Linux 2.6 sources for MPC852T processor

Linux 2.6 sources for MPC852T processor

[Chandrasekhar Nagaraj]

[-- Attachment #1: Type: text/html, Size: 668 bytes --]

Re: Linux 2.6 sources for MPC852T processor

[David Jander]

On Wednesday 26 April 2006 11:50, Chandrasekhar Nagaraj wrote:
> <DIV style="font-family:Arial,sans-serif;"><DIV>Hi,</DIV>
> <DIV>I have a customized board based on the MPC852T based processor.</DIV>
> <DIV>I intend to develop a BSP for this board.</DIV>
> <DIV>&nbsp;</DIV>
> <DIV>Does 2.6.16 from the kernel.org support this processor?</DIV>

First of all, please avoid HTML in e-mail messages. It is hard to read, and 
normally banned on mailing list such as this one.

Yes, MPC852T is supported, although I might add that I have been using 2.6.14 
and 2.6.15 sucessfully with our own MPC852T-based board, but 2.6.16 did not 
boot and as of today I don't know why, or whether this is an issue at all 
with boards other than ours.

> <DIV>&nbsp;</DIV>
> <DIV>In the 2.6.16 sources I found support for CONFIG_8xx. Does this mean
> that 852T processor is also supported?</DIV> <DIV>&nbsp;</DIV>

Yes. Look at BSP stuff for other 8xx boards to learn how to port yours. Keep 
an eye on the new platform_bus stuff, that's currently being implemented for 
different drivers and subsystmes for powerpc (this could be the reason, our 
own BSP stuff stopped working with 2.6.16, btw).
Also a transition from /arch/ppc and /arch/ppc64 towards the 
common /arch/powerpc is in progress, and therefore some things might be in a 
state of flux between released versions of the kernel. As of today (kernel 
2.6.16) the architecture you need to use is still /arch/ppc.
Good luck!

Greetings,

-- 
David Jander

Re: Linux 2.6 sources for MPC852T processor

[Dan Malek]

On Apr 26, 2006, at 7:50 AM, David Jander wrote:

> Yes, MPC852T is supported, although I might add that I have been  
> using 2.6.14
> and 2.6.15 sucessfully with our own MPC852T-based board, but 2.6.16  
> did not
> boot and as of today I don't know why, or whether this is an issue  
> at all
> with boards other than ours.

There is a horrible bug in the 8xx TLB miss handler that is in the  
2.6.16
sources.  I don't know when it appeared.  Enable the CPU6 Errata
workaround to see if that solves the problem and please report back
to me.  I'm working on a solution.

Thanks.

	-- Dan

Re: Linux 2.6 sources for MPC852T processor

[David Jander]

On Thursday 27 April 2006 01:32, Dan Malek wrote:
> There is a horrible bug in the 8xx TLB miss handler that is in the
> 2.6.16
> sources.  I don't know when it appeared.  Enable the CPU6 Errata
> workaround to see if that solves the problem and please report back
> to me.  I'm working on a solution.

I received instructions from Marcelo to reverse two patches he sent me this 
morning, that did solve the problem. In fact those patches changed some 
TLB-related assembly code in head_8xx.S

Greetings,

-- 
David Jander

Re: Linux 2.6 sources for MPC852T processor

[Marcelo Tosatti]

On Wed, Apr 26, 2006 at 07:32:13PM -0400, Dan Malek wrote:
> 
> On Apr 26, 2006, at 7:50 AM, David Jander wrote:
> 
> > Yes, MPC852T is supported, although I might add that I have been  
> > using 2.6.14
> > and 2.6.15 sucessfully with our own MPC852T-based board, but 2.6.16  
> > did not
> > boot and as of today I don't know why, or whether this is an issue  
> > at all
> > with boards other than ours.
> 
> There is a horrible bug in the 8xx TLB miss handler that is in the  
> 2.6.16
> sources.  I don't know when it appeared.  Enable the CPU6 Errata
> workaround to see if that solves the problem and please report back
> to me.  I'm working on a solution.

Hi Dan,

Shame on me! I screwed up.

David, can you please try the following patch on top of vanilla v2.6.16?

The large TLB change introduced in v2.6.16 (which is broken as is for
configurations without CPU6 errata enabled as Dan noted) should boost
performance significantly, so you might want to rerun the lmbench
tests...

diff --git a/arch/ppc/kernel/head_8xx.S b/arch/ppc/kernel/head_8xx.S
index ec53c7d..7a2f205 100644
--- a/arch/ppc/kernel/head_8xx.S
+++ b/arch/ppc/kernel/head_8xx.S
@@ -355,9 +355,7 @@ #endif
 
 	. = 0x1200
 DataStoreTLBMiss:
-#ifdef CONFIG_8xx_CPU6
 	stw	r3, 8(r0)
-#endif
 	DO_8xx_CPU6(0x3f80, r3)
 	mtspr	SPRN_M_TW, r10	/* Save a couple of working registers */
 	mfcr	r10
@@ -417,9 +415,7 @@ #endif
 	lwz	r11, 0(r0)
 	mtcr	r11
 	lwz	r11, 4(r0)
-#ifdef CONFIG_8xx_CPU6
 	lwz	r3, 8(r0)
-#endif
 	rfi
 
 /* This is an instruction TLB error on the MPC8xx.  This could be due

Re: Linux 2.6 sources for MPC852T processor

[David Jander]

Hi,

I did the following:

 $ patch -p1 <tlbfix-mt.patch

(This is the patch you sent me)

 $ cg-seek v2.6.16
 Warning: uncommitted local changes, trying to bring them along
 patching file arch/ppc/8xx_io/Kconfig
 patching file arch/ppc/8xx_io/Makefile
 The next patch would create the file arch/ppc/8xx_io/cpm_spi.c,
 which already exists!  Skipping patch.
 1 out of 1 hunk ignored -- saving rejects to file 
arch/ppc/8xx_io/cpm_spi.c.rej
 patching file arch/ppc/Kconfig
 Hunk #2 succeeded at 927 (offset -16 lines).
 patching file arch/ppc/kernel/head_8xx.S
 patching file arch/ppc/platforms/Makefile
 Hunk #1 succeeded at 37 (offset -1 lines).
 The next patch would create the file arch/ppc/platforms/prtppc.h,
 which already exists!  Skipping patch.
 1 out of 1 hunk ignored -- saving rejects to file 
arch/ppc/platforms/prtppc.h.rej
 The next patch would create the file arch/ppc/platforms/prtppc_serial.h,
 which already exists!  Skipping patch.
 1 out of 1 hunk ignored -- saving rejects to file 
arch/ppc/platforms/prtppc_serial.h.rej
 The next patch would create the file arch/ppc/platforms/prtppc_setup.c,
 which already exists!  Skipping patch.
 1 out of 1 hunk ignored -- saving rejects to file 
arch/ppc/platforms/prtppc_setup.c.rej
 patching file arch/ppc/syslib/m8xx_setup.c
 patching file drivers/input/keyboard/Kconfig
 patching file drivers/input/keyboard/Makefile
 patching file drivers/usb/Kconfig
 patching file drivers/usb/host/isp116x.h
 patching file drivers/video/Kconfig
 patching file drivers/video/Makefile
 The next patch would create the file include/asm-ppc/cpm_spi.h,
 which already exists!  Skipping patch.
 1 out of 1 hunk ignored -- saving rejects to file 
include/asm-ppc/cpm_spi.h.rej
 patching file include/asm-ppc/mpc8xx.h
 patching file include/asm-ppc/ppc_sys.h
 patching file include/asm-ppc/serial.h
 Adding file arch/ppc/8xx_io/cpm_spi.c
 Adding file arch/ppc/platforms/prtppc.h
 Adding file arch/ppc/platforms/prtppc_serial.h
 Adding file arch/ppc/platforms/prtppc_setup.c
 Adding file include/asm-ppc/cpm_spi.h
 On commit 7705a8792b0fc82fd7d4dd923724606bbfd9fb20

All the abve is BSP stuff and drivers for our board, plus macrelo's patch.
To verify, I did:

 $ cg-diff arch/ppc/kernel/head_8xx.S

 diff --git a/arch/ppc/kernel/head_8xx.S b/arch/ppc/kernel/head_8xx.S
 --- a/arch/ppc/kernel/head_8xx.S
 +++ b/arch/ppc/kernel/head_8xx.S
 @@ -357,9 +357,7 @@ InstructionTLBMiss: 
 
         . = 0x1200
  DataStoreTLBMiss:
 -#ifdef CONFIG_8xx_CPU6
         stw     r3, 8(r0)
 -#endif
         DO_8xx_CPU6(0x3f80, r3)
         mtspr   SPRN_M_TW, r10  /* Save a couple of working registers */
         mfcr    r10
 @@ -419,9 +417,7 @@ DataStoreTLBMiss:
         lwz     r11, 0(r0)
         mtcr    r11
         lwz     r11, 4(r0)
 -#ifdef CONFIG_8xx_CPU6
         lwz     r3, 8(r0)
 -#endif
         rfi 
 
  /* This is an instruction TLB error on the MPC8xx.  This could be due
  
Looks good.

 $ make uImage

Kernel doesn't boot, no output produced :-(

Greetings,

-- 
David Jander