From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mail-wr0-x22b.google.com ([2a00:1450:400c:c0c::22b]) by metis.ext.pengutronix.de with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128) (Exim 4.89) (envelope-from ) id 1eg820-0001NY-7x for ptxdist@pengutronix.de; Mon, 29 Jan 2018 12:55:24 +0100 Received: by mail-wr0-x22b.google.com with SMTP id a1so6883139wri.5 for ; Mon, 29 Jan 2018 03:55:23 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <20180129103720.njvzqdjralhiwptl@pengutronix.de> References: <20180126124030.37ehgkax5csjhfuz@falbala.home.lespocky.de> <20180126125819.6i5klyhtxl5ra23c@pengutronix.de> <20180129103720.njvzqdjralhiwptl@pengutronix.de> From: Guillermo Rodriguez Garcia Date: Mon, 29 Jan 2018 12:55:22 +0100 Message-ID: Subject: Re: [ptxdist] Platform-specific kernel patchset not applied to kernel headers package? List-Id: PTXdist Development Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: ptxdist@pengutronix.de Content-Type: multipart/mixed; boundary="===============0915015252==" Errors-To: ptxdist-bounces@pengutronix.de Sender: "ptxdist" To: ptxdist@pengutronix.de --===============0915015252== Content-Type: multipart/alternative; boundary="94eb2c0defe4cb2c850563e8eb1b" --94eb2c0defe4cb2c850563e8eb1b Content-Type: text/plain; charset="UTF-8" Hi, 2018-01-29 11:37 GMT+01:00 Michael Olbrich : > Hi, > > On Mon, Jan 29, 2018 at 10:51:07AM +0100, Guillermo Rodriguez Garcia wrote: > > 2018-01-26 13:58 GMT+01:00 Michael Olbrich : > > > On Fri, Jan 26, 2018 at 01:40:31PM +0100, Alexander Dahl wrote: > > > > On Fri, Jan 26, 2018 at 08:58:52AM +0100, Guillermo Rodriguez Garcia > > > wrote: > > > > > I am creating a ptxdist BSP and have a series. patchset > for a > > > > > given kernel version. I have noticed that this patchset is ignored > > > when the > > > > > KERNEL_HEADER package is built -- instead, a "series" file is > > > automatically > > > > > created and used. > > > > > > > > > > Is this normal/expected? Shouldn't the series. file be > used > > > > > instead? > > > > > > > > It is at least known. I stumbled over this last year: > > > > > > > > https://www.mail-archive.com/ptxdist@pengutronix.de/msg12233.html > > > > > > > > Michael gave some hints for possible workarounds. > > > > > > > > > Right, check that thread. > > > > > > Also, why do you need a 'series.' instead of a normal series > > > file? > > > > > > > I used series. because the BSP currently supports two platforms > > (two boards), and some of the patches are platform-specific. > > > > > > > And why do you need the patches for the kernel headers? > > > > > > > This is because I am building imx6 specific packages which require a set > of > > kernel headers, and these must include imx6-specific bits. > > > > In the thread that Alexander mentioned in his reply you mention the > > following: > > > > 1. Move the patches to configs/platform-XXX/patches/linux-4.9.47/ and > just > > use 'series'. > > In most cases, that's the correct way to handle this. All platform specific > files should be in configs/platform-XXX/ anyways. > > Keeping the patches Toplevel has one use-case: > Some patches are used by both platforms. I'd still move the patches to the > platforms and use symlinks to share: Patches can be in subdirs, so just on > Symlink to the common set of patches is enough. > > > I guess this solves my problem, however, just out of curiosity: If the > > kernel headers package will pick up the correct patchset from > > configs/platform-XXX/patches/linux-4.9.47, shouldn't it also pick it up > if > > it lives in patches/linux-4.9.47/series.XXX ? > > Checking in configs/platform-XXX/patches/ is a generic PTXdist mechanism. > Using 'series.XXX' is a hack in the Kernel patches to work with bad > patches. It's a mechanism that I'd like to get rid of. But there are still > people using it. > It's not something I want to spread to other packages. > Thank you for the explanation, I know understand the background and motivations. Still I would suggest to either get rid of the "series.XXX" hack (perhaps with a check and a warning) or make it applicable to kernel headers as well. The current behaviour is a bit surprising. Anyway my problem is solved now so thank you again. Guillermo Rodriguez Garcia guille.rodriguez@gmail.com --94eb2c0defe4cb2c850563e8eb1b Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Hi,

2018-01-29 11:37 GMT+01:00 Michael Olbrich <m.olbrich@pengut= ronix.de>:
Hi,

On Mon, Jan 29, 2018 at 10:51:07AM +0100, Guillermo Rodriguez Garcia wrote:=
> 2018-01-26 13:58 GMT+01:00 Michael Olbrich <m.olbrich@pengutronix.de>:
> > On Fri, Jan 26, 2018 at 01:40:31PM = +0100, Alexander Dahl wrote:
> > > On Fri, Jan 26, 2018 at 08:58:52AM +0100, Guillermo Rodrigue= z Garcia
> > wrote:
> > > > I am creating a ptxdist BSP and have a series.<platf= orm> patchset for a
> > > > given kernel version. I have noticed that this patchset= is ignored
> > when the
> > > > KERNEL_HEADER package is built -- instead, a "seri= es" file is
> > automatically
> > > > created and used.
> > > >
> > > > Is this normal/expected? Shouldn't the series.<p= latform> file be used
> > > > instead?
> > >
> > > It is at least known. I stumbled over this last year:
> > >
> > > https://www.mail-arc= hive.com/ptxdist@pengutronix.de/msg12233.html
> > >
> > > Michael gave some hints for possible workarounds.
> >
> >
> > Right, check that thread.
> >
> > Also, why do you need a 'series.<platform>' instead= of a normal series
> > file?
> >
>
> I used series.<platform> because the BSP currently supports two = platforms
> (two boards), and some of the patches are platform-specific.
>
>
> > And why do you need the patches for the kernel headers?
> >
>
> This is because I am building imx6 specific packages which require a s= et of
> kernel headers, and these must include imx6-specific bits.
>
> In the thread that Alexander mentioned in his reply you mention the > following:
>
> 1. Move the patches to configs/platform-XXX/patches/linux-4.9.47/= and just
> use 'series'.

In most cases, that's the correct way to handle this. All p= latform specific
files should be in configs/platform-XXX/ anyways.

Keeping the patches Toplevel has one use-case:
Some patches are used by both platforms. I'd still move the patches to = the
platforms and use symlinks to share: Patches can be in subdirs, so just on<= br> Symlink to the common set of patches is enough.

> I guess this solves my problem, however, just out of curiosity: If the=
> kernel headers package will pick up the correct patchset from
> configs/platform-XXX/patches/linux-4.9.47, shouldn't it also = pick it up if
> it lives in patches/linux-4.9.47/series.XXX ?

Checking in configs/platform-XXX/patches/ is a generic PTXdist mecha= nism.
Using 'series.XXX' is a hack in the Kernel patches to work with bad=
patches. It's a mechanism that I'd like to get rid of. But there ar= e still
people using it.
It's not something I want to spread to other packages.
=

Thank you for the explanation, I know understand the ba= ckground and motivations.

Still I would suggest to= either get rid of the "series.XXX" hack (perhaps with a check an= d a warning) or make it applicable to kernel headers as well. The current b= ehaviour is a bit surprising.

Anyway my problem is= solved now so thank you again.
Guillermo Rodriguez Garcia
= guille.rodr= iguez@gmail.com
--94eb2c0defe4cb2c850563e8eb1b-- --===============0915015252== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: base64 Content-Disposition: inline X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX18KcHR4ZGlzdCBt YWlsaW5nIGxpc3QKcHR4ZGlzdEBwZW5ndXRyb25peC5kZQ== --===============0915015252==--