Skip to contentSkip to conversation

kernel-devicetree: allow specification of dtb directory

PATCH In review [email protected] 4 participants 9 messages May 5, 2023 – May 8, 2023 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 kernel-devicetree: allow specification of dtb directory

From: Randolph Sapp <[email protected]>

Fedora/Redhat and Arch are somewhat standardized on their dtb directory structure. Let's add some flags to configure yocto to mimic that behavior.

Add the following variables to the kernel class:
- KERNEL_DTBDEST (controls the destination directory for dtbs)
- KERNEL_DTBVENDORED (controls if vendor subdirectories are to be respected)

Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST and KERNEL_DTBVENDORED is expected to be "true"/"false". This only applies to the package directory structure. The deploydir structure is purposely left untouched for compatibility with existing recipes.

By default this is configured to behave the same as the current recipe and produce a flat dtb directory at KERNEL_IMAGEDEST.

---

Well, suppose I was breaking things by submitting this to kirkstone first. This is just the master version of the following patchset:
https://lists.openembedded.org/g/openembedded-core/message/180754

I'd love to get that series merged as well if this patch is acceptable.

 meta/classes-recipe/kernel-devicetree.bbclass | 22 ++++++++++++++-----
 meta/classes-recipe/kernel.bbclass            |  2 ++
 2 files changed, 19 insertions(+), 5 deletions(-)
meta/classes-recipe/kernel-devicetree.bbclass +17 -5
index 4d0ecb1032..a6c6c5f227 100644
@@ -12,7 +12,12 @@ python () {
             d.appendVar("PACKAGES", " ${KERNEL_PACKAGE_NAME}-image-zimage-bundle")
 }
 
-FILES:${KERNEL_PACKAGE_NAME}-devicetree = "/${KERNEL_IMAGEDEST}/*.dtb /${KERNEL_IMAGEDEST}/*.dtbo"
+FILES:${KERNEL_PACKAGE_NAME}-devicetree = " \
+    /${KERNEL_DTBDEST}/*.dtb \
+    /${KERNEL_DTBDEST}/*.dtbo \
+    /${KERNEL_DTBDEST}/*/*.dtb \
+    /${KERNEL_DTBDEST}/*/*.dtbo \
+"
 FILES:${KERNEL_PACKAGE_NAME}-image-zimage-bundle = "/${KERNEL_IMAGEDEST}/zImage-*.dtb.bin"
 
 # Generate kernel+devicetree bundle
@@ -73,12 +78,16 @@ do_compile:append() {
 }
 
 do_install:append() {
+	install -d ${D}/${KERNEL_DTBDEST}
 	for dtbf in ${KERNEL_DEVICETREE}; do
 		dtb=`normalize_dtb "$dtbf"`
-		dtb_ext=${dtb##*.}
-		dtb_base_name=`basename $dtb .$dtb_ext`
 		dtb_path=`get_real_dtb_path_in_kernel "$dtb"`
-		install -m 0644 $dtb_path ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
+		if [ ${KERNEL_DTBVENDORED} == "false" ]; then
+			dtb_ext=${dtb##*.}
+			dtb_base_name=`basename $dtb .$dtb_ext`
+			dtb=$dtb_base_name.$dtb_ext
+		fi
+		install -Dm 0644 $dtb_path ${D}/${KERNEL_DTBDEST}/$dtb
 	done
 }
 
@@ -88,7 +97,10 @@ do_deploy:append() {
 		dtb_ext=${dtb##*.}
 		dtb_base_name=`basename $dtb .$dtb_ext`
 		install -d $deployDir
-		install -m 0644 ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
+		if [ ${KERNEL_DTBVENDORED} == "false" ]; then
+			dtb=$dtb_base_name.$dtb_ext
+		fi
+		install -m 0644 ${D}/${KERNEL_DTBDEST}/$dtb $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
 		if [ "${KERNEL_IMAGETYPE_SYMLINK}" = "1" ] ; then
 			ln -sf $dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext $deployDir/$dtb_base_name.$dtb_ext
 		fi
meta/classes-recipe/kernel.bbclass +2 -0
index e634eabd49..8f022b234d 100644
@@ -215,6 +215,8 @@ KERNEL_RELEASE ?= "${KERNEL_VERSION}"
 # The directory where built kernel lies in the kernel tree
 KERNEL_OUTPUT_DIR ?= "arch/${ARCH}/boot"
 KERNEL_IMAGEDEST ?= "boot"
+KERNEL_DTBDEST ?= "${KERNEL_IMAGEDEST}"
+KERNEL_DTBVENDORED ?= "false"
 
 #
 # configuration
Signature · 1 line
2.40.1
#1 lore raw
Quoted 44 lines · Randolph Sapp via lists.openembedded.org

From: Randolph Sapp <[email protected]>

Fedora/Redhat and Arch are somewhat standardized on their dtb directory structure. Let's add some flags to configure yocto to mimic that behavior.

Add the following variables to the kernel class:
- KERNEL_DTBDEST (controls the destination directory for dtbs)
- KERNEL_DTBVENDORED (controls if vendor subdirectories are to be respected)

Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST and KERNEL_DTBVENDORED is expected to be "true"/"false". This only applies to the package directory structure. The deploydir structure is purposely left untouched for compatibility with existing recipes.

By default this is configured to behave the same as the current recipe and produce a flat dtb directory at KERNEL_IMAGEDEST.

---

Well, suppose I was breaking things by submitting this to kirkstone first. This is just the master version of the following patchset:
https://lists.openembedded.org/g/openembedded-core/message/180754

I'd love to get that series merged as well if this patch is acceptable.

 meta/classes-recipe/kernel-devicetree.bbclass | 22 ++++++++++++++-----
 meta/classes-recipe/kernel.bbclass            |  2 ++
 2 files changed, 19 insertions(+), 5 deletions(-)
meta/classes-recipe/kernel-devicetree.bbclass +0 -2
index 4d0ecb1032..a6c6c5f227 100644
...
 
 
 
-               dtb_base_name=`basename $dtb .$dtb_ext`
                dtb_path=`get_real_dtb_path_in_kernel "$dtb"`
-               install -m 0644 $dtb_path
Quoted 3 lines

${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

dash doesn't like this:

/bin/dash -c "if [ false == "false" ]; then echo foo; fi"

add quotes or use single '='.

+ dtb_ext=${dtb##*.}

Quoted 12 lines
meta/classes-recipe/kernel-devicetree.bbclass +4 -1
+                       dtb_base_name=`basename $dtb .$dtb_ext`
+                       dtb=$dtb_base_name.$dtb_ext
+               fi
+               install -Dm 0644 $dtb_path ${D}/${KERNEL_DTBDEST}/$dtb
        done
 }
@@ -88,7 +97,10 @@ do_deploy:append() {
                dtb_ext=${dtb##*.}
                dtb_base_name=`basename $dtb .$dtb_ext`
                install -d $deployDir
-               install -m 0644
Quoted 4 lines

${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

Same here

Quoted 9 lines
meta/classes-recipe/kernel-devicetree.bbclass +3 -0
+                       dtb=$dtb_base_name.$dtb_ext
+               fi
+               install -m 0644 ${D}/${KERNEL_DTBDEST}/$dtb

$deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext

                if [ "${KERNEL_IMAGETYPE_SYMLINK}" = "1" ] ; then
                        ln -sf $dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext

$deployDir/$dtb_base_name.$dtb_ext
fi

#2 lore raw

[ Outlook does not support commenting inline for HTML mails, thus I’m top posting…]

The problem is ==, which is a bashism. POSIX shells (like dash) only support =. When it comes to quoting, you typically want to quote shell variables in tests in case they are empty. OTOH, static strings without whitespace or other special characters do not need quoting. So the correct way to write the if statements is:

               if [ "${KERNEL_DTBVENDORED}" = false ]; then

(In this case it is actually a bitbake variable being quoted, but unless you can guarantee it is not empty, the same rule applies.)

//Peter

From: [email protected] <[email protected]> On Behalf Of Martin Jansa
Sent: den 6 maj 2023 11:10
To: [email protected]
Cc: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

On Fri, May 5, 2023 at 6:38 PM Randolph Sapp via lists.openembedded.org<http://lists.openembedded.org> <[email protected]<mailto:[email protected]>> wrote:
From: Randolph Sapp <[email protected]<mailto:[email protected]>>

Fedora/Redhat and Arch are somewhat standardized on their dtb directory structure. Let's add some flags to configure yocto to mimic that behavior.

Add the following variables to the kernel class:
- KERNEL_DTBDEST (controls the destination directory for dtbs)
- KERNEL_DTBVENDORED (controls if vendor subdirectories are to be respected)

Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST and KERNEL_DTBVENDORED is expected to be "true"/"false". This only applies to the package directory structure. The deploydir structure is purposely left untouched for compatibility with existing recipes.

By default this is configured to behave the same as the current recipe and produce a flat dtb directory at KERNEL_IMAGEDEST.

---

Well, suppose I was breaking things by submitting this to kirkstone first. This is just the master version of the following patchset:
https://lists.openembedded.org/g/openembedded-core/message/180754

I'd love to get that series merged as well if this patch is acceptable.

 meta/classes-recipe/kernel-devicetree.bbclass | 22 ++++++++++++++-----
 meta/classes-recipe/kernel.bbclass            |  2 ++
 2 files changed, 19 insertions(+), 5 deletions(-)
meta/classes-recipe/kernel-devicetree.bbclass +0 -0
index 4d0ecb1032..a6c6c5f227 100644

...

- dtb_base_name=basename $dtb .$dtb_ext

                dtb_path=`get_real_dtb_path_in_kernel "$dtb"`

- install -m 0644 $dtb_path ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

dash doesn't like this:

/bin/dash -c "if [ false == "false" ]; then echo foo; fi"

add quotes or use single '='.

+                       dtb_ext=${dtb##*.}
+                       dtb_base_name=`basename $dtb .$dtb_ext`
+                       dtb=$dtb_base_name.$dtb_ext

+ fi
+ install -Dm 0644 $dtb_path ${D}/${KERNEL_DTBDEST}/$dtb
done
}

@@ -88,7 +97,10 @@ do_deploy:append() {
dtb_ext=${dtb##*.} dtb_base_name=basename $dtb .$dtb_ext install -d $deployDir
- install -m 0644 ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

Same here

+ dtb=$dtb_base_name.$dtb_ext
+ fi
+ install -m 0644 ${D}/${KERNEL_DTBDEST}/$dtb $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext

                if [ "${KERNEL_IMAGETYPE_SYMLINK}" = "1" ] ; then
                        ln -sf $dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext $deployDir/$dtb_base_name.$dtb_ext
                fi
#3 lore raw

or
if "${@'true' if oe.types.boolean(d.getVar('KERNEL_DTBVENDORED')) else 'false'}"; then
to avoid test as well (like systemd recipes is using) and I did here as well in:
https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757

Quoted 134 lines · Peter Kjellerstedt

[ Outlook does not support commenting inline for HTML mails, thus I’m top posting…]

The problem is ==, which is a bashism. POSIX shells (like dash) only support =. When it comes to quoting, you typically want to quote shell variables in tests in case they are empty. OTOH, static strings without whitespace or other special characters do not need quoting. So the correct way to write the if statements is:

               if [ "${KERNEL_DTBVENDORED}" = false ]; then

(In this case it is actually a bitbake variable being quoted, but unless you can guarantee it is not empty, the same rule applies.)

//Peter

*From:* [email protected] < [email protected]> *On Behalf Of *Martin Jansa *Sent:* den 6 maj 2023 11:10
*To:* [email protected]
*Cc:* [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected] *Subject:* Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

On Fri, May 5, 2023 at 6:38 PM Randolph Sapp via lists.openembedded.org <[email protected]> wrote:

From: Randolph Sapp <[email protected]>

Fedora/Redhat and Arch are somewhat standardized on their dtb directory structure. Let's add some flags to configure yocto to mimic that behavior.

Add the following variables to the kernel class:
- KERNEL_DTBDEST (controls the destination directory for dtbs)
- KERNEL_DTBVENDORED (controls if vendor subdirectories are to be respected)

Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST and KERNEL_DTBVENDORED is expected to be "true"/"false". This only applies to the package directory structure. The deploydir structure is purposely left untouched for compatibility with existing recipes.

By default this is configured to behave the same as the current recipe and produce a flat dtb directory at KERNEL_IMAGEDEST.

---

Well, suppose I was breaking things by submitting this to kirkstone first. This is just the master version of the following patchset:
https://lists.openembedded.org/g/openembedded-core/message/180754

I'd love to get that series merged as well if this patch is acceptable.

 meta/classes-recipe/kernel-devicetree.bbclass | 22 ++++++++++++++-----
 meta/classes-recipe/kernel.bbclass            |  2 ++
 2 files changed, 19 insertions(+), 5 deletions(-)
meta/classes-recipe/kernel-devicetree.bbclass +0 -2
index 4d0ecb1032..a6c6c5f227 100644
...
 
 
 
-               dtb_base_name=`basename $dtb .$dtb_ext`
                dtb_path=`get_real_dtb_path_in_kernel "$dtb"`
-               install -m 0644 $dtb_path

${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

dash doesn't like this:

/bin/dash -c "if [ false == "false" ]; then echo foo; fi"

add quotes or use single '='.

meta/classes-recipe/kernel-devicetree.bbclass +5 -1
+                       dtb_ext=${dtb##*.}
+                       dtb_base_name=`basename $dtb .$dtb_ext`
+                       dtb=$dtb_base_name.$dtb_ext
+               fi
+               install -Dm 0644 $dtb_path ${D}/${KERNEL_DTBDEST}/$dtb
        done
 }
@@ -88,7 +97,10 @@ do_deploy:append() {
                dtb_ext=${dtb##*.}
                dtb_base_name=`basename $dtb .$dtb_ext`
                install -d $deployDir
-               install -m 0644

${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

Same here

meta/classes-recipe/kernel-devicetree.bbclass +3 -0
+                       dtb=$dtb_base_name.$dtb_ext
+               fi
+               install -m 0644 ${D}/${KERNEL_DTBDEST}/$dtb

$deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext

                if [ "${KERNEL_IMAGETYPE_SYMLINK}" = "1" ] ; then
                        ln -sf $dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext

$deployDir/$dtb_base_name.$dtb_ext
fi

#4 lore raw

That is most certainly a good alternative.

//Peter

From: Martin Jansa <[email protected]>
Sent: den 8 maj 2023 00:20
To: Peter Kjellerstedt <[email protected]>
Cc: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

or
if "${@'true' if oe.types.boolean(d.getVar('KERNEL_DTBVENDORED')) else 'false'}<mailto:$%7b@'true'%20if%20oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))%20else%20'false'%7d>"; then to avoid test as well (like systemd recipes is using) and I did here as well in:
https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757

On Sun, May 7, 2023 at 10:49 PM Peter Kjellerstedt <[email protected]<mailto:[email protected]>> wrote:
[ Outlook does not support commenting inline for HTML mails, thus I’m top posting…]

The problem is ==, which is a bashism. POSIX shells (like dash) only support =. When it comes to quoting, you typically want to quote shell variables in tests in case they are empty. OTOH, static strings without whitespace or other special characters do not need quoting. So the correct way to write the if statements is:

               if [ "${KERNEL_DTBVENDORED}" = false ]; then

(In this case it is actually a bitbake variable being quoted, but unless you can guarantee it is not empty, the same rule applies.)

//Peter

From: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> On Behalf Of Martin Jansa
Sent: den 6 maj 2023 11:10
To: [email protected]<mailto:[email protected]>
Cc: [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>; [email protected]<mailto:[email protected]>
Subject: Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

On Fri, May 5, 2023 at 6:38 PM Randolph Sapp via lists.openembedded.org<http://lists.openembedded.org> <[email protected]<mailto:[email protected]>> wrote:
From: Randolph Sapp <[email protected]<mailto:[email protected]>>

Fedora/Redhat and Arch are somewhat standardized on their dtb directory structure. Let's add some flags to configure yocto to mimic that behavior.

Add the following variables to the kernel class:
- KERNEL_DTBDEST (controls the destination directory for dtbs)
- KERNEL_DTBVENDORED (controls if vendor subdirectories are to be respected)

Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST and KERNEL_DTBVENDORED is expected to be "true"/"false". This only applies to the package directory structure. The deploydir structure is purposely left untouched for compatibility with existing recipes.

By default this is configured to behave the same as the current recipe and produce a flat dtb directory at KERNEL_IMAGEDEST.

---

Well, suppose I was breaking things by submitting this to kirkstone first. This is just the master version of the following patchset:
https://lists.openembedded.org/g/openembedded-core/message/180754

I'd love to get that series merged as well if this patch is acceptable.

 meta/classes-recipe/kernel-devicetree.bbclass | 22 ++++++++++++++-----
 meta/classes-recipe/kernel.bbclass            |  2 ++
 2 files changed, 19 insertions(+), 5 deletions(-)
meta/classes-recipe/kernel-devicetree.bbclass +0 -0
index 4d0ecb1032..a6c6c5f227 100644

...

- dtb_base_name=basename $dtb .$dtb_ext

                dtb_path=`get_real_dtb_path_in_kernel "$dtb"`

- install -m 0644 $dtb_path ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

dash doesn't like this:

/bin/dash -c "if [ false == "false" ]; then echo foo; fi"

add quotes or use single '='.

+                       dtb_ext=${dtb##*.}
+                       dtb_base_name=`basename $dtb .$dtb_ext`
+                       dtb=$dtb_base_name.$dtb_ext

+ fi
+ install -Dm 0644 $dtb_path ${D}/${KERNEL_DTBDEST}/$dtb
done
}

@@ -88,7 +97,10 @@ do_deploy:append() {
dtb_ext=${dtb##*.} dtb_base_name=basename $dtb .$dtb_ext install -d $deployDir
- install -m 0644 ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
+ if [ ${KERNEL_DTBVENDORED} == "false" ]; then

Same here

+ dtb=$dtb_base_name.$dtb_ext
+ fi
+ install -m 0644 ${D}/${KERNEL_DTBDEST}/$dtb $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext

                if [ "${KERNEL_IMAGETYPE_SYMLINK}" = "1" ] ; then
                        ln -sf $dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext $deployDir/$dtb_base_name.$dtb_ext
                fi
#5 lore raw
Quoted 137 lines · Peter Kjellerstedt

That is most certainly a good alternative.

//Peter

*From:*Martin Jansa <[email protected]>
*Sent:* den 8 maj 2023 00:20
*To:* Peter Kjellerstedt <[email protected]>
*Cc:* [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected] *Subject:* Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

or

if "${@'true' if oe.types.boolean(d.getVar('KERNEL_DTBVENDORED')) else 'false'} <mailto:$%7b@'true'%20if%20oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))%20else%20'false'%7d>"; then

to avoid test as well (like systemd recipes is using) and I did here as well in:

https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757 <https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757>

On Sun, May 7, 2023 at 10:49 PM Peter Kjellerstedt <[email protected] <mailto:[email protected]>> wrote:

[ Outlook does not support commenting inline for HTML mails, thus I’m top posting…]

The problem is ==, which is a bashism. POSIX shells (like dash) only support =. When it comes to quoting, you typically want to quote shell variables in tests in case they are empty. OTOH, static strings without whitespace or other special characters do not need quoting. So the correct way to write the if statements is:

               if [ "${KERNEL_DTBVENDORED}" = false ]; then

(In this case it is actually a bitbake variable being quoted, but unless you can guarantee it is not empty, the same rule applies.)

//Peter

*From:*[email protected] <mailto:[email protected]> <[email protected] <mailto:[email protected]>> *On Behalf Of *Martin Jansa
*Sent:* den 6 maj 2023 11:10
*To:* [email protected] <mailto:[email protected]>
*Cc:* [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]> *Subject:* Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

On Fri, May 5, 2023 at 6:38 PM Randolph Sapp via lists.openembedded.org <http://lists.openembedded.org>
<[email protected] <mailto:[email protected]>> wrote:

From: Randolph Sapp <[email protected] <mailto:[email protected]>>

Fedora/Redhat and Arch are somewhat standardized on their dtb directory
structure. Let's add some flags to configure yocto to mimic that behavior.

Add the following variables to the kernel class:
        - KERNEL_DTBDEST (controls the destination directory
for dtbs)
        - KERNEL_DTBVENDORED (controls if vendor subdirectories
are to
          be respected)

Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST
and KERNEL_DTBVENDORED is expected to be "true"/"false". This only applies to the package directory structure. The deploydir structure is
purposely left untouched for compatibility with existing recipes.

By default this is configured to behave the same as the current recipe
and produce a flat dtb directory at KERNEL_IMAGEDEST.

Signed-off-by Randolph Sapp <[email protected] <mailto:[email protected]>>

---

Well, suppose I was breaking things by submitting this to kirkstone
first. This is just the master version of the following patchset:
https://lists.openembedded.org/g/openembedded-core/message/180754 <https://lists.openembedded.org/g/openembedded-core/message/180754>

I'd love to get that series merged as well if this patch is acceptable.

 meta/classes-recipe/kernel-devicetree.bbclass | 22
++++++++++++++-----
 meta/classes-recipe/kernel.bbclass            |  2 ++ 2 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/meta/classes-recipe/kernel-devicetree.bbclass b/meta/classes-recipe/kernel-devicetree.bbclass index 4d0ecb1032..a6c6c5f227 100644
--- a/meta/classes-recipe/kernel-devicetree.bbclass +++ b/meta/classes-recipe/kernel-devicetree.bbclass
...

-               dtb_base_name=basename $dtb .$dtb_ext
                dtb_path=get_real_dtb_path_in_kernel "$dtb"
-               install -m 0644 $dtb_path ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
+               if [ ${KERNEL_DTBVENDORED} == "false" ]; then

dash doesn't like this:

 /bin/dash -c "if [ false == "false" ]; then echo foo; fi"

add quotes or use single '='.

+                       dtb_ext=${dtb##*.}
+                       dtb_base_name=basename $dtb .$dtb_ext +                       dtb=$dtb_base_name.$dtb_ext +               fi
+               install -Dm 0644 $dtb_path ${D}/${KERNEL_DTBDEST}/$dtb
        done
 }

@@ -88,7 +97,10 @@ do_deploy:append() {
                dtb_ext=${dtb##*.} dtb_base_name=basename $dtb .$dtb_ext install -d $deployDir
-               install -m 0644 ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext +               if [ ${KERNEL_DTBVENDORED} == "false" ]; then

Same here

Quoted 10 lines

+                       dtb=$dtb_base_name.$dtb_ext
+               fi
+               install -m 0644 ${D}/${KERNEL_DTBDEST}/$dtb $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
                if [ "${KERNEL_IMAGETYPE_SYMLINK}" = "1" ] ; then ln -sf
$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext $deployDir/$dtb_base_name.$dtb_ext
                fi

Ah, my bad. I need to figure out a way to get shellcheck to run on bash blocks in bb recipes. Would save everyone some time. I'll submit a V2 with your modifications Martin!

#6 lore raw
Quoted 157 lines · Randolph Sapp
Quoted 146 lines · Peter Kjellerstedt

That is most certainly a good alternative.

//Peter

*From:*Martin Jansa <[email protected]>
*Sent:* den 8 maj 2023 00:20
*To:* Peter Kjellerstedt <[email protected]> *Cc:* [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
*Subject:* Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

or

if "${@'true' if oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))
else 'false'} <mailto:$%7b@'true'%20if%20oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))%20else%20'false'%7d>"; then

to avoid test as well (like systemd recipes is using) and I did here as well in:

https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757 <https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757>

On Sun, May 7, 2023 at 10:49 PM Peter Kjellerstedt <[email protected] <mailto:[email protected]>> wrote:

[ Outlook does not support commenting inline for HTML mails, thus I’m top posting…]

The problem is ==, which is a bashism. POSIX shells (like dash) only support =. When it comes to quoting, you typically want to quote shell variables in tests in case they are empty. OTOH, static strings without whitespace or other special characters do not need quoting. So the correct way to write the if statements is:

                   if [ "${KERNEL_DTBVENDORED}" = false ]; then

(In this case it is actually a bitbake variable being quoted, but unless you can guarantee it is not empty, the same rule applies.)

//Peter

*From:*[email protected] <mailto:[email protected]> <[email protected] <mailto:[email protected]>> *On Behalf Of *Martin Jansa
*Sent:* den 6 maj 2023 11:10
*To:* [email protected] <mailto:[email protected]>
*Cc:* [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]>; [email protected] <mailto:[email protected]> *Subject:* Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

On Fri, May 5, 2023 at 6:38 PM Randolph Sapp via lists.openembedded.org <http://lists.openembedded.org>
<[email protected] <mailto:[email protected]>> wrote:

From: Randolph Sapp <[email protected] <mailto:[email protected]>>

Fedora/Redhat and Arch are somewhat standardized on their dtb directory
structure. Let's add some flags to configure yocto to mimic that behavior.

Add the following variables to the kernel class:
        - KERNEL_DTBDEST (controls the destination directory
for dtbs)
        - KERNEL_DTBVENDORED (controls if vendor subdirectories
are to
          be respected)

Currently KERNEL_DTBDEST is expected to be a subdir of KERNEL_IMAGEDEST
and KERNEL_DTBVENDORED is expected to be "true"/"false". This only applies to the package directory structure. The deploydir structure is
purposely left untouched for compatibility with existing recipes.

By default this is configured to behave the same as the current recipe
and produce a flat dtb directory at KERNEL_IMAGEDEST.

Signed-off-by Randolph Sapp <[email protected] <mailto:[email protected]>>

---

Well, suppose I was breaking things by submitting this to kirkstone
first. This is just the master version of the following patchset:
https://lists.openembedded.org/g/openembedded-core/message/180754 <https://lists.openembedded.org/g/openembedded-core/message/180754>

I'd love to get that series merged as well if this patch is acceptable.

 meta/classes-recipe/kernel-devicetree.bbclass | 22
++++++++++++++-----
 meta/classes-recipe/kernel.bbclass            |  2 ++ 2 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/meta/classes-recipe/kernel-devicetree.bbclass b/meta/classes-recipe/kernel-devicetree.bbclass index 4d0ecb1032..a6c6c5f227 100644
--- a/meta/classes-recipe/kernel-devicetree.bbclass +++ b/meta/classes-recipe/kernel-devicetree.bbclass
...

-               dtb_base_name=basename $dtb .$dtb_ext
                dtb_path=get_real_dtb_path_in_kernel "$dtb"
-               install -m 0644 $dtb_path ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
+               if [ ${KERNEL_DTBVENDORED} == "false" ]; then

   dash doesn't like this:

     /bin/dash -c "if [ false == "false" ]; then echo foo; fi"

   add quotes or use single '='.

       +                       dtb_ext=${dtb##*.}
       +                       dtb_base_name=`basename $dtb .$dtb_ext`
       +                       dtb=$dtb_base_name.$dtb_ext
       +               fi
       +               install -Dm 0644 $dtb_path
       ${D}/${KERNEL_DTBDEST}/$dtb
                done
         }

       @@ -88,7 +97,10 @@ do_deploy:append() {
                        dtb_ext=${dtb##*.}
                        dtb_base_name=`basename $dtb .$dtb_ext`
                        install -d $deployDir
       -               install -m 0644
       ${D}/${KERNEL_IMAGEDEST}/$dtb_base_name.$dtb_ext
       $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
       +               if [ ${KERNEL_DTBVENDORED} == "false" ]; then

   Same here

       +                       dtb=$dtb_base_name.$dtb_ext
Quoted 9 lines
       +               fi
       +               install -m 0644 ${D}/${KERNEL_DTBDEST}/$dtb
       $deployDir/$dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
                        if [ "${KERNEL_IMAGETYPE_SYMLINK}" = "1" ] ; then
                                ln -sf
       $dtb_base_name-${KERNEL_DTB_NAME}.$dtb_ext
       $deployDir/$dtb_base_name.$dtb_ext
                        fi
Quoted 3 lines

Ah, my bad. I need to figure out a way to get shellcheck to run on bash blocks in bb recipes. Would save everyone some time. I'll submit a V2 with your modifications Martin!

No V2 - it already got merged into master. Moreover, due to this oversight, it defaults to "vendored" DT install in some builds, where bash is not the system shell...

We should merge Martin's fix from his contrib tree ASAP.

Signature · 1 line
Denys

A message that is not in the archive (sent off-list or not yet received): [email protected]

#8 lore raw
Quoted 45 lines · Denys Dmytriyenko
Quoted 44 lines · Randolph Sapp
Quoted 43 lines · Peter Kjellerstedt

That is most certainly a good alternative.

//Peter

*From:*Martin Jansa <[email protected]>
*Sent:* den 8 maj 2023 00:20
*To:* Peter Kjellerstedt <[email protected]> *Cc:* [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
*Subject:* Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

or

if "${@'true' if oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))
else 'false'} <mailto:$%7b@'true'%20if%20oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))%20else%20'false'%7d>"; then

to avoid test as well (like systemd recipes is using) and I did here as well in:

https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757 <https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757>

On Sun, May 7, 2023 at 10:49 PM Peter Kjellerstedt <[email protected] <mailto:[email protected]>> wrote:

[ Outlook does not support commenting inline for HTML mails, thus I’m top posting…]

The problem is ==, which is a bashism. POSIX shells (like dash) only support =. When it comes to quoting, you typically want to quote shell variables in tests in case they are empty. OTOH, static strings without whitespace or other special characters do not need quoting. So the correct way to write the if statements is:

                   if [ "${KERNEL_DTBVENDORED}" = false ]; then

(In this case it is actually a bitbake variable being quoted, but unless you can guarantee it is not empty, the same rule applies.)

   //Peter

<snip>

Quoted 9 lines
Quoted 3 lines

Ah, my bad. I need to figure out a way to get shellcheck to run on bash blocks in bb recipes. Would save everyone some time. I'll submit a V2 with your modifications Martin!

No V2 - it already got merged into master. Moreover, due to this oversight, it defaults to "vendored" DT install in some builds, where bash is not the system shell...

We should merge Martin's fix from his contrib tree ASAP.

BTW, there are other nuisances with this change:

https://patchwork.yoctoproject.org/project/ti/patch/[email protected]/

Not sure how common the use of third level in KERNEL_DTBDEST dir structure is, maybe we should add a third level generic glob to OE-Core to be safe...

Signature · 1 line
Denys
#9 lore raw
Quoted 58 lines · Denys Dmytriyenko
Quoted 45 lines · Denys Dmytriyenko
Quoted 44 lines · Randolph Sapp
Quoted 43 lines · Peter Kjellerstedt

That is most certainly a good alternative.

//Peter

*From:*Martin Jansa <[email protected]>
*Sent:* den 8 maj 2023 00:20
*To:* Peter Kjellerstedt <[email protected]> *Cc:* [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
*Subject:* Re: [OE-core][PATCH] kernel-devicetree: allow specification of dtb directory

or

if "${@'true' if oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))
else 'false'} <mailto:$%7b@'true'%20if%20oe.types.boolean(d.getVar('KERNEL_DTBVENDORED'))%20else%20'false'%7d>"; then

to avoid test as well (like systemd recipes is using) and I did here as well in:

https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757 <https://git.openembedded.org/openembedded-core-contrib/commit/?h=jansa/master&id=e2a6da5202a6671113758f9746ddbd8141a75757>

On Sun, May 7, 2023 at 10:49 PM Peter Kjellerstedt <[email protected] <mailto:[email protected]>> wrote:

[ Outlook does not support commenting inline for HTML mails, thus I’m top posting…]

The problem is ==, which is a bashism. POSIX shells (like dash) only support =. When it comes to quoting, you typically want to quote shell variables in tests in case they are empty. OTOH, static strings without whitespace or other special characters do not need quoting. So the correct way to write the if statements is:

               if [ "${KERNEL_DTBVENDORED}" = false ]; then

(In this case it is actually a bitbake variable being quoted, but unless you can guarantee it is not empty, the same rule applies.)

//Peter

<snip>

Quoted 8 lines
Quoted 3 lines

Ah, my bad. I need to figure out a way to get shellcheck to run on bash blocks in bb recipes. Would save everyone some time. I'll submit a V2 with your modifications Martin!

No V2 - it already got merged into master. Moreover, due to this oversight, it defaults to "vendored" DT install in some builds, where bash is not the system shell...

Quoted 9 lines
Quoted 1 line

We should merge Martin's fix from his contrib tree ASAP.

BTW, there are other nuisances with this change:

https://patchwork.yoctoproject.org/project/ti/patch/[email protected]/

Not sure how common the use of third level in KERNEL_DTBDEST dir structure is, maybe we should add a third level generic glob to OE-Core to be safe...

May just be a better idea to have a python / bash function to find all related files recursively. Technically there is no set depth to the kernel dtb structure. They just tend to idle around 2/3.