which gcc version?

which gcc version?

[Manoj Sontakke]

Hi
	I am getting linker error "undefined reference to __divdi3".
This is because c = a/b; where a,b,c are of type "long long"
I understand this is gcc problem.
	I am doing this on a pentium with gcc -v = egcs-2.91.66

Thanks for all the help.

Manoj

Re: which gcc version?

[Alexander Viro]

On Thu, 5 Apr 2001, Manoj Sontakke wrote:

> Hi
> 	I am getting linker error "undefined reference to __divdi3".
> This is because c = a/b; where a,b,c are of type "long long"
> I understand this is gcc problem.
> 	I am doing this on a pentium with gcc -v = egcs-2.91.66

Don't do it in the kernel. It has nothing to gcc version.

Re: which gcc version?

[Manoj Sontakke]

hi

On Thu, 5 Apr 2001, Alexander Viro wrote:
> On Thu, 5 Apr 2001, Manoj Sontakke wrote:
> 
> > Hi
> > 	I am getting linker error "undefined reference to __divdi3".
> > This is because c = a/b; where a,b,c are of type "long long"
> > I understand this is gcc problem.
> > 	I am doing this on a pentium with gcc -v = egcs-2.91.66
> 
> Don't do it in the kernel. It has nothing to gcc version.

Addition and subtraction works fine. The problem is with multiplication
and division. I am doing this to avoid floating point calculation and
doing fixed point calculation. The rage is large enough to need "long
long" Any other way to achieve this?

thanks
manoj

Re: which gcc version?

[Bart Trojanowski]

On Thu, 5 Apr 2001, Manoj Sontakke wrote:

> Addition and subtraction works fine. The problem is with multiplication
> and division. I am doing this to avoid floating point calculation and
> doing fixed point calculation. The rage is large enough to need "long
> long" Any other way to achieve this?

gcc requires a function call to do a mul/div on a long long.  There is no
easy way to do a 64bit op of this type on a 32 bit CPU...

IFF you are making a module that will never be part of the kernel tree you
can get away with adding `gcc -print-libgcc-file-name` to your link line.
And by that I mean:

	${LD} ${LDFLAGS} -o $@ $< `gcc -print-libgcc-file-name`

This is a hack that will link in the __div3 function from glibc into your
module.  NOTE that this will never be done for you in the kernel... so if
you can don't use div or mul on long long.

Cheers,
Bart.

-- 
	WebSig: http://www.jukie.net/~bart/sig/

Re: which gcc version?

[Matti Aarnio]

On Thu, Apr 05, 2001 at 07:37:03AM -0400, Bart Trojanowski wrote:
> On Thu, 5 Apr 2001, Manoj Sontakke wrote:
> gcc requires a function call to do a mul/div on a long long.  There is no
> easy way to do a 64bit op of this type on a 32 bit CPU...

	Actually there is -- presuming your CPU has lots of registers,
	which i386 does not have...

	This rule -- "no -lgcc in kernel code" -- is to trap code which
	in fast-paths would call those slow routines.
	Fastpaths should either not grok such values, or at most use
	shifts.

	To think of it, there really should be explicitely callable
	versions of these with LinuxKernel names for them, not gcc
	builtins.   That way people would *know* they are doing
	something, which is potentially very slow.
	(And the API would not change from underneath them.)

> Cheers,
> Bart.
> -- 
> 	WebSig: http://www.jukie.net/~bart/sig/

/Matti Aarnio

Re: which gcc version?

[David Woodhouse]

matti.aarnio@zmailer.org said:
> 	To think of it, there really should be explicitely callable
> 	versions of these with LinuxKernel names for them, not gcc
> 	builtins.   That way people would *know* they are doing
> 	something, which is potentially very slow.
> 	(And the API would not change from underneath them.) 

Like include/asm-*/div64.h::do_div()?

--
dwmw2

Re: which gcc version?

[Matti Aarnio]

On Thu, Apr 05, 2001 at 01:26:32PM +0100, David Woodhouse wrote:
> matti.aarnio@zmailer.org said:
> > 	To think of it, there really should be explicitely callable
> > 	versions of these with LinuxKernel names for them, not gcc
> > 	builtins.   That way people would *know* they are doing
> > 	something, which is potentially very slow.
> > 	(And the API would not change from underneath them.) 
> 
> Like include/asm-*/div64.h::do_div()?

	Like that (*) - but that has limited value spaces,
	the divider can be at most 32 bits, probably far less.

> --
> dwmw2

/Matti Aarnio

(*) Trying to recall when I, and few others, created that beast.
    The divisor code in itself isn't mine, but the idea of supporting
    %Ld at printk() most definitely is..

Re: which gcc version?

[Richard B. Johnson]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: TEXT/PLAIN; charset=US-ASCII, Size: 1362 bytes --]

On Thu, 5 Apr 2001, David Woodhouse wrote:

> 
> matti.aarnio@zmailer.org said:
> > 	To think of it, there really should be explicitely callable
> > 	versions of these with LinuxKernel names for them, not gcc
> > 	builtins.   That way people would *know* they are doing
> > 	something, which is potentially very slow.
> > 	(And the API would not change from underneath them.) 
> 
> Like include/asm-*/div64.h::do_div()?
> 

Some, perhaps all, of the recent requirements for a 64 bit division
can be handled with the (Intel) 32-bit division because only the
dividend is 64 bits, and the division can be unsigned, i.e.

HIGHLONG = 0x08		; Depends upon caller's stack.
LOWLONG  = 0x0C
DIVISOR  = 0x10

		movl	HIGHLONG(%esp), %edx
		movl	LOWLLONG(%esp), %eax
		movl	DIVISOR(%esp), %ecx
		divl	%ecx,%eax

... returns a longword in %eax

This could be an __inline__ function. Most other stuff can be reduced
with shifts before a 32-bit divide.

64 bit signed division is a bitch in any 32-bit machine because it's
done just line 4th grade long division. There are remainders that have
to carried, etc.


Cheers,
Dick Johnson

Penguin : Linux version 2.4.1 on an i686 machine (799.53 BogoMips).

"Memory is like gasoline. You use it up when you are running. Of
course you get it all back when you reboot..."; Actual explanation
obtained from the Micro$oft help desk.

Re: which gcc version?

[Matti Aarnio]

On Thu, Apr 05, 2001 at 07:39:14PM +0530, Manoj Sontakke wrote:
> hi
> 
> On Thu, 5 Apr 2001, Alexander Viro wrote:
> > On Thu, 5 Apr 2001, Manoj Sontakke wrote:
> > 
> > > Hi
> > > 	I am getting linker error "undefined reference to __divdi3".
> > > This is because c = a/b; where a,b,c are of type "long long"
> > > I understand this is gcc problem.
> > > 	I am doing this on a pentium with gcc -v = egcs-2.91.66
> > 
> > Don't do it in the kernel. It has nothing to gcc version.
> 
> Addition and subtraction works fine. The problem is with multiplication
> and division. I am doing this to avoid floating point calculation and
> doing fixed point calculation. The rage is large enough to need "long
> long" Any other way to achieve this?

	How arbitrary is the divisor ?
	Not by change a power-of-two ?

	If it is arbitrary, how large ?
	The printk() contains support routine doing  longlong/int
	division when a) speed is *not* important, b) divisor is
	sufficiently small, and I think c) divisor is positive.

	When you look at that code, you should begin to grasp,
	what kind of monster is the longlong/longlong code at -lgcc ...
	(and why gcc doesn't produce inline code for it.)

> thanks
> manoj

/Matti Aarnio

Re: which gcc version?

[christophe barbe]

Which kernel are you using ?
GFS use this kind of computation. And with kernel 2.2, a module divdi3.o provides all missing symbol like __divdi3.

Christophe

On jeu, 05 avr 2001 16:09:14 Manoj Sontakke wrote:
> hi
> 
> On Thu, 5 Apr 2001, Alexander Viro wrote:
> > On Thu, 5 Apr 2001, Manoj Sontakke wrote:
> > 
> > > Hi
> > > 	I am getting linker error "undefined reference to __divdi3".
> > > This is because c = a/b; where a,b,c are of type "long long"
> > > I understand this is gcc problem.
> > > 	I am doing this on a pentium with gcc -v = egcs-2.91.66
> > 
> > Don't do it in the kernel. It has nothing to gcc version.
> 
> Addition and subtraction works fine. The problem is with multiplication
> and division. I am doing this to avoid floating point calculation and
> doing fixed point calculation. The rage is large enough to need "long
> long" Any other way to achieve this?
> 
> thanks
> manoj
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 
-- 
Christophe Barbé
Software Engineer
Lineo High Availability Group
42-46, rue Médéric
92110 Clichy - France
phone (33).1.41.40.02.12
fax (33).1.41.40.02.01
www.lineo.com

Which gcc version?

[Roy Sigurd Karlsbakk]

hi all

I just wonder...
With a clean rh72 install, I've got two gcc versions installed in parllel, 
2.96 and 3.0.2. Which one should I use to compile the kernel?

thanks

roy
--
Roy Sigurd Karlsbakk, MCSE, MCNE, CLS, LCA

Computers are like air conditioners.
They stop working when you open Windows.

Re: Which gcc version?

[Arjan van de Ven]

Roy Sigurd Karlsbakk wrote:
> 
> hi all
> 
> I just wonder...
> With a clean rh72 install, I've got two gcc versions installed in parllel,
> 2.96 and 3.0.2. Which one should I use to compile the kernel?

> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

See this last line.
That URL has a nice description..... Also just see Documentation/Changes

Re: Which gcc version?

[war]

You should use gcc-2.95.3.

Roy Sigurd Karlsbakk wrote:

> hi all
>
> I just wonder...
> With a clean rh72 install, I've got two gcc versions installed in parllel,
> 2.96 and 3.0.2. Which one should I use to compile the kernel?
>
> thanks
>
> roy
> --
> Roy Sigurd Karlsbakk, MCSE, MCNE, CLS, LCA
>
> Computers are like air conditioners.
> They stop working when you open Windows.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Which gcc version?

[J Sloan]

Actually there is no need to go download yet
another compiler -

The gcc-2.96 shipped with redhat will work
perfectly for compiling your kernel -

See Documentation/Changes, line 87 for a
heads-up

cu

jjs

war wrote:

> You should use gcc-2.95.3.
>
> Roy Sigurd Karlsbakk wrote:
>
> > hi all
> >
> > I just wonder...
> > With a clean rh72 install, I've got two gcc versions installed in parllel,
> > 2.96 and 3.0.2. Which one should I use to compile the kernel?

Re: Which gcc version?

[Anton Altaparmakov]

At 13:51 23/11/01, war wrote:
>You should use gcc-2.95.3.

That's not true. gcc-2.96 as provided with RedHat 7.2 is perfectly fine.

gcc-3x OTOH is not a good idea at the moment.

Anton


>Roy Sigurd Karlsbakk wrote:
>
> > hi all
> >
> > I just wonder...
> > With a clean rh72 install, I've got two gcc versions installed in parllel,
> > 2.96 and 3.0.2. Which one should I use to compile the kernel?
> >
> > thanks
> >
> > roy
> > --
> > Roy Sigurd Karlsbakk, MCSE, MCNE, CLS, LCA
> >
> > Computers are like air conditioners.
> > They stop working when you open Windows.
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/
>
>-
>To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>Please read the FAQ at  http://www.tux.org/lkml/

-- 
   "I've not lost my mind. It's backed up on tape somewhere." - Unknown
-- 
Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
Linux NTFS Maintainer / WWW: http://linux-ntfs.sf.net/
ICQ: 8561279 / WWW: http://www-stu.christs.cam.ac.uk/~aia21/

Re: Which gcc version?

[war]

#1) The compiler from redhat (gcc-2.96) is not an official GNU release.
#2) http://www.atnf.csiro.au/~rgooch/linux/docs/kernel-newsflash.html/
      "the reccomend compiler is now gcc-2.95.3, rather than gcc-2.91.66"


Anton Altaparmakov wrote:

> At 13:51 23/11/01, war wrote:
> >You should use gcc-2.95.3.
>
> That's not true. gcc-2.96 as provided with RedHat 7.2 is perfectly fine.
>
> gcc-3x OTOH is not a good idea at the moment.
>
> Anton
>
> >Roy Sigurd Karlsbakk wrote:
> >
> > > hi all
> > >
> > > I just wonder...
> > > With a clean rh72 install, I've got two gcc versions installed in parllel,
> > > 2.96 and 3.0.2. Which one should I use to compile the kernel?
> > >
> > > thanks
> > >
> > > roy
> > > --
> > > Roy Sigurd Karlsbakk, MCSE, MCNE, CLS, LCA
> > >
> > > Computers are like air conditioners.
> > > They stop working when you open Windows.
> > > -
> > > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > > the body of a message to majordomo@vger.kernel.org
> > > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > > Please read the FAQ at  http://www.tux.org/lkml/
> >
> >-
> >To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> >the body of a message to majordomo@vger.kernel.org
> >More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >Please read the FAQ at  http://www.tux.org/lkml/
>
> --
>    "I've not lost my mind. It's backed up on tape somewhere." - Unknown
> --
> Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
> Linux NTFS Maintainer / WWW: http://linux-ntfs.sf.net/
> ICQ: 8561279 / WWW: http://www-stu.christs.cam.ac.uk/~aia21/
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Which gcc version?

[Arjan van de Ven]

war wrote:
> 
> #1) The compiler from redhat (gcc-2.96) is not an official GNU release.
> #2) http://www.atnf.csiro.au/~rgooch/linux/docs/kernel-newsflash.html/
>       "the reccomend compiler is now gcc-2.95.3, rather than gcc-2.91.66"

>From Documentation/Changes in 2.4.15:

The Red Hat gcc 2.96 compiler subtree can also be used to build this
tree.
You should ensure you use gcc-2.96-74 or later. gcc-2.96-54 will not
build
the kernel correctly.

Re: Which gcc version?

[Gábor Lénárt]

On Fri, Nov 23, 2001 at 03:14:48PM +0000, Arjan van de Ven wrote:
> war wrote:
> > 
> > #1) The compiler from redhat (gcc-2.96) is not an official GNU release.
> > #2) http://www.atnf.csiro.au/~rgooch/linux/docs/kernel-newsflash.html/
> >       "the reccomend compiler is now gcc-2.95.3, rather than gcc-2.91.66"
> 
> From Documentation/Changes in 2.4.15:
> 
> The Red Hat gcc 2.96 compiler subtree can also be used to build this
> tree.
> You should ensure you use gcc-2.96-74 or later. gcc-2.96-54 will not
> build
> the kernel correctly.

True, but as it's known, gcc-2.96 is NOT an official gcc release by the gcc
team. It was RedHat's fault to fetch a development CVS gcc snapshot and
release it as gcc 2.96 in RedHat distributions, while object format used by
2.96 is not compatible with 2.95 nor 3.0.x at least according information
can be found on site of gcc. It was very ROTFL RedHat to release kgcc to be
able to compile kernel. And these type of distributions are marked as even
enterprise-ready and likes by RedHat :) Sorry for the flame, but IMHO it's
very funny :) [Also, while developing MPlayer we had got problems with even
newer 2.96's, so we do not recommend it in the dox, and ./configure won't
able you to use 2.96 without a special configure switch ...]

But for being more serious: do newer gcc 2.96 versions have these type of
errors? I mean mainly incompatible object format ...

- Gabor

Re: Which gcc version?

[Arjan van de Ven]

On Fri, Nov 23, 2001 at 04:30:19PM +0100, Gábor Lénárt wrote:
> True, but as it's known, gcc-2.96 is NOT an official gcc release by the gcc
> team. It was RedHat's fault to fetch a development CVS gcc snapshot and
> release it as gcc 2.96 in RedHat distributions, while object format used by
> 2.96 is not compatible with 2.95 nor 3.0.x at least according information
> can be found on site of gcc.

That is only the case for C++. And the same is true for egcs -> 2.95.x and
from 2.95.x -> 3.0.x and from 3.0.x -> 3.1

> It was very ROTFL RedHat to release kgcc to be
> able to compile kernel.

gcc 3.0 doesn't compile 2.2 kernels. "2.96" is like 3.0 in that respect. All
2.4 kernel RPMs from Red Hat are compiled with the "2.96" gcc.

> [Also, while developing MPlayer we had got problems with even
> newer 2.96's, so we do not recommend it in the dox, and ./configure won't
> able you to use 2.96 without a special configure switch ...]

I noticed yes. Most likely a case of broken inline asm..... 
gcc 3 will break just as much

Re: Which gcc version?

[J Sloan]

This is all silly FUD - time for the pointer again -

http://www.bero.org/gcc296.html

cu

jjs

Gábor Lénárt wrote:

> True, but as it's known, gcc-2.96 is NOT an official gcc release by the gcc
> team. It was RedHat's fault to fetch a development CVS gcc snapshot and
> release it as gcc 2.96 in RedHat distributions, while object format used by
> 2.96 is not compatible with 2.95 nor 3.0.x at least according information
> can be found on site of gcc. It was very ROTFL RedHat to release kgcc to be
> able to compile kernel. And these type of distributions are marked as even
> enterprise-ready and likes by RedHat :) Sorry for the flame, but IMHO it's
> very funny :) [Also, while developing MPlayer we had got problems with even
> newer 2.96's, so we do not recommend it in the dox, and ./configure won't
> able you to use 2.96 without a special configure switch ...]
>

Re: Which gcc version?

[John Jasen]

On Fri, 23 Nov 2001, J Sloan wrote:

> This is all silly FUD - time for the pointer again -
>
> http://www.bero.org/gcc296.html

Redhat apologists are given the due they deserve.

Their first cut at a released 2.96 should never have made it out of the
labs alive, and for that matter, neither should have 7.0.

That said, the versions in 7.1, and the patches available have done
reasonably well for everything that I've tried to compile (kernels
included).

What are they going to do about gcc 3.0.x? It'll be amusing to watch.

--
-- John E. Jasen (jjasen1@umbc.edu)
-- In theory, theory and practise are the same. In practise, they aren't.

Re: Which gcc version?

[J Sloan]

John Jasen wrote:

> Their first cut at a released 2.96 should never have made it out of the
> labs alive, and for that matter, neither should have 7.0.

Mistakes are made by all - redhat fixed theirs,
that's good enough for me.

> That said, the versions in 7.1, and the patches available have done
> reasonably well for everything that I've tried to compile (kernels
> included).

indeed -

> What are they going to do about gcc 3.0.x? It'll be amusing to watch.

Not sure you mean "what are they going to do"
since gcc 3.0.x is included in redhat 7.2 - could
you explain about the amusing part?

;-)

cu

jjs

Re: Which gcc version?

[John Jasen]

On Fri, 23 Nov 2001, J Sloan wrote:

> > What are they going to do about gcc 3.0.x? It'll be amusing to watch.
>
> Not sure you mean "what are they going to do"
> since gcc 3.0.x is included in redhat 7.2 - could
> you explain about the amusing part?

I got into a long argument with someone from Redhat, who suggested that I
read too many ./ postings, and gained my opinion of gcc 2.96 as shipped in
7.0 from that.

Where my opinion of it was formed by running it on my laptop for T&E for
two weeks, and having about a 25% success ratio of compiling code under
it. Well, okay ... 50% for compilation, and of that, 50% running without
catching on fire.

In that conversation, he assured me that RH would take every measure to
ensure that future releases were compatible with code generated under
2.96.

So, it could be amusing ...

--
-- John E. Jasen (jjasen1@umbc.edu)
-- In theory, theory and practise are the same. In practise, they aren't.

Re: Which gcc version?

[Paul G. Allen]

John Jasen wrote:
> 
> On Fri, 23 Nov 2001, J Sloan wrote:
> 
> > This is all silly FUD - time for the pointer again -
> >
> > http://www.bero.org/gcc296.html
> 
> Redhat apologists are given the due they deserve.
> 
> Their first cut at a released 2.96 should never have made it out of the
> labs alive, and for that matter, neither should have 7.0.
> 
> That said, the versions in 7.1, and the patches available have done
> reasonably well for everything that I've tried to compile (kernels
> included).
> 
> What are they going to do about gcc 3.0.x? It'll be amusing to watch.
> 

I installed RH 7.0 when it was released and was sorely dissappointed.
The compiler never worked for anything I tried to compile (and seeing as
how programming was 50% of my job, THAT was a MAJOR issue). I quickly
went back to RH 6.2.

That said, I now run RH 7.1 and 7.2 (depending upon the box, this one
runs 7.1 w/kernel 2.4.14).

[root@keroon /root]# gcc -v
Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/2.96/specs
gcc version 2.96 20000731 (Red Hat Linux 7.1 2.96-85)
[root@keroon /root]# 

Works just fine for everything. The problem with the RH 7.1 release is
with make, not gcc. make 3.79.x has bugs that cause, among other things,
internal assertion errors. Though it works fine for the kernel, it will
not work for many other programs (see disscussion groups, e.g. -
gnu.org, for more info). So I back-rev'd that to 3.78.1-4:

[root@keroon /root]# make -v
GNU Make version 3.78.1, by Richard Stallman and Roland McGrath.
Built for i386-redhat-linux-gnu
Copyright (C) 1988, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99
	Free Software Foundation, Inc.
This is free software; see the source for copying conditions.
There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE.

Report bugs to <bug-make@gnu.org>.

[root@keroon /root]# 


and everything works just peachy. I compile everything for Athlon, and
it still works fine.

PGA
-- 
Paul G. Allen
UNIX Admin II/Network Security
Akamai Technologies, Inc.
www.akamai.com

Re: Which gcc version?

[Alan Cox]

> That said, the versions in 7.1, and the patches available have done
> reasonably well for everything that I've tried to compile (kernels
> included).
> 
> What are they going to do about gcc 3.0.x? It'll be amusing to watch.

Its shipped as an additional compiler as well. The only real question now is
how long before 3.x becomes really solid. That should be in time for the
next RH major update I hope

RE: Which gcc version?

[Ahmed Masud]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> > That said, the versions in 7.1, and the patches available have
> > done  reasonably well for everything that I've tried to compile
> > (kernels  included).
> > 
> > What are they going to do about gcc 3.0.x? It'll be amusing 
> to watch.
> 
> Its shipped as an additional compiler as well. The only real 
> question now is how long before 3.x becomes really solid. 
> That should be in time for the next RH major update I hope

In case this is pertinent: I have been using gcc 3.0.2 since its
inception to compile the kernel on various CPUs (Pentium II, Pentium
III, AMD Athelon, and also the experimental stuff for PA/RISC 2.0)
and have not had any trouble with it.   Seems to be fairly solid and
trouble free.  RH has to get its act together though ... The gcc 2.96
and a kgcc for the kernel was just too hilarious and rather sad at
the same time.

Cheers.

-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 6.5.3 for non-commercial use <http://www.pgp.com>

iQA/AwUBPANR7eA+WVFT6/r4EQK/OwCg+X2va7QIzbwDvRBv27Kf/AXmy3UAn1/h
AvFgqGDvS0iv7RQXMprUIJB8
=Hs5v
-----END PGP SIGNATURE-----

Re: Which gcc version?

[Anton Altaparmakov]

At 15:30 23/11/01, =?iso-8859-2?B?R+Fib3IgTOlu4XJ0?= wrote:
>On Fri, Nov 23, 2001 at 03:14:48PM +0000, Arjan van de Ven wrote:
> > war wrote:
> > >
> > > #1) The compiler from redhat (gcc-2.96) is not an official GNU release.
> > > #2) http://www.atnf.csiro.au/~rgooch/linux/docs/kernel-newsflash.html/
> > >       "the reccomend compiler is now gcc-2.95.3, rather than gcc-2.91.66"
> >
> > From Documentation/Changes in 2.4.15:
> >
> > The Red Hat gcc 2.96 compiler subtree can also be used to build this
> > tree.
> > You should ensure you use gcc-2.96-74 or later. gcc-2.96-54 will not
> > build
> > the kernel correctly.
>
>True, but as it's known, gcc-2.96 is NOT an official gcc release by the gcc
>team. It was RedHat's fault to fetch a development CVS gcc snapshot and
>release it as gcc 2.96 in RedHat distributions, while object format used by
>2.96 is not compatible with 2.95 nor 3.0.x at least according information
>can be found on site of gcc. It was very ROTFL RedHat to release kgcc to be
>able to compile kernel. And these type of distributions are marked as even
>enterprise-ready and likes by RedHat :) Sorry for the flame, but IMHO it's
>very funny :) [Also, while developing MPlayer we had got problems with even
>newer 2.96's, so we do not recommend it in the dox, and ./configure won't
>able you to use 2.96 without a special configure switch ...]

Oh, I have to look at your configure.in for how you do that so I can add 
the opposite check to linux-ntfs and refuse to compile if gcc-2.96 or 3.x 
is NOT used. (-:

>But for being more serious: do newer gcc 2.96 versions have these type of
>errors? I mean mainly incompatible object format ...

I don't know about that. All I can say is that since RedHat 7.0 and the 
updated gcc-2.96 (the first release was indeed broken) I have been using 
gcc-2.96 for everything and I haven't experienced any problems with it 
(compiling both kernels and user space code with it but YMMV as I don't 
compile that much user space code myself except for my own code which 
obviously works considering I write it for gcc-2.96...). I even installed 
gcc-2.96 on my SuSE 7.2 system and that has been happy with it, too.

Anton


-- 
   "I've not lost my mind. It's backed up on tape somewhere." - Unknown
-- 
Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
Linux NTFS Maintainer / WWW: http://linux-ntfs.sf.net/
ICQ: 8561279 / WWW: http://www-stu.christs.cam.ac.uk/~aia21/

Re: Which gcc version?

[Anton Altaparmakov]

At 14:11 23/11/01, war wrote:
>#1) The compiler from redhat (gcc-2.96) is not an official GNU release.

And anyone should care because...?

>#2) http://www.atnf.csiro.au/~rgooch/linux/docs/kernel-newsflash.html/
>       "the reccomend compiler is now gcc-2.95.3, rather than gcc-2.91.66"

Several main kernel developers use gcc-2.96 for their kernel development 
work and according to Alan Cox using gcc-2.96 for the kernel is fine (from 
a certain version onwards, can't remember the minimum release number he 
said but the one in RH 7.2 is fine in any case).

Considering the person asking was giving gcc-2.96 and gcc-3.x as choices, 
and was saying he is using RH 7.2, answering he should be using a third 
choice is IMO not a very constructive answer, considering one of the 
original choices is fine!

And anyway, the ultimate guide is the Documentation/Changes document in the 
kernel tree which states (cut'n'paste from kernel 2.4.15):

"The Red Hat gcc 2.96 compiler subtree can also be used to build this tree.
You should ensure you use gcc-2.96-74 or later. gcc-2.96-54 will not build
the kernel correctly."

So while you are right that gcc-2.95.3 or .4 is the recommended compiler, 
gcc-2.96 if an officially supported compiler, too. (-;

Best regards,

Anton

>Anton Altaparmakov wrote:
>
> > At 13:51 23/11/01, war wrote:
> > >You should use gcc-2.95.3.
> >
> > That's not true. gcc-2.96 as provided with RedHat 7.2 is perfectly fine.
> >
> > gcc-3x OTOH is not a good idea at the moment.
> >
> > Anton
> >
> > >Roy Sigurd Karlsbakk wrote:
> > >
> > > > hi all
> > > >
> > > > I just wonder...
> > > > With a clean rh72 install, I've got two gcc versions installed in 
> parllel,
> > > > 2.96 and 3.0.2. Which one should I use to compile the kernel?
> > > >
> > > > thanks
> > > >
> > > > roy
> > > > --
> > > > Roy Sigurd Karlsbakk, MCSE, MCNE, CLS, LCA
> > > >
> > > > Computers are like air conditioners.
> > > > They stop working when you open Windows.
> > > > -
> > > > To unsubscribe from this list: send the line "unsubscribe 
> linux-kernel" in
> > > > the body of a message to majordomo@vger.kernel.org
> > > > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > > > Please read the FAQ at  http://www.tux.org/lkml/
> > >
> > >-
> > >To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > >the body of a message to majordomo@vger.kernel.org
> > >More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > >Please read the FAQ at  http://www.tux.org/lkml/
> >
> > --
> >    "I've not lost my mind. It's backed up on tape somewhere." - Unknown
> > --
> > Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
> > Linux NTFS Maintainer / WWW: http://linux-ntfs.sf.net/
> > ICQ: 8561279 / WWW: http://www-stu.christs.cam.ac.uk/~aia21/
> >
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/

-- 
   "I've not lost my mind. It's backed up on tape somewhere." - Unknown
-- 
Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
Linux NTFS Maintainer / WWW: http://linux-ntfs.sf.net/
ICQ: 8561279 / WWW: http://www-stu.christs.cam.ac.uk/~aia21/

Re: Which gcc version?

[Jeff Garzik]

On Fri, 23 Nov 2001, Anton Altaparmakov wrote:

> At 14:11 23/11/01, war wrote:
> >#1) The compiler from redhat (gcc-2.96) is not an official GNU release.
> 
> And anyone should care because...?
> 
> >#2) http://www.atnf.csiro.au/~rgooch/linux/docs/kernel-newsflash.html/
> >       "the reccomend compiler is now gcc-2.95.3, rather than gcc-2.91.66"
> 
> Several main kernel developers use gcc-2.96 for their kernel development 
> work and according to Alan Cox using gcc-2.96 for the kernel is fine (from 
> a certain version onwards, can't remember the minimum release number he 
> said but the one in RH 7.2 is fine in any case).

Yes, 2.96 has a lot of bug fixes and is very well tested and stable at
this point.  I trust it more than 2.95.3, but not more than egcs-1.1.2 :)

	Jeff

Re: Which gcc version?

[Daniel Phillips]

On November 23, 2001 02:59 pm, Anton Altaparmakov wrote:
> At 13:51 23/11/01, war wrote:
> >You should use gcc-2.95.3.
> 
> That's not true. gcc-2.96 as provided with RedHat 7.2 is perfectly fine.
> 
> gcc-3x OTOH is not a good idea at the moment.

Do you have any particular reason for saying that?

Just for the record, I've been building kernels exclusively with 3.0x 
(currently 3.02) for more than a month without problems.  At some point you 
just have to take a deep breath and take the plunge ;-)

I agree that it's a good idea for comercial distributions, large corporations 
and to like, to be as a year or two behind the compiler curve, but for a 
kernel developer or unstable kernel addict to be paranoid about gcc 3.0 is in 
my mind, just plain silly.  If you are building the latest kernel you might 
as well build it with the latest compiler, the risks are roughly the same.

Through widespread use, whatever wrinkles still remain in gcc3 will be ironed 
out.  In fact, I'd say it's something of a moral responsibility to adopt the 
new compiler sooner, rather than later.  It helps make our tool chain 
stronger.

--
Daniel

Re: Which gcc version?

[Anton Altaparmakov]

At 18:30 23/11/01, Daniel Phillips wrote:
>On November 23, 2001 02:59 pm, Anton Altaparmakov wrote:
> > At 13:51 23/11/01, war wrote:
> > >You should use gcc-2.95.3.
> >
> > That's not true. gcc-2.96 as provided with RedHat 7.2 is perfectly fine.
> >
> > gcc-3x OTOH is not a good idea at the moment.
>
>Do you have any particular reason for saying that?

I haven't done any measurements myself but from what I have read, gcc-3.x 
produces significantly slower code than gcc-2.96. I know I should try 
myself some time... but if that is indeed true that is a very good reason 
to stick with gcc-2.96.

Anton


-- 
   "I've not lost my mind. It's backed up on tape somewhere." - Unknown
-- 
Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
Linux NTFS Maintainer / WWW: http://linux-ntfs.sf.net/
ICQ: 8561279 / WWW: http://www-stu.christs.cam.ac.uk/~aia21/

Re: Which gcc version?

[Daniel Phillips]

On November 23, 2001 07:56 pm, Anton Altaparmakov wrote:
> At 18:30 23/11/01, Daniel Phillips wrote:
> >On November 23, 2001 02:59 pm, Anton Altaparmakov wrote:
> > > At 13:51 23/11/01, war wrote:
> > > >You should use gcc-2.95.3.
> > >
> > > That's not true. gcc-2.96 as provided with RedHat 7.2 is perfectly fine.
> > >
> > > gcc-3x OTOH is not a good idea at the moment.
> >
> >Do you have any particular reason for saying that?
> 
> I haven't done any measurements myself but from what I have read, gcc-3.x 
> produces significantly slower code than gcc-2.96. I know I should try 
> myself some time... but if that is indeed true that is a very good reason 
> to stick with gcc-2.96.

If it does I certainly haven't noticed it.  I think we managed to get 2.4 
running slower than 2.2 for a while, and we all ran those kernels anyway, 
whether we had to or not, right?

If there is a performance hit, it's not enough to worry about.

--
Daniel

Re: Which gcc version?

[Andrew Morton]

Daniel Phillips wrote:
> 
> If there is a performance hit, it's not enough to worry about.

except egcs-1.1.2 (2.91.6) compiles stuff at almost twice the speed
of gcc3.  The person who breaks egcs-1.1.2 for kernel builds owes
me a quad Xeon, thanks very much.

-

Re: Which gcc version?

[Anton Altaparmakov]

Hi Andrew,

At 20:14 23/11/01, Andrew Morton wrote:
>Daniel Phillips wrote:
> >
> > If there is a performance hit, it's not enough to worry about.
>
>except egcs-1.1.2 (2.91.6) compiles stuff at almost twice the speed
>of gcc3.  The person who breaks egcs-1.1.2 for kernel builds owes
>me a quad Xeon, thanks very much.

Have you read the current Documentation/Changes? It says "the 2.5 tree is 
likely to drop egcs-1.1.2 workarounds". Whoever wrote that seems to be 
wanting to break it in the near future...

Anton


-- 
   "I've not lost my mind. It's backed up on tape somewhere." - Unknown
-- 
Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
Linux NTFS Maintainer / WWW: http://linux-ntfs.sf.net/
ICQ: 8561279 / WWW: http://www-stu.christs.cam.ac.uk/~aia21/

Re: Which gcc version?

[Andrew Morton]

Anton Altaparmakov wrote:
> 
> Hi Andrew,
> 
> At 20:14 23/11/01, Andrew Morton wrote:
> >Daniel Phillips wrote:
> > >
> > > If there is a performance hit, it's not enough to worry about.
> >
> >except egcs-1.1.2 (2.91.6) compiles stuff at almost twice the speed
> >of gcc3.  The person who breaks egcs-1.1.2 for kernel builds owes
> >me a quad Xeon, thanks very much.
> 
> Have you read the current Documentation/Changes? It says "the 2.5 tree is
> likely to drop egcs-1.1.2 workarounds". Whoever wrote that seems to be
> wanting to break it in the near future...

Well that's great news.  To whom do I send my shipping address?

Actually, I have negligible interest in working on something which
won't be useful to real people for three years, so that works out,
doesn't it?

-

Re: Which gcc version?

[David Weinehall]

On Fri, Nov 23, 2001 at 12:35:12PM -0800, Andrew Morton wrote:
> Anton Altaparmakov wrote:
> > 
> > Hi Andrew,
> > 
> > At 20:14 23/11/01, Andrew Morton wrote:
> > >Daniel Phillips wrote:
> > > >
> > > > If there is a performance hit, it's not enough to worry about.
> > >
> > >except egcs-1.1.2 (2.91.6) compiles stuff at almost twice the speed
> > >of gcc3.  The person who breaks egcs-1.1.2 for kernel builds owes
> > >me a quad Xeon, thanks very much.
> > 
> > Have you read the current Documentation/Changes? It says "the 2.5 tree is
> > likely to drop egcs-1.1.2 workarounds". Whoever wrote that seems to be
> > wanting to break it in the near future...
> 
> Well that's great news.  To whom do I send my shipping address?
> 
> Actually, I have negligible interest in working on something which
> won't be useful to real people for three years, so that works out,
> doesn't it?

Dropping workarounds for egcs-1.1.2 doesn't mean that gcc-2.95.3+ or
gcc-2.96-x (x > whatever the infamed version nr was) will stop working.

Thus, there's a perfectly fine alternative to gcc3. gcc3 is too broken
to use on many platforms. Then again, it's the only alternative for
others... Sigh.

As long as gcc-2.95.3 is the recommended minimum version, I'm
happy.


Regards: David Weinehall
  _                                                                 _
 // David Weinehall <tao@acc.umu.se> /> Northern lights wander      \\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\>  http://www.acc.umu.se/~tao/    </   Full colour fire           </

Re: Which gcc version?

[Luigi Genoni]

On Fri, 23 Nov 2001, Anton Altaparmakov wrote:

> At 18:30 23/11/01, Daniel Phillips wrote:
> >On November 23, 2001 02:59 pm, Anton Altaparmakov wrote:
> > > At 13:51 23/11/01, war wrote:
> > > >You should use gcc-2.95.3.
> > >
> > > That's not true. gcc-2.96 as provided with RedHat 7.2 is perfectly fine.
> > >
> > > gcc-3x OTOH is not a good idea at the moment.
> >
> >Do you have any particular reason for saying that?
>
> I haven't done any measurements myself but from what I have read, gcc-3.x
> produces significantly slower code than gcc-2.96. I know I should try
> myself some time... but if that is indeed true that is a very good reason
> to stick with gcc-2.96.
>
I did some serious bench.
On all my codes, using eavilly floating point computation, binaries
built with gcc 3.0.2 are about 5% slower that the ones built with 2.95.3
on athlon processor with athlon optimizzations.
On the other side, on sparclinux, same codes compiled with gcc 3.0.2 are
really faster, about 20%, that with 2.95.3

Luigi

Re: Which gcc version?

[Daniel Phillips]

On November 24, 2001 05:01 pm, Luigi Genoni wrote:
> On Fri, 23 Nov 2001, Anton Altaparmakov wrote:
> > At 18:30 23/11/01, Daniel Phillips wrote:
> > >On November 23, 2001 02:59 pm, Anton Altaparmakov wrote:
> > > > gcc-3x OTOH is not a good idea at the moment.
> > >
> > >Do you have any particular reason for saying that?
> >
> > I haven't done any measurements myself but from what I have read, gcc-3.x
> > produces significantly slower code than gcc-2.96. I know I should try
> > myself some time... but if that is indeed true that is a very good reason
> > to stick with gcc-2.96.
> 
> I did some serious bench.
> On all my codes, using eavilly floating point computation, binaries
> built with gcc 3.0.2 are about 5% slower that the ones built with 2.95.3
> on athlon processor with athlon optimizzations.
> On the other side, on sparclinux, same codes compiled with gcc 3.0.2 are
> really faster, about 20%, that with 2.95.3

Interesting, but not as interesting as knowing what the results are for 
non-fp code, since we are talking about kernel compilation.

--
Daniel

Re: Which gcc version?

[Luigi Genoni]

On Sat, 24 Nov 2001, Daniel Phillips wrote:

> On November 24, 2001 05:01 pm, Luigi Genoni wrote:
> > On Fri, 23 Nov 2001, Anton Altaparmakov wrote:
> > > At 18:30 23/11/01, Daniel Phillips wrote:
> > > >On November 23, 2001 02:59 pm, Anton Altaparmakov wrote:
> > > > > gcc-3x OTOH is not a good idea at the moment.
> > > >
> > > >Do you have any particular reason for saying that?
> > >
> > > I haven't done any measurements myself but from what I have read, gcc-3.x
> > > produces significantly slower code than gcc-2.96. I know I should try
> > > myself some time... but if that is indeed true that is a very good reason
> > > to stick with gcc-2.96.
> >
> > I did some serious bench.
> > On all my codes, using eavilly floating point computation, binaries
> > built with gcc 3.0.2 are about 5% slower that the ones built with 2.95.3
> > on athlon processor with athlon optimizzations.
> > On the other side, on sparclinux, same codes compiled with gcc 3.0.2 are
> > really faster, about 20%, that with 2.95.3
>
> Interesting, but not as interesting as knowing what the results are for
> non-fp code, since we are talking about kernel compilation.
>
on sparc64 nor gcc 2.95.3 nor 3.0.2 can be used to compile the kernel, at
less for what i know, you have to stay with egcs 64 bit compiler.

Re: Which gcc version?

[RaúlNúñez de Arenas Coronado]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 856 bytes --]

    Hi Roy :)))

>Which one should I use to compile the kernel?

    Dunno about RH, but I compile the kernel using gcc-3.0.1 and
works fine. Well, take into account that this is NOT the recommended
choice, since gcc 3.x series is not considered stable for the kernel.

    Anyway, I'm tired to having a compiler for binaries, another
compiler for the kernel, another one for libraries, etc... and I
compile all using gcc 3.0.1.

    Sooner or later the kernel will need to be ported to gcc 3.x
series, so, the sooner it gets tested with this compiler, the better.

    Anyway, if you have gcc 2.95.x installed onto your distro, use
that for the kernel for maximum stability.

    My compiled kernel works with gcc 3.x but it's a kernel with no
so many drivers. I don't know what kind of weird things can happen
with a heluvalota drivers kernel.

    Raúl

Re: Which gcc version?

[M. Edward (Ed) Borasky]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: TEXT/PLAIN; charset=X-UNKNOWN, Size: 1312 bytes --]

On Fri, 23 Nov 2001, [ISO-8859-1] Raúl[ISO-8859-1] Núñez de Arenas  Coronado wrote:

>     Sooner or later the kernel will need to be ported to gcc 3.x
> series, so, the sooner it gets tested with this compiler, the better.

One of the regression tests for gcc is to compile *a* Linux kernel, although
I have no clue which kernel they use, or if they just haul down the latest one
from the Internet.

>     Anyway, if you have gcc 2.95.x installed onto your distro, use
> that for the kernel for maximum stability.

I haven't seen it on RH 7.x; I had to download and install it separately. And
it is a *royal* pain to get all the symbols/paths/libraries, etc., correct if
you want to use *anything* other than the default compiler/libraries/binutils
with Red Hat.  One would need extreme motivation and extreme care.

For you (fellow) Athlon geeks, 2.95.x generates better code than either 2.96.x
or 3.0.x on the Atlas high-speed numerical linear algebra library. The reverse
is true for IA64; 3.0.x is about the only thing that generates decent code. If
that's motivating enough, have at it. YMMV.
--
znmeb@aracnet.com (M. Edward Borasky) http://www.aracnet.com/~znmeb
Relax! Run Your Own Brain with Neuro-Semantics!
http://www.aracnet.com/~znmeb/Flyer.htm

When puns are outlawed, only inlaws will have gnus.

Re: Which gcc version?

[Thorsten Glaser]

Dixitur de M. Edward (Ed) Borasky respondebo ad:

> On Fri, 23 Nov 2001, [ISO-8859-1] Ra?l[ISO-8859-1] N??ez de Arenas  Coronado wrote:
>
> >     Sooner or later the kernel will need to be ported to gcc 3.x
> > series, so, the sooner it gets tested with this compiler, the better.
>
> One of the regression tests for gcc is to compile *a* Linux kernel, although
> I have no clue which kernel they use, or if they just haul down the latest one
> from the Internet.

It's one of the 2.2 series IIRC. Go http://gcc.gnu.org/

> >     Anyway, if you have gcc 2.95.x installed onto your distro, use
> > that for the kernel for maximum stability.

I've run 2.4.3-ac7 plus andrea's rwsem, compiled on a gcc-3 beta
of April 2001, since then (not in 24/7 though) with _no_ problems.

-mirabilos
-- 
| This message body is covered by Germanic and International | OpenBSD30
| Copyright law. Modification of any kind and redistribution | centericq
| via AOL or the Microsoft network are strictly prohibited!! | UIN seems
| Scientific-style quotation permitted if due credits given. | 132315236