[ECOS] Re: Little Endian GDB stub for VRC4373

[ECOS] Re: Little Endian GDB stub for VRC4373

[Pengyu Xue]

Yes, we are working on  little-endian mode in the platform 
vr4300-vrc4373 for a long time, it really works.
You can compile a little-endian mode gdb stub using -EL in your compile 
flag, and also you must modify the CYGPKG_HAL_MIPS_MSBFIRST to 
CYGPKG_HAL_MIPS_LSBFIRST, and -O2 is necessary for compiling gdb stub.
If you still have some problems, maybe I can help you, because we have 
almost our eCos project in this platform. ^_^




pyxue

Re: [ECOS] Re: Little Endian GDB stub for VRC4373

[Ling Su]

Hi, Dear Pengyu and others,

With the enlightment of Penyu's email, I decided to go ahead to bring up the
Little Endian GDB stub. I also take a look at the platform dependent code, I
didn't find any specific part to be modified. The LSBFIRST and MSBFIRST
macro definiation guards all the related type definition and address
mapping. I changed the ecos.ecc for the CFLAG after I generate it by apply
the "stubs" Template. Then I go ahead change the
./install/include/pkgconf/hal_mips_vr4300.h make the LSBFIRST become default
setting. Then I execute make for the gdb_module.bin, it generated
successfully. Unfortnately, it doesn't work after I replace it with my only
stub on board. I trace the PCI bus by using LA, no signal pin is active, so
it doesn't execute at all. I am wondering what happened, so I think it would
be better to make a Big Endian stub first before I jump for the little
endian one. In my memory, I have never make a good stubs before by using my
own configuration. I tried "ecosconfig new vrc4373 stubs" on the new CVS
repository and some old ones which I am working on. To my surprise, no one
works. (I did remember to change the JP66 for the endian issue).

Now, I am quite confused on how to make my own GDBstubs work. Since looks to
me, even the default big endian version doesn't work at all. I am using the
binary one provided in ecos-1.3.1 release till now due to my failure on
building it. Have anyone experienced the same problem? or how to resolve it?

Pengyu, could you please do me a favor to send a little endian stub to my
private mailbox? I hope I could get a working one and take a try.

Thanks for everyone prvoiding valuable hints!

Regards,
-Ling

----- Original Message -----
From: "Pengyu Xue" <pyxue@zh.t2-design.com>
To: <ecos-discuss@sourceware.cygnus.com>
Sent: Thursday, January 11, 2001 8:06 PM
Subject: [ECOS] Re: Little Endian GDB stub for VRC4373


> Yes, we are working on  little-endian mode in the platform
> vr4300-vrc4373 for a long time, it really works.
> You can compile a little-endian mode gdb stub using -EL in your compile
> flag, and also you must modify the CYGPKG_HAL_MIPS_MSBFIRST to
> CYGPKG_HAL_MIPS_LSBFIRST, and -O2 is necessary for compiling gdb stub.
> If you still have some problems, maybe I can help you, because we have
> almost our eCos project in this platform. ^_^
>
>
>
>
> pyxue
>
>

Re: [ECOS] Re: Little Endian GDB stub for VRC4373

[Jesper Skov]

>>>>> "Ling" == Ling Su <lingsu@palmmicro.com> writes:

Ling> some old ones which I am working on. To my surprise, no one
Ling> works. (I did remember to change the JP66 for the endian issue).

Wow wide is the flash part? If it's 16/32 bits, your problems could be
caused by the prommer using the wrong endianess.

Jesper

Re: [ECOS] Re: Little Endian GDB stub for VRC4373

[Ling Su]

> Ling> some old ones which I am working on. To my surprise, no one
> Ling> works. (I did remember to change the JP66 for the endian issue).
>
> Wow wide is the flash part? If it's 16/32 bits, your problems could be
> caused by the prommer using the wrong endianess.
>
Hi,  Jesper

If the problem is caused by programmer, but why it is correct when I use the
gdb_moudle.bin provided in eCos-1.3.1 release. I doubt if I reproduce one by
using the public available source codes.

Regards,
-Ling

Re: [ECOS] Re: Little Endian GDB stub for VRC4373

[Jesper Skov]

>>>>> "Ling" == Ling Su <lingsu@palmmicro.com> writes:

Ling> If the problem is caused by programmer, but why it is correct
Ling> when I use the gdb_moudle.bin provided in eCos-1.3.1 release. I
Ling> doubt if I reproduce one by using the public available source
Ling> codes.

Then it's not the prommer, obviously.

Jesper