[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <c7e4e7b9-c5e8-9af9-6c16-2a8e0e7e61f7@xilinx.com>
Date: Tue, 17 May 2022 22:02:04 -0700
From: Lizhi Hou <lizhi.hou@...inx.com>
To: Frank Rowand <frowand.list@...il.com>,
<devicetree@...r.kernel.org>, <robh@...nel.org>
CC: <linux-kernel@...r.kernel.org>, <yilun.xu@...el.com>,
<maxz@...inx.com>, <sonal.santan@...inx.com>, <larryliu@....com>,
<michal.simek@...inx.com>, <stefanos@...inx.com>,
<trix@...hat.com>, <mdf@...nel.org>, <dwmw2@...radead.org>,
Max Zhen <max.zhen@...inx.com>
Subject: Re: [PATCH V2 Create empty OF root 1/1] of: create empty of root
On 5/16/22 20:22, Frank Rowand wrote:
> CAUTION: This message has originated from an External Source. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.
>
>
> On 2/16/22 23:00, Lizhi Hou wrote:
>> Xilinx Alveo PCIe accelerator cards use flattened device tree to describe
>> describe apertures in its PCIe BARs. Each device tree node represents an
>> aperture and each aperture is an hardware unit which requires a driver.
>> The product detail:
>> https://www.xilinx.com/products/boards-and-kits/alveo.html
>>
>> Thus a base device tree is required. Then the Alveo card driver can load
>> its flattened device tree and overlay it to the base tree. However device
>> tree does not always exist. To resolve this, add OF_EMPTY_ROOT config.
>> When it is selected and there is not a device tree, create an empty device
>> tree root node.
>>
> Please run scripts/get_maintainer on your patches to see who to put in
> the distribution list.
>
> I recently stumbled across this patch series and the previous related
> patch series (currently up to
> "[PATCH V3 XRT Alveo Infrastructure 0/8] XRT Alveo driver infrastructure overview")
> when I noticed it in an email archive.
Sorry about this. We will use scripts/get_maintainer to make sure the
required maintainers are included next time.
>
> A similar need of creating an of_root if otherwise empty is being discussed
> in the thread "[PATCH 0/3] add dynamic PCI device of_node creation for overlay"
> at https://lore.kernel.org/r/20220427094502.456111-1-clement.leger@bootlin.com
Thanks a lot for your information. We investigated on adding dt node for
PCI device as well and got some questions.
https://lore.kernel.org/lkml/79e4c876-e5a4-861f-cfbc-c75ed1a07ddd@xilinx.com/#t
Thanks,
Lizhi
>
> -Frank
>
>> Signed-off-by: Sonal Santan <sonal.santan@...inx.com>
>> Signed-off-by: Max Zhen <max.zhen@...inx.com>
>> Signed-off-by: Lizhi Hou <lizhi.hou@...inx.com>
>> ---
>> drivers/of/Kconfig | 5 ++++
>> drivers/of/Makefile | 1 +
>> drivers/of/of_empty_root.c | 51 ++++++++++++++++++++++++++++++++++++++
>> 3 files changed, 57 insertions(+)
>> create mode 100644 drivers/of/of_empty_root.c
>>
>> diff --git a/drivers/of/Kconfig b/drivers/of/Kconfig
>> index 80b5fd44ab1c..452d2316b47b 100644
>> --- a/drivers/of/Kconfig
>> +++ b/drivers/of/Kconfig
>> @@ -94,4 +94,9 @@ config OF_DMA_DEFAULT_COHERENT
>> # arches should select this if DMA is coherent by default for OF devices
>> bool
>>
>> +config OF_EMPTY_ROOT
>> + # driver which requires at least the empty base device tree to
>> + # overlay its own device nodes should select this.
>> + bool
>> +
>> endif # OF
>> diff --git a/drivers/of/Makefile b/drivers/of/Makefile
>> index e0360a44306e..c65364f32935 100644
>> --- a/drivers/of/Makefile
>> +++ b/drivers/of/Makefile
>> @@ -12,6 +12,7 @@ obj-$(CONFIG_OF_RESERVED_MEM) += of_reserved_mem.o
>> obj-$(CONFIG_OF_RESOLVE) += resolver.o
>> obj-$(CONFIG_OF_OVERLAY) += overlay.o
>> obj-$(CONFIG_OF_NUMA) += of_numa.o
>> +obj-$(CONFIG_OF_EMPTY_ROOT) += of_empty_root.o
>>
>> ifdef CONFIG_KEXEC_FILE
>> ifdef CONFIG_OF_FLATTREE
>> diff --git a/drivers/of/of_empty_root.c b/drivers/of/of_empty_root.c
>> new file mode 100644
>> index 000000000000..5c429c7a27bd
>> --- /dev/null
>> +++ b/drivers/of/of_empty_root.c
>> @@ -0,0 +1,51 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * Copyright (C) 2022 Xilinx, Inc.
>> + */
>> +
>> +#include <linux/of.h>
>> +#include <linux/slab.h>
>> +
>> +#include "of_private.h"
>> +
>> +static int __init of_root_init(void)
>> +{
>> + struct property *prop = NULL;
>> + struct device_node *node;
>> + __be32 *val = NULL;
>> +
>> + if (of_root)
>> + return 0;
>> +
>> + pr_info("Create empty OF root node\n");
>> + node = kzalloc(sizeof(*node), GFP_KERNEL);
>> + if (!node)
>> + return -ENOMEM;
>> + of_node_init(node);
>> + node->full_name = "/";
>> +
>> + prop = kcalloc(2, sizeof(*prop), GFP_KERNEL);
>> + if (!prop)
>> + return -ENOMEM;
>> +
>> + val = kzalloc(sizeof(*val), GFP_KERNEL);
>> + if (!val)
>> + return -ENOMEM;
>> + *val = cpu_to_be32(sizeof(void *) / sizeof(u32));
>> +
>> + prop->name = "#address-cells";
>> + prop->value = val;
>> + prop->length = sizeof(u32);
>> + of_add_property(node, prop);
>> + prop++;
>> + prop->name = "#size-cells";
>> + prop->value = val;
>> + prop->length = sizeof(u32);
>> + of_add_property(node, prop);
>> + of_root = node;
>> + for_each_of_allnodes(node)
>> + __of_attach_node_sysfs(node);
>> +
>> + return 0;
>> +}
>> +pure_initcall(of_root_init);
Powered by blists - more mailing lists