gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[Joel Soete]

Quoting John David Anglin <dave@hiauly1.hia.nrc.ca>:

> > > Your problem appears to be with arch/parisc/kernel/parisc_ksyms.c.
> > > It imports the symbol.
> > > 
> > 
> > Simply comment it out will broken to compile with gcc-3.0.x, I
> suppose?
> 
> I suspect that you can just delete
> 
>   extern void $$mulU(void);
> 
> and
> 
>   #ifndef __LP64__
>   EXPORT_SYMBOL_NOVERS($$mulU);
>   #endif
> 
> I didn't see any assembly usage of this function in linux and I don't
> believe gcc uses it (even 3.0.x), so probably the lines can just be
> deleted.
> 

To summarise:
This clean should be applied without problem (test with gcc-3.2 and gcc-3.0.4)

This allow me to produce an operational kernel with gcc-3.2 (bootstrap an x);
telnet, ftp and ssh connection OK

But with gcc-3.1.1 still crashing at the first remote connection?

I will try a more recent CVS update.

Joel




-------------------------------------------------
This mail sent through Tiscali Webmail (http://webmail.tiscali.be)

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[John David Anglin]

> But with gcc-3.1.1 still crashing at the first remote connection?

If 64-bit, this could be the tcp header alignment problem.

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

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[joel.soete]

Quoting John David Anglin <dave@hiauly1.hia.nrc.ca>:

> > But with gcc-3.1.1 still crashing at the first remote connection?
> 
> If 64-bit, this could be the tcp header alignment problem.
> 
Sorry it is on a 712 (32-bit)

Joel 


-------------------------------------------------
This mail sent through Tiscali Webmail (http://webmail.tiscali.be)

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[Matthew Wilcox]

On Wed, Jul 17, 2002 at 07:09:54PM +0200, Joel Soete wrote:
> To summarise:
> This clean should be applied without problem (test with gcc-3.2 and gcc-3.0.4)
> 
> This allow me to produce an operational kernel with gcc-3.2 (bootstrap an x);
> telnet, ftp and ssh connection OK

but did you test modules?

-- 
Revolutions do not require corporate support.

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[joel.soete]

Quoting Matthew Wilcox <willy@debian.org>:

> On Wed, Jul 17, 2002 at 07:09:54PM +0200, Joel Soete wrote:
> > To summarise:
> > This clean should be applied without problem (test with gcc-3.2 and
> gcc-3.0.4)
> > 
> > This allow me to produce an operational kernel with gcc-3.2 (bootstrap
> an x);
> > telnet, ftp and ssh connection OK
> 
> but did you test modules?
> 
They were all together build with 'make oldconfig' after a distclean, so yes
with modules support. A monolithic test should be interesting? (I already do
this test for gcc-3.1 not 3.1.1 it does not bring success)

Joel



-------------------------------------------------
This mail sent through Tiscali Webmail (http://webmail.tiscali.be)

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[Matthew Wilcox]

On Wed, Jul 17, 2002 at 07:44:04PM +0200, joel.soete@freebel.net wrote:
> Quoting Matthew Wilcox <willy@debian.org>:
> > but did you test modules?
> They were all together build with 'make oldconfig' after a distclean, so yes
> with modules support. A monolithic test should be interesting? (I already do
> this test for gcc-3.1 not 3.1.1 it does not bring success)

build is not sufficient; the EXPORT_MODULE stuff can only be tested by
loading modules at runtime.

-- 
Revolutions do not require corporate support.

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[joel.soete]

Quoting Matthew Wilcox <willy@debian.org>:

> On Wed, Jul 17, 2002 at 07:44:04PM +0200, joel.soete@freebel.net wrote:
> > Quoting Matthew Wilcox <willy@debian.org>:
> > > but did you test modules?
> > They were all together build with 'make oldconfig' after a distclean,
> so yes
> > with modules support. A monolithic test should be interesting? (I
> already do
> > this test for gcc-3.1 not 3.1.1 it does not bring success)
> 
> build is not sufficient; the EXPORT_MODULE stuff can only be tested by
> loading modules at runtime.


Is it just to test to load a module? (no relation with TCP problem?)

Joel



-------------------------------------------------
This mail sent through Tiscali Webmail (http://webmail.tiscali.be)

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[joel.soete]

Quoting joel.soete@freebel.net:

> Quoting Matthew Wilcox <willy@debian.org>:
> 
> > On Wed, Jul 17, 2002 at 07:44:04PM +0200, joel.soete@freebel.net
> wrote:
> > > Quoting Matthew Wilcox <willy@debian.org>:
> > > > but did you test modules?
> > > They were all together build with 'make oldconfig' after a
> distclean,
> > so yes
> > > with modules support. A monolithic test should be interesting? (I
> > already do
> > > this test for gcc-3.1 not 3.1.1 it does not bring success)
> > 
> > build is not sufficient; the EXPORT_MODULE stuff can only be tested
> by
> > loading modules at runtime.
> 
> 
> Is it just to test to load a module? (no relation with TCP problem?)
> 
> Joel
> 
Sorry for delay but I would like to re-compile some more modules. I just test
lvm-mod as well as nfs and nfsd. No problem: insmod, lsmod, rsmod works fine for
gcc-3.1.1 and for gcc-3.2. (do you know some problem with modules with those
release of gcc compiler?)

Joel



-------------------------------------------------
This mail sent through Tiscali Webmail (http://webmail.tiscali.be)

Re: gcc-3.2 Ok gcc-3.1.1 Ko (was Re: [parisc-linux] gcc-3.2 bootstrap?)

[joel.soete]

Quoting Joel Soete <joel.soete@freebel.net>:

> Quoting John David Anglin <dave@hiauly1.hia.nrc.ca>:
> 
> > > > Your problem appears to be with
> arch/parisc/kernel/parisc_ksyms.c.
> > > > It imports the symbol.
> > > > 
> > > 
> > > Simply comment it out will broken to compile with gcc-3.0.x, I
> > suppose?
> > 
> > I suspect that you can just delete
> > 
> >   extern void $$mulU(void);
> > 
> > and
> > 
> >   #ifndef __LP64__
> >   EXPORT_SYMBOL_NOVERS($$mulU);
> >   #endif
> > 
> > I didn't see any assembly usage of this function in linux and I don't
> > believe gcc uses it (even 3.0.x), so probably the lines can just be
> > deleted.
> > 
> 
> To summarise:
> This clean should be applied without problem (test with gcc-3.2 and
> gcc-3.0.4)
> 
> This allow me to produce an operational kernel with gcc-3.2 (bootstrap
> an x);
> telnet, ftp and ssh connection OK
> 
> But with gcc-3.1.1 still crashing at the first remote connection?
> 
> I will try a more recent CVS update.

Same problem with the last CVS update today (20020718)

> 
> Joel
> 

PS: A new baby is ongoing to be born ? (2.5 :) or :( for 2.4)
I better understand on which important job Paul was working :)

-------------------------------------------------
This mail sent through Tiscali Webmail (http://webmail.tiscali.be)