Re: [PATCH 1/3] of: always populate a root node

From: Frank Rowand <frowand.list@gmail.com>
To: "Rob Herring" <robh@kernel.org>,
	"Clément Léger" <clement.leger@bootlin.com>
Cc: Pantelis Antoniou <pantelis.antoniou@konsulko.com>,
	Bjorn Helgaas <bhelgaas@google.com>,
	Allan Nielsen <allan.nielsen@microchip.com>,
	Horatiu Vultur <horatiu.vultur@microchip.com>,
	Steen Hegelund <steen.hegelund@microchip.com>,
	Thomas Petazzoni <thomas.petazonni@bootlin.com>,
	Alexandre Belloni <alexandre.belloni@bootlin.com>,
	Mark Brown <broonie@kernel.org>,
	Andy Shevchenko <andriy.shevchenko@linux.intel.com>,
	Jakub Kicinski <kuba@kernel.org>,
	Hans de Goede <hdegoede@redhat.com>, Andrew Lunn <andrew@lunn.ch>,
	devicetree@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-pci@vger.kernel.org
Subject: Re: [PATCH 1/3] of: always populate a root node
Date: Mon, 16 May 2022 23:11:03 -0400	[thread overview]
Message-ID: <d356acbe-daff-1c66-6511-aab97a171c82@gmail.com> (raw)
In-Reply-To: <YnEx5/ni1ddIFCj9@robh.at.kernel.org>

On 5/3/22 08:45, Rob Herring wrote:
> On Wed, Apr 27, 2022 at 11:45:00AM +0200, Clément Léger wrote:
>> When enabling CONFIG_OF on a platform where of_root is not populated by
>> firmware, we end up without a root node. In order to apply overlays and
>> create subnodes of the root node, we need one. This commit creates an
>> empty root node if not present.
> 
> The existing unittest essentially does the same thing for running the 
> tests on non-DT systems. It should be modified to use this support 
> instead. Maybe that's just removing the unittest code that set of_root.
> 
> I expect Frank will have some comments.

My preference would be for unflatten_and_copy_device_tree() to
use a compiled in FDT that only contains a root node, in the
case that no valid device tree is found (in other words,
"if (!initial_boot_params)".

unflatten_and_copy_device_tree() calls unittest_unflatten_overlay_base()
after unflattening the device tree passed into the booting kernel.  This
step is needed for a specific portion of the unittests.

I'm still looking at the bigger picture of using overlays for the PCIe
card, so more comments will be coming about that bigger picture.

-Frank

> 
>> Co-developed-by: Rob Herring <robh@kernel.org>
>> Signed-off-by: Rob Herring <robh@kernel.org>
>> Signed-off-by: Clément Léger <clement.leger@bootlin.com>
>> ---
>>  drivers/of/base.c | 16 ++++++++++++++--
>>  1 file changed, 14 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/of/base.c b/drivers/of/base.c
>> index e7d92b67cb8a..6b8584c39f73 100644
>> --- a/drivers/of/base.c
>> +++ b/drivers/of/base.c
>> @@ -177,6 +177,19 @@ void __init of_core_init(void)
>>  		pr_err("failed to register existing nodes\n");
>>  		return;
>>  	}
>> +
>> +	if (!of_root) {
>> +		of_root = kzalloc(sizeof(*of_root), GFP_KERNEL);
>> +		if (!of_root) {
>> +			mutex_unlock(&of_mutex);
>> +			pr_err("failed to create root node\n");
>> +			return;
>> +		}
>> +
>> +		of_root->full_name = "/";
>> +		of_node_init(of_root);
>> +	}
>> +
>>  	for_each_of_allnodes(np) {
>>  		__of_attach_node_sysfs(np);
>>  		if (np->phandle && !phandle_cache[of_phandle_cache_hash(np->phandle)])
>> @@ -185,8 +198,7 @@ void __init of_core_init(void)
>>  	mutex_unlock(&of_mutex);
>>  
>>  	/* Symlink in /proc as required by userspace ABI */
>> -	if (of_root)
>> -		proc_symlink("device-tree", NULL, "/sys/firmware/devicetree/base");
>> +	proc_symlink("device-tree", NULL, "/sys/firmware/devicetree/base");
>>  }
>>  
>>  static struct property *__of_find_property(const struct device_node *np,
>> -- 
>> 2.34.1
>>
>>