[Letux-kernel] JZ4730 kernel support

H. Nikolaus Schaller hns at goldelico.com
Mon Oct 21 13:10:29 CEST 2019


> Am 21.10.2019 um 13:01 schrieb Paul Boddie <paul at boddie.org.uk>:
> 
> On Monday 21. October 2019 09.37.32 H. Nikolaus Schaller wrote:
>>> Am 19.10.2019 um 13:50 schrieb Paul Boddie <paul at boddie.org.uk>:
>>> 
>>> commit 48beba8932f2d296cd0a031003cf295096c9dc40 (grafted,
>>> origin/letux/jz4730)
>> Yes, that did work.
> 
> Great!
> 
>>> There should be five patches in this bundle, three of which you will have
>>> seen before. Sorry not to be able to prepare these in a nicer way, but
>>> git generally isn't happy running properly on my hardware with these
>>> repositories.
>> No problem. A patch set sent like this just needs knowledge about the
>> merge base - if the merge base itself changes. This seems to be the cas
>> with -rc for the ingenic subtree...
> 
> Is the history being rewritten or something?

Yes. All letux/ branches are rebased regularly on top of the latest
linus/master version.

This makes sure that (after long time) commits posted from a branch
to the LKML are "eaten up" from the bottom when they are merged by
Linus. And if all such commits are in linus/master the feature branch
becomes empty.

And in the mean-time we have a working system with latest upstream
changes.

> That is the only explanation for 
> why patches might suddenly not be applicable, at least if I am making them for 
> the right branch.
> 
>>> Maybe preparing them all separately instead of in combined form is better.
>>> I've attached an archive to this message containing the individual
>>> patches.
>> 
>> That doesn't make a big difference - except that some patches may apply
>> cleanly while others still fail.
> 
> Yes, the intention is to permit partial failure rather than always experience 
> complete failure.
> 
>> But what is your git repo problem? I have not seen a problem cloning it on
>> quite different systems: e.g. macOS or a RasPi 3B+ with a 200GB µSD card
>> running ifself with LetuxOS.
> 
> I did look at my complaints about this in the archives, but I don't think 
> there was a successful resolution. You did write something about this, but I 
> didn't find that message.
> 
> It is possible that some tuning is needed with git, given that I am using 
> machines with only 1GB RAM,

Well, the RasPi 3 also has only 1GB. Yes, it is slow but seems to work.

> but as I noted before, the advice about such 
> things is all over the place: people suggest changing some settings, other 
> people claimed that it worked for them, still other people claim that it 
> didn't work but something else did. And so on.
> 
>> And maybe it would be good if you would become "Happy Crew" member of the
>> letux kernel team and get write-access to the repo:
>> 
>> http://projects.goldelico.com/p/gta04-kernel/
>> 
>> I have already added you. To get ssh write access you should add your
>> key to the user-config page and then I have to run a script to copy it
>> to the ssh/git daemon's permission files.
> 
> I did manage to submit patches to the gta04-kernel before now. Navigating 
> branches was slow, but it did work, more or less.
> 
> My workaround currently involves a shallow clone of the appropriate branch in 
> the letux-kernel repository, which is the one piece of advice I found that 
> actually seems to work for people in my situation. Unfortunately, switching 
> branches seems to cause git to want to fetch all the history, even though the 
> (awful) man pages suggest that this shouldn't happen for shallow clones. And 
> it is this history fetching that leads to the "index pack failed" error.

Hm. Do you use git or http access? AFAIR git access is not really reliable.
https or ssh+git can retry in some cases.

Or can you do a test if it is really on your side:

git clone git at github.com:goldelico/letux-kernel.git

If that works it is more a network issue (which may lead to a timeout somewhere).

> I'll try and think of other approaches, but I don't have so many ideas at the 
> moment. Then again, I don't think there are so many other changes to be made 
> at the moment, either.

Ok, fine.

BR and thanks,
Nikolaus



More information about the Letux-kernel mailing list