[Bug 1000313] unable to execute linux kernel with Redboot

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313


jifl@ecoscentric.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|gary@mlbassoc.com           |msalter@redhat.com




------- Additional Comments From jifl@ecoscentric.com  2006-09-14 13:38 -------
I don't know then sorry, on the basis of the info you've given - I'm not sure
what would have differed between those releases since I don't know what they
contain (not being ecos-sourceware.org release). The only thing that occurs to
me is: have you checked the endianness? Both of the Linux kernel and RedBoot.

Alternatively, as I said, you could try from ecos.sourceware.org cvs.

I'll assign this to msalter, who may be able to help you further.


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313


msalter@redhat.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |NOTABUG




------- Additional Comments From msalter@redhat.com  2006-09-22 14:43 -------
That's where RedBoot places and executes the code which copies and executes
the kernel image. In your case, with a compressed image (zImage), no copying
gets done, but the tramoline code is the code that jumps to the zImage.
Look at redboot_linux_exec.c:do_exec to see the trampoline code.

I'm closing this bug because your problem only occurs on your custom
hardware. Any further discussion should be taken to the ecos-discuss list.


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-22 14:00 -------
What is it CYGHWR_REDBOOT_ARM_TRAMPOLINE_ADDRESS=0x00001f00?

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-21 15:15 -------
I build kernel 2.6.16 with flag CONFIG_DEBUG_LL, but olny "Uncompressing
Linux.... done, booting the kernel." 

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From msalter@redhat.com  2006-09-21 15:07 -------
Okay, so now we are back where we started. :)

Once you get the "Uncompressing..." message, RedBoot is largely
out of the picture. The kernel has taken over the hardware at
this point. Some thoughts:

  * you probably want to pass a command line to the kernel telling the
    kernel to use one of the serial ports as the serial console and the
    baudrate to use.

  * Configure the kernel with CONFIG_DEBUG_LL turned on. This will force
    some characters out of the serial port early in the kernel boot process.
    This should give you an indication that the kernel code is actually
    executing.

As I mentioned before, there could be a RedBoot problem, but you will most
likely need to debug it from the kernel side.

>From my point of view, there doesn't seem to be a RedBoot bug with the IXDP425
port. You're porting to custom hardware. If you can't reproduce a problem on the
IXDP425, then you should be using the ecos-discuss list and/or the ARM linux
mailing list to ask for help.


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-21 14:49 -------
Trying NPE-B...success. Using NPE-B with PHY 0.
Ethernet eth0: MAC address 00:03:47:df:32:a8
IP: 192.168.0.1/255.255.255.0, Gateway: 192.168.0.1
Default server: 192.168.0.10

RedBoot(tm) bootstrap and debug environment [ROM]
Non-certified release, version UNKNOWN - built 12:08:38, Sep 21 2006

Platform: IXDP425 Development Platform (IXP42X 266MHz) BE
Copyright (C) 2000, 2001, 2002, 2003, 2004 Red Hat, Inc.
Copyright (C) 2003, 2004, 2005, 2006 eCosCentric Limited

RAM: 0x00000000-0x02000000, [0x00029930-0x01fc1000] available
FLASH: 0x50000000 - 0x51000000, 128 blocks of 0x00020000 bytes each.
RedBoot> load -v -r -m tftp -b 0x1000000 /linux/kernel
/
Raw file loaded 0x01000000-0x010fffff, assumed entry at 0x01000000
RedBoot> exec -b 0x1000000 -l 0 0x1000000
entry=10008000 target=10008000
Uncompressing Linux........................................................
done, booting the kernel.

and hang silently

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-21 07:30 -------
I have made all as you have written but loading kernel to asm code in
redboot_linux_boot.c and hang silently. This id part of asm code -

    asm volatile (
        CYGARC_HAL_MMU_OFF(%5)
        "__tramp_start__:\n"
        " cmp %1,%4;\n"       // Default kernel load address. Relocate
        " beq 2f;\n"          // kernel image there if necessary, and
        " cmp %2,#0;\n"       // if size is non-zero
        " beq 2f;\n"
....

> Just using "exec" by itself is not going to work for you
> because it will assume you are booting a raw kernel image
> without the decompressor wrapper. from previous messages it
> looks like you are booting a compressed image. You should
> try something like:
> 
> RedBoot> load -v -r -m tftp -b 0x1000000 /linux/kernel
> RedBoot> exec -b 0x1000000 -l 0 0x1000000
> 
> The "-l 0" part is important to keep "exec" from moving
> the image down to where the final kernel code will reside.
> The 0x1000000 address is mostly arbitrary and just needs to
> be somewhere not used by RedBoot.
> 
> As far as the memory controller setup goes, you'll need to
> decide for yourself if its working for you or not. I do
> believe the 566Mhz clock used in the refresh calculation is
> wrong. You need to use the memory bus clock, not the cpu
> core clock, IIRC, the memory bus runs at (4 * 33MHz).
> 



-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From msalter@redhat.com  2006-09-19 16:14 -------
> RedBoot> load -v -r -m tftp -b %{FREEMEMLO} /linux/kernel
> Raw file loaded 0x00029c00-0x00129bff, assumed entry at 0x00029c00
> RedBoot> exec
> Using base address 0x00029c00 and length 0x00100000
> $T050f:06000000;0d:0000b6a8;

Just using "exec" by itself is not going to work for you
because it will assume you are booting a raw kernel image
without the decompressor wrapper. from previous messages it
looks like you are booting a compressed image. You should
try something like:

RedBoot> load -v -r -m tftp -b 0x1000000 /linux/kernel
RedBoot> exec -b 0x1000000 -l 0 0x1000000

The "-l 0" part is important to keep "exec" from moving
the image down to where the final kernel code will reside.
The 0x1000000 address is mostly arbitrary and just needs to
be somewhere not used by RedBoot.

As far as the memory controller setup goes, you'll need to
decide for yourself if its working for you or not. I do
believe the 566Mhz clock used in the refresh calculation is
wrong. You need to use the memory bus clock, not the cpu
core clock, IIRC, the memory bus runs at (4 * 33MHz).


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-19 12:23 -------
When running redboot ROM - 
RedBoot(tm) bootstrap and debug environment [ROM]
Non-certified release, version UNKNOWN - built 14:24:42, Sep 19 2006

Platform: IXDP425 Development Platform (IXP42X 533MHz) BE
Copyright (C) 2000, 2001, 2002, 2003, 2004 Red Hat, Inc.
Copyright (C) 2003, 2004, 2005, 2006 eCosCentric Limited

RAM: 0x00000000-0x04000000, [0x00029900-0x03fc1000] available
FLASH: 0x50000000 - 0x50800000, 64 blocks of 0x00020000 bytes each.
RedBoot> load -v -r -m tftp -b %{FREEMEMLO} /linux/kernel
/
Raw file loaded 0x00029c00-0x00129bff, assumed entry at 0x00029c00
RedBoot> exec
Using base address 0x00029c00 and length 0x00100000
$T050f:06000000;0d:0000b6a8;

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-19 10:40 -------
RedBoot(tm) bootstrap and debug environment [RAM]
Non-certified release, version UNKNOWN - built 13:01:49, Sep 19 2006

Platform: IXDP425 Development Platform (IXP42X 533MHz) BE
Copyright (C) 2000, 2001, 2002, 2003, 2004 Red Hat, Inc.
Copyright (C) 2003, 2004, 2005, 2006 eCosCentric Limited
With this configuration the board works but to not be loaded a kernel

RAM: 0x00000000-0x02000000, [0x0016cc98-0x01fc1000] available
FLASH: 0x50000000 - 0x50800000, 64 blocks of 0x00020000 bytes each.
RedBoot> mfill -b 0x0016d000 -l 0x1000 -p 1
RedBoot> mfill -b 0x0016d000 -l 0x10000 -p 1
RedBoot> mfill -b 0x0016d000 -l 0x100000 -p 2
RedBoot> mfill -b 0x0016d000 -l 0xf00000 -p 5
RedBoot> mfill -b 0x0016d000 -l 0x1000000 -p 0
RedBoot> mfill -b 0x0016d000 -l 0x1a00000 -p 1
RedBoot>
RedBoot> load -r -v -b 0x01600000 /linux/kernel
Using default protocol (TFTP)
/
Raw file loaded 0x01600000-0x016fffff, assumed entry at 0x01600000
RedBoot> fis create -b 0x01600000 -l 0x1000000 -e 0x00800000 -r 0x01600000 zimage
** WARNING: RAM address: 0x01600000 may be invalid
   valid range is 0x00000000-0x02000000
Can't locate 1000000(16777216) bytes free in FLASH
RedBoot> fis create -b 0x01600000 -l 0xc0000 -e 0x00800000 -r 0x01600000 zimage
... Erase from 0x50080000-0x50140000: ......
... Program from 0x01600000-0x016c0000 at 0x50080000: ......
... Unlock from 0x507e0000-0x50800000: .
... Erase from 0x507e0000-0x50800000: .
... Program from 0x01fe0000-0x02000000 at 0x507e0000: .
... Lock from 0x507e0000-0x50800000: .
RedBoot> fis load zimage
RedBoot> exec -b 0x01600000
Length required for non-standard base address
RedBoot> exec
Using base address 0x01600000 and length 0x000c0000

it hang silently
=((

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 841 bytes --]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-19 09:56 -------
On my board SDRAM is 2 chip 2*256Mb
MT48LC16M16A2 – 4 Meg x 16 x 4 banks
I set define
#define SDRAM_SIZE                         0x04000000  // 64MB
#define IXP425_SDRAM_CONFIG_INIT  (SDRAM_CONFIG_CAS_3 | SDRAM_CONFIG_2x16Mx16)
The IXDP SDRAM controller has to refresh the 8192 rows each 64 ms which
means one refresh each 7.81us. At 533Mhz it means ~ 4162 cycles which is
0x1042. 
#define IXP425_SDRAM_REFRESH_CNT  0x1042
#define IXP425_SDRAM_SET_MODE_CMD SDRAM_IR_MODE_SET_CAS3

It is correct?



-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From msalter@redhat.com  2006-09-15 16:13 -------
You'll need to consult the IXP425 developer's manual for details on
the memory controller, but the configuration bits are fairly simple.

The SDRAM controller is configured as one of the first steps during
a poweron boot. This in done in:

  hal/arm/xscale/ixdp425/current/include/hal_platform_setup.h

The code in there is generic and differences in memory hardware
are controlled by #defines in:

  hal/arm/xscale/ixdp425/current/include/ixdp425.h

What you'll need to look at are:

  SDRAM_SIZE   - This must already be set to 32MiB as indicated
                 by the RedBoot startup banner.

  IXP425_SDRAM_CONFIG_INIT - This tells the SDRAM layout (number of banks, etc)
  IXP425_SDRAM_REFRESH_CNT - See manual on how to calculate
  IXP425_SDRAM_SET_MODE_CMD - See manual for description of mode



-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-15 15:26 -------
 
When mfill crashed, it hang silently.
How to adjust a memory controller setup? On this board - RAM 32M and 16M flash 



-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From msalter@redhat.com  2006-09-15 13:46 -------
I'd say you need to find the problem with mfill before worrying about
booting linux. I notice from the startup message that your board has
32MiB of ram. This is not the standard IXDP425 board which has 256MiB.
I see no such problem with the standard IXDP425 and the prebuilt
RedBoot 2.02 image. I would experiment with mfill to see if there is
a particular area of memory that has the problem. Also, I would 
recheck the memory controller setup to make sure everything is okay
there. The only time I've seen mfill fail like that is when the
memory controller was incorrectly configured.

Also, you say mfill crashed, but you didn't say how. Did it hang
silently or was there a message?


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-14 18:29 -------
Version my kernel is 2.6.16. This kernel image has work on old version of redboot
(e.g. 1.94). 

Command mfill -b 0x01000000 -l 0x00100000 crashed in redboot 2.02

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From msalter@redhat.com  2006-09-14 15:49 -------
What version of the kernel are you using? Are you sure the kernel is configured 
for a serial console. Once you get to the "booting kernel" stage, the decompressor
jumps into the kernel-proper. If something goes wrong before the console gets
initilized you'll not see any messages. In the kernel, you can turn on:

  CONFIG_DEBUG_LL

which will force some characters out the uart early in the boot process.

Basically, this may be a RedBoot problem, but it'll have to be debugged
from the kernel side.

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-14 15:29 -------
Please you could not tell as to build redboot from cvs under ixp425 with npe_eth?

-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From jifl@ecoscentric.com  2006-09-14 15:10 -------
In case it helps, this is specifically what has worked for me before with the
eCosCentric RedBoot, which is more similar to redboot 2.02 than 1.94. (Before
you ask, unfortunately I can't provide that redboot unless you wish to purchse
an eCosPro Developer's Kit).

But this is what has worked (loaded via TFTP):
RedBoot> load -r -m tftp -b %{freememlo} /zImage
Raw file loaded 0x0007ac00-0x001feaf7, assumed entry at 0x0007ac00
RedBoot> exec
Using base address 0x0007ac00 and length 0x00183ef8
Uncompressing
Linux.........................................................................................................
done, booting the kernel.
Linux version 2.6.12 (root@andy) (gcc version 3.4.4) #43 Tue Nov 15 16:40:04 GMT
2005
CPU: XScale-IXP42x Family [690541c1] revision 1 (ARMv5TE)
CPU0: D VIVT undefined 5 cache
CPU0: I cache: 32768 bytes, associativity 32, 32 byte lines, 32 sets
CPU0: D cache: 32768 bytes, associativity 32, 32 byte lines, 32 sets
Machine: Intel IXDP425 Development Platform
[snip many more lines of output]


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-14 13:59 -------
Thanks, hat you have answered me!
Kernel  and redboot - big endian. 

This is output from console:
Platform: IXDP425 Development Platform (IXP42X 266MHz) BE
Copyright (C) 2000, 2001, 2002, 2003, 2004 Red Hat, Inc.

RAM: 0x00000000-0x02000000, [0x0016c338-0x01fc1000] available
FLASH: 0x50000000 - 0x51000000, 128 blocks of 0x00020000 bytes each.
RedBoot> load -v -r -b 0x01600000 /linux/kernel
Using default protocol (TFTP)
/
Raw file loaded 0x01600000-0x016fffff, assumed entry at 0x01600000
RedBoot> exec
exec start // my comment
entry_address=1600000 entry=8000 CYGHWR_REDBOOT_ARM_LINUX_EXEC_ADDRESS=8000 //
my comment
wait=0 // my comment
base_addr=0 // my comment
Using base address 0x01600000 and length 0x00100000
ETH stop...yes // my comment
begin asm code... yes //my comment
Uncompressing Linux........................................................
done, booting the kernel.

and end. =((


I can send my .ecm script.


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From lexaiu4@mail.ru  2006-09-14 13:27 -------
(In reply to comment #1)
> I have used these successfully:
> 
> CYGHWR_REDBOOT_ARM_TRAMPOLINE_ADDRESS should be 0x00001f00
> CYGHWR_REDBOOT_ARM_LINUX_EXEC_ADDRESS_DEFAULT should be 0x00008000
> 
> If those are the same as yours, your problem is elsewhere. You don't say whether
> "old version" refers to Linux or RedBoot. Incidentally, 2.02 is an Intel
> release, not one from ecos.sourceware.org.
> 
> You could try rebuilding redboot from current ecos.sourceware.org cvs.
> 

Yes old version is Redboot(e.g. 1.94) and 2.02 - Intel realease
What can be problem then?



-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.

[Bug 1000313] unable to execute linux kernel with Redboot

[bugzilla-daemon]

http://bugzilla.ecoscentric.com/show_bug.cgi?id=1000313





------- Additional Comments From jifl@ecoscentric.com  2006-09-14 11:18 -------
I have used these successfully:

CYGHWR_REDBOOT_ARM_TRAMPOLINE_ADDRESS should be 0x00001f00
CYGHWR_REDBOOT_ARM_LINUX_EXEC_ADDRESS_DEFAULT should be 0x00008000

If those are the same as yours, your problem is elsewhere. You don't say whether
"old version" refers to Linux or RedBoot. Incidentally, 2.02 is an Intel
release, not one from ecos.sourceware.org.

You could try rebuilding redboot from current ecos.sourceware.org cvs.


-- 
Configure bugmail: http://bugzilla.ecoscentric.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.