Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: [PATCH V11 3/5] kbuild: Allow .dtso format for overlay source
 files
To:     Viresh Kumar <viresh.kumar@linaro.org>
Cc:     Masahiro Yamada <masahiroy@kernel.org>,
        Michal Marek <michal.lkml@markovi.net>,
        Vincent Guittot <vincent.guittot@linaro.org>,
        David Gibson <david@gibson.dropbear.id.au>,
        Michal Simek <michal.simek@xilinx.com>,
        Geert Uytterhoeven <geert@linux-m68k.org>,
        Anmar Oueja <anmar.oueja@linaro.org>,
        Bill Mills <bill.mills@linaro.org>,
        Geert Uytterhoeven <geert+renesas@glider.be>,
        Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
References: <cover.1615354376.git.viresh.kumar@linaro.org>
 <170e086a5fa076869e7b37de8eea850fa7c39118.1615354376.git.viresh.kumar@linaro.org>
 <CAK7LNASACr5EaG9j5c-eD3bYxKgrisb60Z3Qy7UsyS-i9YjORg@mail.gmail.com>
 <20210312044712.srmqfuie7fae55pb@vireshk-i7>
 <17c65559-865f-f742-660f-0ab30ed45d90@gmail.com>
 <4d9bee7a-416e-50a1-65a5-0674ae83d42e@gmail.com>
 <20210312071325.zosmlttse4ym7sit@vireshk-i7>
From:   Frank Rowand <frowand.list@gmail.com>
Message-ID: <6f093bb1-1a80-a906-fb4c-3f6fdeed4838@gmail.com>
Date:   Fri, 12 Mar 2021 23:11:25 -0600
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.10.0
MIME-Version: 1.0
In-Reply-To: <20210312071325.zosmlttse4ym7sit@vireshk-i7>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 3/12/21 1:13 AM, Viresh Kumar wrote:
> On 12-03-21, 01:09, Frank Rowand wrote:
>> I suggested having the .dtso files include the .dts file because that is a relatively
>> small and easy change to test.  What would probably make more sense is the rename
>> the existing overlay .dts files to be .dtso files and then for each overlay .dtso
>> file create a new .dts file that #includes the corresponding .dtso file.  This is
>> more work and churn, but easier to document that the .dts files are a hack that is
>> needed so that the corresponding .dtb.S files will be generated.
> 
> What about creating links instead then ?
> 

I don't really like the idea of using links here.

Maybe it is best to make the changes needed to allow the unittest
overlays to be .dtso instead of .dts.

Off the top of my head:

  scripts/Makefile.lib:
     The rule for %.dtb.S invokes cmd_dt_S_dtb, which puts the
     overlay data in section .dtb.init.rodata, with a label
     pointing to the beginning of the overlay __dtb_XXX_begin and
     a label pointing to the end of the overlay __dtb_XXX_end,
     for the overlay named XXX.  I _think_ that you could simply
     add a corresponding rule for %.dtbo.S using a new command
     cmd_dt_S_dtbo (the same as cmd_dt_S_dtb, except use labels
     __dtbo_XXX_begin and __dtbo_XXX_end).

  drivers/of/unittest.o:
     would need to have the #define of OVERLAY_INFO() changed to
     reflect the changed label names (use __dtbo_##overlayname##begin
     and __dtb_##overlay_name##_end).

  drivers/of/unittest-data/Makefile:
     In obj-$(CONFIG_OF_OVERLAY) change the *.dtb.o names to *.dtbo.o

     I'm not sure how the DTC_FLAGS_... += -@ differentiates between
     .dts / .dtb and .dtso / .dtbo  That is worth looking at.

-Frank