that other 
compilers started adding  gcc2_compiled symbols too, defeating the 
whole purpose of it in the first place.

>
> zw
>

Re: why no gcc3_compiled symbol

[Zack Weinberg]

On Wed, Jul 31, 2002 at 03:23:07PM -0400, Jack Howarth wrote:
> ...what I was actually wondering was if anyone knew the original
> intent of the gcc2_compiled symbol?  Was its presence intended to
> keep object files from being empty of symbols or was there some
> other reason for it?

Its purpose was to indicate to GDB that an object file had been
compiled by GCC 2 as opposed to some other compiler.  It turned
out that GDB did not need this information, and the presence of
the symbol caused other problems, so we removed it.

zw

Re: why no gcc3_compiled symbol

[Daniel Jacobowitz]

On Wed, Jul 31, 2002 at 03:23:07PM -0400, Jack Howarth wrote:
> Andreas,
>     Thats fine as I am currently testing proposed patches to the
> XFree86 loader for HJ. However what I was actually wondering was
> if anyone knew the original intent of the gcc2_compiled symbol?
> Was its presence intended to keep object files from being empty
> of symbols or was there some other reason for it?

Debuggers and related tools use it to identify the object.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer

Re: why no gcc3_compiled symbol

[Jack Howarth]

Andreas,
    Thats fine as I am currently testing proposed patches to the
XFree86 loader for HJ. However what I was actually wondering was
if anyone knew the original intent of the gcc2_compiled symbol?
Was its presence intended to keep object files from being empty
of symbols or was there some other reason for it?
                        Jack

Re: why no gcc3_compiled symbol

[Andreas Schwab]

Jack Howarth <howarth@bromo.msbb.uc.edu> writes:

|>      Does anyone know why gcc >= 3.0 doesn't have a gcc3_compiled
|> symbol like gcc 2.9x had a gcc2_compiled symbol? I ask because in
|> tracking down a problem with a gcc 3.1.1 built XFree86 loading its
|> libGLcore.a xserver module we found gcc 3.1.1 will create completely
|> empty object files. This compares to gcc 2.95.4 or 2.96 which would
|> always leave a gcc2_compiled placekeeper symbol in an empty object
|> file. This is arose because when you strip a gcc 3.1.1 built 
|> static lib that has such an empty object file linked in you get
|> a object file with no symbols at all. This breaks the current
|> XFree86 loader.

Then the XFree86 loader is broken, IMHO.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

why no gcc3_compiled symbol

[Jack Howarth]

     Does anyone know why gcc >= 3.0 doesn't have a gcc3_compiled
symbol like gcc 2.9x had a gcc2_compiled symbol? I ask because in
tracking down a problem with a gcc 3.1.1 built XFree86 loading its
libGLcore.a xserver module we found gcc 3.1.1 will create completely
empty object files. This compares to gcc 2.95.4 or 2.96 which would
always leave a gcc2_compiled placekeeper symbol in an empty object
file. This is arose because when you strip a gcc 3.1.1 built 
static lib that has such an empty object file linked in you get
a object file with no symbols at all. This breaks the current
XFree86 loader. Thanks in advance for any explanation.
                          Jack
ps This can be seen by...

#touch x.c
#gcc-2.95 -c x.c
#nm x.o
00000000 t gcc2_compiled.
#strip --strip-debug x.o
#nm x.o
00000000 t gcc2_compiled.

compared to

#touch x.c
#gcc-3.1 -c x.c
#nm x.o
#
#strip --strip-debug x.o
#nm x.o
nm: x.o: no symbols

Do we really want to create completely empty object files in gcc 3.1?