of/fdt: export fdt blob as /sys/firmware/fdt

Create a new /sys entry '/sys/firmware/fdt' to export the FDT blob that was passed to the kernel by the bootloader. This allows userland applications such as kexec to access the raw binary.

The fact that this node does not reside under /sys/firmware/device-tree is deliberate: FDT is also used on arm64 UEFI/ACPI systems to communicate just the UEFI and ACPI entry points, but the FDT is never unflattened and used to configure the system.

A CRC32 checksum is calculated over the entire FDT blob, and verified at late_initcall time. The sysfs entry is instantiated only if the checksum is valid, i.e., if the FDT blob has not been modified in the mean time. Otherwise, a warning is printed.

Signed-off-by Ard Biesheuvel <[email protected]>

---

v5.5: based on Grant's changes
"""
Actually, I made a couple of changes when I merged it. I removed the older debugfs interface since it overlaps, and I added tests for initial_boot_params to make sure it doesn't try to run on an invalid FDT
"""
but I removed the second fdt_check_header() again in the late init code, as it would result in a corrupted FDT to be silently ignored. Note that initial_boot_params can only be set if fdt_check_header() succeeded the first time, so a failure occurring the second time should produce a warning, and the CRC check will catch it anyway. The CRC check itself is fixed to use the API as it is supposed to. I also moved the CRC variable definition inside the #ifdef OF_EARLY_FLATTREE region.

v4: use pr_warn() instead of WARN()
v3: keep checksum instead of copying the entire blob, and WARN on mismatch

 drivers/of/Kconfig |  1 +
 drivers/of/fdt.c   | 43 +++++++++++++++++++++++++++----------------
 2 files changed, 28 insertions(+), 16 deletions(-)

drivers/of/Kconfig +1 -0

index 1a13f5b722c5..0348c208343c 100644
@@ -23,6 +23,7 @@ config OF_FLATTREE
 	bool
 	select DTC
 	select LIBFDT
+	select CRC32
 
 config OF_EARLY_FLATTREE
 	bool

drivers/of/fdt.c +27 -16

index d1ffca8b34ea..2bbda0775f57 100644
@@ -9,6 +9,7 @@
  * version 2 as published by the Free Software Foundation.
  */
 
+#include <linux/crc32.h>
 #include <linux/kernel.h>
 #include <linux/initrd.h>
 #include <linux/memblock.h>

@@ -22,6 +23,7 @@
 #include <linux/libfdt.h>
 #include <linux/debugfs.h>
 #include <linux/serial_core.h>
+#include <linux/sysfs.h>
 
 #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */
 #include <asm/page.h>

@@ -425,6 +427,8 @@ void *initial_boot_params;
 
 #ifdef CONFIG_OF_EARLY_FLATTREE
 
+static u32 of_fdt_crc32;
+
 /**
  * res_mem_reserve_reg() - reserve all memory described in 'reg' property
  */

@@ -996,6 +1000,8 @@ bool __init early_init_dt_verify(void *params)
 
 	/* Setup flat device-tree pointer */
 	initial_boot_params = params;
+	of_fdt_crc32 = crc32_be(~0, initial_boot_params,
+				fdt_totalsize(initial_boot_params));
 
 	/* check device tree validity */
 	if (fdt_check_header(params)) {

@@ -1080,27 +1086,32 @@ void __init unflatten_and_copy_device_tree(void)
 	unflatten_device_tree();
 }
 
-#if defined(CONFIG_DEBUG_FS) && defined(DEBUG)
-static struct debugfs_blob_wrapper flat_dt_blob;
-
-static int __init of_flat_dt_debugfs_export_fdt(void)
+#ifdef CONFIG_SYSFS
+static ssize_t of_fdt_raw_read(struct file *filp, struct kobject *kobj,
+			       struct bin_attribute *bin_attr,
+			       char *buf, loff_t off, size_t count)
 {
-	struct dentry *d = debugfs_create_dir("device-tree", NULL);
-
-	if (!d)
-		return -ENOENT;
+	memcpy(buf, initial_boot_params + off, count);
+	return count;
+}
 
-	flat_dt_blob.data = initial_boot_params;
-	flat_dt_blob.size = fdt_totalsize(initial_boot_params);
+static int __init of_fdt_raw_init(void)
+{
+	static struct bin_attribute of_fdt_raw_attr =
+		__BIN_ATTR(fdt, S_IRUSR, of_fdt_raw_read, NULL, 0);
 
-	d = debugfs_create_blob("flat-device-tree", S_IFREG | S_IRUSR,
-				d, &flat_dt_blob);
-	if (!d)
-		return -ENOENT;
+	if (!initial_boot_params)
+		return 0;
 
-	return 0;
+	if (of_fdt_crc32 != crc32_be(~0, initial_boot_params,
+				     fdt_totalsize(initial_boot_params))) {
+		pr_warn("fdt: not creating '/sys/firmware/fdt': CRC check failed\n");
+		return 0;
+	}
+	of_fdt_raw_attr.size = fdt_totalsize(initial_boot_params);
+	return sysfs_create_bin_file(firmware_kobj, &of_fdt_raw_attr);
 }
-module_init(of_flat_dt_debugfs_export_fdt);
+late_initcall(of_fdt_raw_init);
 #endif
 
 #endif /* CONFIG_OF_EARLY_FLATTREE */

Signature · 1 line

1.8.3.2

Quoted 9 lines · Ard Biesheuvel

Create a new /sys entry '/sys/firmware/fdt' to export the FDT blob that was passed to the kernel by the bootloader. This allows userland applications such as kexec to access the raw binary.

The fact that this node does not reside under /sys/firmware/device-tree is deliberate: FDT is also used on arm64 UEFI/ACPI systems to communicate just the UEFI and ACPI entry points, but the FDT is never unflattened and used to configure the system.

Quoted 6 lines

A CRC32 checksum is calculated over the entire FDT blob, and verified at late_initcall time. The sysfs entry is instantiated only if the checksum is valid, i.e., if the FDT blob has not been modified in the mean time. Otherwise, a warning is printed.

Signed-off-by Ard Biesheuvel <[email protected]>

So, I have a doubt...

When this patch is merged, we have two separate ABIs for extracting the DT from the kernel. /proc/device-tree and /sys/firmware/fdt. Merging it into mainline means we're committed to supporting both interfaces.

The whole purpose of this patch is to support kexec. Specifically, kexec on ACPI platforms that don't currently unflatten the tree.* Would it not be better to always unflatten the tree and have only one interface that kexec needs to use to obtain the tree?

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

g.

(side comment: I just realized that if I do merge this patch, it needs to include documentation of the ABI in /Documentation/ABI/*/sysfs-8)

Quoted 126 lines

---

v5.5: based on Grant's changes
"""
Actually, I made a couple of changes when I merged it. I removed the older debugfs interface since it overlaps, and I added tests for initial_boot_params to make sure it doesn't try to run on an invalid FDT
"""
but I removed the second fdt_check_header() again in the late init code, as it would result in a corrupted FDT to be silently ignored. Note that initial_boot_params can only be set if fdt_check_header() succeeded the first time, so a failure occurring the second time should produce a warning, and the CRC check will catch it anyway. The CRC check itself is fixed to use the API as it is supposed to. I also moved the CRC variable definition inside the #ifdef OF_EARLY_FLATTREE region.

v4: use pr_warn() instead of WARN()
v3: keep checksum instead of copying the entire blob, and WARN on mismatch

 drivers/of/Kconfig |  1 +
 drivers/of/fdt.c   | 43 +++++++++++++++++++++++++++----------------
 2 files changed, 28 insertions(+), 16 deletions(-)

drivers/of/Kconfig +1 -0

index 1a13f5b722c5..0348c208343c 100644
@@ -23,6 +23,7 @@ config OF_FLATTREE
 	bool
 	select DTC
 	select LIBFDT
+	select CRC32
 
 config OF_EARLY_FLATTREE
 	bool

drivers/of/fdt.c +27 -16

index d1ffca8b34ea..2bbda0775f57 100644
@@ -9,6 +9,7 @@
  * version 2 as published by the Free Software Foundation.
  */
 
+#include <linux/crc32.h>
 #include <linux/kernel.h>
 #include <linux/initrd.h>
 #include <linux/memblock.h>

@@ -22,6 +23,7 @@
 #include <linux/libfdt.h>
 #include <linux/debugfs.h>
 #include <linux/serial_core.h>
+#include <linux/sysfs.h>
 
 #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */
 #include <asm/page.h>

@@ -425,6 +427,8 @@ void *initial_boot_params;
 
 #ifdef CONFIG_OF_EARLY_FLATTREE
 
+static u32 of_fdt_crc32;
+
 /**
  * res_mem_reserve_reg() - reserve all memory described in 'reg' property
  */

@@ -996,6 +1000,8 @@ bool __init early_init_dt_verify(void *params)
 
 	/* Setup flat device-tree pointer */
 	initial_boot_params = params;
+	of_fdt_crc32 = crc32_be(~0, initial_boot_params,
+				fdt_totalsize(initial_boot_params));
 
 	/* check device tree validity */
 	if (fdt_check_header(params)) {

@@ -1080,27 +1086,32 @@ void __init unflatten_and_copy_device_tree(void)
 	unflatten_device_tree();
 }
 
-#if defined(CONFIG_DEBUG_FS) && defined(DEBUG)
-static struct debugfs_blob_wrapper flat_dt_blob;
-
-static int __init of_flat_dt_debugfs_export_fdt(void)
+#ifdef CONFIG_SYSFS
+static ssize_t of_fdt_raw_read(struct file *filp, struct kobject *kobj,
+			       struct bin_attribute *bin_attr,
+			       char *buf, loff_t off, size_t count)
 {
-	struct dentry *d = debugfs_create_dir("device-tree", NULL);
-
-	if (!d)
-		return -ENOENT;
+	memcpy(buf, initial_boot_params + off, count);
+	return count;
+}
 
-	flat_dt_blob.data = initial_boot_params;
-	flat_dt_blob.size = fdt_totalsize(initial_boot_params);
+static int __init of_fdt_raw_init(void)
+{
+	static struct bin_attribute of_fdt_raw_attr =
+		__BIN_ATTR(fdt, S_IRUSR, of_fdt_raw_read, NULL, 0);
 
-	d = debugfs_create_blob("flat-device-tree", S_IFREG | S_IRUSR,
-				d, &flat_dt_blob);
-	if (!d)
-		return -ENOENT;
+	if (!initial_boot_params)
+		return 0;
 
-	return 0;
+	if (of_fdt_crc32 != crc32_be(~0, initial_boot_params,
+				     fdt_totalsize(initial_boot_params))) {
+		pr_warn("fdt: not creating '/sys/firmware/fdt': CRC check failed\n");
+		return 0;
+	}
+	of_fdt_raw_attr.size = fdt_totalsize(initial_boot_params);
+	return sysfs_create_bin_file(firmware_kobj, &of_fdt_raw_attr);
 }
-module_init(of_flat_dt_debugfs_export_fdt);
+late_initcall(of_fdt_raw_init);
 #endif
 
 #endif /* CONFIG_OF_EARLY_FLATTREE */

Signature · 1 line

1.8.3.2

Quoted 21 lines · Grant Likely

Quoted 9 lines · Ard Biesheuvel

Create a new /sys entry '/sys/firmware/fdt' to export the FDT blob that was passed to the kernel by the bootloader. This allows userland applications such as kexec to access the raw binary.

The fact that this node does not reside under /sys/firmware/device-tree is deliberate: FDT is also used on arm64 UEFI/ACPI systems to communicate just the UEFI and ACPI entry points, but the FDT is never unflattened and used to configure the system.

Quoted 6 lines

A CRC32 checksum is calculated over the entire FDT blob, and verified at late_initcall time. The sysfs entry is instantiated only if the checksum is valid, i.e., if the FDT blob has not been modified in the mean time. Otherwise, a warning is printed.

Signed-off-by Ard Biesheuvel <[email protected]>

So, I have a doubt...

Quoted 8 lines

When this patch is merged, we have two separate ABIs for extracting the DT from the kernel. /proc/device-tree and /sys/firmware/fdt. Merging it into mainline means we're committed to supporting both interfaces.

The whole purpose of this patch is to support kexec. Specifically, kexec on ACPI platforms that don't currently unflatten the tree.* Would it not be better to always unflatten the tree and have only one interface that kexec needs to use to obtain the tree?

The whole reasoning for not unflattening the DTB in the presence of ACPI is that it's unlikely we can stop drivers using DTB information in addition to ACPI, leaving us with a completely non-standard mess. We really don't want people mixing the two, and it's going to be a struggle if the information is readily available (even preventing probing based on compatible string isn't enough because of some drivers which probe from an initcall or look up magic paths).

So if we were to unflatten the tree in the ACPI case, at a minimum it needs to be unflattened to a separate root that only exists for sysfs, and is not at exposed to the mercy of the rest of the kernel.

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Can't we expose the header fields under something like /sys/firmware/devicetree/dtb-header/, parallel to the usual /sys/firmware/devicetree/base for nodes?

Quoted 7 lines · Mark Rutland

Quoted 6 lines · Grant Likely

Quoted 3 lines · Ard Biesheuvel

Create a new /sys entry '/sys/firmware/fdt' to export the FDT blob that was passed to the kernel by the bootloader. This allows userland applications such as kexec to access the raw binary.

Quoted 7 lines

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Deprecated, not removed or disabled. It will still work pretty much forever, but users should be encouraged to move to the reserve-memory tree.

Quoted 3 lines

Can't we expose the header fields under something like /sys/firmware/devicetree/dtb-header/, parallel to the usual /sys/firmware/devicetree/base for nodes?

We could do that too.

Honestly though, I'm just unsure of what the best thing to do is. If you and a few others tell me that, "no, exporting the raw dtb is the right thing to do", then I'll be okay, merge the patch and sleep properly.

Quoted 10 lines · Grant Likely

Quoted 7 lines · Mark Rutland

Quoted 6 lines · Grant Likely

Quoted 3 lines · Ard Biesheuvel

Create a new /sys entry '/sys/firmware/fdt' to export the FDT blob that was passed to the kernel by the bootloader. This allows userland applications such as kexec to access the raw binary.

[...]

Quoted 11 lines

Quoted 7 lines

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Deprecated, not removed or disabled. It will still work pretty much forever, but users should be encouraged to move to the reserve-memory tree.

I thought you had said reserve map was still the right way for memory the kernel should never touch.

Quoted 9 lines

Quoted 3 lines

Can't we expose the header fields under something like /sys/firmware/devicetree/dtb-header/, parallel to the usual /sys/firmware/devicetree/base for nodes?

We could do that too.

Honestly though, I'm just unsure of what the best thing to do is. If you and a few others tell me that, "no, exporting the raw dtb is the right thing to do", then I'll be okay, merge the patch and sleep properly.

I always sleep better when others can take the blame.

What happens when we rev the dtb format? Is the ABI the blob or the format of the blob?

I lean towards we should add this. This is providing what is "in the firmware" while /proc/devicetree provides the live tree state including overlays.

Quoted 39 lines · Rob Herring

Quoted 10 lines · Grant Likely

Quoted 7 lines · Mark Rutland

Quoted 6 lines · Grant Likely

Quoted 3 lines · Ard Biesheuvel

Create a new /sys entry '/sys/firmware/fdt' to export the FDT blob that was passed to the kernel by the bootloader. This allows userland applications such as kexec to access the raw binary.

[...]

Quoted 11 lines

Quoted 7 lines

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Deprecated, not removed or disabled. It will still work pretty much forever, but users should be encouraged to move to the reserve-memory tree.

I thought you had said reserve map was still the right way for memory the kernel should never touch.

Quoted 9 lines

Quoted 3 lines

Can't we expose the header fields under something like /sys/firmware/devicetree/dtb-header/, parallel to the usual /sys/firmware/devicetree/base for nodes?

We could do that too.

Honestly though, I'm just unsure of what the best thing to do is. If you and a few others tell me that, "no, exporting the raw dtb is the right thing to do", then I'll be okay, merge the patch and sleep properly.

Quoted 9 lines

I always sleep better when others can take the blame.

What happens when we rev the dtb format? Is the ABI the blob or the format of the blob?

I lean towards we should add this. This is providing what is "in the firmware" while /proc/devicetree provides the live tree state including overlays.

Well, my pov is that FDT != devicetree ever since we started (ab)using the FDT container format to pass just the UEFI entry points to the kernel.
The boot protocol describes what should be passed in x0, and /that/ is what we expose in /sys/firmware/fdt, regardless of how the kernel decided to configure itself.
(perhaps we need to fix the wording in the document to refer to FDT not dtb)

So that also means I don't care about memreserve vs reserved-memory or other DT specific details: as long as x0 points to something libfdt understands at boot, we expose it and not interpret it any further.

Quoted 7 lines

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Deprecated, not removed or disabled. It will still work pretty much forever, but users should be encouraged to move to the reserve-memory tree.

Quoted 8 lines · Grant Likely

Quoted 7 lines

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Quoted 3 lines

Deprecated, not removed or disabled. It will still work pretty much forever, but users should be encouraged to move to the reserve-memory tree.

I'm curious why that should be for the "OS should never touch this, here be dragons" type memory, what are the benefits of the new scheme in that case?

Quoted 8 lines · Grant Likely

Quoted 7 lines

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Quoted 3 lines

Deprecated, not removed or disabled. It will still work pretty much forever, but users should be encouraged to move to the reserve-memory tree.

I'm curious why that should be for the "OS should never touch this, here be dragons" type memory, what are the benefits of the new scheme in that case?

Hi Ian,

Quoted 13 lines · Ian Campbell

Quoted 11 lines · Grant Likely

Quoted 7 lines

Quoted 3 lines

* It also helps with exposing the reserved map to userspace, but kexec has done without that feature for years, and it is in the process of being deprecated in favour of /reserved-memory anyway.

This is the first I'd heard of the reserve map being deprecated, and we're going to have DTs with reserved map entries for a long time going forwards.

Deprecated, not removed or disabled. It will still work pretty much forever, but users should be encouraged to move to the reserve-memory tree.

Quoted 3 lines

I'm curious why that should be for the "OS should never touch this, here be dragons" type memory, what are the benefits of the new scheme in that case?

/memreserve/ device tree entries are not available in /proc/device-tree, but reserved-memory nodes are. Some solution is needed to get all reserved memory info into the dtb passed to the second stage kernel during a kexec re-boot.