[ECOS] Problems with RedBoot on x86 generic pc...

[ECOS] Problems with RedBoot on x86 generic pc...

[Matthew M. DeLoera]

BTW, I didn't thank some of you for responding to my previous problem
with building for a pc target. With Jifl's CVS update, it now builds.

Anyway, my current problem is with RedBoot. I built the x86 version and
booted with the diskette. It seems to come up, and I see a RedBoot
prompt on my screen. I'm assuming that even though I can see it on the
local console, the local console isn't active, right? So, I have to use
a serial port or network, right?

It's not working with my particular ethernet card, so I'm trying with
the serial port. I've tested my cable with my host machine and another
Linux machine, and they talk successfully with minicom. But, when I
connect the same cable between my host and my target, and boot the
diskette on the target, and attempt to connect with gdb, I can't get a
connection. I'm assuming I'd get the RedBoot prompt, right?

So, I'm stuck at the moment. I'm using the June 8, 2001 build of
RedBoot.

Any suggestions?

- matt

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Jonathan Larmour]

"Matthew M. DeLoera" wrote:
> 
> BTW, I didn't thank some of you for responding to my previous problem
> with building for a pc target. With Jifl's CVS update, it now builds.
> 
> Anyway, my current problem is with RedBoot. I built the x86 version and
> booted with the diskette. It seems to come up, and I see a RedBoot
> prompt on my screen. I'm assuming that even though I can see it on the
> local console, the local console isn't active, right? So, I have to use
> a serial port or network, right?

Yes it should be active. Is it not?
 
> It's not working with my particular ethernet card,

It only supports the i82559 (intel etherexpress pro).

> so I'm trying with
> the serial port. I've tested my cable with my host machine and another
> Linux machine, and they talk successfully with minicom. But, when I
> connect the same cable between my host and my target, and boot the
> diskette on the target, and attempt to connect with gdb, I can't get a
> connection. I'm assuming I'd get the RedBoot prompt, right?

On the console, what happens if you type "+$k#6b" (without the quotes).

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine
Come to the Red Hat TechWorld open source conference in Brussels!
Keynotes, techie talks and exhibitions    http://www.redhat-techworld.com/

RE: [ECOS] Problems with RedBoot on x86 generic pc...

[Trenton D. Adams]

Since mine isn't working either, I figured I'd chime in.  

  > "Matthew M. DeLoera" wrote:
  > >
  > > BTW, I didn't thank some of you for responding to my previous
problem
  > > with building for a pc target. With Jifl's CVS update, it now
builds.
  > >
  > > Anyway, my current problem is with RedBoot. I built the x86
version
  > and
  > > booted with the diskette. It seems to come up, and I see a RedBoot
  > > prompt on my screen. I'm assuming that even though I can see it on
the
  > > local console, the local console isn't active, right? So, I have
to
  > use
  > > a serial port or network, right?
  > 
  > Yes it should be active. Is it not?
  > 

Nope, mine isn't active.

  > 
  > On the console, what happens if you type "+$k#6b" (without the
quotes).
  > 

I assume you mean the local console.  If so, it does nothing for me.


The redboot I'm using is the June 8th binary on the website, redboot.bin

p.s.
I have the same problems as Matthew except I don't have a compatible
network card in the machine.  I have an NE2000, but it's set to PNP mode
at the moment.

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Jonathan Larmour]

"Trenton D. Adams" wrote:
> 
> Since mine isn't working either, I figured I'd chime in.
> 
>   > "Matthew M. DeLoera" wrote:
>   > >
>   > > Anyway, my current problem is with RedBoot. I built the x86
> version
>   > and
>   > > booted with the diskette. It seems to come up, and I see a RedBoot
>   > > prompt on my screen. I'm assuming that even though I can see it on
> the
>   > > local console, the local console isn't active, right? So, I have
> to
>   > use
>   > > a serial port or network, right?
>   >
>   > Yes it should be active. Is it not?
>   >
> 
> Nope, mine isn't active.
> 
>   >
>   > On the console, what happens if you type "+$k#6b" (without the
> quotes).
>   >
> 
> I assume you mean the local console.  If so, it does nothing for me.
> 
> The redboot I'm using is the June 8th binary on the website, redboot.bin

Although I tested this binary before, I just retested it just to be sure
and it Works For Me(tm).

One possibility is that there is noise on the serial line. RedBoot listens
to all input channels, and as soon as it gets something on one, it uses
that from then on. So if it got a glitch on the serial line, it might
switch to that and stay there. How many serial ports does the machine have?
 
> p.s.
> I have the same problems as Matthew except I don't have a compatible
> network card in the machine.  I have an NE2000, but it's set to PNP mode
> at the moment.

If an i82559 card isn't detected it should still continue.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

RE: [ECOS] Problems with RedBoot on x86 generic pc...

[Trenton D. Adams]

  > > The redboot I'm using is the June 8th binary on the website,
  > redboot.bin
  > 
  > Although I tested this binary before, I just retested it just to be
sure
  > and it Works For Me(tm).
  > 

Me(tm) LOL?  Like windows me?

  > One possibility is that there is noise on the serial line. RedBoot
  > listens
  > to all input channels, and as soon as it gets something on one, it
uses
  > that from then on. So if it got a glitch on the serial line, it
might
  > switch to that and stay there. How many serial ports does the
machine
  > have?

It doesn't work on the serial line either.  I don't see how it could
have noise on the serial line when nothing is connected to it.  I tried
it with and without the serial line plugged in.  Also, I tried it on
three separate and completely different computers.  In particular, a
Dell, an epox motherboard, and another motherboard that I've forgotten
the brand name of.  One computer has a single serial port, the other one
has only one enabled, and the one I tried today has a serial line
connected to it that is connected to my EDB7xxx board while the board is
off.

One other thing I forgot to mention before is that I do get the redboot
prompt if I connect to the serial line from the host machine before
booting the RedBoot server.  Although I do get the prompt, it won't
respond to anything.

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Jonathan Larmour]

"Trenton D. Adams" wrote:
> 
>   > > The redboot I'm using is the June 8th binary on the website,
>   > redboot.bin
>   >
>   > Although I tested this binary before, I just retested it just to be
> sure
>   > and it Works For Me(tm).
>   >
> 
> Me(tm) LOL?  Like windows me?

"Works For Me" is a common idiom for this type of situation :).
 
> One other thing I forgot to mention before is that I do get the redboot
> prompt if I connect to the serial line from the host machine before
> booting the RedBoot server.  Although I do get the prompt, it won't
> respond to anything.

I think you'll have to see if you can debug it then. Try compiling your own
redboot without any of the net support for a start, to see if that helps.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

RE: [ECOS] Problems with RedBoot on x86 generic pc...

[Trenton D. Adams]

  > >
  > > Me(tm) LOL?  Like windows me?
  > 
  > "Works For Me" is a common idiom for this type of situation :).
  > 

Oh, I think I get it! ;)

  > > One other thing I forgot to mention before is that I do get the
  > redboot
  > > prompt if I connect to the serial line from the host machine
before
  > > booting the RedBoot server.  Although I do get the prompt, it
won't
  > > respond to anything.
  > 
  > I think you'll have to see if you can debug it then. Try compiling
your
  > own
  > redboot without any of the net support for a start, to see if that
  > helps.
  > 

Ok, thanks.  I'll do that.

I have to finish setting up Windows 2000 on my other machine at home
first, and then I'll try it.  The config tool crashes every time in
windows 98.  I suppose I could setup the linux tools on my server, but
since I haven't used them yet, I'm not sure if it's worth my time.  If I
had a linux machine from the start of the project it would have been
better.

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Jonathan Larmour]

"Trenton D. Adams" wrote:
> 
> I have to finish setting up Windows 2000 on my other machine at home
> first, and then I'll try it.  The config tool crashes every time in
> windows 98.  I suppose I could setup the linux tools on my server, but
> since I haven't used them yet, I'm not sure if it's worth my time.  If I
> had a linux machine from the start of the project it would have been
> better.

The config tool v2 is also available for Windows, not just Linux.
http://sources.redhat.com/ecos/ct2.html

It should work with win 98.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Adam Agnew]

Hi, I'm using the same image, getting the same results. Did anyone ever
have a solution to this problem?

On Tue, 31 Jul 2001, Jonathan Larmour wrote:

> "Trenton D. Adams" wrote:
> >
> > Since mine isn't working either, I figured I'd chime in.
> >
> >   > "Matthew M. DeLoera" wrote:
> >   > >
> >   > > Anyway, my current problem is with RedBoot. I built the x86
> > version
> >   > and
> >   > > booted with the diskette. It seems to come up, and I see a RedBoot
> >   > > prompt on my screen. I'm assuming that even though I can see it on
> > the
> >   > > local console, the local console isn't active, right? So, I have
> > to
> >   > use
> >   > > a serial port or network, right?
> >   >
> >   > Yes it should be active. Is it not?
> >   >
> >
> > Nope, mine isn't active.
> >
> >   >
> >   > On the console, what happens if you type "+$k#6b" (without the
> > quotes).
> >   >
> >
> > I assume you mean the local console.  If so, it does nothing for me.
> >
> > The redboot I'm using is the June 8th binary on the website, redboot.bin
>
> Although I tested this binary before, I just retested it just to be sure
> and it Works For Me(tm).
>
> One possibility is that there is noise on the serial line. RedBoot listens
> to all input channels, and as soon as it gets something on one, it uses
> that from then on. So if it got a glitch on the serial line, it might
> switch to that and stay there. How many serial ports does the machine have?
>
> > p.s.
> > I have the same problems as Matthew except I don't have a compatible
> > network card in the machine.  I have an NE2000, but it's set to PNP mode
> > at the moment.
>
> If an i82559 card isn't detected it should still continue.
>
> Jifl
> --
> Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
> Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine
>

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Adam Agnew]

I've tried the provided image for a floppy on several x86 boxen, and it
works on some, doesn't on others. As a test, i tried setting

cdl_option CYGNUM_HAL_VIRTUAL_VECTOR_COMM_CHANNELS { user_value 1 };
and
cdl_option CYGSEM_HAL_I386_PC_DIAG_SCREEN { user_value 0 };

That certainly cleared up the problem, but of course gets rid of the PC's
screen, which was nice. Maybe someone can use this info to track down the
bug, I'm not that familiar with the code base yet. But that's a temporary
fix for poor folks like me who it wasn't working for. Perhaps this is
just a problem with motherboards with only one serial port. Perhaps its
deeper, *shrug*.

- adam agnew

On Fri, 7 Sep 2001, Adam Agnew wrote:

>
> Hi, I'm using the same image, getting the same results. Did anyone ever
> have a solution to this problem?
>
> On Tue, 31 Jul 2001, Jonathan Larmour wrote:
>
> > "Trenton D. Adams" wrote:
> > >
> > > Since mine isn't working either, I figured I'd chime in.
> > >
> > >   > "Matthew M. DeLoera" wrote:
> > >   > >
> > >   > > Anyway, my current problem is with RedBoot. I built the x86
> > > version
> > >   > and
> > >   > > booted with the diskette. It seems to come up, and I see a RedBoot
> > >   > > prompt on my screen. I'm assuming that even though I can see it on
> > > the
> > >   > > local console, the local console isn't active, right? So, I have
> > > to
> > >   > use
> > >   > > a serial port or network, right?
> > >   >
> > >   > Yes it should be active. Is it not?
> > >   >
> > >
> > > Nope, mine isn't active.
> > >
> > >   >
> > >   > On the console, what happens if you type "+$k#6b" (without the
> > > quotes).
> > >   >
> > >
> > > I assume you mean the local console.  If so, it does nothing for me.
> > >
> > > The redboot I'm using is the June 8th binary on the website, redboot.bin
> >
> > Although I tested this binary before, I just retested it just to be sure
> > and it Works For Me(tm).
> >
> > One possibility is that there is noise on the serial line. RedBoot listens
> > to all input channels, and as soon as it gets something on one, it uses
> > that from then on. So if it got a glitch on the serial line, it might
> > switch to that and stay there. How many serial ports does the machine have?
> >
> > > p.s.
> > > I have the same problems as Matthew except I don't have a compatible
> > > network card in the machine.  I have an NE2000, but it's set to PNP mode
> > > at the moment.
> >
> > If an i82559 card isn't detected it should still continue.
> >
> > Jifl
> > --
> > Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
> > Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine
> >
>
>
>

[ECOS] RedBoot IDE access on x86 without BIOS

[Adam Agnew]

Greetings,
Right now I have RedBoot running from a ram based elf image from the
rom with the use of LinuxBIOS (www.linuxbios.org) . It's pretty swell.
But, it's not detecting the presence of any of the hard drives. I looked
at the code, and it doesn't seem to be using any BIOS int calls. Anyone
have any idea as to what could be wrong? I don't think it's a matter of
the hard drives not having time to spin up, because a reboot doesn't make
it suddenly work either. Everything looks OK to me, except perhaps these
defines in ecos/packages/hal/i386/pcmb/current/include/pcmb_io.h :

#define __PCMB_IDE_PRI_CMD   0x1f0
#define __PCMB_IDE_PRI_CTL   0x3f4
#define __PCMB_IDE_SEC_CMD   0x170
#define __PCMB_IDE_SEC_CTL   0x374

Shouldn't __PCMB_IDE_PRI_CTL be 0x3f6 and __PCMB_IDE_SEC_CTL 0x376? And if
not, could you explain why please?

Thank you for your time.

Sincerely, Adam Agnew

Re: [ECOS] RedBoot IDE access on x86 without BIOS

[Mark Salter]

>>>>> Adam Agnew writes:

> Greetings,
> Right now I have RedBoot running from a ram based elf image from the
> rom with the use of LinuxBIOS (www.linuxbios.org) . It's pretty swell.
> But, it's not detecting the presence of any of the hard drives. I looked
> at the code, and it doesn't seem to be using any BIOS int calls. Anyone
> have any idea as to what could be wrong? I don't think it's a matter of
> the hard drives not having time to spin up, because a reboot doesn't make
> it suddenly work either. Everything looks OK to me, except perhaps these
> defines in ecos/packages/hal/i386/pcmb/current/include/pcmb_io.h :

> #define __PCMB_IDE_PRI_CMD   0x1f0
> #define __PCMB_IDE_PRI_CTL   0x3f4
> #define __PCMB_IDE_SEC_CMD   0x170
> #define __PCMB_IDE_SEC_CTL   0x374

> Shouldn't __PCMB_IDE_PRI_CTL be 0x3f6 and __PCMB_IDE_SEC_CTL 0x376? And if
> not, could you explain why please?

I think you're right, but the odd thing is it sees the drives on my pc.
The above defines are correct for the base of the control registers.
Its the macros for HAL_IDE_READ_ALTSTATUS and HAL_IDE_WRITE_CONTROL
which need to add 2 to the base to get the status/ctl regs.

--Mark

Re: [ECOS] RedBoot IDE access on x86 without BIOS

[Adam Agnew]

On Thu, 13 Sep 2001, Mark Salter wrote:

> I think you're right, but the odd thing is it sees the drives on my pc.
> The above defines are correct for the base of the control registers.
> Its the macros for HAL_IDE_READ_ALTSTATUS and HAL_IDE_WRITE_CONTROL
> which need to add 2 to the base to get the status/ctl regs.
>

Alright, I tried that change. However, it didn't fix my ram from rom
boot's problem. Any other ideas? Thanks!

Adam

Re: [ECOS] RedBoot IDE access on x86 without BIOS

[Mark Salter]

>>>>> Adam Agnew writes:

> On Thu, 13 Sep 2001, Mark Salter wrote:

>> I think you're right, but the odd thing is it sees the drives on my pc.
>> The above defines are correct for the base of the control registers.
>> Its the macros for HAL_IDE_READ_ALTSTATUS and HAL_IDE_WRITE_CONTROL
>> which need to add 2 to the base to get the status/ctl regs.
>> 

> Alright, I tried that change. However, it didn't fix my ram from rom
> boot's problem. Any other ideas? Thanks!

The IDE code in RedBoot assumes that the IDE controller/interface is
initialized. If you aren't using the normal PC bios, perhaps the
chipset has not been sufficiently initialized to support IDE.

--Mark

Re: [ECOS] RedBoot IDE access on x86 without BIOS

[Adam Agnew]

On Thu, 13 Sep 2001, Mark Salter wrote:

> The IDE code in RedBoot assumes that the IDE controller/interface is
> initialized. If you aren't using the normal PC bios, perhaps the
> chipset has not been sufficiently initialized to support IDE.

Hrmm.. Interesting, yes, you're probably right! In fact, I found the patch
that makes Linux force the IDE controllers on in the LinuxBIOS -> kernel
configuration. I forgot about that and assumed RedBoot would just work
because the Linux kernel just worked. Alright, I'll get something in and
working real soon, and if there's a good generic way to do so on all x86s,
maybe it'd be worth it to fill in that HAL_IDE_INIT macro.

Thanks again, you probably just saved me days :)

- Adam

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Jonathan Larmour]

Adam Agnew wrote:
> 
> I've tried the provided image for a floppy on several x86 boxen, and it
> works on some, doesn't on others. As a test, i tried setting
> 
> cdl_option CYGNUM_HAL_VIRTUAL_VECTOR_COMM_CHANNELS { user_value 1 };
> and
> cdl_option CYGSEM_HAL_I386_PC_DIAG_SCREEN { user_value 0 };
> 
> That certainly cleared up the problem, but of course gets rid of the PC's
> screen, which was nice. Maybe someone can use this info to track down the
> bug, I'm not that familiar with the code base yet. But that's a temporary
> fix for poor folks like me who it wasn't working for. Perhaps this is
> just a problem with motherboards with only one serial port. Perhaps its
> deeper, *shrug*.

One way to find out if there's a problem with boards with one serial
port.... go to hal/i386/pcmb/current/src/pcmb_serial.c and comment out the
stuff for channel 1 in cyg_hal_plf_serial_init.

Then go to pcmb_screen.c and in cyg_hal_plf_screen_init change all the "2"s
to "1"s.

Go to /hal_diag.c and remove the middle entry of pc_ser_channels and change
the 3 to a 2.

Finally go to hal/i386/pc/current/cdl_hal_i386_pc.cdl and change
CYGNUM_HAL_VIRTUAL_VECTOR_COMM_CHANNELS to have:

       default_value   { CYGSEM_HAL_I386_PC_DIAG_SCREEN ? 2 : 1 }

Then build and try a redboot on a PC with one serial port. If it works
where the old image didn't, let us know.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Adam Agnew]

I finally found time to try that, and yes it worked on machines where it
wasn't working when there were two serial ports specified. Thanks, and I
hope this info is helpfull in tracking the bug down.
- adam agnewx

On Thu, 13 Sep 2001, Jonathan Larmour wrote:

> Adam Agnew wrote:
> >
> > I've tried the provided image for a floppy on several x86 boxen, and it
> > works on some, doesn't on others. As a test, i tried setting
> >
> > cdl_option CYGNUM_HAL_VIRTUAL_VECTOR_COMM_CHANNELS { user_value 1 };
> > and
> > cdl_option CYGSEM_HAL_I386_PC_DIAG_SCREEN { user_value 0 };
> >
> > That certainly cleared up the problem, but of course gets rid of the PC's
> > screen, which was nice. Maybe someone can use this info to track down the
> > bug, I'm not that familiar with the code base yet. But that's a temporary
> > fix for poor folks like me who it wasn't working for. Perhaps this is
> > just a problem with motherboards with only one serial port. Perhaps its
> > deeper, *shrug*.
>
> One way to find out if there's a problem with boards with one serial
> port.... go to hal/i386/pcmb/current/src/pcmb_serial.c and comment out the
> stuff for channel 1 in cyg_hal_plf_serial_init.
>
> Then go to pcmb_screen.c and in cyg_hal_plf_screen_init change all the "2"s
> to "1"s.
>
> Go to /hal_diag.c and remove the middle entry of pc_ser_channels and change
> the 3 to a 2.
>
> Finally go to hal/i386/pc/current/cdl_hal_i386_pc.cdl and change
> CYGNUM_HAL_VIRTUAL_VECTOR_COMM_CHANNELS to have:
>
>        default_value   { CYGSEM_HAL_I386_PC_DIAG_SCREEN ? 2 : 1 }
>
> Then build and try a redboot on a PC with one serial port. If it works
> where the old image didn't, let us know.
>
> Jifl
> --
> Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
> Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine
>

Re: [ECOS] Problems with RedBoot on x86 generic pc...

[Matthew DeLoera]

To all concerned....

I thought I'd post a note to say that I haven't had a chance to work
with ecos any further in quite a while, because I've been swamped with
other projects.

I'm looking forward to getting back to this project, but for now it's on
hiatus...

Sorry...
- Matt