GCC 3.2.2 build error on Solaris 9 x86

GCC 3.2.2 build error on Solaris 9 x86

[J.D. Bronson]

I was able to build gcc 3.2.1 on solaris 9x86 w/o any trouble.
However, the same cannot be said for gcc 3.2.2:

# gcc -v
Reading specs from /usr/local/lib/gcc-lib/i386-pc-solaris2.9/3.2.1/specs
Configured with: ../configure --with-as=/usr/local/bin/as 
--with-ld=/usr/local/bin/ld --disable-nls
Thread model: posix
gcc version 3.2.1

I use the same build concept and try building gcc 3.2.2:
<snip>

gcc -c -DIN_GCC    -g  -W -Wall -Wwrite-strings -Wstrict-prototypes 
-Wmissing-prototypes -Wtraditional -pedantic 
-Wno-long-long  -DHAVE_CONFIG_H    -I. -I. -I../../gcc -I../../gcc/. 
-I../../gcc/config -I../../gcc/../include \
         ../../gcc/config/i386/i386.c -o i386.o
../../gcc/config/i386/i386.c: In function `override_options':
../../gcc/config/i386/i386.c:927: `DEFAULT_PCC_STRUCT_RETURN' undeclared 
(first use in this function)
../../gcc/config/i386/i386.c:927: (Each undeclared identifier is reported 
only once
../../gcc/config/i386/i386.c:927: for each function it appears in.)
../../gcc/config/i386/i386.c: In function `ix86_save_reg':
../../gcc/config/i386/i386.c:3982: warning: comparison between signed and 
unsigned
make[2]: *** [i386.o] Error 1
make[2]: Leaving directory `/usr/temp/gcc-3.2.2/obj/gcc'
make[1]: *** [stage1_build] Error 2
make[1]: Leaving directory `/usr/temp/gcc-3.2.2/obj/gcc'
make: *** [bootstrap] Error 2
#

Any thoughts?







-- 
J.D. Bronson
Aurora Health Care // Information Systems // Milwaukee, WI USA
Office: 414.978.8282 // Fax: 414.328.8282 // Pager: 414.603.8282

Re: GCC 3.2.2 build error on Solaris 9 x86

[Gerald Pfeifer]

On Fri, 14 Feb 2003, J.D. Bronson wrote:
> I was able to build gcc 3.2.1 on solaris 9x86 w/o any trouble.
> However, the same cannot be said for gcc 3.2.2:

Yes, this is, unfortunately a regression introduced by a bug fix. :-(

I believe it has been fixed now, and you might want to try and use CVS
to access the most current version on the gcc-3.2 branch.


(Perhaps this and the missing info files[1} alone would warrant a GCC
3.2.2a release?)

Gerald

[1] Has someone already updated the release scripts to avoid/detect this
in the future?
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

Re: GCC 3.2.2 build error on Solaris 9 x86

[Gerald Pfeifer]

J.D./postmaster@xpec.com, in case you read this somehow:

You are apparently using
  http://www.spamlist.org/html/the_list.html
which blocked my originally message because -- my server is in Austria
(ccTLD .at), one of the three countries on this planet with the strongest
anti-SPAM laws.

Reading the list above, really makes one wonder.  Some folks really see
themselves as a kind of world police. <sigh>

Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

Re: GCC 3.2.2 build error on Solaris 9 x86

[Gabriel Dos Reis]

Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at> writes:

[...]

| (Perhaps this and the missing info files[1} alone would warrant a GCC
| 3.2.2a release?)

Hi Gerald,

  Has the SC decided on any future release from the gcc-3_2-branch?

Thanks,

-- Gaby

Re: GCC 3.2.2 build error on Solaris 9 x86

[Kaveh R. Ghazi]

 > On Fri, 14 Feb 2003, J.D. Bronson wrote:
 > > I was able to build gcc 3.2.1 on solaris 9x86 w/o any trouble.
 > > However, the same cannot be said for gcc 3.2.2:
 > 
 > Yes, this is, unfortunately a regression introduced by a bug fix. :-(
 > 
 > I believe it has been fixed now, and you might want to try and use CVS
 > to access the most current version on the gcc-3.2 branch.

I checked CVS with --target=i386-pc-solaris2.9 and it still fails on
all three branches (trunk, 3.3 and 3.2).

There was a lot of discussion but no patch was ever checked in. :-(


 > (Perhaps this and the missing info files[1} alone would warrant a GCC
 > 3.2.2a release?)
 > Gerald

My recollection of the "a" release convention is that it is for
packaging problems only, no code changes.  The missing info files
qualifies, but not the bugfix.  We'll need to release a 3.2.3.

		--Kaveh
--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu

Re: GCC 3.2.2 build error on Solaris 9 x86 (patch?)

[J.D. Bronson]

At 10:27 AM 2/17/2003, Kaveh R. Ghazi wrote:
>  > On Fri, 14 Feb 2003, J.D. Bronson wrote:
>  > > I was able to build gcc 3.2.1 on solaris 9x86 w/o any trouble.
>  > > However, the same cannot be said for gcc 3.2.2:
>  >
>  > Yes, this is, unfortunately a regression introduced by a bug fix. :-(
>  >
>  > I believe it has been fixed now, and you might want to try and use CVS
>  > to access the most current version on the gcc-3.2 branch.
>
>I checked CVS with --target=i386-pc-solaris2.9 and it still fails on
>all three branches (trunk, 3.3 and 3.2).
>
>There was a lot of discussion but no patch was ever checked in. :-(
>
>
>  > (Perhaps this and the missing info files[1} alone would warrant a GCC
>  > 3.2.2a release?)
>  > Gerald
>
>My recollection of the "a" release convention is that it is for
>packaging problems only, no code changes.  The missing info files
>qualifies, but not the bugfix.  We'll need to release a 3.2.3.
>
>                 --Kaveh
>--
>Kaveh R. Ghazi                  ghazi@caip.rutgers.edu



This is what fixed it for me:

------------------- clip ------------------------
*** sysv4.h.orig Tue Jan 22 21:59:30 2002
--- sysv4.h Sun Feb 16 17:54:24 2003
***************
*** 27,32 ****
--- 27,35 ----
   /* The svr4 ABI for the i386 says that records and unions are returned
      in memory.  */

+ #undef DEFAULT_PCC_STRUCT_RETURN
+ #define DEFAULT_PCC_STRUCT_RETURN 0
+
   #undef RETURN_IN_MEMORY
   #define RETURN_IN_MEMORY(TYPE) \
     (TYPE_MODE (TYPE) == BLKmode \
------------------- clip ------------------------







-- 
J.D. Bronson
Aurora Health Care // Information Systems // Milwaukee, WI USA
Office: 414.978.8282 // Fax: 414.328.8282 // Pager: 414.603.8282

spam list ...was GCC 3.2.2 build error on Solaris 9 x86

[J.D. Bronson]

Sorry about this....but in using this list for over 4mos, this is the 
*first* false positive we received. Really....the very first!

It has killed off tons of spam (along with using RBL).

We dont know anyone in virtually all the countries listed (like .ru example)
and therefore dont plan on accepting email from them at this time...

I do have bypass email addresses in place, and of course can allow any 
legit tld or domain in...

I can certainly edit the file and allow .at in :)

the spam list may be a radical solution, but it has seriously stopped 
spammmers. I am very thankful to them for this list.

Sorry about the rejected email.

Jeff

At 11:59 AM 2/18/2003, Joe Buck wrote:
>On Sat, Feb 15, 2003 at 06:17:21PM +0100, Gerald Pfeifer wrote:
> > J.D./postmaster@xpec.com, in case you read this somehow:
> >
> > You are apparently using
> >   http://www.spamlist.org/html/the_list.html
> > which blocked my originally message because -- my server is in Austria
> > (ccTLD .at), one of the three countries on this planet with the strongest
> > anti-SPAM laws.
> >
> > Reading the list above, really makes one wonder.  Some folks really see
> > themselves as a kind of world police. <sigh>
>
>Astounding: these guys block all mail from Switzerland, where the Web
>was first invented.  In all good conscience they should shut down their
>web site, as clearly anything Swiss cannot be trusted. :-)
>
>They also block China, Korea, Japan, Italy, Poland, Russia, Singapore,
>Spain, Columbia, Hungary, Hong Kong, France, the Czech Republic,
>all Swedish class C networks, and Greece.
>
>These guys are evil: we should ask them to add gcc.gnu.org to their block
>list, so that no one who subscribes to their list can get our mail.

Re: GCC 3.2.2 build error on Solaris 9 x86

[J.D. Bronson]

At 11:51 AM 2/18/2003, Joe Buck wrote:
>On Mon, Feb 17, 2003 at 08:16:01AM +0100, Gabriel Dos Reis wrote:
> >   Has the SC decided on any future release from the gcc-3_2-branch?
>
>There has been some discussion.  It does appear that we're going to need
>3.2.3.  I had been resisting this because I've been worried about how
>well we can execute with three branches going at once, but clearly 3.3
>is going to have to slip more; we need to improve its compile speed for
>-O0 at least some for it to be releasable.  Also, clearly there are some
>significant problems still with 3.2.2.  If it were only packaging problems
>(missing info files), then 3.2.2a (just adding info files) would be
>a possibility, but we have a few serious fixes that should be applied.


So is there a recommended fix for solaris?

>

Re: spam list ...was GCC 3.2.2 build error on Solaris 9 x86

[Allen Gwinn]

Greetings,

While the solution is rather draconian, very little legitimate mail originates
from the listed domains.  Those that resolve ptrs to a .com or .org, in those
countries, are not affected (unless we have a block of address space listed).  As
everyone knows, Korea and China are extremely friendly toward spammers.  We did
an exhaustive audit over the period of several months on several different
servers.  We did not blacklist anything from which we had gotten legitimate
email.

Given Gerald's complaint, I went back and checked logs for the past month.  So
far, we've received about 70 pieces of spam from .at all destined for our domreg
account.  Virtually all of it originated from one of the following 3 domains:

.liwest.at
.surfer.at
.teleweb.at

In the interest of maintaining good international relations with "one of three
countries on the planet with the strongest anti-SPAM laws" (sorry, :) couldn't
resist the jab--you know us filthy Americans), we will revise the list to reflect
those domains.

In all seriousness, I would welcome help with the list.  Excuse me: I would
WELCOME help!  We created this list to solve a problem at a handful of servers.
It has worked out well--with a miniscule 3 specific exceptions.  Even so, I would
welcome assistance in fine tuning it (or coming up with procedures to maintain
it).

Thanks,

Allen

"J.D. Bronson" wrote:

> Sorry about this....but in using this list for over 4mos, this is the
> *first* false positive we received. Really....the very first!
>
> It has killed off tons of spam (along with using RBL).
>
> We dont know anyone in virtually all the countries listed (like .ru example)
> and therefore dont plan on accepting email from them at this time...
>
> I do have bypass email addresses in place, and of course can allow any
> legit tld or domain in...
>
> I can certainly edit the file and allow .at in :)
>
> the spam list may be a radical solution, but it has seriously stopped
> spammmers. I am very thankful to them for this list.
>
> Sorry about the rejected email.
>
> Jeff
>
> At 11:59 AM 2/18/2003, Joe Buck wrote:
> >On Sat, Feb 15, 2003 at 06:17:21PM +0100, Gerald Pfeifer wrote:
> > > J.D./postmaster@xpec.com, in case you read this somehow:
> > >
> > > You are apparently using
> > >   http://www.spamlist.org/html/the_list.html
> > > which blocked my originally message because -- my server is in Austria
> > > (ccTLD .at), one of the three countries on this planet with the strongest
> > > anti-SPAM laws.
> > >
> > > Reading the list above, really makes one wonder.  Some folks really see
> > > themselves as a kind of world police. <sigh>
> >
> >Astounding: these guys block all mail from Switzerland, where the Web
> >was first invented.  In all good conscience they should shut down their
> >web site, as clearly anything Swiss cannot be trusted. :-)
> >
> >They also block China, Korea, Japan, Italy, Poland, Russia, Singapore,
> >Spain, Columbia, Hungary, Hong Kong, France, the Czech Republic,
> >all Swedish class C networks, and Greece.
> >
> >These guys are evil: we should ask them to add gcc.gnu.org to their block
> >list, so that no one who subscribes to their list can get our mail.

Re: spam list ...was GCC 3.2.2 build error on Solaris 9 x86

[Allen Gwinn]

Hi Joe,

That is absolutely not true.  I've blacklisted top level domains from which we've
never received a legitimate piece of email.  Most of it is dialups.  Most legitimate
mail originates from .com's, or .org's, or other top levels--same as in the U.S.

I wish I had a way to blacklist U.S. dialups.  We try to do it with the
mail-abuse.org DUL.  However, they pop up quicker than the issue can be addressed.

Thanks for the note, and we love the Swiss!  We just don't want spam from your
dialups.

Allen

Joe Buck wrote:

> On Wed, Feb 19, 2003 at 06:45:40AM -0600, Allen Gwinn wrote:
> > While the solution is rather draconian, very little legitimate mail originates
> > from the listed domains.
>
> Oh, come on.  You've blacklisted all of Switzerland, Italy, France,
> and many other European countries.  I've seen no evidence that a mail
> sent by someone in
> those countries is more likely to be spam than one sent from the US.
>
> If you are convinced that nothing of value can come out of Switzerland, you
> should stop using the web immediately.

Re: spam list ...was GCC 3.2.2 build error on Solaris 9 x86

[Allen Gwinn]

Hi Joe,

Joe Buck wrote:

> But you block anyone in the whole country who chooses to use the national
> domain, not just dialups.  This would be OK if you didn't publish the list
> for others to use with the false claim that it is a good spam-blocking
> list.  Many of our active developers use domains that you are blocking.
>
> For that reason, I strongly recommend that people interested in
> spam-blocking lists use someone else's list, not yours.

But it *is* a good spam blocking list, and I don't force anyone to use it.  People are
always free to tweak it (as we suggest) for their own needs.

I've been giving alot of thought about breaking the list up into sections, and producing
them in a text-based format where downloads can be automated.  That way, people can pick
and choose what they want.  If they only want to block China and Korea, they'll be able to
do that.

I'm sorry we have the disagreement, but we'll just have to agree to disagree!

Allen

Re: GCC 3.2.2 build error on Solaris 9 x86 (patch?)

[Kaveh R. Ghazi]

 > From: "J.D. Bronson" <jeff@xpec.com>
 > 
 > At 10:27 AM 2/17/2003, Kaveh R. Ghazi wrote:
 > >  > On Fri, 14 Feb 2003, J.D. Bronson wrote:
 > >  > > I was able to build gcc 3.2.1 on solaris 9x86 w/o any trouble.
 > >  > > However, the same cannot be said for gcc 3.2.2:
 > >  >
 > >  > Yes, this is, unfortunately a regression introduced by a bug fix. :-(
 > >  >
 > >  > I believe it has been fixed now, and you might want to try and use CVS
 > >  > to access the most current version on the gcc-3.2 branch.
 > >
 > >I checked CVS with --target=i386-pc-solaris2.9 and it still fails on
 > >all three branches (trunk, 3.3 and 3.2).
 > >
 > >There was a lot of discussion but no patch was ever checked in. :-(

I think the patch that went with the ChangeLog below may have fixed
this.  (At least I was able to build cc1 with a cross-compiler
targeted to i386-pc-solaris2.7.)  Would you please try out full
bootstrap on your system with a newly updated cvs checkout from the
3.2.x branch (and hopefully post testsuite results) ?

		Thanks,
		--Kaveh


2003-02-28  Joel Sherrill <joel@OARcorp.com>

	PR 9638/other
	* config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Ensure the 
	this constant defaults to 1.

--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu