RE: [ECOS] Redboot image load via PCI bus?

RE: [ECOS] Redboot image load via PCI bus?

[Snider, Marc]

Makes sense, Gary.  Thanks for the pointers...  Assuming I'm otherwise
entirely comfortable with what version 1.3.1 provides, will I need to go to
(do you advise) version 2.0 before making these mods or is 1.3.1 adequate?  


Thanks again,
Marc


-----Original Message-----
From: Gary Thomas [mailto:gary@mlbassoc.com]
Sent: Thursday, November 06, 2003 12:42 PM
To: Snider, Marc
Cc: 'ecos-discuss@sources.redhat.com'
Subject: RE: [ECOS] Redboot image load via PCI bus?


On Thu, 2003-11-06 at 10:37, Snider, Marc wrote:
> My thought was to provide the standard I/O device driver routines (open,
> close, read...) on top of the PCI accesses and thus shield the higher
layer
> loader code.  I'd make it look like the image data was coming from
somewhere
> else like the flash.  It looks like do_load doesn't care where the image
> data comes from as long as the native format (in my case likely SREC) is
> maintained...
> 
> Yes, the host does have access to the memory on the target system, but it
> would be preferable to pull the data from the host as opposed to pushing
it
> from there to the target...
> 

Understood.  If you look carefully at how the "load" command works,
you won't need to make *any* changes directly to it, but just add a
new I/O method.  Also, I'd suggest that you put your new PCI method
in your platform HAL - there is no need to mess about with the core
of RedBoot.

note: a lot of care has been taken (by me) to allow such extensions
without any [direct] changes to the RedBoot tree.  

... Ah, the joy of tables.

> Marc
> 
> 
> -----Original Message-----
> From: Gary Thomas [mailto:gary@mlbassoc.com]
> Sent: Thursday, November 06, 2003 11:46 AM
> To: Snider, Marc
> Cc: 'ecos-discuss@sources.redhat.com'
> Subject: Re: [ECOS] Redboot image load via PCI bus?
> 
> 
> On Thu, 2003-11-06 at 09:45, Snider, Marc wrote:
> > Is there a canned method implemented somewhere for loading an image via
> the
> > PCI bus (from another host), as opposed to using the network or loading
> from
> > flash?   I'd like to load an image from a control host across the PCI
bus
> > and then burn it to flash...   I'm assuming that if there isn't then
I'll
> > need to modify the do_load command and provide a set of I/O routines
that
> > map PCI accesses into standard I/O convention...
> > 
> > I'm on a PowerPC440 platform, if the platform matters.
> 
> What method would you use to access the image via the PCI?
> Can your host access the memory on the board?
> 
> -- 
> Gary Thomas <gary@mlbassoc.com>
> MLB Associates
-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss

RE: [ECOS] Redboot image load via PCI bus?

[Gary Thomas]

On Thu, 2003-11-06 at 11:13, Snider, Marc wrote:
> Makes sense, Gary.  Thanks for the pointers...  Assuming I'm otherwise
> entirely comfortable with what version 1.3.1 provides, will I need to go to
> (do you advise) version 2.0 before making these mods or is 1.3.1 adequate?  
> 

1.3.1 is too out of date - don't even think about it.

Use the latest from CVS - that's the best approach.

> 
> Thanks again,
> Marc
> 
> 
> -----Original Message-----
> From: Gary Thomas [mailto:gary@mlbassoc.com]
> Sent: Thursday, November 06, 2003 12:42 PM
> To: Snider, Marc
> Cc: 'ecos-discuss@sources.redhat.com'
> Subject: RE: [ECOS] Redboot image load via PCI bus?
> 
> 
> On Thu, 2003-11-06 at 10:37, Snider, Marc wrote:
> > My thought was to provide the standard I/O device driver routines (open,
> > close, read...) on top of the PCI accesses and thus shield the higher
> layer
> > loader code.  I'd make it look like the image data was coming from
> somewhere
> > else like the flash.  It looks like do_load doesn't care where the image
> > data comes from as long as the native format (in my case likely SREC) is
> > maintained...
> > 
> > Yes, the host does have access to the memory on the target system, but it
> > would be preferable to pull the data from the host as opposed to pushing
> it
> > from there to the target...
> > 
> 
> Understood.  If you look carefully at how the "load" command works,
> you won't need to make *any* changes directly to it, but just add a
> new I/O method.  Also, I'd suggest that you put your new PCI method
> in your platform HAL - there is no need to mess about with the core
> of RedBoot.
> 
> note: a lot of care has been taken (by me) to allow such extensions
> without any [direct] changes to the RedBoot tree.  
> 
> ... Ah, the joy of tables.
> 
> > Marc
> > 
> > 
> > -----Original Message-----
> > From: Gary Thomas [mailto:gary@mlbassoc.com]
> > Sent: Thursday, November 06, 2003 11:46 AM
> > To: Snider, Marc
> > Cc: 'ecos-discuss@sources.redhat.com'
> > Subject: Re: [ECOS] Redboot image load via PCI bus?
> > 
> > 
> > On Thu, 2003-11-06 at 09:45, Snider, Marc wrote:
> > > Is there a canned method implemented somewhere for loading an image via
> > the
> > > PCI bus (from another host), as opposed to using the network or loading
> > from
> > > flash?   I'd like to load an image from a control host across the PCI
> bus
> > > and then burn it to flash...   I'm assuming that if there isn't then
> I'll
> > > need to modify the do_load command and provide a set of I/O routines
> that
> > > map PCI accesses into standard I/O convention...
> > > 
> > > I'm on a PowerPC440 platform, if the platform matters.
> > 
> > What method would you use to access the image via the PCI?
> > Can your host access the memory on the board?
> > 
> > -- 
> > Gary Thomas <gary@mlbassoc.com>
> > MLB Associates
-- 
Gary Thomas <gary@mlbassoc.com>
MLB Associates


-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss

[ECOS] ECOS development model

[Adrian Caceres]

I was a tad surprised by the answer below regarding what version of ECOS 
should an
app developer use so I was wondering what did I miss:

- What is open source community development model for  ECOS?

Many open source projects have some form of a stable release model
where some form of testing has been performed and it is somewhat 
"blessed" by
the maintainers.

I thought/assumed that ECOS had a similar model and that 2.0 was a stable
version.  And that it would be expected for an ECOS user to base a new
project off such stable version.  And that going forwards there
would be other stable releases such as 2.1 or 3.0, etc.

So I was a tad surprised that Gary recommended the use of the CVS tree
directly .  I would have expected that the CVS tree was really the ECOS
development source tree, that it would suffer from the instability such
a source would have, and thus would not be a recommended choice
for someone developing an app for it.

Did I miss something?  Was the context of the thread below regarding a
feature that is going into the ECOS source tree and I missed it?

thanks

adrian

Gary Thomas wrote:

>On Thu, 2003-11-06 at 11:13, Snider, Marc wrote:
>  
>
>>Makes sense, Gary.  Thanks for the pointers...  Assuming I'm otherwise
>>entirely comfortable with what version 1.3.1 provides, will I need to go to
>>(do you advise) version 2.0 before making these mods or is 1.3.1 adequate?  
>>
>>    
>>
>
>1.3.1 is too out of date - don't even think about it.
>
>Use the latest from CVS - that's the best approach.
>
>  
>
>>Thanks again,
>>Marc
>>
>>
>>-----Original Message-----
>>From: Gary Thomas [mailto:gary@mlbassoc.com]
>>Sent: Thursday, November 06, 2003 12:42 PM
>>To: Snider, Marc
>>Cc: 'ecos-discuss@sources.redhat.com'
>>Subject: RE: [ECOS] Redboot image load via PCI bus?
>>
>>
>>On Thu, 2003-11-06 at 10:37, Snider, Marc wrote:
>>    
>>
>>>My thought was to provide the standard I/O device driver routines (open,
>>>close, read...) on top of the PCI accesses and thus shield the higher
>>>      
>>>
>>layer
>>    
>>
>>>loader code.  I'd make it look like the image data was coming from
>>>      
>>>
>>somewhere
>>    
>>
>>>else like the flash.  It looks like do_load doesn't care where the image
>>>data comes from as long as the native format (in my case likely SREC) is
>>>maintained...
>>>
>>>Yes, the host does have access to the memory on the target system, but it
>>>would be preferable to pull the data from the host as opposed to pushing
>>>      
>>>
>>it
>>    
>>
>>>from there to the target...
>>>
>>>      
>>>
>>Understood.  If you look carefully at how the "load" command works,
>>you won't need to make *any* changes directly to it, but just add a
>>new I/O method.  Also, I'd suggest that you put your new PCI method
>>in your platform HAL - there is no need to mess about with the core
>>of RedBoot.
>>
>>note: a lot of care has been taken (by me) to allow such extensions
>>without any [direct] changes to the RedBoot tree.  
>>
>>... Ah, the joy of tables.
>>
>>    
>>
>>>Marc
>>>
>>>
>>>-----Original Message-----
>>>From: Gary Thomas [mailto:gary@mlbassoc.com]
>>>Sent: Thursday, November 06, 2003 11:46 AM
>>>To: Snider, Marc
>>>Cc: 'ecos-discuss@sources.redhat.com'
>>>Subject: Re: [ECOS] Redboot image load via PCI bus?
>>>
>>>
>>>On Thu, 2003-11-06 at 09:45, Snider, Marc wrote:
>>>      
>>>
>>>>Is there a canned method implemented somewhere for loading an image via
>>>>        
>>>>
>>>the
>>>      
>>>
>>>>PCI bus (from another host), as opposed to using the network or loading
>>>>        
>>>>
>>>from
>>>      
>>>
>>>>flash?   I'd like to load an image from a control host across the PCI
>>>>        
>>>>
>>bus
>>    
>>
>>>>and then burn it to flash...   I'm assuming that if there isn't then
>>>>        
>>>>
>>I'll
>>    
>>
>>>>need to modify the do_load command and provide a set of I/O routines
>>>>        
>>>>
>>that
>>    
>>
>>>>map PCI accesses into standard I/O convention...
>>>>
>>>>I'm on a PowerPC440 platform, if the platform matters.
>>>>        
>>>>
>>>What method would you use to access the image via the PCI?
>>>Can your host access the memory on the board?
>>>
>>>-- 
>>>Gary Thomas <gary@mlbassoc.com>
>>>MLB Associates
>>>      
>>>


-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss

Re: [ECOS] ECOS development model

[Andrew Lunn]

On Mon, Nov 10, 2003 at 05:19:03PM -0800, Adrian Caceres wrote:
> I was a tad surprised by the answer below regarding what version of ECOS 
> should an
> app developer use so I was wondering what did I miss:
> 
> - What is open source community development model for  ECOS?

This thread kind of answers your question. 

http://sources.redhat.com/ml/ecos-discuss/2003-09/msg00388.html

        Andrew

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss