wireless lan PCMCIA card advice

2003-01-07 Thread Wolfgang Denk

Dear Curt,

in message <20030106183517.E24490 at curtisb.com> you wrote:
>
> I am running ELDK 2.0.2 which does have RPMs for pcmcia-cs and
> wireless-tools.  I also see my target installation has the "iwconfig"
> program installed in /sbin.  The target also has /etc/pcmcia/* and
> /etc/rc.d/init.d/pcmcia.

OK.

> >From your email I gather all I need to do is recompile my kernel to
> include support for the wireless LAN (CONFIG_NET_RADIO) .  Is that
> correct ?

Right, but you need a  recent  kernel  (from  our  CVS).  The  kernel
specific  parts  (stuff like drivers/pcmcia/m8xx_pcmcia.c) were added
after the release of  ELDK  2.0.2  [and  some  more  extensions  like
simultaneous  support  for  both  PCMCIA  slots on MPC860 systems was
added even later].

Don't forget to run

bash$ make modules
bash$ make modules_install INSTALL_MOD_PATH=/opt/eldk/ppc_8xx

when building your kernel.

Best regards,

Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd at denx.de
Of course there's no reason for it, it's just our policy.

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





Whats the "standard" embedded C lib?

2003-01-07 Thread James Don

Hi there,

I am just looking at compiling "busybox". And am now at a point where I need
to consider more than just the kernel and my bootloader.

I have build my gcc 2.95.3 using newlib (from redhat I believe) ...

I am having some troubles building busy box ... as I need to sort our which
'headers' to include etc ...

I think I am on the right track to completing this but I read something in
the busy box makefile that kind of stirred my confidence ... it said uClib
is best with busy box ... is this the same as newlib?

What embedded C lib do you folks recommend? (for PPC 860 if that matters)

Regards,
Jim

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





Whats the "standard" embedded C lib?

2003-01-07 Thread Mark Hatle

James Don wrote:
> Hi there,
>
> I am just looking at compiling "busybox". And am now at a point where I need
> to consider more than just the kernel and my bootloader.
>
> I have build my gcc 2.95.3 using newlib (from redhat I believe) ...

newlib is only used as a bootstrap mechanism before you build a "real" C 
library.

> I am having some troubles building busy box ... as I need to sort our which
> 'headers' to include etc ...
>
> I think I am on the right track to completing this but I read something in
> the busy box makefile that kind of stirred my confidence ... it said uClib
> is best with busy box ... is this the same as newlib?

busybox will work with uclibc or glibc.  I recommend glibc, but due to size
limitations you may want to use uclibc.  (Be aware that uclib isn't as well
supported on PPC.  At least wasn't last time I looked..)

> What embedded C lib do you folks recommend? (for PPC 860 if that matters)

glibc is much larger then uclibc, but is a "full featured" c library (w/ shared
library support).  There many ways to shrink down the size of glibc to something
more manageable for an embedded project, but it will probably still not be as
small as uclibc.

--Mark


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





Whats the "standard" embedded C lib?

2003-01-07 Thread James Don

Thanks for the info ...

I now think I remember reading that about newlib from when I built the gcc
cross-compilers in september ...

I will do some reading on glibc and uclibc to see what I should pick tonight
...

Thanks again for the tip,
Jim



-Original Message-
From: Mark Hatle [mailto:[EMAIL PROTECTED]
Sent: Tuesday, January 07, 2003 4:28 PM
To: James Don
Cc: 'linuxppc-embedded at lists.linuxppc.org'
Subject: Re: Whats the "standard" embedded C lib?

James Don wrote:
> Hi there,
>
> I am just looking at compiling "busybox". And am now at a point where I
need
> to consider more than just the kernel and my bootloader.
>
> I have build my gcc 2.95.3 using newlib (from redhat I believe) ...

newlib is only used as a bootstrap mechanism before you build a "real" C
library.

> I am having some troubles building busy box ... as I need to sort our
which
> 'headers' to include etc ...
>
> I think I am on the right track to completing this but I read something in
> the busy box makefile that kind of stirred my confidence ... it said uClib
> is best with busy box ... is this the same as newlib?

busybox will work with uclibc or glibc.  I recommend glibc, but due to size
limitations you may want to use uclibc.  (Be aware that uclib isn't as well
supported on PPC.  At least wasn't last time I looked..)

> What embedded C lib do you folks recommend? (for PPC 860 if that matters)

glibc is much larger then uclibc, but is a "full featured" c library (w/
shared
library support).  There many ways to shrink down the size of glibc to
something
more manageable for an embedded project, but it will probably still not be
as
small as uclibc.

--Mark

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





Compiling ULibc and gcc questions

2003-01-07 Thread James Don

Hello,
I have been looking at compiling uClibc ...

I just got it to compile ...

But now I have some questions regard gcc ... up until now I have been able
to get along fine with a boot strapped version of gcc ...

I now have these questions:

1.) After I build uClibc ... should I recompile gcc with uClibc ?
2.) When I built my boot strap gcc is the "--enable-shared" a good idea to
have? Can some explain if have this feature is good for embedded systems ...
I have some old emails saying "no" ...

Regards,
Jim





I got uClibc to build once I removed the PIC support from the config ... and
pointed the build to my correct kernel source ;-)

Is it importanted to build with " --enable-shared" ?

Jim

-Original Message-
From: James Don
Sent: Tuesday, January 07, 2003 6:54 PM
To: 'uclibc at uclibc.org'
Subject: RE: Help: Simple newbie build problem

Ok I think I know what this problem is ... I think it has to do with the way
I built my compiler ... I believe it has to do with the fact I did not build
gcc with the " --enable-shared" option ...

Any one know for sure?

I will keep reading ...

Jim



-Original Message-
From: James Don
Sent: Tuesday, January 07, 2003 6:31 PM
To: 'uclibc at uclibc.org'
Subject: Help: Simple newbie build problem

Hello,

Just hoping to find some help with a simple build problem I an having with
uClib.

First off here's some facts:
1.) I have built cross compilers 2.95.3 ... powerpc-linux-gcc with newlib
... etc
2.) I have build bootloader and kernel for embedded board ...
3.) Now I am building uClibc ... I changed the rules.mak file to have the
right CROSS=powerpc-linux- ...

Now here's my problem:
1.) I do this make defconfig
2.) Then I do this make
3.) Then I see this:
powerpc-linux-ld: cannot open
/home/jgdon/ppcgcc/lib/gcc-lib/powerpc-
linux/2.95.3/pic/crtbeginS.o: No such file or directory
make[1]: *** [shared] Error 1
make[1]: Leaving directory `/home/jgdon/ppcUClibc/uClibc
0.9.16/libpthread'

I have found the file crtbeginS.o in home/jgdon/ppcgcc/lib/gcc-lib/powerpc-
linux/2.95.3/ ... or one directory back of where my build is looking ...

Any advice would be much appreciated,
Jim


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





BDI-2000

2003-01-07 Thread Muaddi, Cecilia

Hello,

I finally got my BDI-2000 for PPC 860 this week, and tried to set it up to
debug my MMU problem.
This is my first time using the BDI-2000, and really hope someone here can
give me a hand.

Here is what I am trying to do:

I have a proven hardware platform with
PPC860, IMMR is mapped to 0xf000.
16 MB of SRAM in CS3 starts at physical locatio 0x0
1 MB of BootROM in CS1 starts at physical location 0x2800
16 MB of code flash in CS5 starts at physical location 0x0800
some other IO mapped to miscellaneous addresses.

The hardware platform is running vxWorks with vxWorks bootrom.  I am trying
to
port the Linux to this custom hardware platform.  I was successful building
a zImage (no intrd yet)
with the compress vmLinux.gz.  The zImage.srec is then downloaded to the
target
platform using the vxWorks bootrom (I wrote a little program to take
srecords).  The following
information was dump out on the console

boot device  : motfec
unit number  : 0
processor number : 0
host name: linux
file name: linux\images\zImage.hex
inet on ethernet (e) : 192.168.0.238
host inet (h): 192.168.0.211
user (u) : me
ftp password (pw): me
flags (f): 0x8
target name (tn) : onu
other (o): motfec

Attached TCP/IP interface to motfec0.
Attaching network interface lo0... done.
Loading... S-Record:
Module Name: ../images/zImage.srec
Entry Location: 0x40
Starting at 0x40...

loaded at: 0040 0040C30C
board data at: 004001C0 004001E4
relocated to:  0040C0E8 0040C10C
zimage at: 0040C30C 004BC445
avail ram: 004BD000 0080

Linux/PPC load: console=ttyS0,9600
nfsroot=192.168.0.211:/exports/rpx860/2.4.7-t
imesys-3.1.254 ip=192.168.0.238
Uncompressing Linux... done.
Now booting the kernel

I was able to follow the execution from the power reset until the point
where the MMU is enabled
in the function kernel\arch\ppc\kernel\head_8xx.S:turn_on_mmu with visionICE
II (yes, I know there
is an update for the firmware to get visionICE to work with MMU, but I am
waiting for that also).

I don't think my MMU table is setup properly, therefore, I have aquired a
Abatron BDI-2000.  I have to
modified the .cfg files to use my UPM A table.  Furthermore, I modified the
registers defined in the
sample .cfg file to the values defined in my firmware.  I was successful in
powering up the BDI-2000,
connecting it to my target board, and telnet into the BDI.

I am able to check and modify the SRAM (at physical address location 0x0).
and view the bootROM memory
(at physical address of 0x2800).  I enable the MMU XLAT and set the
PTBASE to 0xf0 as suggested in the
user manual.
Here is the output of the BDI

BDI>reset
- TARGET: processing user reset request
- TARGET: resetting target passed
- TARGET: processing target init list 
- TARGET: processing target init list passed
BDI>bi 0x0c00 0x0c00
Breakpoint identification is 0
BDI>go 0x28000100
- TARGET: target has entered debug mode
BDI>ci
BDI>info
Target state  : debug mode
Debug entry cause : entering check stop state
Current PC: 0x0220
BDI>

I never seem to get the break at the location of 0x0c00.
The same problem if I set the break at location of start_here grep from
System.map

However I do get the break if I set at address location of 0 (start of the
head_8xx.S)
Can someone help me out here??

Thanks

Cecilia

P.S. here is a brief information of my .cfg file which I loaded to BDI

; bdiGDB configuration file for TQM855L/TQM860L Modules
; -
;
[INIT]
; init core register
WREGMSR 0x1002  ;MSR  : ME,RI
WSPR27  0x1002  ;SRR1 : ME,RI
;;WSPR  149 0x2002000F  ;DER  : set debug enable register
WSPR149 0x2006000F  ;DER  : enable SYSIE for BDI flash
progr.
WSPR638 0xF000  ;IMMR : internal memory at
0xF000
WSPR158 0x0007  ;ICTRL:

; init SIU register
WM320xF000  0x70630200  ;SIUMCR
WM320xF004  0xFF80  ;SYPCR

;WM32   0xFFF00284  0x  ;PLPRCR no need to change (1:1 clock
mode)

; Init Memory Controller:
;
; BR0 and OR0 (BOOT FLASH)
WM320xF104  0xFFF00550  ; 1 MB
WM320xF100  0x28000401  ; @ 0x2800 - 0x280F, 8Bit,
GPCM

; BR3 and OR 3 (SDRAM)
WM320xF11C  0xFF000800  ; 16 MB
WM320xF118  0x0081  ; @ 0x - 0x00FF, 32Bit,
UPMA

; BR5 and OR5 (MAIN FLASH & MDIO GLUE)
WM320xF12C  0xFD000550  ; 16 MB FLASH / 16 MB GLUE
WM320xF