2.3.3 compile error on ppc

2.3.3 compile error on ppc

[Mark Lin]

gcc -D__KERNEL__ -I/usr/src/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -D__powerpc__ -fsigned-char -msoft-float -pipe -fno-builtin -ffixed-r2 -Wno-uninitialized -mmultiple -mstring   -c -o fork.o fork.c
fork.c: In function `mm_alloc':
fork.c:306: warning: implicit declaration of function `init_MUTEX_LOCKED'
fork.c: In function `do_fork':
fork.c:539: warning: implicit declaration of function `DECLARE_MUTEX_LOCKED'
fork.c:539: `sem' undeclared (first use in this function)
fork.c:539: (Each undeclared identifier is reported only once
fork.c:539: for each function it appears in.)
make[2]: *** [fork.o] Error 1
make[2]: Leaving directory `/usr/src/linux/kernel'
make[1]: *** [first_rule] Error 2
make[1]: Leaving directory `/usr/src/linux/kernel'
make: *** [_dir_kernel] Error 2

Where's the site for the ppc patched kernels?

--
Mark Lin
MPI Software Technology, Inc.


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Hollis R Blanchard]

On Wed, 19 May 1999, Mark Lin wrote:
> 
> gcc -D__KERNEL__ -I/usr/src/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -D__powerpc__ -fsigned-char -msoft-float -pipe -fno-builtin -ffixed-r2 -Wno-uninitialized -mmultiple -mstring   -c -o fork.o fork.c
> fork.c: In function `mm_alloc':
> fork.c:306: warning: implicit declaration of function `init_MUTEX_LOCKED'
> fork.c: In function `do_fork':
> fork.c:539: warning: implicit declaration of function `DECLARE_MUTEX_LOCKED'
> fork.c:539: `sem' undeclared (first use in this function)
> fork.c:539: (Each undeclared identifier is reported only once
> fork.c:539: for each function it appears in.)
> make[2]: *** [fork.o] Error 1
> make[2]: Leaving directory `/usr/src/linux/kernel'
> make[1]: *** [first_rule] Error 2
> make[1]: Leaving directory `/usr/src/linux/kernel'
> make: *** [_dir_kernel] Error 2
> 
> Where's the site for the ppc patched kernels?

I think it was Cort who said to stay away from 2.3.x for now, because it will
be broken for us for a little. Were you having problems with 2.2.x?

-Hollis


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Andreas Bogk]

Hollis R Blanchard <hollis+@andrew.cmu.edu> writes:

> be broken for us for a little. Were you having problems with 2.2.x?

There is a problem with 2.2.9, and here's the patch:

diff -r -u linux-2.2.9/arch/ppc/kernel/syscalls.c linux/arch/ppc/kernel/syscalls.c
--- linux-2.2.9/arch/ppc/kernel/syscalls.c	Sat May  8 20:14:01 1999
+++ linux/arch/ppc/kernel/syscalls.c	Wed May 19 18:36:23 1999
@@ -30,6 +30,7 @@
 #include <linux/shm.h>
 #include <linux/stat.h>
 #include <linux/mman.h>
+#include <linux/file.h>
 #include <linux/sys.h>
 #include <linux/ipc.h>
 #include <linux/utsname.h>

I wonder why such trivial problems creep into so-called stable
releases of the Linux kernel again and again. Does nobody care that on
i386, you just go to ftp.kernel.org and download the latest version
and it just works, whereas on PowerPC you have to find out:

- what CVS is
- what vger is
- what CVS tags are
- which tag the one is which gives you a stable version ?

I don't care that the development kernel breaks on PPC. But that the
official stable kernel breaks is a huge problem.

Andreas

-- 
Reality is two's complement. See:
ftp://ftp.netcom.com/pub/hb/hbaker/hakmem/hacks.html#item154

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Tom Rini]

On 19 May 1999, Andreas Bogk wrote:

> 
> Hollis R Blanchard <hollis+@andrew.cmu.edu> writes:
> 
> > be broken for us for a little. Were you having problems with 2.2.x?
> 
> There is a problem with 2.2.9, and here's the patch:
> 
> diff -r -u linux-2.2.9/arch/ppc/kernel/syscalls.c linux/arch/ppc/kernel/syscalls.c
> --- linux-2.2.9/arch/ppc/kernel/syscalls.c	Sat May  8 20:14:01 1999
> +++ linux/arch/ppc/kernel/syscalls.c	Wed May 19 18:36:23 1999
> @@ -30,6 +30,7 @@
>  #include <linux/shm.h>
>  #include <linux/stat.h>
>  #include <linux/mman.h>
> +#include <linux/file.h>
>  #include <linux/sys.h>
>  #include <linux/ipc.h>
>  #include <linux/utsname.h>
> 
> I wonder why such trivial problems creep into so-called stable
> releases of the Linux kernel again and again.

Well.  This I assume is because this problem shows up only under certian
configs.  I just compiled a 2.2.9 w/o that patch(vger so it might be
possible something just missed a merge to linus).

> Does nobody care that on i386, you just go to ftp.kernel.org and
> download the latest version and it just works, whereas on PowerPC you
> have to find out:

2.2.0->2.2.8 compile fine.  Various 2.1.1xx's compile fine.  Considering
we haven't been around all that long (in an official manner, pmac wise),
it's not too bad.

> - what CVS is
> - what vger is
> - what CVS tags are
> - which tag the one is which gives you a stable version ?

The last two are only true because 2.3.x is now out.  The first two I'm
almost sure are in a FAQ by now.  x86 has it's own little neuances(sp)
too. (RH6 doesn't ship w/ a Linus kernel approved compiler anymore, I find
that funny, and don't tell me you've never had a problem w/ lilo).

> I don't care that the development kernel breaks on PPC. But that the
> official stable kernel breaks is a huge problem.

The official kernel releases are only ever checked on x86.  So sometimes
they'll be broken (not usually on a stable tree).

---
Tom Rini (TR1265)
http://dobbstown.yeti.edu/


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Martin Costabel]

Hollis R Blanchard wrote:
> 
> On Wed, 19 May 1999, Mark Lin wrote:
> >
> > gcc -D__KERNEL__ -I/usr/src/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -D__powerpc__ -fsigned-char -msoft-float -pipe -fno-builtin -ffixed-r2 -Wno-uninitialized -mmultiple -mstring   -c -o fork.o fork.c
> > fork.c: In function `mm_alloc':
> > fork.c:306: warning: implicit declaration of function `init_MUTEX_LOCKED'
> > fork.c: In function `do_fork':
> > fork.c:539: warning: implicit declaration of function `DECLARE_MUTEX_LOCKED'
> > fork.c:539: `sem' undeclared (first use in this function)
> > fork.c:539: (Each undeclared identifier is reported only once
> > fork.c:539: for each function it appears in.)
> > make[2]: *** [fork.o] Error 1
> > make[2]: Leaving directory `/usr/src/linux/kernel'
> > make[1]: *** [first_rule] Error 2
> > make[1]: Leaving directory `/usr/src/linux/kernel'
> > make: *** [_dir_kernel] Error 2
> >
> > Where's the site for the ppc patched kernels?

See http://cvs.on.openprojects.net/

> 
> I think it was Cort who said to stay away from 2.3.x for now, because it will
> be broken for us for a little. Were you having problems with 2.2.x?

Pretty hard to keep the fingers out of the pot :-)

--
Martin
(running 2.3.3)

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Mark Lin]

What are the modules to check out?  "kernel" doesn't work as suggested
on the web page.


--
Mark Lin
MPI Software Technology, Inc.

On Wed, 19 May 1999, Martin Costabel wrote:

> Hollis R Blanchard wrote:
> > 
> > On Wed, 19 May 1999, Mark Lin wrote:
> > >
> > > gcc -D__KERNEL__ -I/usr/src/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -D__powerpc__ -fsigned-char -msoft-float -pipe -fno-builtin -ffixed-r2 -Wno-uninitialized -mmultiple -mstring   -c -o fork.o fork.c
> > > fork.c: In function `mm_alloc':
> > > fork.c:306: warning: implicit declaration of function `init_MUTEX_LOCKED'
> > > fork.c: In function `do_fork':
> > > fork.c:539: warning: implicit declaration of function `DECLARE_MUTEX_LOCKED'
> > > fork.c:539: `sem' undeclared (first use in this function)
> > > fork.c:539: (Each undeclared identifier is reported only once
> > > fork.c:539: for each function it appears in.)
> > > make[2]: *** [fork.o] Error 1
> > > make[2]: Leaving directory `/usr/src/linux/kernel'
> > > make[1]: *** [first_rule] Error 2
> > > make[1]: Leaving directory `/usr/src/linux/kernel'
> > > make: *** [_dir_kernel] Error 2
> > >
> > > Where's the site for the ppc patched kernels?
> 
> See http://cvs.on.openprojects.net/


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Martin Costabel]

The real culprit in this story is the following broken (or incomplete)
patch that got into the kernel.org kernel version 2.2.8. Fortunately it
did not make its way into the vger tree. (This is from patch-2.2.8):

diff -u --recursive --new-file v2.2.7/linux/arch/ppc/kernel/syscalls.c
linux/arch/ppc/kernel/syscalls.c
--- v2.2.7/linux/arch/ppc/kernel/syscalls.c	Mon Oct  5 13:13:36 1998
+++ linux/arch/ppc/kernel/syscalls.c	Sat May  8 11:14:01 1999
@@ -205,12 +205,15 @@
 
 	lock_kernel();
 	if (!(flags & MAP_ANONYMOUS)) {
-		if (fd >= NR_OPEN || !(file = current->files->fd[fd]))
+		file = fget(fd);
+		if (!file)
 			goto out;
 	}
 	
 	flags &= ~(MAP_EXECUTABLE | MAP_DENYWRITE);
 	ret = do_mmap(file, addr, len, prot, flags, offset);
+	if (file)
+		fput(file);
 out:
 	unlock_kernel();
 	return ret;

--
Martin

Andreas Bogk wrote:
> 
> Hollis R Blanchard <hollis+@andrew.cmu.edu> writes:
> 
> > be broken for us for a little. Were you having problems with 2.2.x?
> 
> There is a problem with 2.2.9, and here's the patch:
> 
> diff -r -u linux-2.2.9/arch/ppc/kernel/syscalls.c linux/arch/ppc/kernel/syscalls.c
> --- linux-2.2.9/arch/ppc/kernel/syscalls.c      Sat May  8 20:14:01 1999
> +++ linux/arch/ppc/kernel/syscalls.c    Wed May 19 18:36:23 1999
> @@ -30,6 +30,7 @@
>  #include <linux/shm.h>
>  #include <linux/stat.h>
>  #include <linux/mman.h>
> +#include <linux/file.h>
>  #include <linux/sys.h>
>  #include <linux/ipc.h>
>  #include <linux/utsname.h>
> 
> I wonder why such trivial problems creep into so-called stable
> releases of the Linux kernel again and again. Does nobody care that on
> i386, you just go to ftp.kernel.org and download the latest version
> and it just works, whereas on PowerPC you have to find out:
> 
> - what CVS is
> - what vger is
> - what CVS tags are
> - which tag the one is which gives you a stable version ?
> 
> I don't care that the development kernel breaks on PPC. But that the
> official stable kernel breaks is a huge problem.

It seems that version-2.2.8 suffered from quite a lot of broken patches,
not only for the ppc architecture. 

--
Martin

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Martin Costabel]

Mark Lin wrote:
> 
> What are the modules to check out?  "kernel" doesn't work as suggested
> on the web page.

linux 
(yes, this gives /cvs/linux/linux as repository)

--
Martin

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: 2.3.3 compile error on ppc

[Hollis R Blanchard]

On Wed, 19 May 1999, Tom Rini wrote:
>
> The official kernel releases are only ever checked on x86.  So sometimes
> they'll be broken (not usually on a stable tree).

I'm beginning to think we need a status page like at
http://cvs-mirror.mozilla.org/webtools/tinderbox/showbuilds.cgi?tree=SeaMonkey&hours=12

You know, where you see the Intel person check something in and then all the
other platform builds start going red?

-Hollis


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]