Re: [coLinux-devel] good news....

[Paolo M. <pao...@gm...>]

Hi Henry,
I have tried to use your devel-64bit packages on my 32bit linux system.
All compiling process is OK.
To understand 64bit internal (IDT and memory management) I'd like to
test a linux native 64bit on a qemu-64bit.
But I need gcc 64bit.

My gcc produce 32bit code. Is there some option ?
I have see that your 64bit package compile linux kernel. But it uses
gcc. And my gcc is 32bit.

Thanks,
Paolo





On Wed, Jun 23, 2010 at 9:52 PM, Henry Nestler <hen...@ar...> wrote:
>  Hello Paolo,
>
> the Windows driver for 64 bit works. I have it. I wrote a simple drive
> example for 64 bit. So, the tools are complete. Yes, ReactOS have the basic
> headers. It was a little puzzle, and I have bug-fixed some small source
> there.
>
> Interestings you can read on my home page:
> http://www.henrynestler.com/colinux/tools/mingw-w64/
>
> If you would use Qemu, you can emulate a 64 bit CPU. If curse boot Windows
> 2008 64 bit would need round about 10 minutes. But, it works. Qemu was my
> first experimence with 64 bit, on my old Pentium 800 MHz. There I have build
> the TAP 64 bit driver for openVPN.
>
> Hard programming all the 64 bit stuff would be one option. An other would be
> to use hardware virtualisation VT-x/AMD-V. VirtualBox use it. The goddy
> would be, that the OS-switch would be very fast. The badly is, I can not
> test it under VirtualBox. So I would need Qemu again for testing coLinux in
> a saver environment.
>
> All what everybody with and without 64 bit CPU can do, is to fix the tons or
> warnings, where "pointers" are casted to "long" and sometimes such "long" is
> casted back to "pointer". This is a warning for compiler, but in many cases
> it is a bug for 64 bit systems. (pointer is 64 bit and long is 32 bit under
> Windows!) One of the typically problems are the macros for queue handling
> (co_queue_get_tail, co_queue_get_prev, ...)
>
> Other problems are not supported functions in DDK for 64 bit. For example
> RtlDosPathNameToNtPathName_U. An alternate function or a code snip to
> replace this function would be helpfully. It's in coLinux the function
> "co_canonize_cobd_path" that needs to adjust. This you can also do under 32
> bit. - This and the queue macros are good parts for you. ;-)
>
> I know the task is complex. If everybody does a small part, it is doable.
> The very big step was to have a basic driver working, and good working set
> of tools. Currently, I have not checked about 64 bit memory management. I
> will do it as next step. So, of curse I will give my try on the harddest
> part. For this I need some time to locate all the missing informations and
> bring it into my little brain.
>
> --
> Henry
>
>
> On 23.06.2010 12:25, Paolo Minazzi wrote:
>>
>> Henry,
>> The task is complex.
>>
>> Surely you already know it .... there is a good book about linux kernel
>> ia64.
>> It explain a little about low level programing ia64.
>> I have already thought to buy it.
>>
>> About win64 driver with mingw64 ... I have read that ReactOs people
>> can generate them. I don't know if you are already able to do it.
>>
>> For now I have no 64bit computer.
>> I have never used linux64bit native.
>>
>> Bye,
>> Paolo
>>
>>
>>
>> On Tue, Jun 22, 2010 at 9:41 PM, Henry Nestler<hen...@ar...>  wrote:
>>>
>>>  On 21.06.2010 08:51, Paolo Minazzi wrote:
>>>>
>>>> ....about 64bit porting....
>>>
>>> Yes, and I hope somebody would help now.
>>>
>>> I totally have no idea for the passage page. My plan is to switch from 64
>>> bit mode into 32 bit mode and use the old passsage code. In this case we
>>> can
>>> use the current Linux kernel code. In the next step we can create a
>>> passage
>>> page for 64 bit to 64 bit later.
>>>
>>> --
>>> Henry N.
>>>
>>>
>