From mboxrd@z Thu Jan 1 00:00:00 1970 From: Heiko =?utf-8?q?St=C3=BCbner?= Subject: Re: [PATCH v4 0/5] move s3c24xx-irq to drivers/irqchip and add dt support Date: Fri, 22 Mar 2013 22:21:17 +0100 Message-ID: <201303222221.18605.heiko@sntech.de> References: <201303221843.37668.heiko@sntech.de> <201303222104.54225.arnd@arndb.de> Mime-Version: 1.0 Content-Type: Text/Plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from gloria.sntech.de ([95.129.55.99]:44418 "EHLO gloria.sntech.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1423269Ab3CVVVb (ORCPT ); Fri, 22 Mar 2013 17:21:31 -0400 In-Reply-To: <201303222104.54225.arnd@arndb.de> Sender: linux-samsung-soc-owner@vger.kernel.org List-Id: linux-samsung-soc@vger.kernel.org To: Arnd Bergmann Cc: Kukjin Kim , Grant Likely , Rob Herring , Thomas Abraham , devicetree-discuss@lists.ozlabs.org, linux-arm-kernel@lists.infradead.org, linux-samsung-soc@vger.kernel.org Am Freitag, 22. M=C3=A4rz 2013, 22:04:54 schrieb Arnd Bergmann: > On Friday 22 March 2013, Heiko St=C3=BCbner wrote: > > As suggested by Rob Herring move back to get the parent releationsh= ip > > from dt like in the first versions, but set the handler from the > > interrupt property rather than from a specific list. > >=20 > > This version also implements (hopefully correctly) an idea from him= and > > Arnd Bergmann to have the parent relationship not described in the > > controller node but the device nodes instead. > >=20 > > Therefore the main controller continues to use a two-cell descripto= r to > > set hwirq and trigger type but the sub-controller switches to a > > three-cell descriptor where the third bit describes the hwirq of it= s > > parent irq in the main controller. > >=20 > > As a result a serial node would then look like: > > serial@50000000 { > > =20 > > compatible =3D "samsung,s3c2410-uart"; > > reg =3D <0x50000000 0x4000>; > > interrupt-parent =3D <&subintc>; > > interrupts =3D <0 4 28>, <1 4 28>; > > =20 > > }; > >=20 > > Tested on a s3c2416-based board. > >=20 > > As it depends on changes already pending for 3.10 it should probabl= y go > > thru the samsung tree. >=20 > Hi Heiko, >=20 > I should probably read the original discussion thread again. I alread= y > commented on the binding here, but I may have missed something import= ant > that led to it being the way it is now. Hi Arnd, Not really ... it's only my current interpretation of your short sugges= tion to=20 handle the parent relationship in the interrupt descriptor of the devic= e node=20 and not in the controller node itself. Your other mail did contain a lot of interesting thoughts on how to imp= rove=20 the current state - thanks for them. Heiko From mboxrd@z Thu Jan 1 00:00:00 1970 From: heiko@sntech.de (Heiko =?utf-8?q?St=C3=BCbner?=) Date: Fri, 22 Mar 2013 22:21:17 +0100 Subject: [PATCH v4 0/5] move s3c24xx-irq to drivers/irqchip and add dt support In-Reply-To: <201303222104.54225.arnd@arndb.de> References: <201303221843.37668.heiko@sntech.de> <201303222104.54225.arnd@arndb.de> Message-ID: <201303222221.18605.heiko@sntech.de> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Am Freitag, 22. M?rz 2013, 22:04:54 schrieb Arnd Bergmann: > On Friday 22 March 2013, Heiko St?bner wrote: > > As suggested by Rob Herring move back to get the parent releationship > > from dt like in the first versions, but set the handler from the > > interrupt property rather than from a specific list. > > > > This version also implements (hopefully correctly) an idea from him and > > Arnd Bergmann to have the parent relationship not described in the > > controller node but the device nodes instead. > > > > Therefore the main controller continues to use a two-cell descriptor to > > set hwirq and trigger type but the sub-controller switches to a > > three-cell descriptor where the third bit describes the hwirq of its > > parent irq in the main controller. > > > > As a result a serial node would then look like: > > serial at 50000000 { > > > > compatible = "samsung,s3c2410-uart"; > > reg = <0x50000000 0x4000>; > > interrupt-parent = <&subintc>; > > interrupts = <0 4 28>, <1 4 28>; > > > > }; > > > > Tested on a s3c2416-based board. > > > > As it depends on changes already pending for 3.10 it should probably go > > thru the samsung tree. > > Hi Heiko, > > I should probably read the original discussion thread again. I already > commented on the binding here, but I may have missed something important > that led to it being the way it is now. Hi Arnd, Not really ... it's only my current interpretation of your short suggestion to handle the parent relationship in the interrupt descriptor of the device node and not in the controller node itself. Your other mail did contain a lot of interesting thoughts on how to improve the current state - thanks for them. Heiko