Skip to contentSkip to conversation

Fixing ethernet driver compilation error for i.MX31 ADS board

PATCH New Ian Lartey 2 participants 11 messages Sep 30, 2010 – Oct 13, 2010

Subsystems (3)
lore
Share & tools
lore

Ian Lartey

author
#0 lore raw

PATCH 1/2 Fixing ethernet driver compilation error for i.MX31 ADS board

This is only a partial revert of "ARM: mx3/mx31ads: fold board header in its only user" [commit ccfa7c269843001077df02d98918c6c9bde91395)]

As some of the the board defines are also used in the cs89x0 ethernet driver by the i.MX31 ADS.

---
 arch/arm/mach-mx3/mach-mx31ads.c               |   43 +++---------------
 arch/arm/plat-mxc/include/mach/board-mx31ads.h |   57 ++++++++++++++++++++++++
 2 files changed, 64 insertions(+), 36 deletions(-)
 create mode 100644 arch/arm/plat-mxc/include/mach/board-mx31ads.h
arch/arm/mach-mx3/mach-mx31ads.c +7 -36
index 94b3e7c..7d13e99 100644
@@ -29,6 +29,7 @@
 #include <asm/memory.h>
 #include <asm/mach/map.h>
 #include <mach/common.h>
+#include <mach/board-mx31ads.h>
 #include <mach/iomux-mx3.h>
 
 #ifdef CONFIG_MACH_MX31ADS_WM1133_EV1
@@ -40,42 +41,12 @@
 #include "devices-imx31.h"
 #include "devices.h"
 
-/* Base address of PBC controller */
-#define PBC_BASE_ADDRESS        MX31_CS4_BASE_ADDR_VIRT
-/* Offsets for the PBC Controller register */
-
-/* PBC Board interrupt status register */
-#define PBC_INTSTATUS           0x000016
-
-/* PBC Board interrupt current status register */
-#define PBC_INTCURR_STATUS      0x000018
-
-/* PBC Interrupt mask register set address */
-#define PBC_INTMASK_SET         0x00001A
-
-/* PBC Interrupt mask register clear address */
-#define PBC_INTMASK_CLEAR       0x00001C
-
-/* External UART A */
-#define PBC_SC16C652_UARTA      0x010000
-
-/* External UART B */
-#define PBC_SC16C652_UARTB      0x010010
-
-#define PBC_INTSTATUS_REG	(PBC_INTSTATUS + PBC_BASE_ADDRESS)
-#define PBC_INTMASK_SET_REG	(PBC_INTMASK_SET + PBC_BASE_ADDRESS)
-#define PBC_INTMASK_CLEAR_REG	(PBC_INTMASK_CLEAR + PBC_BASE_ADDRESS)
-#define EXPIO_PARENT_INT	IOMUX_TO_IRQ(MX31_PIN_GPIO1_4)
-
-#define MXC_EXP_IO_BASE		(MXC_BOARD_IRQ_START)
-#define MXC_IRQ_TO_EXPIO(irq)	((irq) - MXC_EXP_IO_BASE)
-
-#define EXPIO_INT_XUART_INTA	(MXC_EXP_IO_BASE + 10)
-#define EXPIO_INT_XUART_INTB	(MXC_EXP_IO_BASE + 11)
-
-#define MXC_MAX_EXP_IO_LINES	16
-/*
- * This file contains the board-specific initialization routines.
+/*!
+ * @file mx31ads.c
+ *
+ * @brief This file contains the board-specific initialization routines.
+ *
+ * @ingroup System
  */
 
 #if defined(CONFIG_SERIAL_8250) || defined(CONFIG_SERIAL_8250_MODULE)
arch/arm/plat-mxc/include/mach/board-mx31ads.h new file +57 -0

57 lines not on this page · raw message

Signature · 1 line
1.7.0.4
author
#1 lore raw

PATCH 2/2 Adding cs89x0, standard ethernet driver for iMX31 platform.

---
 arch/arm/configs/mx3_defconfig |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
arch/arm/configs/mx3_defconfig +2 -0
index 161f907..b0b3409 100644
@@ -68,6 +68,8 @@ CONFIG_NET_ETHERNET=y
 CONFIG_SMSC911X=y
 CONFIG_DNET=y
 CONFIG_FEC=y
+CONFIG_CS89x0=y
+CONFIG_CS89x0_NONISA_IRQ=y
 # CONFIG_NETDEV_1000 is not set
 # CONFIG_NETDEV_10000 is not set
 # CONFIG_INPUT is not set
Signature · 1 line
1.7.0.4
#8 lore raw

PATCH 1/2 Fixing ethernet driver compilation error for i.MX31 ADS board

In reply to Ian Lartey

Hello,

Quoted 7 lines · Ian Lartey

This is only a partial revert of "ARM: mx3/mx31ads: fold board header in its only user" [commit ccfa7c269843001077df02d98918c6c9bde91395)]

As some of the the board defines are also used in the cs89x0 ethernet driver by the i.MX31 ADS.

It seems you got me wrong. I wanted you to just define the constants in arch/arm/plat-mxc/include/mach/board-mx31ads.h that are needed by the cs89x0 driver. That is PBC_BASE_ADDRESS, PBC_CS8900A_IOBASE and EXPIO_INT_ENET_INT. And I wanted a comment like:

/*
* These symbols are used by drivers/net/cs89x0.c.
* This is ugly as hell, but we have to provide them until
* someone fixed the driver.
*/

And you definitly shouldn't reintroduce these doxygen comments.

Best regards
Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
#2 lore raw

PATCH 2/2 Adding cs89x0, standard ethernet driver for iMX31 platform.

In reply to Ian Lartey
Quoted 10 lines · Ian Lartey
---
 arch/arm/configs/mx3_defconfig |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
arch/arm/configs/mx3_defconfig +0 -0
index 161f907..b0b3409 100644

arch/arm/configs/mx3_defconfig CONFIG_NET_ETHERNET=y · line 75

Quoted 8 lines
arch/arm/configs/mx3_defconfig +2 -0
@@ -68,6 +68,8 @@ CONFIG_NET_ETHERNET=y
 CONFIG_SMSC911X=y
 CONFIG_DNET=y
 CONFIG_FEC=y
+CONFIG_CS89x0=y
+CONFIG_CS89x0_NONISA_IRQ=y
 # CONFIG_NETDEV_1000 is not set
 # CONFIG_NETDEV_10000 is not set
 # CONFIG_INPUT is not set

This change is bad for two reasons:

- I just tested a kernel with CONFIG_CS89x0=y on a Phytec Phycore pcm043 using your updated defconfig and it barfs:

	...
	cs89x0:cs89x0_probe(0x0)
	Unable to handle kernel paging request at virtual address f402030a
	pgd = c0004000
	[f402030a] *pgd=00000000
	Internal error: Oops: 5 [#1] PREEMPT
	last sysfs file: 
	Modules linked in:
	CPU: 0    Not tainted  (2.6.36-rc6-00104-g1374af3 #13)
	PC is at cs89x0_probe1+0xb4/0x6cc
	LR is at kmem_cache_alloc+0x94/0xa4
	pc : [<c00188dc>]    lr : [<c009da7c>]    psr: 60000013
	sp : c781df08  ip : c036d1cc  fp : 00000000
	r10: 00000000  r9 : 00000000  r8 : c0018670
	r7 : f4020300  r6 : f4020300  r5 : c79f2800  r4 : c79f2b20
	r3 : 00000000  r2 : 0000000a  r1 : f402030f  r0 : c79f9cc0
	Flags: nZCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment kernel
	Control: 00c5387d  Table: 80004008  DAC: 00000017
	Process swapper (pid: 1, stack limit = 0xc781c268)
	Stack: (0xc781df08 to 0xc781e000)
	df00:                   00000000 c781df7c c79f2803 00000002 c031f56c ffffffff
	df20: c031f56e c79f2800 3860d7ff c017daac ff0a0210 ffffffff 00000000 c781df40
	df40: ff0a0210 00000000 c79f2800 00000000 00000000 c0018670 00000000 00000000
	df60: c00224ec c79f2800 00000000 00000000 c0018670 00000000 00000000 c0018f80
	df80: 00000000 c00224b8 00000001 c0018634 c0025f2c 00000000 00000001 c0025f2c
	dfa0: 00000013 c00186f0 c001fa44 c001f9b0 c001fa44 c00243fc 00000039 00000000
	dfc0: 00000000 000000f4 c036edfc c001f9b0 c001fa44 c0025f2c 00000013 00000000
	dfe0: 00000000 c0008408 00000000 00000000 c0008370 c0025f2c fd59f476 ffdddbee
	[<c00188dc>] (cs89x0_probe1+0xb4/0x6cc) from [<c0018f80>] (cs89x0_probe+0x8c/0xe4)
	[<c0018f80>] (cs89x0_probe+0x8c/0xe4) from [<c0018634>] (probe_list2+0x34/0x70)
	[<c0018634>] (probe_list2+0x34/0x70) from [<c00186f0>] (net_olddevs_init+0x80/0xc8)
	[<c00186f0>] (net_olddevs_init+0x80/0xc8) from [<c00243fc>] (do_one_initcall+0xc4/0x190)
	[<c00243fc>] (do_one_initcall+0xc4/0x190) from [<c0008408>] (kernel_init+0x98/0x150)
	[<c0008408>] (kernel_init+0x98/0x150) from [<c0025f2c>] (kernel_thread_exit+0x0/0x8)
	Code: eb09cd60 ea000012 e3c77003 e3a0200a (e1d730ba) 
	---[ end trace feac84c64e9ed5aa ]---
	Kernel panic - not syncing: Attempted to kill init!
	[<c002a6b0>] (unwind_backtrace+0x0/0xec) from [<c028bd28>] (panic+0x60/0x18c)
	[<c028bd28>] (panic+0x60/0x18c) from [<c0040ad8>] (do_exit+0x64/0x634)
	[<c0040ad8>] (do_exit+0x64/0x634) from [<c0028bc0>] (die+0x2bc/0x2fc)
	[<c0028bc0>] (die+0x2bc/0x2fc) from [<c002b534>] (__do_kernel_fault+0x64/0x84)
	[<c002b534>] (__do_kernel_fault+0x64/0x84) from [<c002b7b0>] (do_bad_area+0x6c/0x78)
	[<c002b7b0>] (do_bad_area+0x6c/0x78) from [<c002b85c>] (do_translation_fault+0xa0/0xb0)
	[<c002b85c>] (do_translation_fault+0xa0/0xb0) from [<c00242d4>] (do_DataAbort+0x34/0x94)
	[<c00242d4>] (do_DataAbort+0x34/0x94) from [<c0024a4c>] (__dabt_svc+0x4c/0x80)
	Exception stack(0xc781dec0 to 0xc781df08)
	dec0: c79f9cc0 f402030f 0000000a 00000000 c79f2b20 c79f2800 f4020300 f4020300
	dee0: c0018670 00000000 00000000 00000000 c036d1cc c781df08 c009da7c c00188dc
	df00: 60000013 ffffffff
	[<c0024a4c>] (__dabt_svc+0x4c/0x80) from [<c00188dc>] (cs89x0_probe1+0xb4/0x6cc)
	[<c00188dc>] (cs89x0_probe1+0xb4/0x6cc) from [<c0018f80>] (cs89x0_probe+0x8c/0xe4)
	[<c0018f80>] (cs89x0_probe+0x8c/0xe4) from [<c0018634>] (probe_list2+0x34/0x70)
	[<c0018634>] (probe_list2+0x34/0x70) from [<c00186f0>] (net_olddevs_init+0x80/0xc8)
	[<c00186f0>] (net_olddevs_init+0x80/0xc8) from [<c00243fc>] (do_one_initcall+0xc4/0x190)
	[<c00243fc>] (do_one_initcall+0xc4/0x190) from [<c0008408>] (kernel_init+0x98/0x150)
	[<c0008408>] (kernel_init+0x98/0x150) from [<c0025f2c>] (kernel_thread_exit+0x0/0x8)

(Did I already mention that this cs89x0 driver is ugly and needs a general overhaul?)

- adding "CONFIG_CS89x0_NONISA_IRQ=y" is redundant as it is always on for MACH_MX31ADS.

So I suggest to just add CONFIG_CS89x0=m

Best regards
Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
author
#9 lore raw

PATCH 1/2 Fixing ethernet driver compilation error for i.MX31 ADS board

In reply to Uwe Kleine-König
Quoted 15 lines · Uwe Kleine-K?nig

Hello,

Quoted 7 lines · Ian Lartey

This is only a partial revert of "ARM: mx3/mx31ads: fold board header in its only user" [commit ccfa7c269843001077df02d98918c6c9bde91395)]

As some of the the board defines are also used in the cs89x0 ethernet driver by the i.MX31 ADS.

It seems you got me wrong. I wanted you to just define the constants in arch/arm/plat-mxc/include/mach/board-mx31ads.h that are needed by the cs89x0 driver. That is PBC_BASE_ADDRESS, PBC_CS8900A_IOBASE and EXPIO_INT_ENET_IAm I right in assuming NT. And I wanted a comment like:

Quoted 6 lines

/*
* These symbols are used by drivers/net/cs89x0.c.
* This is ugly as hell, but we have to provide them until
* someone fixed the driver.
*/

Ah, yes it is _very ugly which was why I was trying everything else but I can understand your reasoning with the current state of the driver.

Just to be clear. You want just the above defines in board-mx31ads.h and either:

1. Common defines to be removed from board-mx31ads.c to board-mx31ads.h _or_
2. Duplicate defines in both board-mx31ads.c and board-mx31ads.h.

I'm assuming that you want 1. (2. is too ugly and dangerous).

Quoted 1 line

And you definitly shouldn't reintroduce these doxygen comments.

Right. I'll watch this in future too.

Quoted 4 lines

Best regards
Uwe

Best Regards,
Ian.

author
#3 lore raw

PATCH 2/2 Adding cs89x0, standard ethernet driver for iMX31 platform.

In reply to Uwe Kleine-König
Quoted 21 lines · Uwe Kleine-K?nig
Quoted 10 lines · Ian Lartey
---
 arch/arm/configs/mx3_defconfig |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
arch/arm/configs/mx3_defconfig +0 -0
index 161f907..b0b3409 100644

arch/arm/configs/mx3_defconfig CONFIG_NET_ETHERNET=y · line 75

Quoted 8 lines
arch/arm/configs/mx3_defconfig +2 -0
@@ -68,6 +68,8 @@ CONFIG_NET_ETHERNET=y
 CONFIG_SMSC911X=y
 CONFIG_DNET=y
 CONFIG_FEC=y
+CONFIG_CS89x0=y
+CONFIG_CS89x0_NONISA_IRQ=y
 # CONFIG_NETDEV_1000 is not set
 # CONFIG_NETDEV_10000 is not set
 # CONFIG_INPUT is not set

This change is bad for two reasons:

Quoted 4 lines

- I just tested a kernel with CONFIG_CS89x0=y on a Phytec Phycore pcm043 using your updated defconfig and it barfs:

	[.....]

Ouch, I'll see if I can correct that. (Needs to be done before I can add the config file change). Uwe I don't have another i.MX3 (apart from the i.MV31ADS) would you be able to test further changes for me please to help catch this kind of thing ?

Quoted 4 lines

(Did I already mention that this cs89x0 driver is ugly and needs a general overhaul?)

Yes, and I'm getting additional errors from the cs89x0 driver once it is up and running so I was going to start looking for help once my changes were in.... any volunteers ?

Quoted 4 lines

- adding "CONFIG_CS89x0_NONISA_IRQ=y" is redundant as it is always on for MACH_MX31ADS.

So I suggest to just add CONFIG_CS89x0=m

Agreed.

Quoted 4 lines

Best regards
Uwe

Best Regards,
Ian.

#10 lore raw

PATCH 1/2 Fixing ethernet driver compilation error for i.MX31 ADS board

In reply to Ian Lartey

Hello Ian,

Quoted 27 lines · Ian Lartey
Quoted 15 lines · Uwe Kleine-K?nig

Hello,

Quoted 7 lines · Ian Lartey

This is only a partial revert of "ARM: mx3/mx31ads: fold board header in its only user" [commit ccfa7c269843001077df02d98918c6c9bde91395)]

As some of the the board defines are also used in the cs89x0 ethernet driver by the i.MX31 ADS.

It seems you got me wrong. I wanted you to just define the constants in arch/arm/plat-mxc/include/mach/board-mx31ads.h that are needed by the cs89x0 driver. That is PBC_BASE_ADDRESS, PBC_CS8900A_IOBASE and EXPIO_INT_ENET_IAm I right in assuming NT. And I wanted a comment like:

Quoted 6 lines

/*
* These symbols are used by drivers/net/cs89x0.c.
* This is ugly as hell, but we have to provide them until
* someone fixed the driver.
*/

Ah, yes it is _very ugly which was why I was trying everything else but I can understand your reasoning with the current state of the driver.

Quoted 8 lines

Just to be clear. You want just the above defines in board-mx31ads.h and either:

1. Common defines to be removed from board-mx31ads.c to board-mx31ads.h _or_
2. Duplicate defines in both board-mx31ads.c and board-mx31ads.h.

I'm assuming that you want 1. (2. is too ugly and dangerous).

Right, I want 1.

Best regards
Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
#4 lore raw

PATCH 2/2 Adding cs89x0, standard ethernet driver for iMX31 platform.

In reply to Ian Lartey
Quoted 25 lines · Ian Lartey
Quoted 24 lines · Uwe Kleine-K?nig
Quoted 10 lines · Ian Lartey
---
 arch/arm/configs/mx3_defconfig |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
arch/arm/configs/mx3_defconfig +0 -0
index 161f907..b0b3409 100644

arch/arm/configs/mx3_defconfig CONFIG_NET_ETHERNET=y · line 75

Quoted 8 lines
arch/arm/configs/mx3_defconfig +2 -0
@@ -68,6 +68,8 @@ CONFIG_NET_ETHERNET=y
 CONFIG_SMSC911X=y
 CONFIG_DNET=y
 CONFIG_FEC=y
+CONFIG_CS89x0=y
+CONFIG_CS89x0_NONISA_IRQ=y
 # CONFIG_NETDEV_1000 is not set
 # CONFIG_NETDEV_10000 is not set
 # CONFIG_INPUT is not set

This change is bad for two reasons:

- I just tested a kernel with CONFIG_CS89x0=y on a Phytec Phycore pcm043 using your updated defconfig and it barfs:

Quoted 6 lines
Quoted 1 line
	[.....]

Ouch, I'll see if I can correct that. (Needs to be done before I can add the config file change). Uwe I don't have another i.MX3 (apart from the i.MV31ADS) would you be able to test further changes for me please to help catch this kind of thing ?

I'm not sure if I have access to an mx31ads, but we have several i.MX machines in our lab. And I'm willing to give my comments to a general cs89x0 cleanup. (This drivers was one of the first things I saw when I started kernel development. Back then I gave up trying to understand it. :-) So maybe if you don't come around fixing it I might have some energy to do it now with more knowledge about the kernel in general and having seen *much* ugly code.

Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
author
#5 lore raw

PATCH 2/2 Adding cs89x0, standard ethernet driver for iMX31 platform.

Quoted 32 lines · Uwe Kleine-K?nig
Quoted 25 lines · Ian Lartey
Quoted 24 lines · Uwe Kleine-K?nig
Quoted 10 lines · Ian Lartey
---
 arch/arm/configs/mx3_defconfig |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
arch/arm/configs/mx3_defconfig +0 -0
index 161f907..b0b3409 100644

arch/arm/configs/mx3_defconfig CONFIG_NET_ETHERNET=y · line 75

Quoted 8 lines
arch/arm/configs/mx3_defconfig +2 -0
@@ -68,6 +68,8 @@ CONFIG_NET_ETHERNET=y
 CONFIG_SMSC911X=y
 CONFIG_DNET=y
 CONFIG_FEC=y
+CONFIG_CS89x0=y
+CONFIG_CS89x0_NONISA_IRQ=y
 # CONFIG_NETDEV_1000 is not set
 # CONFIG_NETDEV_10000 is not set
 # CONFIG_INPUT is not set

This change is bad for two reasons:

- I just tested a kernel with CONFIG_CS89x0=y on a Phytec Phycore pcm043 using your updated defconfig and it barfs:

Quoted 6 lines
Quoted 1 line
	[.....]

Ouch, I'll see if I can correct that. (Needs to be done before I can add the config file change). Uwe I don't have another i.MX3 (apart from the i.MV31ADS) would you be able to test further changes for me please to help catch this kind of thing ?

Quoted 7 lines

I'm not sure if I have access to an mx31ads, but we have several i.MX machines in our lab. And I'm willing to give my comments to a general cs89x0 cleanup. (This drivers was one of the first things I saw when I started kernel development. Back then I gave up trying to understand it. :-) So maybe if you don't come around fixing it I might have some energy to do it now with more knowledge about the kernel in general and having seen *much* ugly code.

I've had a look round but I cannot get a handle on fixing the issue. I only have an i.mx31ads that failure doesn't occur, sorry I hope you have more success than me.

For now I'll resubmit PATCH 1/2 with the changes to board-mx31ads.h and board-mx31ads.c but I'll not resubmit this patch to mx3_defconfig so nobody will get it by default.

Does this seem OK to you ?

Cheers,
Ian.

#6 lore raw

PATCH 2/2 Adding cs89x0, standard ethernet driver for iMX31 platform.

Hi Ian,

Quoted 41 lines · Ian Lartey
Quoted 32 lines · Uwe Kleine-K?nig
Quoted 25 lines · Ian Lartey
Quoted 24 lines · Uwe Kleine-K?nig
Quoted 10 lines · Ian Lartey
---
 arch/arm/configs/mx3_defconfig |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
arch/arm/configs/mx3_defconfig +0 -0
index 161f907..b0b3409 100644

arch/arm/configs/mx3_defconfig CONFIG_NET_ETHERNET=y · line 75

Quoted 8 lines
arch/arm/configs/mx3_defconfig +2 -0
@@ -68,6 +68,8 @@ CONFIG_NET_ETHERNET=y
 CONFIG_SMSC911X=y
 CONFIG_DNET=y
 CONFIG_FEC=y
+CONFIG_CS89x0=y
+CONFIG_CS89x0_NONISA_IRQ=y
 # CONFIG_NETDEV_1000 is not set
 # CONFIG_NETDEV_10000 is not set
 # CONFIG_INPUT is not set

This change is bad for two reasons:

- I just tested a kernel with CONFIG_CS89x0=y on a Phytec Phycore pcm043 using your updated defconfig and it barfs:

Quoted 6 lines
Quoted 1 line
	[.....]

Ouch, I'll see if I can correct that. (Needs to be done before I can add the config file change). Uwe I don't have another i.MX3 (apart from the i.MV31ADS) would you be able to test further changes for me please to help catch this kind of thing ?

Quoted 7 lines

I'm not sure if I have access to an mx31ads, but we have several i.MX machines in our lab. And I'm willing to give my comments to a general cs89x0 cleanup. (This drivers was one of the first things I saw when I started kernel development. Back then I gave up trying to understand it. :-) So maybe if you don't come around fixing it I might have some energy to do it now with more knowledge about the kernel in general and having seen *much* ugly code.

Quoted 1 line

I've had a look round but I cannot get a handle on fixing the issue.

no time or no clue?

Quoted 2 lines

I only have an i.mx31ads that failure doesn't occur, sorry I hope you have more success than me.

I put in on my todo list, but I don't expect to get that done soon.

Quoted 5 lines

For now I'll resubmit PATCH 1/2 with the changes to board-mx31ads.h and board-mx31ads.c but I'll not resubmit this patch to mx3_defconfig so nobody will get it by default.

Does this seem OK to you ?

OK

Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
author
#7 lore raw

PATCH 2/2 Adding cs89x0, standard ethernet driver for iMX31 platform.

Quoted 44 lines · Uwe Kleine-K?nig

Hi Ian,

Quoted 41 lines · Ian Lartey
Quoted 39 lines · Uwe Kleine-K?nig
Quoted 25 lines · Ian Lartey
Quoted 24 lines · Uwe Kleine-K?nig
Quoted 10 lines · Ian Lartey
---
 arch/arm/configs/mx3_defconfig |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
arch/arm/configs/mx3_defconfig +0 -0
index 161f907..b0b3409 100644

arch/arm/configs/mx3_defconfig CONFIG_NET_ETHERNET=y · line 75

Quoted 8 lines
arch/arm/configs/mx3_defconfig +2 -0
@@ -68,6 +68,8 @@ CONFIG_NET_ETHERNET=y
 CONFIG_SMSC911X=y
 CONFIG_DNET=y
 CONFIG_FEC=y
+CONFIG_CS89x0=y
+CONFIG_CS89x0_NONISA_IRQ=y
 # CONFIG_NETDEV_1000 is not set
 # CONFIG_NETDEV_10000 is not set
 # CONFIG_INPUT is not set

This change is bad for two reasons:

- I just tested a kernel with CONFIG_CS89x0=y on a Phytec Phycore pcm043 using your updated defconfig and it barfs:

Quoted 6 lines
Quoted 1 line
	[.....]

Ouch, I'll see if I can correct that. (Needs to be done before I can add the config file change). Uwe I don't have another i.MX3 (apart from the i.MV31ADS) would you be able to test further changes for me please to help catch this kind of thing ?

I'm not sure if I have access to an mx31ads, but we have several i.MX machines in our lab. And I'm willing to give my comments to a general cs89x0 cleanup. (This drivers was one of the first things I saw when I started kernel development. Back then I gave up trying to understand it. :-) So maybe if you don't come around fixing it I might have some energy to do it now with more knowledge about the kernel in general and having seen *much* ugly code.

Quoted 2 lines
Quoted 1 line

I've had a look round but I cannot get a handle on fixing the issue.

no time or no clue?

A bit of both, but I think no-clue takes priority in this case!

Quoted 4 lines
Quoted 3 lines

I only have an i.mx31ads that failure doesn't occur, sorry I hope you have more success than me.

I put in on my todo list, but I don't expect to get that done soon.

OK.

Quoted 7 lines
Quoted 6 lines

For now I'll resubmit PATCH 1/2 with the changes to board-mx31ads.h and board-mx31ads.c but I'll not resubmit this patch to mx3_defconfig so nobody will get it by default.

Does this seem OK to you ?

OK

Ian.