kernel memory mapping problem

kernel memory mapping problem

[ZhongJun Zhou]

Dear all,

I have a problem that :

the function is :

int func()
{
	unsigned char tmp[10];
	unsigned char * p;

	p = __pa(tmp);
	...
	pI2cRxBD -> bd_addr = p;
	...
	...
}
when run this module( lsmod mymodule.o ), kernel is always dead. Are there any errors of this function ?

Any help is appreciated!

John Zhou

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

RE: kernel memory mapping problem

[John Zhou]

Another strange problem:

in my driver moduler, I can directly access cpu's registers such as port A, B,C,D( my IMMR = 0xF0000000 ), but I can't directly read the SDRAM's value. for example:

int func()
{
	unsigned char tmp[10];
	unsigned char * p;
	unsigned long val;

	val = *( unsigned long * ) 0xF0010D0C; /* Port Open-Drain Registers (PODRA) */
	p = __pa(tmp);
	...
	pI2cRxBD -> bd_addr = p;
	...
	...
}

Thanks in advance!

John

-----Original Message-----
From: ZhongJun Zhou [mailto:zjhzou@newrocktech.com]
Sent: Thursday, February 07, 2002 11:21 AM
To: linuxppc-embedded@lists.linuxppc.org
Subject: kernel memory mapping problem


Dear all,

I have a problem that :

the function is :

int func()
{
	unsigned char tmp[10];
	unsigned char * p;

	p = __pa(tmp);
	...
	pI2cRxBD -> bd_addr = p;
	...
	...
}
when run this module( lsmod mymodule.o ), kernel is always dead. Are there any errors of this function ?

Any help is appreciated!

John Zhou

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

This question may be simple for you, but it doesn't let me sleeping.

[John Zhou]

Dear all,

This question may be simple for you, but it doesn't let me sleeping.

I have linux2.4.1 from Lineo runing on ppc8250, IMMR = 0xF0000000, in my driver function, I can directly access 0xF00xxxxx( internal memory area on PPC's chip), why? I think , in device driver, we also use logical memory, when accessing physical address, it should first map this address to be accessed by hardware. Is it right?

Another, I write a character driver to operate I2C device. When passing a logical address pointer to BD's address pointer that allocated by function kmalloc().  I can't correctly operate I2C device, But when I used free memory located in IMM ( 0xF00037C0 ~ 0xF00037F0  ) and directly pass this address to BD's buffer pointer, I can correctly operate I2C device. I also try to use function __pa() and virt_to_phys()  to map logical address to physical address, The driver still can't work OK.

Any comments and suggestions are appreciated!

John Zhou

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

RE: debugging IBM405GP with abatron bdi 2000

[Steven Blakeslee]

I found a work around that would suggest the opposite.  If I do the
following I can get Linux to boot with an Abatron connected.
1.  Connect the board to the abatron and power up.
2.  Wait for the configuration to complete.
3.  Give the go command, the bootloader now runs.
4.  Give the reset command in the bootloader, this causes a checkstop reset.
A checkstop reset sets most registers to their default values.
5.  The bootloader loads a second time.
6.  Download Linux and boot from bootloader.

This tells me that the abatron is doing some type of initialization other
than the config file.  Whatever it is doing behind the scenes is causing a
problem after the rfi instruction.

-----Original Message-----
From: Chris Hallinan [mailto:clh@net1plus.com]
Sent: Tuesday, August 05, 2003 1:40 PM
To: Steven Blakeslee
Subject: RE: debugging IBM405GP with abatron bdi 2000


It is likely that your kernel is mucking with the debug registers in
the processor core.  This is typically what happens if that is the
case...  Just a thought...

> -----Original Message-----
> From: owner-linuxppc-embedded@lists.linuxppc.org
> [mailto:owner-linuxppc-embedded@lists.linuxppc.org]On
> Behalf Of Steven
> Blakeslee
> Sent: Tuesday, August 05, 2003 11:53 AM
> To: 'linuxppc-embedded@lists.linuxppc.org'
> Subject: debugging IBM405GP with abatron bdi 2000
>
>
>
> Has anyone used an Abatron BDI 2000 to debug Linux on an IBM 405GP
> processor?  I am using an Embedded Planet ep405 with the latest
> linuxppc_2_4_devel tree.  Linux boots fine on the board
> until I connect the
> Abatron.  When the Abatron is connected I can step
> through the code until
> the rfi command in head_4xx.S then I get the following.
>
> - TARGET: target has entered debug mode
>     Target state      : debug mode
>     Debug entry cause : JTAG stop request
>     Current PC        : 0x00000700
>     Current CR        : 0x28004088
>     Current MSR       : 0x00000000
>     Current LR        : 0x000007a8
> # Step timeout detected
>
>
> Below is the config file I am using.  I would appreciate
> any adive anyone
> can offer.
>
>
> [INIT]
> ;
> WSPR	954	0x00000000	;DCWR: Disable data cache write-thru
> WSPR	1018	0x00000000	;DCCR: Disable data cache
> WSPR	1019	0x00000000	;ICCR: Disable instruction cache
> WSPR	982	0x00000000	;EVPR: Exception Vector
> Table @0x00000000
> ;
> ; Setup TCR (Timer Control Register)
> WSPR	986	0x00400000
> ; Setup MMU info
> WM32    0x000000f0  0x00000000  ;invalidate page table base
> ;
> ;
> [TARGET]
> JTAGCLOCK   0			;use 16 MHz JTAG clock
> CPUTYPE	405			;the used target CPU type
> BDIMODE	AGENT			;the BDI working
> mode (LOADONLY | AGENT)
> BREAKMODE   SOFT			;SOFT or HARD, HARD uses PPC
> hardware breakpoint
> STEPMODE    JTAG			;JTAG or HWBP, HWPB
> uses one or two
> hardware breakpoints
> MMU	    XLAT 0xC0000000	;enable virtual address mode
> PTBASE      0x000000f0  	;here is the pointer to the
> page table
> pointers
> ;
> [HOST]
> IP          10.0.0.198     ;Windows host
> FILE        rImage.srec
> FORMAT      SREC
> LOAD        MANUAL              ;load code MANUAL or AUTO
> after reset
> ;
> ;
> [REGS]
> IDCR1	0x010	0x011	;MEMCFGADR and MEMCFGDATA
> IDCR2	0x012	0x013	;EBCCFGADR and EBCCFGDATA
> IDCR3	0x014	0x015	;KIAR and KIDR
> FILE    defs/reg405gp.def
>
>


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

Re: debugging IBM405GP with abatron bdi 2000

[Wolfgang Denk]

In message <D73A25AA6E54D511AD74009027B1110F3C099A@ORION> you wrote:
>
> This tells me that the abatron is doing some type of initialization other
> than the config file.  Whatever it is doing behind the scenes is causing a
> problem after the rfi instruction.

You cannot single step through some instructions, including rfi. This
has been discussed before, you may want to search  the  mailing  list
archive.


Best regards,

Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
Perfection is reached, not when there is no longer anything  to  add,
but when there is no longer anything to take away.
                                           - Antoine de Saint-Exupery

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

MPC8250 MCC driver

[Adam Kaczynski]

I would like to present a very simple driver for MPC8250/8260 MCC
working in transparent and HDLC modes, released under GPL.

It uses one time-slot assigner (128 timeslots) and each individual
channel may be transparent or HDLC. The "read" and "write" functions are
non-blocking and the user should block on "poll". This driver is
prepared to work on our custom-made MPC8250 board but it would be easy
to port it to any evaluation board.

http://www.dgt-lab.com.pl/Serwis/char_dev.tgz

This is rather an example code than a fully polished masterpiece but it
works well and it is being tested in our new product. The sample
programmes are attached to this module.

I would like to thank Omanakuttan for his help and for his precious
documentation:
http://www.ibiblio.org/gferg/ldp/MPC8260-MCC-HOWTO.txt

Adam


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

load_elf_bin problem

[dong in kang]

 Hello,

 The linux kernel boot up well till the last stage to call /bin/sh.
It crashes inside of load_elf_bin.
More specifically

   load_elf_bin()  --> setup_arg_pages() --> put_dirty_page();

Inisde of put_dirty_page() routing, it crashes at
   if (!pte_none(pte)) {

The program accesses pte->pte, but the address of pte->pte is not translated
to physical address properly.
So machine exception occurs, and the kernel crashes.

 Is it paging problem?
I turned L1 and L2 cache off because those have defects.
Does it cause any problem to paging mechanism?

 I'll appreciate your help.

 Dong-In


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

Re: kernel memory mapping problem

[Peter Barada]

>when run this module( lsmod mymodule.o ), kernel is always dead. Are
>there any errors of this function ?
>
>Any help is appreciated!

What are you doing with the physical address of a varaible that is on
the *stack*?  If you let some hardware modify that value, then you can
easily trash your stack if the hardware modifies this value *after*
this function exits since tmp will go out of scope.

I'd suggest that you move tmp outside of func for starters.

--
Peter Barada                             Peter.Barada@motorola.com
Wizard                                   781-852-2768 (direct)
WaveMark Solutions(A Motorola Company)   781-270-0193 (fax)

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

Re: kernel memory mapping problem

[ZhongJun Zhou]

Thanks very much.

But my functions are serial running on PPC82xx, its
means that when I mapped the the address using __pa(
(void *)tmp) to bd_address, then reading the BD's
status and control registers, if OK, then return from
this function. Is it OK with this method? Next, if I
use "kmalloc" to get the buffer and map it to CPM with
__pa() function, that can be run, is right?  any other
functions can finish this mapping action ? or give me
any good documents for my reading?

thanks in advance!

--- Peter Barada <pbarada@mail.wm.sps.mot.com>
> >when run this module( lsmod mymodule.o ), kernel is always dead. Are
> >there any errors of this function ?
>
> What are you doing with the physical address of a varaible that is on
> the *stack*? If you let some hardware modify that value, then you can
> easily trash your stack if the hardware modifies this value *after*
> this function exits since tmp will go out of scope.
>
> I'd suggest that you move tmp outside of func for starters.

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

Re: kernel memory mapping problem

[Wolfgang Denk]

In message <20030807154923.56073.qmail@web105.mail.yahoo.com> you wrote:
>
> But my functions are serial running on PPC82xx, its
> means that when I mapped the the address using __pa(
> (void *)tmp) to bd_address, then reading the BD's
> status and control registers, if OK, then return from

Just a stupid question: what are the alignment requirements of a  BD,
and which alignment is guaranteed for your tmp buffer?


Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
This cultural mystique surrounding the  biological  function  --  you
realize humans are overly preoccupied with the subject.
	-- Kelinda the Kelvan, "By Any Other Name", stardate 4658.9

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

Does Linux kernel use paging or BAT registers before running the first user-level program "/bin/sh"?

[dong in kang]

 Does Linux kernel use paging or BAT registers before executing "/bin/sh"?
How can I test if paging works fine?

  My Linux kernel crashes while it accesses page table entry for argument
page which is called
   load_elf_binary --> setup_arg_pages --> put_dirty_page.
The address of page table entry is 0xb0beffc which is not a valid address.
When the kernel accesses that address, it crashes.
I suspect that paging may not work properly.

Did anybody have this problem?

 Dong-In


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

RE: This question may be simple for you, but it doesn't let me sleeping.

[Jean-Denis Boyer]

> I have linux2.4.1 from Lineo runing on ppc8250, IMMR =
> 0xF0000000, in my driver function, I can directly access
> 0xF00xxxxx( internal memory area on PPC's chip), why?

A BAT is configured to map this region with addressing virtual == physical.
See function m8260_map_io in arch/ppc/kernel/m8260_setup.c

> When passing a logical address pointer to BD's address
> pointer that allocated by function kmalloc().

Memory should probably be uncached (as the IMMR is).
Use function m8260_cpm_hostalloc instead of kmalloc, then use __pa.
See for example arch/ppc/8260_io/fcc_enet.c

Regards,
--------------------------------------------
 Jean-Denis Boyer, Eng.
 Software Designer
 M5T Centre d'Excellence en Télécom Inc.
 4283 Garlock Street
 Sherbrooke (Québec)
 J1L 2C8  CANADA
 (819)829-3972 x241
--------------------------------------------

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

how to use busybox to make a ramdisk?

[John Zhou]

Hi all,

I want to get some detail documents about how to use busybox to make a ramdisk!!  Does anybody know? Is it to use mkinitrd ?

Thanks & Regards
John Zhou

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

Re: how to use busybox to make a ramdisk?

[Dan Kegel]

Brian Padalino wrote:
> ... there is also a good tutorial on how to create ramdisk images.
>
> http://www.linux.org/docs/ldp/howto/Bootdisk-HOWTO/buildroot.html
>
> The only thing that I do differently is that I use a loopback device like
> so:
>   mount -t ext2 filename /mnt/image -o loop
> This stops mount from complaining to me and I like that.

I highly recommend using genext2fs instead of a loopback mount;
that way you don't need to be root at all when making the image.

Stock genext2fs doesn't handle ramdisks bigger than 8MB, but
there are patches that fix this, and I've been using them happily
for about nine months.  (Maybe they're in the mainline now; not sure.)
- Dan

--
Dan Kegel
http://www.kegel.com
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=78045


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

Re: how to use busybox to make a ramdisk?

[Wolfgang Denk]

In message <3F4F6819.1090002@kegel.com> you wrote:
>
> Stock genext2fs doesn't handle ramdisks bigger than 8MB, but
> there are patches that fix this, and I've been using them happily
> for about nine months.  (Maybe they're in the mainline now; not sure.)

A patched & extended version  (based  on  Erik  Andersen's  work)  is
included with the ELDK.

Best regards,

Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
365 Days of drinking Lo-Cal beer.                       = 1 Lite-year

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

RE: how to use busybox to make a ramdisk?

[Brian Padalino]

John,

I kind of cheated and used the DENX ELDK (http://www.denx.de) ramdisk that
came with his stuff (mainly for the /dev entries).

But there is also a good tutorial on how to create ramdisk images.

http://www.linux.org/docs/ldp/howto/Bootdisk-HOWTO/buildroot.html

The only thing that I do differently is that I use a loopback device like
so:
  mount -t ext2 filename /mnt/image -o loop
This stops mount from complaining to me and I like that.

Hope this helps and good luck!

Cheers,
Brian Padalino


-----Original Message-----
From: owner-linuxppc-embedded@lists.linuxppc.org
[mailto:owner-linuxppc-embedded@lists.linuxppc.org]On Behalf Of John
Zhou
Sent: Thursday, August 28, 2003 5:54 AM
To: linuxppc-embedded@lists.linuxppc.org
Subject: how to use busybox to make a ramdisk?



Hi all,

I want to get some detail documents about how to use busybox to make a
ramdisk!!  Does anybody know? Is it to use mkinitrd ?

Thanks & Regards
John Zhou


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