64-bit kernel

64-bit kernel

[Carsten Langgaard]

I don't know if anymore has a interest in the 64-bit kernel, but I just
found this bug (see patch below).
It would be nice to know, how many are interested in the 64-bit kernel
and who actually got something running.
So please rise you voice.

/Carsten

Index: include/asm-mips64/exception.h
===================================================================
RCS file:
/home/repository/sw/linux-2.4.18/include/asm-mips64/exception.h,v
retrieving revision 1.1.1.1
diff -u -r1.1.1.1 exception.h
--- include/asm-mips64/exception.h      4 Mar 2002 11:13:25 -0000
1.1.1.1
+++ include/asm-mips64/exception.h      18 Jun 2002 12:18:40 -0000
@@ -28,7 +28,7 @@

        .macro  __build_clear_fpe
        cfc1    a1, fcr31
-       li      a2, ~(0x3f << 13)
+       li      a2, ~(0x3f << 12)
        and     a2, a1
        ctc1    a2, fcr31
        STI



--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com

Re: 64-bit kernel

[Justin Carlson]

On Tue, 2002-06-18 at 05:33, Carsten Langgaard wrote:
> I don't know if anymore has a interest in the 64-bit kernel, but I just
> found this bug (see patch below).
> It would be nice to know, how many are interested in the 64-bit kernel
> and who actually got something running.
> So please rise you voice.
> 

Been running 64-bit stuff here, but nothing even remotely fpu intensive.
It's quite possible we'd never run into this case.

-Justin

Re: 64-bit kernel

[Jun Sun]

Carsten Langgaard wrote:

> I don't know if anymore has a interest in the 64-bit kernel, but I just
> found this bug (see patch below).
> It would be nice to know, how many are interested in the 64-bit kernel
> and who actually got something running.
> So please rise you voice.
> 


Actually I think there is an increased interestes to 64bit kernel with the 
recent introduction of RM7K and bcm1250 CPUs.

If 64bit kernel gets mature, I think two years down the road we will see 
mid-range CPUs edging into that domain as well.

(Jun is raising his voice....)

Jun


> /Carsten
> 
> Index: include/asm-mips64/exception.h
> ===================================================================
> RCS file:
> /home/repository/sw/linux-2.4.18/include/asm-mips64/exception.h,v
> retrieving revision 1.1.1.1
> diff -u -r1.1.1.1 exception.h
> --- include/asm-mips64/exception.h      4 Mar 2002 11:13:25 -0000
> 1.1.1.1
> +++ include/asm-mips64/exception.h      18 Jun 2002 12:18:40 -0000
> @@ -28,7 +28,7 @@
> 
>         .macro  __build_clear_fpe
>         cfc1    a1, fcr31
> -       li      a2, ~(0x3f << 13)
> +       li      a2, ~(0x3f << 12)
>         and     a2, a1
>         ctc1    a2, fcr31
>         STI
> 
> 
> 
> --
> _    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
> |\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
> | \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
>   TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
>                    Denmark             http://www.mips.com
> 
> 
> 
> 

Re: 64-bit kernel

[Ralf Baechle]

On Tue, Jun 18, 2002 at 02:33:50PM +0200, Carsten Langgaard wrote:

> I don't know if anymore has a interest in the 64-bit kernel, but I just
> found this bug (see patch below).
> It would be nice to know, how many are interested in the 64-bit kernel
> and who actually got something running.
> So please rise you voice.

More and more systems run into the limits of system memory so once it's
practically unavoidable.  Take the existence of highmem as a proof for
it.

  Ralf

Re: 64-bit kernel

[Ralf Baechle]

On Tue, Jun 18, 2002 at 02:33:50PM +0200, Carsten Langgaard wrote:

> X-Mailer: Mozilla 4.77 [en] (X11; U; SunOS 5.8 sun4u)
> 
> I don't know if anymore has a interest in the 64-bit kernel, but I just
> found this bug (see patch below).
> It would be nice to know, how many are interested in the 64-bit kernel
> and who actually got something running.
> So please rise you voice.

The patch is ok but please do not send inline patches with Mozilla - it
does bad things like converting tabs to spaces and wordwrapping to
patches.

  Ralf

Re: 64-bit kernel

[Ralf Baechle]

On Tue, Jun 18, 2002 at 09:03:18AM -0700, Justin Carlson wrote:

> On Tue, 2002-06-18 at 05:33, Carsten Langgaard wrote:
> > I don't know if anymore has a interest in the 64-bit kernel, but I just
> > found this bug (see patch below).
> > It would be nice to know, how many are interested in the 64-bit kernel
> > and who actually got something running.
> > So please rise you voice.
> 
> Been running 64-bit stuff here, but nothing even remotely fpu intensive.
> It's quite possible we'd never run into this case.

At this time probably most 64-bit kernels are running on a certain 64-bit
CPU with it's hardware fp disabled so nobody ever saw this one.

  Ralf

Re: 64-bit kernel

[Carsten Langgaard]

I'm trying to compile glibc natively using my 64-bit kernel, but it fails with
the following message:

Out of Memory: Killed process 641 (qmgr).
Out of Memory: Killed process 642 (tlsmgr).
Out of Memory: Killed process 378 (portmap).
Out of Memory: Killed process 9363 (cc1).
Out of Memory: Killed process 9363 (cc1).

So there may be a memory leak problem in 64-bit kernel.
Has anyone seen this ?

/Carsten


Ralf Baechle wrote:

> On Tue, Jun 18, 2002 at 09:03:18AM -0700, Justin Carlson wrote:
>
> > On Tue, 2002-06-18 at 05:33, Carsten Langgaard wrote:
> > > I don't know if anymore has a interest in the 64-bit kernel, but I just
> > > found this bug (see patch below).
> > > It would be nice to know, how many are interested in the 64-bit kernel
> > > and who actually got something running.
> > > So please rise you voice.
> >
> > Been running 64-bit stuff here, but nothing even remotely fpu intensive.
> > It's quite possible we'd never run into this case.
>
> At this time probably most 64-bit kernels are running on a certain 64-bit
> CPU with it's hardware fp disabled so nobody ever saw this one.
>
>   Ralf

--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com

Re: 64-bit kernel

[Ralf Baechle]

On Wed, Jun 19, 2002 at 01:07:46PM +0200, Carsten Langgaard wrote:

> I'm trying to compile glibc natively using my 64-bit kernel, but it fails with
> the following message:
> 
> Out of Memory: Killed process 641 (qmgr).
> Out of Memory: Killed process 642 (tlsmgr).
> Out of Memory: Killed process 378 (portmap).
> Out of Memory: Killed process 9363 (cc1).
> Out of Memory: Killed process 9363 (cc1).
> 
> So there may be a memory leak problem in 64-bit kernel.
> Has anyone seen this ?

No.  The entire Redhat 7.0 stuff I built was done exclusivly on a 64-bit
kernel (with 2GB RAM however ...) and no such problem ever, even after
long uptimes.

  Ralf

64-bit kernel

[Carsten Langgaard]

[-- Attachment #1: Type: text/plain, Size: 437 bytes --]

Here is a patch, which have a more cosmetic value rather than fixing
anything, but it makes the compiler happier.

/Carsten


--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com



[-- Attachment #2: cosmetic.patch --]
[-- Type: text/plain, Size: 1767 bytes --]

Index: include/asm-mips64/irq.h
===================================================================
RCS file: /cvs/linux/include/asm-mips64/irq.h,v
retrieving revision 1.4
diff -u -r1.4 irq.h
--- include/asm-mips64/irq.h	2001/03/02 03:04:55	1.4
+++ include/asm-mips64/irq.h	2002/08/06 11:35:06
@@ -46,6 +46,13 @@
 struct irqaction;
 extern int i8259_setup_irq(int irq, struct irqaction * new);
 extern void disable_irq(unsigned int);
+
+#ifndef CONFIG_NEW_IRQ
+#define disable_irq_nosync      disable_irq
+#else
+extern void disable_irq_nosync(unsigned int);
+#endif
+
 extern void enable_irq(unsigned int);
 
 /* Machine specific interrupt initialization  */
Index: include/asm-mips64/pgtable.h
===================================================================
RCS file: /cvs/linux/include/asm-mips64/pgtable.h,v
retrieving revision 1.47.2.16
diff -u -r1.47.2.16 pgtable.h
--- include/asm-mips64/pgtable.h	2002/08/05 23:53:39	1.47.2.16
+++ include/asm-mips64/pgtable.h	2002/08/06 11:35:07
@@ -77,10 +77,6 @@
 
 #else
 
-extern void (*_flush_icache_all)(void);
-extern void (*_flush_icache_range)(unsigned long start, unsigned long end);
-extern void (*_flush_icache_page)(struct vm_area_struct *vma, struct page *page);
-
 #define flush_cache_mm(mm)		_flush_cache_mm(mm)
 #define flush_cache_range(mm,start,end)	_flush_cache_range(mm,start,end)
 #define flush_cache_page(vma,page)	_flush_cache_page(vma, page)
@@ -89,12 +85,6 @@
 #define flush_icache_user_range(vma, page, addr, len) \
 	flush_icache_page((vma), (page))
 #define flush_icache_page(vma, page)	_flush_icache_page(vma, page)
-#ifdef CONFIG_VTAG_ICACHE
-#define flush_icache_all()		_flush_icache_all()
-#else
-#define flush_icache_all()		do { } while(0)
-#endif
-
 
 #endif /* !CONFIG_CPU_R10000 */
 

Re: 64-bit kernel

[Ralf Baechle]

On Tue, Aug 06, 2002 at 02:05:25PM +0200, Carsten Langgaard wrote:

> Here is a patch, which have a more cosmetic value rather than fixing
> anything, but it makes the compiler happier.

Happy compilers are good compilers :-)

Applied with changes,

  Ralf