From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?iso-8859-2?Q?Horia_Geant=E3?= Subject: Re: [PATCH] arm64: dts: ls1012a: add crypto node Date: Tue, 28 Mar 2017 07:19:43 +0000 Message-ID: References: <20170322122939.22117-1-horia.geanta@nxp.com> <20170324015606.GC30608@dragon> <20170324073420.GM30608@dragon> <20170324140320.GP30608@dragon> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-2" Content-Transfer-Encoding: quoted-printable Cc: Mark Rutland , "devicetree@vger.kernel.org" , Herbert Xu , Harninder Rai , Catalin Marinas , Bhaskar U , Will Deacon , Dan Douglass , Rob Herring , "linux-crypto@vger.kernel.org" , "David S. Miller" , "linux-arm-kernel@lists.infradead.org" To: Shawn Guo Return-path: Received: from mail-ve1eur01on0058.outbound.protection.outlook.com ([104.47.1.58]:11712 "EHLO EUR01-VE1-obe.outbound.protection.outlook.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1754313AbdC1HUJ (ORCPT ); Tue, 28 Mar 2017 03:20:09 -0400 Content-Language: en-US Sender: linux-crypto-owner@vger.kernel.org List-ID: On 3/24/2017 4:04 PM, Shawn Guo wrote:=0A= > On Fri, Mar 24, 2017 at 08:29:17AM +0000, Horia Geant=E3 wrote:=0A= >> On 3/24/2017 9:35 AM, Shawn Guo wrote:=0A= >>> On Fri, Mar 24, 2017 at 07:17:50AM +0000, Horia Geant=E3 wrote:=0A= >>>>>> + sec_mon: sec_mon@1e90000 {=0A= >>>>>=0A= >>>>> Hyphen is more preferred to be used in node name than underscore.=0A= >>>>>=0A= >>>> This would imply changing the=0A= >>>> Documentation/devicetree/bindings/crypto/fsl-sec4.txt binding and=0A= >>>> dealing with all the consequences, which IIUC is probably not worth.= =0A= >>>=0A= >>> I do not care the bindings doc that much, since I'm not the maintainer= =0A= >>> of it. What are the consequences specifically, if we use a better node= =0A= >>> name in dts than bindings example?=0A= >>>=0A= >> Users relying on finding the sec_mon node will obviously stop working.= =0A= >> I don't see any in-kernel users, however there could be others I am not= =0A= >> aware of and DT bindings should provide for backwards compatibility.=0A= > =0A= > Okay, point taken. You can keep the node name as it is.=0A= > =0A= >> I could deprecate "sec_mon" in the bindings and suggest "sec-mon"=0A= >> instead, while leaving all existing dts files as-is.=0A= >> The risk is breaking LS1012A users relying on "sec_mon".=0A= > =0A= > For existing bindings, I do not care that much. But for new ones, I do= =0A= > hope that we recommend to use hyphen, as that's more idiomatic at least= =0A= > for Linux kernel.=0A= > =0A= >> I see that ePAPR:=0A= >> -allows both for hyphen and underline in case of node names=0A= >> -allows only for hyphen (i.e. forbids underline) in case of alias nodes= =0A= >>=0A= >> In the first case, I understand there's an (undocumented?) agreement to= =0A= >> prefer hyphen over underline.=0A= > =0A= > Both are valid, but hyphen is more idiomatic for Linux kernel.=0A= > =0A= >> For the 2nd one, does this mean I should change alias names?=0A= > =0A= > This is something I see difference between specification and DTC.=0A= > =0A= > aliases {=0A= > alias-name =3D &label_name;=0A= > };=0A= > =0A= > label_name: node-name {=0A= > ...=0A= > };=0A= > =0A= > The spec says that only hyphen is valid for alias name, but DTC works=0A= > happily with underscore too. From my experience with DTC playing, both= =0A= > hyphen and underscore are valid for alias and node name. But for label= =0A= > name, only underscore is valid. Using hyphen in label name will cause=0A= > DTC to report syntax error.=0A= > =0A= Yes indeed, thanks for pointing it out.=0A= =0A= For the sake of current patch, please clarify whether a v2 is needed.=0A= IIUC:=0A= -sec_mon node name could stay the same (existing binding)=0A= -label names are ok, since underline is the only option allowed by DTC=0A= -alias names are out-of-spec but accepted by DTC; if changing underline=0A= to hyphen is requested, I will push out v2=0A= =0A= Thanks,=0A= Horia=0A= =0A= From mboxrd@z Thu Jan 1 00:00:00 1970 From: horia.geanta@nxp.com (=?iso-8859-2?Q?Horia_Geant=E3?=) Date: Tue, 28 Mar 2017 07:19:43 +0000 Subject: [PATCH] arm64: dts: ls1012a: add crypto node References: <20170322122939.22117-1-horia.geanta@nxp.com> <20170324015606.GC30608@dragon> <20170324073420.GM30608@dragon> <20170324140320.GP30608@dragon> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 3/24/2017 4:04 PM, Shawn Guo wrote: > On Fri, Mar 24, 2017 at 08:29:17AM +0000, Horia Geant? wrote: >> On 3/24/2017 9:35 AM, Shawn Guo wrote: >>> On Fri, Mar 24, 2017 at 07:17:50AM +0000, Horia Geant? wrote: >>>>>> + sec_mon: sec_mon at 1e90000 { >>>>> >>>>> Hyphen is more preferred to be used in node name than underscore. >>>>> >>>> This would imply changing the >>>> Documentation/devicetree/bindings/crypto/fsl-sec4.txt binding and >>>> dealing with all the consequences, which IIUC is probably not worth. >>> >>> I do not care the bindings doc that much, since I'm not the maintainer >>> of it. What are the consequences specifically, if we use a better node >>> name in dts than bindings example? >>> >> Users relying on finding the sec_mon node will obviously stop working. >> I don't see any in-kernel users, however there could be others I am not >> aware of and DT bindings should provide for backwards compatibility. > > Okay, point taken. You can keep the node name as it is. > >> I could deprecate "sec_mon" in the bindings and suggest "sec-mon" >> instead, while leaving all existing dts files as-is. >> The risk is breaking LS1012A users relying on "sec_mon". > > For existing bindings, I do not care that much. But for new ones, I do > hope that we recommend to use hyphen, as that's more idiomatic at least > for Linux kernel. > >> I see that ePAPR: >> -allows both for hyphen and underline in case of node names >> -allows only for hyphen (i.e. forbids underline) in case of alias nodes >> >> In the first case, I understand there's an (undocumented?) agreement to >> prefer hyphen over underline. > > Both are valid, but hyphen is more idiomatic for Linux kernel. > >> For the 2nd one, does this mean I should change alias names? > > This is something I see difference between specification and DTC. > > aliases { > alias-name = &label_name; > }; > > label_name: node-name { > ... > }; > > The spec says that only hyphen is valid for alias name, but DTC works > happily with underscore too. From my experience with DTC playing, both > hyphen and underscore are valid for alias and node name. But for label > name, only underscore is valid. Using hyphen in label name will cause > DTC to report syntax error. > Yes indeed, thanks for pointing it out. For the sake of current patch, please clarify whether a v2 is needed. IIUC: -sec_mon node name could stay the same (existing binding) -label names are ok, since underline is the only option allowed by DTC -alias names are out-of-spec but accepted by DTC; if changing underline to hyphen is requested, I will push out v2 Thanks, Horia