OSK5912 and CF and network broken?

OSK5912 and CF and network broken?

[Tom Rini]

Hello.  I'm trying to use current linux-omap-2.6 (source.mvista.com one)
git with the omap_osk_5912_defconfig and with U-Boot 1.1.4 release (+ 2
patches from this list).  I can bring the board up just fine but when I
insert a CF device I see:
pccard: PCMCIA card inserted into slot 0
pcmcia: registering new device pcmcia0.0
NETDEV WATCHDOG: eth0: transmit timed out
...
NETDEV WATCHDOG: eth0: transmit timed out
pccard: card ejected from slot 0

Has anyone seen this before?  Or should I just looking at the interrupt
routes?  I'm assuming that the board does support me using both CF and
network at the same time.  Thanks!

-- 
Tom Rini
http://gate.crashing.org/~trini/

Re: OSK5912 and CF and network broken?

[Tony Lindgren]

* Tom Rini <trini@kernel.crashing.org> [060315 10:31]:
> Hello.  I'm trying to use current linux-omap-2.6 (source.mvista.com one)
> git with the omap_osk_5912_defconfig and with U-Boot 1.1.4 release (+ 2
> patches from this list).  I can bring the board up just fine but when I
> insert a CF device I see:
> pccard: PCMCIA card inserted into slot 0
> pcmcia: registering new device pcmcia0.0
> NETDEV WATCHDOG: eth0: transmit timed out
> ...
> NETDEV WATCHDOG: eth0: transmit timed out
> pccard: card ejected from slot 0
> 
> Has anyone seen this before?  Or should I just looking at the interrupt
> routes?  I'm assuming that the board does support me using both CF and
> network at the same time.  Thanks!

This should not happen... It sounds like an issue with the flash bus timings.
Might be worth grepping for EMIFS_CS values.

Tony

Re: OSK5912 and CF and network broken?

[Tom Rini]

On Wed, Mar 15, 2006 at 10:36:45AM -0800, Tony Lindgren wrote:
> * Tom Rini <trini@kernel.crashing.org> [060315 10:31]:
> > Hello.  I'm trying to use current linux-omap-2.6 (source.mvista.com one)
> > git with the omap_osk_5912_defconfig and with U-Boot 1.1.4 release (+ 2
> > patches from this list).  I can bring the board up just fine but when I
> > insert a CF device I see:
> > pccard: PCMCIA card inserted into slot 0
> > pcmcia: registering new device pcmcia0.0
> > NETDEV WATCHDOG: eth0: transmit timed out
> > ...
> > NETDEV WATCHDOG: eth0: transmit timed out
> > pccard: card ejected from slot 0
> > 
> > Has anyone seen this before?  Or should I just looking at the interrupt
> > routes?  I'm assuming that the board does support me using both CF and
> > network at the same time.  Thanks!
> 
> This should not happen... It sounds like an issue with the flash bus timings.
> Might be worth grepping for EMIFS_CS values.

Er, this?
linux-omap-2.6]$ grep -r EMIFS_CS arch/arm include/asm-arm
arch/arm/mach-omap1/board-osk.c:#define EMIFS_CS3_VAL   (0x88013141)
arch/arm/mach-omap1/board-osk.c:        if (EMIFS_CCS(3) != EMIFS_CS3_VAL)
arch/arm/mach-omap1/board-osk.c:                EMIFS_CCS(3) = EMIFS_CS3_VAL;
include/asm-arm/arch-omap/tc.h:#define EMIFS_CS0_CONFIG (TCMIF_BASE + 0x10)
include/asm-arm/arch-omap/tc.h:#define EMIFS_CS1_CONFIG (TCMIF_BASE + 0x14)
include/asm-arm/arch-omap/tc.h:#define EMIFS_CS2_CONFIG (TCMIF_BASE + 0x18)
include/asm-arm/arch-omap/tc.h:#define EMIFS_CS3_CONFIG (TCMIF_BASE + 0x1c)
include/asm-arm/arch-omap/tc.h:#define EMIFS_CCS(n)             __REG32(EMIFS_CS0_CONFIG + (4 * (n)))

-- 
Tom Rini
http://gate.crashing.org/~trini/

Re: OSK5912 and CF and network broken?

[Tony Lindgren]

* Tom Rini <trini@kernel.crashing.org> [060315 11:28]:
> On Wed, Mar 15, 2006 at 10:36:45AM -0800, Tony Lindgren wrote:
> > * Tom Rini <trini@kernel.crashing.org> [060315 10:31]:
> > > Hello.  I'm trying to use current linux-omap-2.6 (source.mvista.com one)
> > > git with the omap_osk_5912_defconfig and with U-Boot 1.1.4 release (+ 2
> > > patches from this list).  I can bring the board up just fine but when I
> > > insert a CF device I see:
> > > pccard: PCMCIA card inserted into slot 0
> > > pcmcia: registering new device pcmcia0.0
> > > NETDEV WATCHDOG: eth0: transmit timed out
> > > ...
> > > NETDEV WATCHDOG: eth0: transmit timed out
> > > pccard: card ejected from slot 0
> > > 
> > > Has anyone seen this before?  Or should I just looking at the interrupt
> > > routes?  I'm assuming that the board does support me using both CF and
> > > network at the same time.  Thanks!
> > 
> > This should not happen... It sounds like an issue with the flash bus timings.
> > Might be worth grepping for EMIFS_CS values.
> 
> Er, this?
> linux-omap-2.6]$ grep -r EMIFS_CS arch/arm include/asm-arm
> arch/arm/mach-omap1/board-osk.c:#define EMIFS_CS3_VAL   (0x88013141)
> arch/arm/mach-omap1/board-osk.c:        if (EMIFS_CCS(3) != EMIFS_CS3_VAL)
> arch/arm/mach-omap1/board-osk.c:                EMIFS_CCS(3) = EMIFS_CS3_VAL;
> include/asm-arm/arch-omap/tc.h:#define EMIFS_CS0_CONFIG (TCMIF_BASE + 0x10)
> include/asm-arm/arch-omap/tc.h:#define EMIFS_CS1_CONFIG (TCMIF_BASE + 0x14)
> include/asm-arm/arch-omap/tc.h:#define EMIFS_CS2_CONFIG (TCMIF_BASE + 0x18)
> include/asm-arm/arch-omap/tc.h:#define EMIFS_CS3_CONFIG (TCMIF_BASE + 0x1c)
> include/asm-arm/arch-omap/tc.h:#define EMIFS_CCS(n)             __REG32(EMIFS_CS0_CONFIG + (4 * (n)))

Yes, and then also some settings are done in drivers/pcmcia/omap_cf.c.
Maybe omap_cf.c undoes some EMIFS_CS settings done in board-osk.c?

Tony

Re: OSK5912 and CF and network broken?

[Tom Rini]

On Wed, Mar 15, 2006 at 11:41:42AM -0800, Tony Lindgren wrote:
> * Tom Rini <trini@kernel.crashing.org> [060315 11:28]:
> > On Wed, Mar 15, 2006 at 10:36:45AM -0800, Tony Lindgren wrote:
> > > * Tom Rini <trini@kernel.crashing.org> [060315 10:31]:
> > > > Hello.  I'm trying to use current linux-omap-2.6 (source.mvista.com one)
> > > > git with the omap_osk_5912_defconfig and with U-Boot 1.1.4 release (+ 2
> > > > patches from this list).  I can bring the board up just fine but when I
> > > > insert a CF device I see:
> > > > pccard: PCMCIA card inserted into slot 0
> > > > pcmcia: registering new device pcmcia0.0
> > > > NETDEV WATCHDOG: eth0: transmit timed out
> > > > ...
> > > > NETDEV WATCHDOG: eth0: transmit timed out
> > > > pccard: card ejected from slot 0
> > > > 
> > > > Has anyone seen this before?  Or should I just looking at the interrupt
> > > > routes?  I'm assuming that the board does support me using both CF and
> > > > network at the same time.  Thanks!
> > > 
> > > This should not happen... It sounds like an issue with the flash bus timings.
> > > Might be worth grepping for EMIFS_CS values.
> > 
> > Er, this?
> > linux-omap-2.6]$ grep -r EMIFS_CS arch/arm include/asm-arm
> > arch/arm/mach-omap1/board-osk.c:#define EMIFS_CS3_VAL   (0x88013141)
> > arch/arm/mach-omap1/board-osk.c:        if (EMIFS_CCS(3) != EMIFS_CS3_VAL)
> > arch/arm/mach-omap1/board-osk.c:                EMIFS_CCS(3) = EMIFS_CS3_VAL;
> > include/asm-arm/arch-omap/tc.h:#define EMIFS_CS0_CONFIG (TCMIF_BASE + 0x10)
> > include/asm-arm/arch-omap/tc.h:#define EMIFS_CS1_CONFIG (TCMIF_BASE + 0x14)
> > include/asm-arm/arch-omap/tc.h:#define EMIFS_CS2_CONFIG (TCMIF_BASE + 0x18)
> > include/asm-arm/arch-omap/tc.h:#define EMIFS_CS3_CONFIG (TCMIF_BASE + 0x1c)
> > include/asm-arm/arch-omap/tc.h:#define EMIFS_CCS(n)             __REG32(EMIFS_CS0_CONFIG + (4 * (n)))
> 
> Yes, and then also some settings are done in drivers/pcmcia/omap_cf.c.
> Maybe omap_cf.c undoes some EMIFS_CS settings done in board-osk.c?

Yes, it does change CS2:
omap_cf: cs2, previous ccs ff80fff3 acs 00000002
omap_cf: cs2, current ccs 0004a1b3 acs 00000000

With the comment:
        /* NOTE:  better EMIFS setup might support more cards; but the
         * TRM only shows how to affect regular flash signals, not their
         * CF/PCMCIA variants...
         */


-- 
Tom Rini
http://gate.crashing.org/~trini/

Re: OSK5912 and CF and network broken?

[Dirk Behme]

Tom Rini wrote:
> Has anyone seen this before?  Or should I just looking at the interrupt
> routes?  I'm assuming that the board does support me using both CF and
> network at the same time.  Thanks!

Yes, board supports both, CF and network. However, OSK CF 
hardware is limited and is really troublesome.

But using recent git without any special patches I get no 
transmit timeouts:

# pccard: PCMCIA card inserted into slot 0
pcmcia: registering new device pcmcia0.0

# ping 192.0.0.1
PING 192.0.0.1 (192.0.0.1): 56 data bytes
64 bytes from 192.0.0.1: icmp_seq=0 ttl=64 time=1.0 ms
64 bytes from 192.0.0.1: icmp_seq=1 ttl=64 time=0.8 ms

--- 192.0.0.1 ping statistics ---
2 packets transmitted, 2 packets received, 0% packet loss
round-trip min/avg/max = 0.8/0.9/1.0 ms
# cp /mnt/nfs/foo.mp3 /dev/null
...
# pccard: card ejected from slot 0

=> No timeouts

Kernel configuration:

#
# ATA/ATAPI/MFM/RLL support
#
CONFIG_IDE=m
CONFIG_BLK_DEV_IDE=m

#
# Please see Documentation/ide.txt for help/info on IDE drives
#
# CONFIG_BLK_DEV_IDE_SATA is not set
CONFIG_BLK_DEV_IDEDISK=m
# CONFIG_IDEDISK_MULTI_MODE is not set
CONFIG_BLK_DEV_IDECS=m
# CONFIG_BLK_DEV_IDECD is not set
# CONFIG_BLK_DEV_IDETAPE is not set
# CONFIG_BLK_DEV_IDEFLOPPY is not set
# CONFIG_BLK_DEV_IDESCSI is not set
# CONFIG_IDE_TASK_IOCTL is not set

#
# PCCARD (PCMCIA/CardBus) support
#
CONFIG_PCCARD=m
# CONFIG_PCMCIA_DEBUG is not set
CONFIG_PCMCIA=m
CONFIG_PCMCIA_LOAD_CIS=y
CONFIG_PCMCIA_IOCTL=y

#
# PC-card bridges
#
CONFIG_OMAP_CF=m

However, seems that my cardmgr is configured wrong again, 
but this is another issue:

# cardmgr
cardmgr[978]: watching 1 socket
cardmgr[978]: could not adjust resource: IO ports 
0xc00-0xcff: Function not implemented
cardmgr[978]: could not adjust resource: IO ports 
0x800-0x8ff: Function not implemented
cardmgr[978]: could not adjust resource: IO ports 
0x100-0x4ff: Function not implemented
cardmgr[978]: could not adjust resource: memory 
0xc0000-0xfffff: Function not implemented
cardmgr[978]: could not adjust resource: memory 
0x60000000-0x60ffffff: Function not implemented
cardmgr[978]: could not adjust resource: memory 
0xa0000000-0xa0ffffff: Function not implemented
cardmgr[978]: could not adjust resource: IO ports 
0xa00-0xaff: Function not implemented
# pccard: PCMCIA card inserted into slot 0
cardmgr[979]: socket 0: Anonymous Memory
cardmgr[979]: executing: 'modprobe memory_cs 2>&1'
cardmgr[979]: + modprobe: module memory_cs not found.
cardmgr[979]: + modprobe: failed to load module memory_cs
cardmgr[979]: modprobe exited with status 1
cardmgr[979]: module 
/lib/modules/2.6.16-rc5-omap1/pcmcia/memory_cs.o not available
cardmgr[979]: bind 'memory_cs' to socket 0 failed: Invalid 
argument

Best regards

Dirk

Re: OSK5912 and CF and network broken?

[Tom Rini]

On Thu, Mar 16, 2006 at 06:03:02PM +0100, Dirk Behme wrote:
> Tom Rini wrote:
> >Has anyone seen this before?  Or should I just looking at the interrupt
> >routes?  I'm assuming that the board does support me using both CF and
> >network at the same time.  Thanks!
> 
> Yes, board supports both, CF and network. However, OSK CF 
> hardware is limited and is really troublesome.

I wonder if my board is just more troublesome than most.  If I turned off
network (and used a ramdisk for userspace) I then got to the point of:

pccard: PCMCIA card inserted into slot 0                                      
pcmcia: registering new device pcmcia0.0                                        
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
ide-cs: ide_register() at 0xc2852000 & 0xc285200e, irq 222 failed

with an old 16MB CF card.  With my 2.2GB microdrive I get more of the
messages, but the same result.  FWIW, u-boot won't load if the MD is
inserted, but will if the CF is.

But perhaps I need a better cardmgr options file?  I don't have anything
board-specific in mine.  Thanks!

-- 
Tom Rini
http://gate.crashing.org/~trini/

Re: OSK5912 and CF and network broken?

[tony]

* Tom Rini <trini@kernel.crashing.org> [060316 09:20]:
> On Thu, Mar 16, 2006 at 06:03:02PM +0100, Dirk Behme wrote:
> > Tom Rini wrote:
> > >Has anyone seen this before?  Or should I just looking at the interrupt
> > >routes?  I'm assuming that the board does support me using both CF and
> > >network at the same time.  Thanks!
> > 
> > Yes, board supports both, CF and network. However, OSK CF 
> > hardware is limited and is really troublesome.
> 
> I wonder if my board is just more troublesome than most.  If I turned off
> network (and used a ramdisk for userspace) I then got to the point of:
> 
> pccard: PCMCIA card inserted into slot 0                                      
> pcmcia: registering new device pcmcia0.0                                        
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> hdb: probing with STATUS(0x00) instead of ALTSTATUS(0x51)                       
> ide-cs: ide_register() at 0xc2852000 & 0xc285200e, irq 222 failed
> 
> with an old 16MB CF card.  With my 2.2GB microdrive I get more of the
> messages, but the same result.  FWIW, u-boot won't load if the MD is
> inserted, but will if the CF is.
> 
> But perhaps I need a better cardmgr options file?  I don't have anything
> board-specific in mine.  Thanks!

I believe there may be some different timings needed depending on the
OSK revision.

Tony

Re: OSK5912 and CF and network broken?

[Dirk Behme]

tony@atomide.com wrote:
> * Tom Rini <trini@kernel.crashing.org> [060316 09:20]:
>>
>>>Tom Rini wrote:
>>But perhaps I need a better cardmgr options file?  I don't have anything
>>board-specific in mine.  Thanks!
> 
> I believe there may be some different timings needed depending on the
> OSK revision.

Yes, that may be possible. But this doesn't explain why on 
my board (where hopefully timing doesn't change ;) ) it 
worked in the past, then was partly broken and now doesn't 
work any more ;(

Dirk