/DEBUGPORT=SCREEN works, requires /MININT on liveCD

[Mna.]

Hi,

Only nearly a month ago, (now I can't remember those build numbers),
booting livecd with /DEBUGPORT=SCREEN works, I mean ReactOS successfully loads and runs,
at least on real hardware.

But now it doesn't work, not on real hardware, not in VMware, nowhere!

Now, ROS starts OK when booting with /DEBUGPORT=COM1
but doesn't start with /DEBUGPORT=SCREEN !

//Could someone retest this?
===
UPD.: Changed the subject line: It was working. But LiveCD always requires /MININT key as kernel parameter

[vicmarcal]

This is a nice place to ask for retesting, but you should report to our Bugzilla too, as our Official Testers ad Devs usually looks Bugzilla more than Forum.
Also try to find (via Vmware) in which revision it stopped working. Use a Binary search to find it:
Binary search

[Mna.]

I know what is binary search.)

But when tried to find latest working version with VMware it turns out I can't find such simple one.
Will try later with real hardware since 1. I remember that number for R.H. and 2. frankly hardware is much more important for me

PS Just wondering (rhetorically) do ROS people live in virtual world to use only virtual machines?...

[EmuandCo]

Real hw is completely different all the time and vms are not. This is the reason of using a vm.

[Mna.]

I see... ReactOS is way too weak to swim in the ocean of real hardware, and that is why it is only swimming in indoor swimming pool with kickboard and under coach supervision

[vicmarcal]

I see... ReactOS is way too weak to swim in the ocean of real hardware, and that is why it is only swimming in indoor swimming pool with kickboard and under coach supervision

It is called development.
As you know how "development os" is done, i won't give you a link

[Mna.]

> As you know how "development os" is done, i won't give you a link
Did you mean something like "Revolution OS" movie?

In fact I'd read some part of "linux device drivers" book, particularly how pressing special buttons on stalled machine brings some info about crush - this doesn't require serial cable
I think this functionality would be very useful for ROS, now, but at least I need working SCREEN debugport, to be sure the similar thing would work with buttons.

My this machine hasn't serial port, and I can't add it, so I can't debug "the main ROS crush"

[hto]

Mna., try to find the bug yourself. ;)

[Mna.]

Mna., try to find the bug yourself.

That's much easier for me as I'm such kind of person.)

Well,
I realized I need some debugging tips help.

The problem is that ROS doesn't react on tab+k so I can't stop it before crush occurs. And trace further.

How this key could be enabled or the problem circumvented other way?

[Haos]

@Mna

ReactOS can swim on real hardware just fine, but very often this task is beyond scope of usual tester want-to-be, especially if that one had no previous experience with testing ReactOS. Using VM allows a nice transition period to catch up with quite steep learning curve. If one wishes to do something new and go for hardware testing, we are all happy about it.

On the other hand, real hardware testing requires additional equipment as well as more work to get it done.

We are always here or on irc to help out anyone determined to proceed with testing our system.

[hto]

The problem is that ROS doesn't react on tab+k so I can't stop it before crush occurs. And trace further.

How this key could be enabled or the problem circumvented other way?

It looks like this known bug. Maybe you can help to fix it?

Can you reproduce these bugs in QEMU? It allows to attach a debugger (GDB) to it.

[Mna.]

ReactOS can swim on real hardware just fine,

I disagree, reactos is fine only for virtual machines.
Since it's tested and run only there. these are closely linked concerns: it works where it tested, and it is tested on where it works.

but very often this task is beyond scope of usual tester want-to-be, especially if that one had no previous experience with testing ReactOS.

What could I say, I'm not going to be only tester and I had previous experience working, I mean fixing bugs of, ROS nearly in 2000,
but now I see it's on the same place, and even more far from making it run, on Real hardware, just run, possibly with bugs.

Using VM allows a nice transition period to catch up with quite steep learning curve.

I see. that's ok. but this is just intermediate aim. The main aim is to run ROS on real hardware.

If one wishes to do something new and go for hardware testing, we are all happy about it.

I'm for ROS on hardware

On the other hand, real hardware testing requires additional equipment as well as more work to get it done.

If you mean null-modem cable and second machine for it, I have it, and that all can't help me to test any modern notebook, they just has no any serial port.
So we need new ways to get logger work, other types of "hardware logging".

Here I outlined/compared cabling options of WinDBG and ROS KDB debuggers:
http://www.reactos.org/forum/viewtopic.php?f=2&t=8795
The only one ROS supported is yet serial cable.

We are always here or on irc to help out anyone determined to proceed with testing our system.

Here in forum I get guaranteed answers, I mean answers are coming. Not the option in IRC,
On the other hand when IRC people help, sometimes it is very fast help.
So I appreciate both options

[Mna.]

Reading the debugger (kdbg) source code I've found SCREEN logger debugging possibility,
and described it on wiki.
here: http://www.reactos.org/wiki/Debugging#A ... bug_logger
Wow!

[Z98]

To suggest that ROS is only tested in virtual machines is rather disingenuous to the several testers that consistently run regression test on real hardware. As Haos is one of them, I'd say he knows what he's talking about when he says ROS can work on real hardware.

[Mna.]

To suggest that ROS is only tested in virtual machines

Z98, please don't try to embroil us, nobody said about "only tested". My point was "mainly"
And I'm speaking about what I see: ROS works in VM almost always, in the same time rare real machine could get such luck.
And the same configuration of hardware could occasionally break to work, or even boot up. (as with the bug of this topic)

As Haos is one of them, I'd say he knows what he's talking about when he says ROS can work on real hardware.

That's good, but I haven't ever said it can't. Even at times when it really couldn't
In year 2000 it requires to deeply debug the code, find bugs, correct them, and voila, you get the Blue screen

My current "ROS workability index" is almost 2/4, http://www.reactos.org/forum/viewtopic. ... 780#p76828
i.e. 2 machines of my 4 boot up almost OK with ReactOS, so I'm *pretty sure* it works.
(almost Ok because of HDD problems, one - ATA and second - SATA)

What about we are speaking about here, at least I tend to, that is about RH debugging difficulties.
And these difficulties have such structure, it is very problematic to debug real hardware that has no serial port.
that is why I need SCREEN debugport, and now, SCREEN debugport is broken.

Any suggestions?

***
I have even found a guy/company that had produced (only one on the planet?!) the SDIO<->Serial port cable adapter,
http://mdatabg.com/index.php?page=SDRS&lang=en
it could make possible to emulate serial port through SDcard slot (What adapter mainly does, it converts synchronous serial signal(SDIO) to asynchronous(RS232 standard)). And it would be very useful, as would allow debug almost every laptop, SD card is very popular on them - a radical solution! well, that way would require some "real programming" too... to deal with SDcard pci interface. anyway...
But, unfortunately I didn't get any answer from MData Systems of mdatabg.com
***
So well main topic here let will be SCREEN debugport