Re: ZMII Code Panics on 440GP Ebony

2 messages, 2 authors, 2002-05-28 · open the first message on its own page

Re: ZMII Code Panics on 440GP Ebony

From: Matt Porter <hidden>
Date: 2002-05-24 20:47:31 

On Fri, May 24, 2002 at 12:17:47PM -0700, Roland Dreier wrote:
quoted
quoted
quoted
quoted
quoted
"Matt" == Matt Porter [off-list ref] writes:
    Matt> I started some things but have been busy with functionality
    Matt> for some time now.  I'd be interested in any cleanup patches
    Matt> you would produce.  The idea is to separate ibm440.h,
    Matt> ibm440gp.h, and <board>.h as appropriate.  As you've
    Matt> noticed, I took some shortcuts in the interest of getting
    Matt> the core code in front of everybody quickly. :)

OK, I will send you some trivial stuff soon.  I do have one question:
what belongs in ibm440.h (as opposed to ibm440gp.h)?  The only
documentation I have is specifically for the 440GP (other than a
presentation on the 440GX) -- how do we know what is common to the 440
family?
Well, as a starting point, most of what is in ibm440.h should be
in ibm440gp.h.  The board stuff will go to ebony.h, the dcr macros
to ibm440.h, and all the on-chip defs (UIC, MEM controller, MAL,
PLB, PCI-X, and peripheral nums) go to ibm440gp.h.  You can
look at the 440 core UM to see what is part of the core.  Considering
the nature of a Book E part, the core defines the existence of
DCRs, but they will vary between implementations.  Looking at
the 405 implementations should give a clue as to what kind
of 440 implementations we can expect to see.

For now, we can ignore the 440GX except keeping in mind that it
is a different set of peripherals.  It's safe to assume that
things like the UIC and memory controller will have similar
if not identical definitions in ibm440gx.h but there's no
point worrying too much until the docs are available. :)
To start off we'll probably copy ibm440gp.h, make additions
and then try to abstract into a shared file later...that's
some time down the road yet though.

Regards,
--
Matt Porter
porter@cox.net
This is Linux Country. On a quiet night, you can hear Windows reboot.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Ebony crash fix & 440GP cleanup step one

From: Roland Dreier <hidden>
Date: 2002-05-28 22:04:22 

This email includes two patches.  First, a very small change to
ocp.h.  Somewhere in the recent cleanup, paddr got changed from a
phys_addr_t to a void *; this doesn't work on 440GP because we have
36-bit physical addresses.  (You get a machine check when the kernel
tries to access the ZMII)

Second, the beginning of moving generic IBM 440GP stuff out of
ebony.c/ebony.h and into platforms/ibm440gp.h.  I tested this on my
Ebony and it boots and runs fine.  I didn't go too far because I want
to make sure everyone feels what I'm doing is the right way to go.
Please let me know what you think.

Best,
Roland

# This is a BitKeeper generated patch for the following project:
# Project Name: Linux 2.4 for PowerPC development tree
# This patch format is intended for GNU patch command version 2.5 or higher.
# This patch includes the following deltas:
#	           ChangeSet	1.1039  -> 1.1040
#	include/asm-ppc/ocp.h	1.16    -> 1.17
#
# The following is the BitKeeper ChangeSet Log
# --------------------------------------------
# 02/05/28	roland@gold.topspincom.com	1.1040
# Fix paddr in ocp.h
# --------------------------------------------
#
diff -Nru a/include/asm-ppc/ocp.h b/include/asm-ppc/ocp.h
--- a/include/asm-ppc/ocp.h	Tue May 28 14:57:19 2002
+++ b/include/asm-ppc/ocp.h	Tue May 28 14:57:19 2002
@@ -117,7 +117,7 @@
 	char name[16];
 	u16 num;
 	enum ocp_type type;	/* OCP devive type */
-	void *paddr;
+        phys_addr_t paddr;
 	void *vaddr;
 	u32 flags;
 	struct irq_resources irq_resource[MAX_EMACS][OCP_MAX_IRQS];


# This is a BitKeeper generated patch for the following project:
# Project Name: Linux 2.4 for PowerPC development tree
# This patch format is intended for GNU patch command version 2.5 or higher.
# This patch includes the following deltas:
#	           ChangeSet	1.1025  -> 1.1025.1.1
#	include/asm-ppc/ibm4xx.h	1.15    -> 1.16
#	include/asm-ppc/ibm440.h	1.8     -> 1.9
#	arch/ppc/platforms/ebony.h	1.8     -> 1.9
#	arch/ppc/platforms/ibm440gp.c	1.11    -> 1.12
#	arch/ppc/platforms/ebony.c	1.9     -> 1.10
#	               (new)	        -> 1.2     arch/ppc/platforms/ibm440gp.h
#
# The following is the BitKeeper ChangeSet Log
# --------------------------------------------
# 02/05/27	dgibson@zax.zax	1.1026
# Change OCP ENET driver to use the proper PCI_DMA_* direction constants
# as arguments to consistent_sync(), instead of defining its own.  TODO:
# change consistent_sync() and/or other code to remove the need for the
# PCI constants.
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1027
# Minor 4xx serial fixups from Scott Anderson [off-list ref].
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1028
# We don't really want to be using PCI_DMA_* directional constants for consistent_sync() since it is not necessarily PCI.
# Cset exclude: dgibson@zax.zax|ChangeSet|20020527053847|07881
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1029
# Convert IBM OCP IDE to use the DMA_* directional constants.
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1030
# Odd.  bk didn't fix this in the cset -x.  Bring back include/asm-ppc/ocp-dma.h
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1031
# Well, that's what I get for not reading my email before working.. Backout my last changes.
# Cset exclude: trini@bill-the-cat.bloom.county|ChangeSet|20020527160528|10008
# Cset exclude: trini@bill-the-cat.bloom.county|ChangeSet|20020527160119|11671
# Cset exclude: trini@bill-the-cat.bloom.county|ChangeSet|20020527155917|11666
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1032
# ...and bk didn't catch the re-deletion again either.  Larry, are you
# snooping this? :)
# --------------------------------------------
# 02/05/27	trini@opus.bloom.county	1.118.1.516
# Critical fix from Paul (from linux-kernel):
# The patch below fixes two bugs in the PPC code.  The first bug was
# caused by an exception table entry pointing to the wrong instruction.
# As a result of this bug, a copy_to_user() with an unmapped or
# inaccessible destination address could cause an oops.  The second bug
# was that clear_user on PPC was returning -EFAULT rather than the
# number of bytes not cleared.
#
# Thanks to Rusty Russell for demonstrating the existence of these bugs
# with his test-read.c program.
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1033
# Merge in recent linuxppc_2_4 changes.
# --------------------------------------------
# 02/05/27	trini@bill-the-cat.bloom.county	1.1034
# Actually remove the right file.  It'd be nice if cset -x did this right
# to start with...
# --------------------------------------------
# 02/05/28	trini@opus.bloom.county	1.1035
# Add support for the Davicom DM9131 PHY to the 8260 FCC driver.  From
# Murray Jensen [off-list ref]
# --------------------------------------------
# 02/05/28	trini@opus.bloom.county	1.1036
# Better describe (and modify slightly) how we come up with 3 ethernet
# addresses from 1 in the 8260 FCC enet driver.  From
# Murray Jensen [off-list ref]
# --------------------------------------------
# 02/05/28	trini@bill-the-cat.bloom.county	1.1037
# Lots of OCP updates from Armin:
# Rename functions to <class>_<action>_<object>.
# Change ocp_driver to ocp_dev.
# Added type_info struct for the name/desc.
# --------------------------------------------
# 02/05/28	trini@bill-the-cat.bloom.county	1.1038
# Add in OCP documentation, from Armin.
# --------------------------------------------
# 02/05/28	roland@gold.topspincom.com	1.1025.1.1
# Begin reorganization of IBM 440GP code out of Ebony-specific files.
# --------------------------------------------
#
diff -Nru a/arch/ppc/platforms/ebony.c b/arch/ppc/platforms/ebony.c
--- a/arch/ppc/platforms/ebony.c	Tue May 28 14:57:14 2002
+++ b/arch/ppc/platforms/ebony.c	Tue May 28 14:57:14 2002
@@ -44,15 +44,8 @@
 #endif /* CONFIG_PPC_RTC */
 #include <asm/bootinfo.h>
 #include <asm/ppc4xx_pic.h>
-#include <asm/ocp.h>
-
-#include <platforms/ibm_ocp.h>
-#include <platforms/ebony.h>

 extern void abort(void);
-extern struct emac_regs *EMAC_ADDR[];
-extern struct zmii_regs *ZMII_ADDR[];
-extern struct ocp_def core_ocp[];

 /* Global Variables */
 unsigned char __res[sizeof (bd_t)];
@@ -208,8 +201,6 @@
 	ppc_md.pci_map_irq = ebony_map_irq;
 }

-extern char cmd_line[];
-
 #ifdef CONFIG_PPC_RTC
 TODC_ALLOC();
 #endif /* CONFIG_PPC_RTC */
@@ -234,16 +225,6 @@
 			ioremap64(EBONY_RTC_ADDR, EBONY_RTC_SIZE),
 			8);
 #endif /* CONFIG_PPC_RTC */
-
-	/* Setup EMAC access */
-	EMAC_ADDR[0] = (struct emac_regs *)
-		ioremap64(EBONY_EMAC0_ADDR, EBONY_EMAC_SIZE);
-	EMAC_ADDR[1] = (struct emac_regs *)
-		ioremap64(EBONY_EMAC1_ADDR, EBONY_EMAC_SIZE);
-
-	/* Setup ZMII access */
-	ZMII_ADDR[0] = (struct zmii_regs *)
-		ioremap64(EBONY_ZMII_ADDR, EBONY_ZMII_SIZE);

 	/* init to some ~sane value until calibrate_delay() runs */
         loops_per_jiffy = 50000000/HZ;
diff -Nru a/arch/ppc/platforms/ebony.h b/arch/ppc/platforms/ebony.h
--- a/arch/ppc/platforms/ebony.h	Tue May 28 14:57:14 2002
+++ b/arch/ppc/platforms/ebony.h	Tue May 28 14:57:14 2002
@@ -21,7 +21,7 @@
 #define __ASM_EBONY_H__

 #include <linux/config.h>
-#include <asm/ibm440.h>
+#include <platforms/ibm440gp.h>

 /* F/W TLB mapping used in bootloader glue to reset EMAC */
 #define EBONY_EMAC0_MR0		0xE0000800
@@ -37,23 +37,6 @@
 /* Default clock rates for Rev. B and Rev. C silicon */
 #define EBONY_440GP_RB_SYSCLK	33000000
 #define EBONY_440GP_RC_SYSCLK	400000000
-
-/* EMAC location */
-#define EBONY_EMAC0_ADDR	0x0000000140000800
-#define EBONY_EMAC1_ADDR	0x0000000140000900
-#define EBONY_EMAC_SIZE		0x70
-
-/* ZMII location */
-#define EBONY_ZMII_ADDR		0x0000000140000780
-#define EBONY_ZMII_SIZE		0x0c
-
-/* I2C location */
-#define EBONY_IIC0_ADDR		0x40000400
-#define EBONY_IIC1_ADDR		0x40000500
-
-/* RTC/NVRAM location */
-#define EBONY_RTC_ADDR		0x0000000148000000
-#define EBONY_RTC_SIZE		0x2000

 /*
  * Serial port defines
diff -Nru a/arch/ppc/platforms/ibm440gp.c b/arch/ppc/platforms/ibm440gp.c
--- a/arch/ppc/platforms/ibm440gp.c	Tue May 28 14:57:14 2002
+++ b/arch/ppc/platforms/ibm440gp.c	Tue May 28 14:57:14 2002
@@ -61,23 +61,21 @@
 struct ocp_def core_ocp[] = {
 	{UART, UART0_IO_BASE, UART0_INT},
 	{UART, UART1_IO_BASE, UART1_INT},
-	{IIC, EBONY_IIC0_ADDR, IIC0_IRQ},
-	{IIC, EBONY_IIC1_ADDR, IIC1_IRQ},
+	{IIC, PPC440GP_IIC0_ADDR, IIC0_IRQ},
+	{IIC, PPC440GP_IIC1_ADDR, IIC1_IRQ},
 #if 0 /* FIXME: Undefined */
 	{GPIO, GPIO0_BASE, OCP_IRQ_NA},
 	{PCI, PCI0_BASE, OCP_IRQ_NA},
 	{OPB, OPB0_BASE, OCP_IRQ_NA},
 #endif
-	{EMAC, EBONY_EMAC0_ADDR, OCP_IRQ_NA},
-	{EMAC, EBONY_EMAC1_ADDR, OCP_IRQ_NA},
-	{ZMII, EBONY_ZMII_ADDR, OCP_IRQ_NA},
+	{EMAC, PPC440GP_EMAC0_ADDR, OCP_IRQ_NA},
+	{EMAC, PPC440GP_EMAC1_ADDR, OCP_IRQ_NA},
+	{ZMII, PPC440GP_ZMII_ADDR, OCP_IRQ_NA},
 	{OCP_NULL_TYPE, 0x0, OCP_IRQ_NA},

 };

-struct emac_regs *EMAC_ADDR[EMAC_NUMS];
-struct zmii_regs *ZMII_ADDR[ZMII_NUMS];
 struct iic_regs *IIC_ADDR[] = {
-	(struct iic_regs *) EBONY_IIC0_ADDR,
-	(struct iic_regs *) EBONY_IIC1_ADDR
+	(struct iic_regs *) PPC440GP_IIC0_ADDR,
+	(struct iic_regs *) PPC440GP_IIC1_ADDR
 };
diff -Nru a/arch/ppc/platforms/ibm440gp.h b/arch/ppc/platforms/ibm440gp.h
--- /dev/null	Wed Dec 31 16:00:00 1969
+++ b/arch/ppc/platforms/ibm440gp.h	Tue May 28 14:57:14 2002
@@ -0,0 +1,58 @@
+/*
+ * arch/ppc/platforms/ibm440gp.h
+ *
+ * PPC440GP definitions
+ *
+ * Roland Dreier <roland@digitalvampire.org>
+ *
+ * Copyright 2002 Roland Dreier
+ *
+ * This program is free software; you can redistribute  it and/or modify it
+ * under  the terms of  the GNU General  Public License as published by the
+ * Free Software Foundation;  either version 2 of the  License, or (at your
+ * option) any later version.
+ *
+ * This file contains code that was originally in the files ibm440.h
+ * and ebony.h, which were written by Matt Porter of MontaVista Software Inc.
+ *
+ */
+
+#ifdef __KERNEL__
+#ifndef __PPC_PLATFORMS_IBM440GP_H
+#define __PPC_PLATFORMS_IBM440GP_H
+
+#include <linux/config.h>
+
+#define EMAC_NUMS		2
+#define UART_NUMS		2
+#define ZMII_NUMS		1
+#define IIC_NUMS		2
+#define IIC0_IRQ		2
+#define IIC1_IRQ		3
+
+/* EMAC location */
+#define PPC440GP_EMAC0_ADDR	0x0000000140000800
+#define PPC440GP_EMAC1_ADDR	0x0000000140000900
+#define PPC440GP_EMAC_SIZE	0x70
+
+/* ZMII location */
+#define PPC440GP_ZMII_ADDR	0x0000000140000780
+#define PPC440GP_ZMII_SIZE	0x0c
+
+/* I2C location */
+#define PPC440GP_IIC0_ADDR	0x40000400
+#define PPC440GP_IIC1_ADDR	0x40000500
+
+/* RTC/NVRAM location */
+#define PPC440GP_RTC_ADDR	0x0000000148000000
+#define PPC440GP_RTC_SIZE	0x2000
+
+/*
+ * Serial port defines
+ */
+#define RS_TABLE_SIZE	2
+
+#include <asm/ibm440.h>
+
+#endif /* __PPC_PLATFORMS_IBM440GP_H */
+#endif /* __KERNEL__ */
diff -Nru a/include/asm-ppc/ibm440.h b/include/asm-ppc/ibm440.h
--- a/include/asm-ppc/ibm440.h	Tue May 28 14:57:14 2002
+++ b/include/asm-ppc/ibm440.h	Tue May 28 14:57:14 2002
@@ -77,26 +77,6 @@
  * into the kernel.
  */
 #ifndef __ASSEMBLY__
-extern unsigned char __res[];
-
-/* Device Control Registers */
-
-#define stringify(s)	tostring(s)
-#define tostring(s)	#s
-
-#define mfdcr(rn) mfdcr_or_dflt(rn, 0)
-
-#define mfdcr_or_dflt(rn,default_rval) \
-	({unsigned int rval;						\
-	if (rn == 0)							\
-		rval = default_rval;					\
-	else								\
-		asm volatile("mfdcr %0," stringify(rn) : "=r" (rval));	\
-	rval;})
-
-#define mtdcr(rn, v)  \
-	{if (rn != 0) \
-		asm volatile("mtdcr " stringify(rn) ",%0" : : "r" (v));}

 /*
  * DCRN definitions
@@ -318,13 +298,6 @@
 #define PCIX0_MSGOH		0x10c
 #define PCIX0_IM		0x1f8

-#define EMAC_NUMS		2
-#define UART_NUMS		2
-#define ZMII_NUMS		1
-#define IIC_NUMS		2
-#define IIC0_IRQ		2
-#define IIC1_IRQ		3
-
 #define IIC_OWN			0x55
 #define IIC_CLOCK		50

@@ -333,6 +306,8 @@
 #define UIC_CASCADE_MASK	0x0003		/* bits 30 & 31 */

 #define BD_EMAC_ADDR(e,i) bi_enetaddr[e][i]
+
+#include <asm/ibm4xx.h>

 #endif /* __ASSEMBLY__ */
 #endif /* __ASM_IBM440_H__ */
diff -Nru a/include/asm-ppc/ibm4xx.h b/include/asm-ppc/ibm4xx.h
--- a/include/asm-ppc/ibm4xx.h	Tue May 28 14:57:14 2002
+++ b/include/asm-ppc/ibm4xx.h	Tue May 28 14:57:14 2002
@@ -16,6 +16,34 @@

 #include <linux/config.h>

+#ifdef CONFIG_4xx
+
+#ifndef __ASSEMBLY__
+extern unsigned char __res[];
+
+/* Device Control Registers */
+
+#define stringify(s)	tostring(s)
+#define tostring(s)	#s
+
+#define mfdcr(rn) mfdcr_or_dflt(rn, 0)
+
+#define mfdcr_or_dflt(rn,default_rval) \
+	({unsigned int rval;						\
+	if (rn == 0)							\
+		rval = default_rval;					\
+	else								\
+		asm volatile("mfdcr %0," stringify(rn) : "=r" (rval));	\
+	rval;})
+
+#define mtdcr(rn, v)  \
+	{if (rn != 0) \
+		asm volatile("mtdcr " stringify(rn) ",%0" : : "r" (v));}
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* CONFIG_4xx */
+
 #ifdef CONFIG_40x

 #if defined(CONFIG_ASH)
@@ -74,32 +102,8 @@
  * The "residual" board information structure the boot loader passes
  * into the kernel.
  */
-#ifndef __ASSEMBLY__
-extern unsigned char __res[];
-
-/* Device Control Registers */
-
-#define stringify(s)	tostring(s)
-#define tostring(s)	#s
-
-#define mfdcr(rn) mfdcr_or_dflt(rn, 0)
-
-#define mfdcr_or_dflt(rn,default_rval) \
-	({unsigned int rval;						\
-	if (rn == 0)							\
-		rval = default_rval;					\
-	else								\
-		asm volatile("mfdcr %0," stringify(rn) : "=r" (rval));	\
-	rval;})
-
-#define mtdcr(rn, v)  \
-	{if (rn != 0) \
-		asm volatile("mtdcr " stringify(rn) ",%0" : : "r" (v));}
-
-#endif /* __ASSEMBLY__ */

 #elif CONFIG_440
-#include <asm/ibm440.h>

 #if defined(CONFIG_EBONY)
 #include <platforms/ebony.h>

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help