[RFC] Re: BestComm and fec

[RFC] Re: BestComm and fec

[Andrey Volkov]

Hi all,

> Andrey: Gabor has a working fec.c implementation.

> Apparently, you've both worked out a version of bestcomm for 2.6. But
> you both based it on old code. (even non GPL).

Terrible, may be its time to create special project for MPC5200 (if it
doesn't exist in somewhere in I-net)? First intention - sourceforge.
Comments/suggestions?

Sylvain, Wolfgang I know you already have projects. But in case of
Sylvain, as I understand, his project on his homepage.
In case of Wolfgang, I'm not sure that DENX will happy to share their
internal CVS with all I-net for write access. Am I right?

> I've just tried just taking the newer one and stick it in 2.6 with the
> fec Gabor sent, but it just crashes so it need more work ;) Dale
> Farnsworth just said it'll look into it and porting new bestcomm as well.

My fec port work ok.

> To resume the "close" future I'm planning :

> I'm busy for two weeks ( I was planning on finishing other stuff earlier
> and didn't work out finally ;( ). So I work less on the MPC but after
> that, that's a full time assignment.

>  - USB: Dale's patch are on their way to the official kernel. Once they
> are in, He or I will send the necessary MPC5200 dependent stuff.
>  - PCI: I'm working on that now, just brushing it up then I'll put it on
> my tree and require testing
>  - I2C: Waiting Adrian Cox patches to be in official tree ( they are in
> -mm ), then apply mpc5200 specifics.
>  - BestComm : Dale (the original fec writing among other things) is
> adapting the newer bestcommcode and cleaning the fec driver as well.
>  - IDE: Once  I get PCI in my tree, I'll look at IDE, a DMA-less version
> at first.

And CAN, RTC, and possible CRC32/16 (if it will fast enough) :).

--
Best regards,
 Andrey Volkov


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: [RFC] Re: BestComm and fec

[Sylvain Munaut]

Hi

>>Apparently, you've both worked out a version of bestcomm for 2.6. But
>>you both based it on old code. (even non GPL).
>>
>>
>
>Terrible, may be its time to create special project for MPC5200 (if it
>doesn't exist in somewhere in I-net)? First intention - sourceforge.
>Comments/suggestions?
>
>Sylvain, Wolfgang I know you already have projects. But in case of
>Sylvain, as I understand, his project on his homepage.
>In case of Wolfgang, I'm not sure that DENX will happy to share their
>internal CVS with all I-net for write access. Am I right?
>
>
I personnaly use BitKeeper to maintain my trees. I can't provide write
access to them because I don't administrate the server they are on,
To write, you would need a ssh account and I juste have one for me and
can't create others.

I try to publish what I'm doing in www.246tNt.com/mpc52xx/state.txt

>>I've just tried just taking the newer one and stick it in 2.6 with the
>>fec Gabor sent, but it just crashes so it need more work ;) Dale
>>Farnsworth just said it'll look into it and porting new bestcomm as well.
>>
>>
>
>My fec port work ok.
>
>
Based on old bestcomm or new ?

>And CAN, RTC, and possible CRC32/16 (if it will fast enough) :).
>
>
CAN is not on my personnal project list since I have nothing that uses
so I couldn't test it anyway.

RTC ? The MPC5200 can't save time AFAIK. And Linux should already
'count' time correctly I think.

By CRC32/16 I guess you mean to use the BestComm DMA engine to implement
the kernel CRC Library ? I'm not sure


I've pushed PCI on my tree ( bk://bkbits.246tNt.com/linux-2.5-mpc52xx )
yesterday night. So if you're in position to test it please report ;)



Sylvain

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: [RFC] Re: BestComm and fec

[David Woodhouse]

On Tue, 2004-08-24 at 12:07 +0400, Andrey Volkov wrote:
> Terrible, may be its time to create special project for MPC5200 (if it
> doesn't exist in somewhere in I-net)? First intention - sourceforge.
> Comments/suggestions?

Don't start a new project. Just put your changes into the official 2.6
tree.

The goal of _all_ external trees should be to make themselves obsolete
by getting all changes to Linus as quickly as possible. Why shoot
yourself in the foot?

--
dwmw2


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re[2]: [RFC] Re: BestComm and fec

[Andrey Volkov]

Hello David,

Tuesday, August 24, 2004, 12:54:46 PM, you wrote:


> On Tue, 2004-08-24 at 12:07 +0400, Andrey Volkov wrote:
>> Terrible, may be its time to create special project for MPC5200 (if it
>> doesn't exist in somewhere in I-net)? First intention - sourceforge.
>> Comments/suggestions?

> Don't start a new project. Just put your changes into the official 2.6
> tree.

> The goal of _all_ external trees should be to make themselves obsolete
> by getting all changes to Linus as quickly as possible. Why shoot
> yourself in the foot?

Indeed, BUT:
 1) IMHO mail list better approach for
    along developer <-> maintainer/many tester
 scheme of development. Cvs/bk will be better for team work isn't it?

 2) So somewhere must exist place for coordination of job, 'cause
 I see as minimum 5 person doing SAME work (terrible).

 3) And I wish have common place for experiments and testing. AFAIK, stable
 kernel is not appropriate for it. (If I wrong - why exist -mm -ac
 etc branches? :)

--
Best regards,
 Andrey Volkov


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: [RFC] Re: BestComm and fec

[Dale Farnsworth]

On Tue, Aug 24, 2004 at 08:07:07AM +0000, Andrey Volkov wrote:
>
> Hi all,
>
> > Andrey: Gabor has a working fec.c implementation.
>
> > Apparently, you've both worked out a version of bestcomm for 2.6. But
> > you both based it on old code. (even non GPL).
>
> Terrible, may be its time to create special project for MPC5200 (if it
> doesn't exist in somewhere in I-net)? First intention - sourceforge.
> Comments/suggestions?

Yes, useless duplication of work is sad, especially if it is using
obsolete code.  However, the use of special source tree for MPC5200
is neither necessary nor sufficient to avoid that.  It could have been
avoided by talking with Sylvain or me.  I have learned that it
is almost essential to discuss linux ppc development on #mklinux
on IRC.

BTW, the code you ported looks like an early development version that
I never intended for distribution.  Motorola begged me to let them use
it for a "demo", then distributed it without notifying me.

> Sylvain, Wolfgang I know you already have projects. But in case of
> Sylvain, as I understand, his project on his homepage.
> In case of Wolfgang, I'm not sure that DENX will happy to share their
> internal CVS with all I-net for write access. Am I right?

Write access is overrated and not necessary, particularly with
bitkeeper's ability to pull changesets from other trees.  Sylvain
and I have each maintainend our own list of patches pending to the
official trees, and have shared those patches back and forth.

However, with more than 2 active developers, it may be useful to have
a temporary tree in which pending support for the various mpc5200
peripherals is coalesced.  If so, I think it makes sense for Sylvain
to do that if he's willing, since he's the mpc5200 maintainer.

> > I've just tried just taking the newer one and stick it in 2.6 with the
> > fec Gabor sent, but it just crashes so it need more work ;) Dale
> > Farnsworth just said it'll look into it and porting new bestcomm as well.
>
> My fec port work ok.

It may appear to work, but from previous discussions here and on
linuxppc-devel, the fec and bestcomm code is not good enough to be
accepted into the main linux trees.  The BestComm API code, in
particular, is hopeless and a complete rewrite will be required.

I'm addressing that now.  I have separated out fec code that interfaces
to the BestComm API code in preparation for replacement of the BestComm
API code.  I'm now separating out and cleaning up the PHY support code.
Next, I plan to replace the BestComm API code, but I may hand that back
to Sylvain if he's available and willing.

Anyway, the fec code will be a lot cleaner in a couple of weeks.
I guarantee I'll post something before then.

> > To resume the "close" future I'm planning :
>
> > I'm busy for two weeks ( I was planning on finishing other stuff earlier
> > and didn't work out finally ;( ). So I work less on the MPC but after
> > that, that's a full time assignment.
>
> >  - USB: Dale's patch are on their way to the official kernel. Once they
> > are in, He or I will send the necessary MPC5200 dependent stuff.
> >  - PCI: I'm working on that now, just brushing it up then I'll put it on
> > my tree and require testing
> >  - I2C: Waiting Adrian Cox patches to be in official tree ( they are in
> > -mm ), then apply mpc5200 specifics.
> >  - BestComm : Dale (the original fec writing among other things) is
> > adapting the newer bestcommcode and cleaning the fec driver as well.
> >  - IDE: Once  I get PCI in my tree, I'll look at IDE, a DMA-less version
> > at first.
>
> And CAN, RTC, and possible CRC32/16 (if it will fast enough) :).
>
> Best regards,
>  Andrey Volkov

Welcome to the team Andrey.  There is plenty of work to go around.
Check with Sylvain if there's something not being done that you'd
like to take on.

-Dale Farnsworth

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

[RFC] MPC5200 dev (was Re: [RFC] Re: BestComm and fec)

[Sylvain Munaut]

Since It's clear that with more peoples working on it, there is a need of
synchro.

Here's what I'm proposing :

I'll a tree :
   bk://bkbits.246tNt.com/linux-mpc52xx-devel
       This tree would contain all the latest code to be tested by
others. Just send patch to the ml and cc me, I'll put them in this
tree. Once every one is happy with it (or part of it), I'll take
care of pushing it upstream.
	Note that since all current work with bestcomm is based on
non-GPL code, It won't be in until Dale send me the version he's
working on (based on latest code).


To avoid duplicate work, you can post here what you're doing, I'll try
to keep
http://www.246tNt.com/mpc52xx/state-devel.txt  up-to-date with who's doing what.

The point is to centralize the testing code and having a 'single' contact
to merge the work upstream to ease the work of upstream maintainers.


This tree is being setup right now, please allow a few hours ;)


Comments, suggestions are of course welcome.


	Sylvain




Andrey Volkov wrote:

>Hello David,
>
>Tuesday, August 24, 2004, 12:54:46 PM, you wrote:
>
>
>
>
>>On Tue, 2004-08-24 at 12:07 +0400, Andrey Volkov wrote:
>>
>>
>>>Terrible, may be its time to create special project for MPC5200 (if it
>>>doesn't exist in somewhere in I-net)? First intention - sourceforge.
>>>Comments/suggestions?
>>>
>>>
>
>
>
>>Don't start a new project. Just put your changes into the official 2.6
>>tree.
>>
>>
>
>
>
>>The goal of _all_ external trees should be to make themselves obsolete
>>by getting all changes to Linus as quickly as possible. Why shoot
>>yourself in the foot?
>>
>>
>
>Indeed, BUT:
> 1) IMHO mail list better approach for
>    along developer <-> maintainer/many tester
> scheme of development. Cvs/bk will be better for team work isn't it?
>
> 2) So somewhere must exist place for coordination of job, 'cause
> I see as minimum 5 person doing SAME work (terrible).
>
> 3) And I wish have common place for experiments and testing. AFAIK, stable
> kernel is not appropriate for it. (If I wrong - why exist -mm -ac
> etc branches? :)
>
>--
>Best regards,
> Andrey Volkov
>
>
>
>
>


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: [RFC] MPC5200 dev (was Re: [RFC] Re: BestComm and fec)

[Andrey Volkov]

Hello Sylvain,

Tuesday, August 24, 2004, 8:17:26 PM, you wrote:


> Since It's clear that with more peoples working on it, there is a need of
> synchro.

> Here's what I'm proposing :

> I'll a tree :
>    bk://bkbits.246tNt.com/linux-mpc52xx-devel
>        This tree would contain all the latest code to be tested by
> others. Just send patch to the ml and cc me, I'll put them in this
> tree. Once every one is happy with it (or part of it), I'll take
> care of pushing it upstream.
> 	Note that since all current work with bestcomm is based on
> non-GPL code, It won't be in until Dale send me the version he's
> working on (based on latest code).

> To avoid duplicate work, you can post here what you're doing, I'll try
> to keep
> http://www.246tNt.com/mpc52xx/state-devel.txt  up-to-date with who's doing what.
May be index.html?

> The point is to centralize the testing code and having a 'single' contact
> to merge the work upstream to ease the work of upstream maintainers.
OK, I accept.

> This tree is being setup right now, please allow a few hours ;)


> Comments, suggestions are of course welcome.

Also don't forget add yourself and ml to MAINTAINERS file, 'cause
if this info were there, then I stopping senselessly :) work early (or
doesn't started at all).

--
Best regards,
 Andrey Volkov


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/