Date   

Re: libquadmath

Konrad Weihmann <kweihmann@...>
 

The following I just found in local.conf.sample.extended

# Enabling FORTRAN
# Note this is not officially supported and is just illustrated here to
# show an example of how it can be done
# You'll also need your fortran recipe to depend on libgfortran
FORTRAN:forcevariable = ",fortran"

guess that will solve your issue without the need for a bbappend.
Just inject that line via distro or local.conf

On 02.01.22 18:31, staticd wrote:
I think to clarify...what I need are these files:
--- snipped from gcc-runtime.inc ---
```
FILES_libquadmath = "${libdir}/libquadmath*.so.*"
SUMMARY_libquadmath = "GNU quad-precision math library"
FILES_libquadmath-dev = "\
    ${libdir}/${TARGET_SYS}/${BINV}/include/quadmath* \
    ${libdir}/libquadmath*.so \
    ${libdir}/libquadmath.la <http://libquadmath.la> \
"
```
---
to be in my $STAGING_LIBDIR
So far, I have been unable to figure out how to do that, unfortunately.
On Sun, Jan 2, 2022 at 12:23 PM staticd <staticd@... <mailto:staticd@...>> wrote:
Thank you, Michael.
I have `DEPENDS = "virtual/arm-oe-linux-gnueabi-compilerlibs"`
and `do_compile[depends] +=
"virtual/arm-oe-linux-gnueabi-compilerlibs:do_check"`
However, libquadmath.so is still not present in `recipe-sysroot/lib`
I used the `do_check` task from gcc_runtime but that still isn't
providing `libquadmath.so` in my $STAGING_LIBDIR, which is where I
need it.
Many thanks.
On Sun, Jan 2, 2022 at 12:17 PM Michael Ho <michael.ho@...
<mailto:michael.ho@...>> wrote:
Not familiar with Fortran but maybe it helps to know that this
is normally handled with the DEPENDS mechanism. When you add
other recipes as dependencies to your recipe, the task
do_prepare_recipe_sysroot (run before do_compile) will make hard
links to the files from those dependency recipes into that
recipe-sysroot directory.
See:
https://docs.yoctoproject.org/singleindex.html#do-prepare-recipe-sysroot
<https://docs.yoctoproject.org/singleindex.html#do-prepare-recipe-sysroot>
Kind regards,
Michael
On Sun, 2 Jan 2022, 6:06 pm staticd, <staticd@...
<mailto:staticd@...>> wrote:
okay...I think I have a more interesting question now...
In the package I am building I have some Fortran code that
requires `libquadmath`
I see that `gcc-runtime` provides the library but I need the
library present in `recipe-sysroot/lib` when my `do_compile`
runs
Is there a way for me to do that?
My current approach is to build my image, copy the
libraries/includes to my recipe and `install` them in
`recipe-sysroot` before `do_compile`
This doesn't seem like the correct approach but I am not
sure how else to do it at this point
Any help would be greatly appreciated.


Re: libquadmath

Konrad Weihmann <kweihmann@...>
 

I think the reason of that is that fortran support isn't enabled in default gcc-runtime recipe.

from gcc-runtime.inc:

RUNTIMETARGET = "${RUNTIMELIBSSP} libstdc++-v3 libgomp libatomic ${RUNTIMELIBITM} \
${@bb.utils.contains_any('FORTRAN', [',fortran',',f77'], 'libquadmath', '', d)} \
"

activates the configuration of libquadmath if either ',fortran' or ',f77' is set in 'FORTRAN' variable.
Unfortunately this variable seems to be set to "" since gcc version 9 update

By just adding

FORTRAN = ",fortran"

to a gcc-runtime_%.bbappend I could create the needed files in the sysroot-components directory

gcc-runtime/usr/lib/libstdc++.so
gcc-runtime/usr/lib/libitm.spec
->> gcc-runtime/usr/lib/libquadmath.so.0.0.0
gcc-runtime/usr/lib/libstdc++.so.6
gcc-runtime/usr/lib/libitm.so.1.0.0
gcc-runtime/usr/lib/libatomic.so.1
gcc-runtime/usr/lib/libgomp.spec
gcc-runtime/usr/lib/libitm.a

now as you already have set `DEPENDS = "virtual/arm-oe-linux-gnueabi-compilerlibs"` it should automatically make the lib available to your recipe workspace.

On 02.01.22 18:31, staticd wrote:
I think to clarify...what I need are these files:
--- snipped from gcc-runtime.inc ---
```
FILES_libquadmath = "${libdir}/libquadmath*.so.*"
SUMMARY_libquadmath = "GNU quad-precision math library"
FILES_libquadmath-dev = "\
    ${libdir}/${TARGET_SYS}/${BINV}/include/quadmath* \
    ${libdir}/libquadmath*.so \
    ${libdir}/libquadmath.la <http://libquadmath.la> \
"
```
---
to be in my $STAGING_LIBDIR
So far, I have been unable to figure out how to do that, unfortunately.
On Sun, Jan 2, 2022 at 12:23 PM staticd <staticd@... <mailto:staticd@...>> wrote:
Thank you, Michael.
I have `DEPENDS = "virtual/arm-oe-linux-gnueabi-compilerlibs"`
and `do_compile[depends] +=
"virtual/arm-oe-linux-gnueabi-compilerlibs:do_check"`
However, libquadmath.so is still not present in `recipe-sysroot/lib`
I used the `do_check` task from gcc_runtime but that still isn't
providing `libquadmath.so` in my $STAGING_LIBDIR, which is where I
need it.
Many thanks.
On Sun, Jan 2, 2022 at 12:17 PM Michael Ho <michael.ho@...
<mailto:michael.ho@...>> wrote:
Not familiar with Fortran but maybe it helps to know that this
is normally handled with the DEPENDS mechanism. When you add
other recipes as dependencies to your recipe, the task
do_prepare_recipe_sysroot (run before do_compile) will make hard
links to the files from those dependency recipes into that
recipe-sysroot directory.
See:
https://docs.yoctoproject.org/singleindex.html#do-prepare-recipe-sysroot
<https://docs.yoctoproject.org/singleindex.html#do-prepare-recipe-sysroot>
Kind regards,
Michael
On Sun, 2 Jan 2022, 6:06 pm staticd, <staticd@...
<mailto:staticd@...>> wrote:
okay...I think I have a more interesting question now...
In the package I am building I have some Fortran code that
requires `libquadmath`
I see that `gcc-runtime` provides the library but I need the
library present in `recipe-sysroot/lib` when my `do_compile`
runs
Is there a way for me to do that?
My current approach is to build my image, copy the
libraries/includes to my recipe and `install` them in
`recipe-sysroot` before `do_compile`
This doesn't seem like the correct approach but I am not
sure how else to do it at this point
Any help would be greatly appreciated.


Re: libquadmath

staticd
 

I think to clarify...what I need are these files:

--- snipped from gcc-runtime.inc ---
```
FILES_libquadmath = "${libdir}/libquadmath*.so.*"
SUMMARY_libquadmath = "GNU quad-precision math library"
FILES_libquadmath-dev = "\
    ${libdir}/${TARGET_SYS}/${BINV}/include/quadmath* \
    ${libdir}/libquadmath*.so \
    ${libdir}/libquadmath.la \
"
```
---
to be in my $STAGING_LIBDIR

So far, I have been unable to figure out how to do that, unfortunately.

On Sun, Jan 2, 2022 at 12:23 PM staticd <staticd@...> wrote:
Thank you, Michael.

I have `DEPENDS = "virtual/arm-oe-linux-gnueabi-compilerlibs"`

and `do_compile[depends] += "virtual/arm-oe-linux-gnueabi-compilerlibs:do_check"`

However, libquadmath.so is still not present in `recipe-sysroot/lib`

I used the `do_check` task from gcc_runtime but that still isn't providing `libquadmath.so` in my $STAGING_LIBDIR, which is where I need it.

Many thanks.

On Sun, Jan 2, 2022 at 12:17 PM Michael Ho <michael.ho@...> wrote:
Not familiar with Fortran but maybe it helps to know that this is normally handled with the DEPENDS mechanism. When you add other recipes as dependencies to your recipe, the task do_prepare_recipe_sysroot (run before do_compile) will make hard links to the files from those dependency recipes into that recipe-sysroot directory. 


Kind regards,
Michael

On Sun, 2 Jan 2022, 6:06 pm staticd, <staticd@...> wrote:
okay...I think I have a more interesting question now...

In the package I am building I have some Fortran code that requires `libquadmath`

I see that `gcc-runtime` provides the library but I need the library present in `recipe-sysroot/lib` when my `do_compile` runs

Is there a way for me to do that?

My current approach is to build my image, copy the libraries/includes to my recipe and `install` them in `recipe-sysroot` before `do_compile`

This doesn't seem like the correct approach but I am not sure how else to do it at this point

Any help would be greatly appreciated.




Re: libquadmath

staticd
 

Thank you, Michael.

I have `DEPENDS = "virtual/arm-oe-linux-gnueabi-compilerlibs"`

and `do_compile[depends] += "virtual/arm-oe-linux-gnueabi-compilerlibs:do_check"`

However, libquadmath.so is still not present in `recipe-sysroot/lib`

I used the `do_check` task from gcc_runtime but that still isn't providing `libquadmath.so` in my $STAGING_LIBDIR, which is where I need it.

Many thanks.

On Sun, Jan 2, 2022 at 12:17 PM Michael Ho <michael.ho@...> wrote:
Not familiar with Fortran but maybe it helps to know that this is normally handled with the DEPENDS mechanism. When you add other recipes as dependencies to your recipe, the task do_prepare_recipe_sysroot (run before do_compile) will make hard links to the files from those dependency recipes into that recipe-sysroot directory. 


Kind regards,
Michael

On Sun, 2 Jan 2022, 6:06 pm staticd, <staticd@...> wrote:
okay...I think I have a more interesting question now...

In the package I am building I have some Fortran code that requires `libquadmath`

I see that `gcc-runtime` provides the library but I need the library present in `recipe-sysroot/lib` when my `do_compile` runs

Is there a way for me to do that?

My current approach is to build my image, copy the libraries/includes to my recipe and `install` them in `recipe-sysroot` before `do_compile`

This doesn't seem like the correct approach but I am not sure how else to do it at this point

Any help would be greatly appreciated.




Re: libquadmath

Michael Ho
 

Not familiar with Fortran but maybe it helps to know that this is normally handled with the DEPENDS mechanism. When you add other recipes as dependencies to your recipe, the task do_prepare_recipe_sysroot (run before do_compile) will make hard links to the files from those dependency recipes into that recipe-sysroot directory. 


Kind regards,
Michael


On Sun, 2 Jan 2022, 6:06 pm staticd, <staticd@...> wrote:
okay...I think I have a more interesting question now...

In the package I am building I have some Fortran code that requires `libquadmath`

I see that `gcc-runtime` provides the library but I need the library present in `recipe-sysroot/lib` when my `do_compile` runs

Is there a way for me to do that?

My current approach is to build my image, copy the libraries/includes to my recipe and `install` them in `recipe-sysroot` before `do_compile`

This doesn't seem like the correct approach but I am not sure how else to do it at this point

Any help would be greatly appreciated.




libquadmath

staticd
 

okay...I think I have a more interesting question now...

In the package I am building I have some Fortran code that requires `libquadmath`

I see that `gcc-runtime` provides the library but I need the library present in `recipe-sysroot/lib` when my `do_compile` runs

Is there a way for me to do that?

My current approach is to build my image, copy the libraries/includes to my recipe and `install` them in `recipe-sysroot` before `do_compile`

This doesn't seem like the correct approach but I am not sure how else to do it at this point

Any help would be greatly appreciated.


[auto-upgrade-helper][PATCH] devtool: return both stdout and stderr

Alexander Kanavin
 

At some point devtool started printing to stderr, which means
patch rebase issues weren't anymore correctly caught, and bogus
updates were being sent out.

Signed-off-by: Alexander Kanavin <alex.kanavin@...>
---
modules/utils/devtool.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/modules/utils/devtool.py b/modules/utils/devtool.py
index 0d27bb0..c26061f 100644
--- a/modules/utils/devtool.py
+++ b/modules/utils/devtool.py
@@ -19,7 +19,7 @@ class Devtool(object):
raise DevtoolError("The following devtool command failed: " + operation,
e.stdout, e.stderr)

- return stdout
+ return stdout + stderr

def upgrade(self, recipe, version = None, revision = None):
cmd = " upgrade " + recipe
--
2.33.1


How to add WIFI debug tool to custom image #raspberrypi #ubuntu #yocto #devtool #distro

mihirdave36@...
 

Hello,

How can I add any generic WiFi debug tool to Yocto imgae?

Thanks
Mihir


How to add Alsa loopback / Alsa lib support to custom image #linux #raspberrypi #ubuntu #yocto #devtool

mihirdave36@...
 

Hello,
Happy New Year

I am trying to learn Yocto, & implement it on NXP & RPI 4 dev boards.
so, could you please guide me how can i add,
1- Alsa loopback or Alsa library.
to my custom Yocto image.

Thanks
Mihir


[meta-security][PATCH] google-authenticator-libpam: update to 1.09

Armin Kuster
 

Signed-off-by: Armin Kuster <akuster808@...>
---
...cator-libpam_1.08.bb => google-authenticator-libpam_1.09.bb} | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
rename recipes-security/google-authenticator-libpam/{google-authenticator-libpam_1.08.bb => google-authenticator-libpam_1.09.bb} (92%)

diff --git a/recipes-security/google-authenticator-libpam/google-authenticator-libpam_1.08.bb b/recipes-security/google-authenticator-libpam/google-authenticator-libpam_1.09.bb
similarity index 92%
rename from recipes-security/google-authenticator-libpam/google-authenticator-libpam_1.08.bb
rename to recipes-security/google-authenticator-libpam/google-authenticator-libpam_1.09.bb
index e8ddf29..8a0b1ee 100644
--- a/recipes-security/google-authenticator-libpam/google-authenticator-libpam_1.08.bb
+++ b/recipes-security/google-authenticator-libpam/google-authenticator-libpam_1.09.bb
@@ -4,7 +4,7 @@ LIC_FILES_CHKSUM = "file://LICENSE;md5=3b83ef96387f14655fc854ddc3c6bd57"
LICENSE = "Apache-2.0"

SRC_URI = "git://github.com/google/google-authenticator-libpam.git;branch=master;protocol=https"
-SRCREV = "2c7415d950fb0b4a7f779f045910666447b100ef"
+SRCREV = "962f353aac6cfc7b804547319db40f8b804f0b6c"

DEPENDS = "libpam"

--
2.25.1


[meta-cgl][PATCH 3/3] cluster-glue: add UPSTREAM_CHECK_GITTAGREGEX

Yi Zhao
 

Add UPSTREAM_CHECK_GITTAGREGEX to check the correct latest version.

Before the patch:
$ devtool latest-version cluster-glue
INFO: Current version: 1.0.12
INFO: Latest version: 10
INFO: Latest version's commit: 62fac3c02affab05abe39c185dad3a91367d3f58

After the patch:
$ devtool latest-version cluster-glue
INFO: Current version: 1.0.12
INFO: Latest version: 1.0.12
INFO: Latest version's commit: 1bc77825c0cfb0c80f9c82a061af7ede68676cb4

Signed-off-by: Yi Zhao <yi.zhao@...>
---
meta-cgl-common/recipes-cgl/cluster-glue/cluster-glue_1.0.12.bb | 2 ++
1 file changed, 2 insertions(+)

diff --git a/meta-cgl-common/recipes-cgl/cluster-glue/cluster-glue_1.0.12.bb b/meta-cgl-common/recipes-cgl/cluster-glue/cluster-glue_1.0.12.bb
index acd3149..c9d1bac 100644
--- a/meta-cgl-common/recipes-cgl/cluster-glue/cluster-glue_1.0.12.bb
+++ b/meta-cgl-common/recipes-cgl/cluster-glue/cluster-glue_1.0.12.bb
@@ -22,6 +22,8 @@ SRC_URI:append_libc-uclibc = " file://kill-stack-protector.patch"

SRCREV = "fd5a3befacd23d056a72cacd2b8ad6bba498e56b"

+UPSTREAM_CHECK_GITTAGREGEX = "glue-(?P<pver>\d+(\.\d+)+)"
+
inherit autotools useradd pkgconfig systemd multilib_script multilib_header

SYSTEMD_SERVICE:${PN} = "logd.service"
--
2.25.1


[meta-cgl][PATCH 2/3] ocfs2-tools: add UPSTREAM_CHECK_GITTAGREGEX

Yi Zhao
 

Add UPSTREAM_CHECK_GITTAGREGEX to check the correct latest version.

Before the patch:
$ devtool latest-version ocfs2-tools
INFO: Current version: 1.8.6
INFO: Latest version: 2
INFO: Latest version's commit: 80a60d697d9052d3f196a932df3d1fb5f0df52d6

After the patch:
$ devtool latest-version ocfs2-tools
INFO: Current version: 1.8.6
INFO: Latest version: 1.8.7
INFO: Latest version's commit: 80a60d697d9052d3f196a932df3d1fb5f0df52d6

Signed-off-by: Yi Zhao <yi.zhao@...>
---
meta-cgl-common/recipes-cgl/ocfs2-tools/ocfs2-tools_1.8.6.bb | 3 +++
1 file changed, 3 insertions(+)

diff --git a/meta-cgl-common/recipes-cgl/ocfs2-tools/ocfs2-tools_1.8.6.bb b/meta-cgl-common/recipes-cgl/ocfs2-tools/ocfs2-tools_1.8.6.bb
index 7c32c54..28ef4b0 100644
--- a/meta-cgl-common/recipes-cgl/ocfs2-tools/ocfs2-tools_1.8.6.bb
+++ b/meta-cgl-common/recipes-cgl/ocfs2-tools/ocfs2-tools_1.8.6.bb
@@ -22,6 +22,9 @@ SRC_URI = "git://github.com/markfasheh/ocfs2-tools;branch=master;protocol=https
file://0001-o2cb.init.sh-Remove-unneeded-lib-lsb-init-functions.patch \
"
SRCREV = "4d76ceb4aa7aaa1fd595368089e99575d708f719"
+
+UPSTREAM_CHECK_GITTAGREGEX = "ocfs2-tools-(?P<pver>\d+(\.\d+)+)"
+
S = "${WORKDIR}/git"

inherit autotools-brokensep pkgconfig systemd
--
2.25.1


[meta-cgl][PATCH 1/3] resource-agents: add UPSTREAM_CHECK_GITTAGREGEX

Yi Zhao
 

Add UPSTREAM_CHECK_GITTAGREGEX to check the correct latest version.

Before the patch:
$ devtool latest-version resource-agents
INFO: Current version: 4.5.0
INFO: Latest version: 10
INFO: Latest version's commit: 5cf93783eafa22e8b82d50834bfb76b9e8434783

After the patch:
$ devtool latest-version resource-agents
INFO: Current version: 4.5.0
INFO: Latest version: 4.10.0
INFO: Latest version's commit: fd0720f73a06042ad0a5475a3398096b2912cf5f

Signed-off-by: Yi Zhao <yi.zhao@...>
---
.../cluster-resource-agents/resource-agents_4.5.0.bb | 2 ++
1 file changed, 2 insertions(+)

diff --git a/meta-cgl-common/recipes-cgl/cluster-resource-agents/resource-agents_4.5.0.bb b/meta-cgl-common/recipes-cgl/cluster-resource-agents/resource-agents_4.5.0.bb
index 00ddd22..c8b2fbd 100644
--- a/meta-cgl-common/recipes-cgl/cluster-resource-agents/resource-agents_4.5.0.bb
+++ b/meta-cgl-common/recipes-cgl/cluster-resource-agents/resource-agents_4.5.0.bb
@@ -24,6 +24,8 @@ SRC_URI = "git://github.com/ClusterLabs/resource-agents;branch=main;protocol=htt

SRCREV = "fee181320547365d7f8c88cca2b32801412b933d"

+UPSTREAM_CHECK_GITTAGREGEX = "v(?P<pver>\d+(\.\d+)+)"
+
S="${WORKDIR}/git"

LIC_FILES_CHKSUM = "file://COPYING;md5=751419260aa954499f7abaabaa882bbe \
--
2.25.1


Re: building image for Realtek RTS3916N mips SoC using vendor provided prebuilt external uClibc toolchain

davis roman <davis.roman84@...>
 

On Wed, Dec 29, 2021 at 6:15 PM Davis Roman <davis.roman84@...> wrote:

On Wed, Dec 29, 2021 at 5:30 PM Khem Raj <raj.khem@...> wrote:



On Wed, Dec 29, 2021 at 2:20 PM Davis Roman <davis.roman84@...> wrote:

On Wed, Dec 29, 2021 at 2:21 PM Anders Montonen <Anders.Montonen@...> wrote:

Hi,

On 29 Dec 2021, at 9:53, davis roman <davis.roman84@...> wrote:

I generated an internal mips toolchain built against musl and I tried
to compile u-boot but unfortunately, I'm getting "opcode not
supported" error messages. https://pastebin.com/QdcLxy69
If instead I use the realtek provided prebuilt toolchain then u-boot
compiles successfully. https://pastebin.com/zcQ5kc20

I'm thinking that Realtek's toolchain has patches specific to their
SoC that have not been pushed upstream. Could this be the reason I'm
unable to compile uboot?
I’m guessing that your U-Boot config doesn’t set the correct MIPS architecture revision. The compiler error shows that you’re trying to assemble a MIPS32r1 instruction, but the compiler is targeting the original MIPS1 architecture. The Realtek toolchain may have set the default architecture to match the SoC, but the fix is to update the config to match the hardware.
You're right. I didn't realize the RX5281 core on the RTS3916N only
supports mips1 or mips16 (https://pasteboard.co/IpsqN6GkBYAs.png).

I happened to have a mips sourcery toolchain installed on my machine
(https://sourcery.sw.siemens.com/GNUToolchain/package12797/public/mips-linux-gnu/mips-2014.05-27-mips-linux-gnu-i686-pc-linux-gnu.tar.bz2)
so I pointed that to u-boot without modifying anything else and it was
able to compile u-boot successfully. Woot!
https://pastebin.com/ySPFae5u

I suppose the next step would be to generate a mips1 yocto toolchain
however according to the available tune values it appears only mips32
or mips64 is available

Any suggestions on how to generate a mips1 yocto toolchain or if
that's even supported?

Yes it’s supported although it’s not default for qemumips so the simple trick you can do is change the DEFAULTTUNE setting in the qemumips.conf away from mips32r2
Sorry Khem, I'm not quite following you. I tried the following patch
(https://pastebin.com/rkmQ3t6P) thinking perhaps this is what you
meant however my build configuration still shows the tune set to
mips32r2 https://pastebin.com/izP9thVW

What am I missing?
Nevermind, I scraped the last patch, recreated my BUILD_DIR, set
MACHINE = "qemumips" and added DEFAULTTUNE = "mips" in my local.conf.
Now the build configuration correct https://pastebin.com/pkkRVL58 and
I'm now waiting for the toolchain to finish so I can attempt to build
realtek's u-boot.





Regards,
Anders Montonen


Re: building image for Realtek RTS3916N mips SoC using vendor provided prebuilt external uClibc toolchain

davis roman <davis.roman84@...>
 

On Wed, Dec 29, 2021 at 5:30 PM Khem Raj <raj.khem@...> wrote:



On Wed, Dec 29, 2021 at 2:20 PM Davis Roman <davis.roman84@...> wrote:

On Wed, Dec 29, 2021 at 2:21 PM Anders Montonen <Anders.Montonen@...> wrote:

Hi,

On 29 Dec 2021, at 9:53, davis roman <davis.roman84@...> wrote:

I generated an internal mips toolchain built against musl and I tried
to compile u-boot but unfortunately, I'm getting "opcode not
supported" error messages. https://pastebin.com/QdcLxy69
If instead I use the realtek provided prebuilt toolchain then u-boot
compiles successfully. https://pastebin.com/zcQ5kc20

I'm thinking that Realtek's toolchain has patches specific to their
SoC that have not been pushed upstream. Could this be the reason I'm
unable to compile uboot?
I’m guessing that your U-Boot config doesn’t set the correct MIPS architecture revision. The compiler error shows that you’re trying to assemble a MIPS32r1 instruction, but the compiler is targeting the original MIPS1 architecture. The Realtek toolchain may have set the default architecture to match the SoC, but the fix is to update the config to match the hardware.
You're right. I didn't realize the RX5281 core on the RTS3916N only
supports mips1 or mips16 (https://pasteboard.co/IpsqN6GkBYAs.png).

I happened to have a mips sourcery toolchain installed on my machine
(https://sourcery.sw.siemens.com/GNUToolchain/package12797/public/mips-linux-gnu/mips-2014.05-27-mips-linux-gnu-i686-pc-linux-gnu.tar.bz2)
so I pointed that to u-boot without modifying anything else and it was
able to compile u-boot successfully. Woot!
https://pastebin.com/ySPFae5u

I suppose the next step would be to generate a mips1 yocto toolchain
however according to the available tune values it appears only mips32
or mips64 is available

Any suggestions on how to generate a mips1 yocto toolchain or if
that's even supported?

Yes it’s supported although it’s not default for qemumips so the simple trick you can do is change the DEFAULTTUNE setting in the qemumips.conf away from mips32r2
Sorry Khem, I'm not quite following you. I tried the following patch
(https://pastebin.com/rkmQ3t6P) thinking perhaps this is what you
meant however my build configuration still shows the tune set to
mips32r2 https://pastebin.com/izP9thVW

What am I missing?





Regards,
Anders Montonen


Re: building image for Realtek RTS3916N mips SoC using vendor provided prebuilt external uClibc toolchain

Khem Raj
 



On Wed, Dec 29, 2021 at 2:20 PM Davis Roman <davis.roman84@...> wrote:
On Wed, Dec 29, 2021 at 2:21 PM Anders Montonen <Anders.Montonen@...> wrote:
>
> Hi,
>
> > On 29 Dec 2021, at 9:53, davis roman <davis.roman84@...> wrote:
> >
> > I generated an internal mips toolchain built against musl and I tried
> > to compile u-boot but unfortunately, I'm getting "opcode not
> > supported" error messages. https://pastebin.com/QdcLxy69
> > If instead I use the realtek provided prebuilt toolchain then u-boot
> > compiles successfully. https://pastebin.com/zcQ5kc20
> >
> > I'm thinking that Realtek's toolchain has patches specific to their
> > SoC that have not been pushed upstream. Could this be the reason I'm
> > unable to compile uboot?
>
> I’m guessing that your U-Boot config doesn’t set the correct MIPS architecture revision. The compiler error shows that you’re trying to assemble a MIPS32r1 instruction, but the compiler is targeting the original MIPS1 architecture. The Realtek toolchain may have set the default architecture to match the SoC, but the fix is to update the config to match the hardware.

You're right. I didn't realize the RX5281 core on the RTS3916N only
supports mips1 or mips16 (https://pasteboard.co/IpsqN6GkBYAs.png).

I happened to have a mips sourcery toolchain installed on my machine
(https://sourcery.sw.siemens.com/GNUToolchain/package12797/public/mips-linux-gnu/mips-2014.05-27-mips-linux-gnu-i686-pc-linux-gnu.tar.bz2)
so I pointed that to u-boot without modifying anything else and it was
able to compile u-boot successfully. Woot!
https://pastebin.com/ySPFae5u

I suppose the next step would be to generate a mips1 yocto toolchain
however according to the available tune values it appears only mips32
or mips64 is available

Any suggestions on how to generate a mips1 yocto toolchain or if
that's even supported?

Yes it’s supported although it’s not default for qemumips so the simple trick you can do is change the DEFAULTTUNE setting in the qemumips.conf away from mips32r2


>
> Regards,
> Anders Montonen


Re: building image for Realtek RTS3916N mips SoC using vendor provided prebuilt external uClibc toolchain

davis roman <davis.roman84@...>
 

On Wed, Dec 29, 2021 at 2:21 PM Anders Montonen <Anders.Montonen@...> wrote:

Hi,

On 29 Dec 2021, at 9:53, davis roman <davis.roman84@...> wrote:

I generated an internal mips toolchain built against musl and I tried
to compile u-boot but unfortunately, I'm getting "opcode not
supported" error messages. https://pastebin.com/QdcLxy69
If instead I use the realtek provided prebuilt toolchain then u-boot
compiles successfully. https://pastebin.com/zcQ5kc20

I'm thinking that Realtek's toolchain has patches specific to their
SoC that have not been pushed upstream. Could this be the reason I'm
unable to compile uboot?
I’m guessing that your U-Boot config doesn’t set the correct MIPS architecture revision. The compiler error shows that you’re trying to assemble a MIPS32r1 instruction, but the compiler is targeting the original MIPS1 architecture. The Realtek toolchain may have set the default architecture to match the SoC, but the fix is to update the config to match the hardware.
You're right. I didn't realize the RX5281 core on the RTS3916N only
supports mips1 or mips16 (https://pasteboard.co/IpsqN6GkBYAs.png).

I happened to have a mips sourcery toolchain installed on my machine
(https://sourcery.sw.siemens.com/GNUToolchain/package12797/public/mips-linux-gnu/mips-2014.05-27-mips-linux-gnu-i686-pc-linux-gnu.tar.bz2)
so I pointed that to u-boot without modifying anything else and it was
able to compile u-boot successfully. Woot!
https://pastebin.com/ySPFae5u

I suppose the next step would be to generate a mips1 yocto toolchain
however according to the available tune values it appears only mips32
or mips64 is available

Any suggestions on how to generate a mips1 yocto toolchain or if
that's even supported?


Regards,
Anders Montonen


Re: building image for Realtek RTS3916N mips SoC using vendor provided prebuilt external uClibc toolchain

Anders Montonen
 

Hi,

On 29 Dec 2021, at 9:53, davis roman <davis.roman84@...> wrote:

I generated an internal mips toolchain built against musl and I tried
to compile u-boot but unfortunately, I'm getting "opcode not
supported" error messages. https://pastebin.com/QdcLxy69
If instead I use the realtek provided prebuilt toolchain then u-boot
compiles successfully. https://pastebin.com/zcQ5kc20

I'm thinking that Realtek's toolchain has patches specific to their
SoC that have not been pushed upstream. Could this be the reason I'm
unable to compile uboot?
I’m guessing that your U-Boot config doesn’t set the correct MIPS architecture revision. The compiler error shows that you’re trying to assemble a MIPS32r1 instruction, but the compiler is targeting the original MIPS1 architecture. The Realtek toolchain may have set the default architecture to match the SoC, but the fix is to update the config to match the hardware.

Regards,
Anders Montonen


Re: [meta-raspberrypi][PATCH] xserver-xorg: remove xshmfence configure option

Andreas Müller
 

On Thu, Dec 9, 2021 at 6:37 AM Khem Raj <raj.khem@...> wrote:



On Wed, Dec 8, 2021 at 7:03 PM Yu, Mingli <mingli.yu@...> wrote:

From: Mingli Yu <mingli.yu@...>

After the commit [1] introduced in openembedded-core layer,
some configure options is't carried over include xshmfence
option, so remove the xshmfence configure option to silence
the below warning.
WARNING: xserver-xorg-2_21.1.1-r0 do_configure: QA Issue: xserver-xorg: invalid PACKAGECONFIG: xshmfence [invalid-packageconfig]

That’s ok to remove it but more importantly does it work now without this option
Looked into xserver's meson.build: As soon as dri3 is added - which
we do - libxshmfence turns into a mandatory dependency. So this patch
is correct.
Somehow libxshmfence is added to sysroot - maybe we should add it
explicitly to dri3 PACKAGECONFIG in oe-core to be more explicit.

Slightly off-topic but we are on xserver on Raspi: Has anybody tested
xserver recently? I see nasty issues on raspi4-64 / XFCE (haven't
tried other desktops yet):

* As soon as a window is opened, screen turns black for 1-2 seconds.
Have not yet found helpful in logs yet.
* There are two mouse devices detected with name 'vc4'. What?

Cheers

Andreas


Re: building image for Realtek RTS3916N mips SoC using vendor provided prebuilt external uClibc toolchain

davis roman <davis.roman84@...>
 

On Tue, Dec 28, 2021 at 10:09 PM Khem Raj <raj.khem@...> wrote:



On 12/28/21 2:53 PM, davis roman wrote:
Hi all,

I'm working on a project utilizing a Realtek RTS3916N mips SoC and as
part of Realtek's bsp offering, they provide their own fork of
buildroot packaged with their prebuilt binary toolchain.
I would much rather use yocto instead however Realtek made it clear
that no yocto support is available from their end and they have no
plans to provide it in the future.

In theory, I have Realtek's u-boot and kernel source so it should be
possible to build an image so long as poky is instructed to use an
external toolchain. This is where I hit my first road block because
Realtek only supports uClibc and I know the yocto community moved way
from uClibc, in favor of musl, since the Morty release. As a result, I
decided to use Krogoth-15.0.0 (knowing it's EOL) only because it
supports uClibc.

Now that I had picked a specific version of poky, I continued forward
with the EXTERNAL_TOOLCHAIN feature. The documentation references the
meta-sourcery layer as the typical use case. Using a sourcery
toolchain tarball ( 2014.05) I was able to build qemuarm/glibc
therefore as far as I can tell the EXTERNAL_TOOLCHAIN feature works as
documented.

Unfortunately, the issue I have now arises when I use the Realtek
toolchain. meta-sourcery informs me that building against uClibc is
not allowed as it has been blacklisted. Only glibc appears to be
supported.

I have looked around to see what other projects are similar and the
closest I can find is the creator CI20 however due to the Realtek
toolchain being based on uClibc, I haven't been able to find anything
suitable for my usecase.

I would greatly appreciate any suggestions regarding how to proceed
forward with yocto integration, using a mips uClibc based toolchain.
We have dropped uclibc support long time ago so any effort to support
uClibc based system will be a bit of work and we have also dropped stale
patches from metadata so even if you are able to get external toolchain
bolted in, you will unfurl next set of problems for yourself So I don't
have a better answer for you here, other than drop the idea of using
uclibc if you want to use modern yocto baseline. We did support it long
time ago so if you like you can use thise EOLed releases but there wont
be much support for it here on community mailing lists.
Thanks Khem for confirming my suspicions.

Is there any other course of action that I can try?
Since you have bootloader and kernel available to you, it might be less
work to put together a BSP layer and machine definition for this SOC
easily and it could be forward looking but there could be some
portability issues w.r.t. toolchain etc. which are manageable
here you will use internal mips toolchain to build your system and you
can use musl instead of uclibc to build the system or glibc is other
supported option.
I generated an internal mips toolchain built against musl and I tried
to compile u-boot but unfortunately, I'm getting "opcode not
supported" error messages. https://pastebin.com/QdcLxy69
If instead I use the realtek provided prebuilt toolchain then u-boot
compiles successfully. https://pastebin.com/zcQ5kc20

I'm thinking that Realtek's toolchain has patches specific to their
SoC that have not been pushed upstream. Could this be the reason I'm
unable to compile uboot?


Thank you,

Davis