Re: CVS Update@ftp.linux-mips.org: linux

Re: CVS Update@ftp.linux-mips.org: linux

[Atsushi Nemoto]

It seems some necessary codes for non-r4k CPUs were lost by this change.

> CVSROOT:	/home/cvs
> Module name:	linux
> Changes by:	ralf@ftp.linux-mips.org	02/09/29 03:49:20
> 
> Modified files:
> 	arch/mips/kernel: Tag: linux_2_4 traps.c 
> 	arch/mips/mm   : Tag: linux_2_4 c-sb1.c tlb-sb1.c 
> 	arch/mips64/kernel: Tag: linux_2_4 traps.c 
> 	arch/mips64/mm : Tag: linux_2_4 Makefile c-sb1.c loadmmu.c 
> 	                 tlb-r4k.c tlb-sb1.c tlbex-r4k.S 
> Added files:
> 	arch/mips64/mm : Tag: linux_2_4 c-andes.c c-r4k.c pg-andes.c 
> 	                 pg-r4k.c tlb-andes.c 
> Removed files:
> 	arch/mips64/mm : Tag: linux_2_4 andes.c r4xx0.c 
> 
> Log message:
> 	Reorganize arch/mips64/mm along the line of it's 32-bit equivalent.

This is a patch to revert the change.

diff -ur linux-mips-cvs/arch/mips/kernel/traps.c linux.new/arch/mips/kernel/traps.c
--- linux-mips-cvs/arch/mips/kernel/traps.c	Sun Sep 29 19:45:07 2002
+++ linux.new/arch/mips/kernel/traps.c	Mon Sep 30 13:41:23 2002
@@ -1015,6 +1015,30 @@
 			memcpy((void *)(KSEG0 + 0x180), &except_vec3_r4000,
 			       0x80);
 		}
+	} else switch (mips_cpu.cputype) {
+	case CPU_SB1:
+		/*
+		 * XXX - This should be folded in to the "cleaner" handling,
+		 * above
+		 */
+		memcpy((void *)(KSEG0 + 0x180), &except_vec3_r4000, 0x80);
+		break;
+	case CPU_R6000:
+	case CPU_R6000A:
+	case CPU_R2000:
+	case CPU_R3000:
+	case CPU_R3000A:
+	case CPU_R3041:
+	case CPU_R3051:
+	case CPU_R3052:
+	case CPU_R3081:
+	case CPU_R3081E:
+	case CPU_TX3912:
+	case CPU_TX3922:
+	case CPU_TX3927:
+	case CPU_TX39XX:
+		memcpy((void *)(KSEG0 + 0x80), &except_vec3_generic, 0x80);
+		break;
 	}
 
 	if (mips_cpu.cputype == CPU_R6000 || mips_cpu.cputype == CPU_R6000A) {
---
Atsushi Nemoto

Re: CVS Update@ftp.linux-mips.org: linux

[Ralf Baechle]

On Mon, Sep 30, 2002 at 01:57:17PM +0900, Atsushi Nemoto wrote:

> It seems some necessary codes for non-r4k CPUs were lost by this change.

Mercyfully sent to /dev/null you wanted to say.

> +	case CPU_SB1:
> +		/*
> +		 * XXX - This should be folded in to the "cleaner" handling,
> +		 * above
> +		 */
> +		memcpy((void *)(KSEG0 + 0x180), &except_vec3_r4000, 0x80);
> +		break;

This was a bug. Except_vec3_r4000 is only intended for the R4000/R4400, the
only MIPS CPUs which support the virtual coherency exception.  SB1 should
use except_vec4_generic instead.

> +	case CPU_TX39XX:
> +		memcpy((void *)(KSEG0 + 0x80), &except_vec3_generic, 0x80);

Eeek, will fix.  Consider the switch gone however.

  Ralf

Re: CVS Update@ftp.linux-mips.org: linux

[Mike Nugent]

Suggestion for this bug:  Since the except_vec3_r4000 is a fix for the
4k series only, add 'default' statement as below:


On Sun, 2002-09-29 at 21:57, Atsushi Nemoto wrote:
> It seems some necessary codes for non-r4k CPUs were lost by this change.
> 
> > CVSROOT:	/home/cvs
> > Module name:	linux
> > Changes by:	ralf@ftp.linux-mips.org	02/09/29 03:49:20
> > 
> > Modified files:
> > 	arch/mips/kernel: Tag: linux_2_4 traps.c 
> > 	arch/mips/mm   : Tag: linux_2_4 c-sb1.c tlb-sb1.c 
> > 	arch/mips64/kernel: Tag: linux_2_4 traps.c 
> > 	arch/mips64/mm : Tag: linux_2_4 Makefile c-sb1.c loadmmu.c 
> > 	                 tlb-r4k.c tlb-sb1.c tlbex-r4k.S 
> > Added files:
> > 	arch/mips64/mm : Tag: linux_2_4 c-andes.c c-r4k.c pg-andes.c 
> > 	                 pg-r4k.c tlb-andes.c 
> > Removed files:
> > 	arch/mips64/mm : Tag: linux_2_4 andes.c r4xx0.c 
> > 
> > Log message:
> > 	Reorganize arch/mips64/mm along the line of it's 32-bit equivalent.
> 
> This is a patch to revert the change.
> 
> diff -ur linux-mips-cvs/arch/mips/kernel/traps.c linux.new/arch/mips/kernel/traps.c
> --- linux-mips-cvs/arch/mips/kernel/traps.c	Sun Sep 29 19:45:07 2002
> +++ linux.new/arch/mips/kernel/traps.c	Mon Sep 30 13:41:23 2002
> @@ -1015,6 +1015,30 @@
>  			memcpy((void *)(KSEG0 + 0x180), &except_vec3_r4000,
>  			       0x80);
>  		}
> +	} else switch (mips_cpu.cputype) {
> +	case CPU_SB1:
> +		/*
> +		 * XXX - This should be folded in to the "cleaner" handling,
> +		 * above
> +		 */
> +		memcpy((void *)(KSEG0 + 0x180), &except_vec3_r4000, 0x80);
> +		break;
> +	case CPU_R6000:
> +	case CPU_R6000A:
> +	case CPU_R2000:
> +	case CPU_R3000:
> +	case CPU_R3000A:
> +	case CPU_R3041:
> +	case CPU_R3051:
> +	case CPU_R3052:
> +	case CPU_R3081:
> +	case CPU_R3081E:
> +	case CPU_TX3912:
> +	case CPU_TX3922:
> +	case CPU_TX3927:
> +	case CPU_TX39XX:
+     default:
> +		memcpy((void *)(KSEG0 + 0x80), &except_vec3_generic, 0x80);
> +		break;
>  	}
>  
>  	if (mips_cpu.cputype == CPU_R6000 || mips_cpu.cputype == CPU_R6000A) {
> ---
> Atsushi Nemoto
> 
> 
-- 
Mike Nugent
Programmer/Author
mike@illuminatus.org
"I believe the use of noise to make music will increase until we reach a
music produced through the aid of electrical instruments which will make
available for musical purposes any and all sounds that can be heard."
 -- composer John Cage, 1937

Re: CVS Update@ftp.linux-mips.org: linux

[Ralf Baechle]

On Mon, Sep 30, 2002 at 06:46:41PM -0700, Mike Nugent wrote:

> Suggestion for this bug:  Since the except_vec3_r4000 is a fix for the
> 4k series only, add 'default' statement as below:

The whole switch statement there is ever inflating and covers only a tiny
aspect of the CPU so I simply killed it.

  Ralf

Re: CVS Update@ftp.linux-mips.org: linux

[Atsushi Nemoto]

>>>>> On Mon, 30 Sep 2002 18:53:47 +0200, ralf@linux-mips.org said:
> Log message:
> 	More cache code cleanup.

This commit contains following change.  It seems 'addr' argument is
not used.  Isn't this a mistake?

> @@ -123,15 +72,14 @@ static inline void protected_flush_icach
>  	__asm__ __volatile__(
>  		".set noreorder\n\t"
>  		".set mips3\n"
> -		"1:\tcache %1,(%0)\n"
> +		"1:\tcache %0,(%1)\n"
>  		"2:\t.set mips0\n\t"
>  		".set reorder\n\t"
>  		".section\t__ex_table,\"a\"\n\t"
>  		STR(PTR)"\t1b,2b\n\t"
>  		".previous"
>  		:
> -		: "r" (addr),
> -		  "i" (Hit_Invalidate_I));
> +		: "i" (Hit_Invalidate_I), "i" (Hit_Invalidate_I));
>  }

---
Atsushi Nemoto

Re: CVS Update@ftp.linux-mips.org: linux

[Ralf Baechle]

On Thu, Oct 03, 2002 at 12:17:05PM +0900, Atsushi Nemoto wrote:

> This commit contains following change.  It seems 'addr' argument is
> not used.  Isn't this a mistake?

Yes, it is.  Fixed,

  Ralf

Re: CVS Update@ftp.linux-mips.org: linux

[Maciej W. Rozycki]

On Sat, 26 Oct 2002 ralf@linux-mips.org wrote:

> Log message:
> 	N32 ...

 Do you really think it's the way to go???

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +