On 19 September 2016 at 12:36, Y.T. Jiang <yutang.ji...@nxp.com> wrote:
> Thank you for your review and suggestion.

Sure. One more note: please take a look at your mailer configuration.
It should keep all quotes prefixed with "> " to keep discussion clear.

> -----Original Message-----
> From: Rafał Miłecki [mailto:zaj...@gmail.com]
> Sent: Monday, September 19, 2016 4:01 PM
> To: John Crispin
> Cc: LEDE Development List; Y.T. Jiang
> Subject: Re: [LEDE-DEV] merging the layerscape target
> On 18 September 2016 at 14:24, John Crispin <j...@phrozen.org> wrote:
>> i have just spent some time reviewing the layerscape PR [1] and
>> started a full build of it. its starting to look good and i cannot see
>> any blockers. if anyone has any hold on this please let me know in the
>> next couple of days. if i dont get any vetos i will merge it.
> I can see following Copyright line over and over:
> Copyright (C) 2016 OpenWrt.org
> Yutang: did you really sign a contract with OpenWrt that included passing 
> your copyrights to the OpenWrt project? If not, you should just keep 
> Copyrights assigned to yourself.
> I really would like assigning copyrights to projects where it doesn't apply.
> [I do not sign a contract with OpenWrt indeed. I refer to some others target 
> while developing/backporting layerscape, I find almost of targets included 
> OpenWrt.org Copyright, so I also put it in my code files. Now should I 
> replace " Copyright (C) 2016 OpenWrt.org" with " Copyright (C) 2016 Jiang 
> Yutang <yutang.ji...@nxp.com>" ? or retain the both copyright: "Copyright (C) 
> LEDE project, Jiang Yutang <yutang.ji...@nxp.com>" ?]

You're correct, current sources are messy about this. I'm trying to
stop adding mode incorrectly copyrighted code.

You should only have something like:
Copyright (C) 2016 Jiang Yutang <yutang.ji...@nxp.com>
for the code you have written.

> What about using some generic profile only and then using DEVICE_TITLE 
> DEVICE_PACKAGES to specify modules that should be included on rootfs?
> [I will try to use the two variables.]

Thanks! This will allow building images for customized boards with a
single "make" call. It's part of recently introduced
TARGET_PER_DEVICE_ROOTFS system. You may take a look at
target/linux/bcm53xx/image/Makefile as an example. There is only 1
subtarget, but it should give you a hint anyway.

> Would that be possible to split patches into accepted ones (backports) and 
> LEDE-specific ones?
> [The kernel patches: dpaa/qbman/fman/etc. it is really too big and 
> interference review LEDE-specially code. I will split those kernel patches in 
> folder patches-4.4 as the second, and keep the rest as fist LEDE-specific,  
> what do you think about it?]

For generic patches we have a following guide:

You may try to follow this, if possible. E.g. you could use
0xxx prefix for upstream accepted patches
and some other prefix 1xxx, 2xxx, or whatever applicable for other ones.

It isn't a strict rule for targets, but it should make your target
easier to maintain I believe.

> Please refresh all target patches, right now I can see they contain all these 
> things like:
> diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c index 
> 4cb98aa..a8a97bd 100644
> [I found it have conflicts in current kernel version with two 
> patches(arm64/mm related, 0060 and 0061) while backporting the 
> dpaa/qbman/fman driver, but I'm unacquainted with both mm and dpaa, our dpaa 
> team are engaged in do upstream work and can't help me. So I revert the two 
> patch to bypass this issue temporary, I would like to wait for more leisure 
> time then to thorough investigate and solve it.]

I think you misunderstood me. I don't have anything against your
patches, just the format. Please call
make target/linux/refresh V=s
and that will convert all your patches to the expected format :)


Lede-dev mailing list

Reply via email to