From mboxrd@z Thu Jan 1 00:00:00 1970 From: pebolle@tiscali.nl (Paul Bolle) Date: Fri, 26 Sep 2014 16:54:53 +0200 Subject: [PATCHv3 2/7] ARM: at91: introduce basic SAMA5D4 support In-Reply-To: <20140926141326.GB4408@piout.net> References: <1410797759-18675-1-git-send-email-alexandre.belloni@free-electrons.com> <1410797759-18675-3-git-send-email-alexandre.belloni@free-electrons.com> <1411732023.7866.15.camel@x220> <20140926141326.GB4408@piout.net> Message-ID: <1411743293.7866.62.camel@x220> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi Alexandre, On Fri, 2014-09-26 at 16:13 +0200, Alexandre Belloni wrote: > On 26/09/2014 at 13:47:03 +0200, Paul Bolle wrote : > > On Mon, 2014-09-15 at 18:15 +0200, Alexandre Belloni wrote: > > > From: Nicolas Ferre > > > > > > Signed-off-by: Nicolas Ferre > > > > This landed in next-20140926, as commit 2dc850b62e5b ("ARM: at91: > > introduce basic SAMA5D4 support"). > > > > > diff --git a/arch/arm/Kconfig.debug b/arch/arm/Kconfig.debug > > > index b11ad54f8d17..7b44db208f10 100644 > > > --- a/arch/arm/Kconfig.debug > > > +++ b/arch/arm/Kconfig.debug > > > @@ -101,6 +101,10 @@ choice > > > bool "Kernel low-level debugging on 9263 and 9g45" > > > depends on HAVE_AT91_DBGU1 > > > > > > + config AT91_DEBUG_LL_DBGU2 > > > + bool "Kernel low-level debugging on sama5d4" > > > + depends on HAVE_AT91_DBGU2 > > > + > > > > Why is this Kconfig symbol needed? Nothing in next-20140926 uses it. Are > > future users perhaps queued somewhere (say, in commits that didn't > > survive today's rather lively linux-next merges)? > > There is no user because it conflicts with the previous one that is > selected by default by the sama5_defconfig. This is for early print and > it means that you can get earlyprintk either for sama5d3 or sama5d4. > > My guess is that it will never be included in any defconfig. I missed that this is inside a "choice" (which complicates matters quite a bit)! There are a number of cases where one of the configs inside the choice only serve to _not_ set anything. In those cases the Kconfig symbol is used nowhere else. I'll have to think about this particular symbol a bit more to see if it actually makes sense to not use it anywhere else. Thinking about the "Kernel low-level debugging port" choice is complicated by the fact that it holds about a gazillion "config" entries, many of which only show up only if some dependency is met. So it might take me some time to say anything sensible about it, if I ever manage to do that at all... Thanks, Paul Bolle