[RFC PATCH 0/3] kbuild: Make sure to generate config and System.map

[RFC PATCH 0/3] kbuild: Make sure to generate config and System.map

[Tiezhu Yang]

This series is based on 6.19-rc6.

Tiezhu Yang (3):
  MAINTAINERS: Add scripts/install.sh into Kbuild entry
  kbuild: Make sure to generate config file
  kbuild: Make sure to generate System.map file

 MAINTAINERS        |  1 +
 scripts/install.sh | 12 ++++++++++++
 2 files changed, 13 insertions(+)

-- 
2.42.0

[RFC PATCH 1/3] MAINTAINERS: Add scripts/install.sh into Kbuild entry

[Tiezhu Yang]

The common installation code is in scripts/install.sh, add it into
Kbuild entry to reflect the reality.

With this updated entry, folks running get_maintainer.pl on patches
that touch scripts/install.sh will know to send it to linux-kbuild@
as well.

Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
 MAINTAINERS | 1 +
 1 file changed, 1 insertion(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index da9dbc1a4019..8effca525b71 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -13681,6 +13681,7 @@ F:	scripts/include/
 F:	scripts/mk*
 F:	scripts/mod/
 F:	scripts/package/
+F:	scripts/install.sh
 F:	usr/
 
 KERNEL HARDENING (not covered by other areas)
-- 
2.42.0

Re: (subset) [RFC PATCH 1/3] MAINTAINERS: Add scripts/install.sh into Kbuild entry

[Nathan Chancellor]

On Tue, 20 Jan 2026 20:37:28 +0800, Tiezhu Yang wrote:
> The common installation code is in scripts/install.sh, add it into
> Kbuild entry to reflect the reality.
> 
> With this updated entry, folks running get_maintainer.pl on patches
> that touch scripts/install.sh will know to send it to linux-kbuild@
> as well.
> 
> [...]

Applied to

  https://git.kernel.org/pub/scm/linux/kernel/git/kbuild/linux.git kbuild-next

Thanks!

[1/3] MAINTAINERS: Add scripts/install.sh into Kbuild entry
      https://git.kernel.org/kbuild/c/1f77593d304e7

Please look out for regression or issue reports or other follow up
comments, as they may result in the patch/series getting dropped or
reverted. Patches applied to an "unstable" branch are accepted pending
wider testing in -next and any post-commit review; they will generally
be moved to the main branch in a week if no issues are found.

Best regards,
-- 
Nathan Chancellor <nathan@kernel.org>

[RFC PATCH 2/3] kbuild: Make sure to generate config file

[Tiezhu Yang]

After commit 75cffd392bfa ("LoongArch: Using generic scripts/install.sh
in `make install`"), arch/loongarch/boot/install.sh is usually not used,
either /root/bin/installkernel or /sbin/installkernel is used if found.

Then it can not generate the config file in most cases, just copy it to
the installation path. Otherwise there may be an error when testing bpf,
for example.

  $ ./test_verifier
  gzopen /boot/config-6.19.0-rc6: No such file or directory

Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
 scripts/install.sh | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/scripts/install.sh b/scripts/install.sh
index 05d62ac513ee..ecf354d8f4f1 100755
--- a/scripts/install.sh
+++ b/scripts/install.sh
@@ -34,6 +34,12 @@ do
 		continue
 	fi
 
+	# Install kernel config file
+	if [ -f ${INSTALL_PATH}/config-${KERNELRELEASE} ]; then
+	  mv ${INSTALL_PATH}/config-${KERNELRELEASE} ${INSTALL_PATH}/config-${KERNELRELEASE}.old
+	fi
+	cp .config ${INSTALL_PATH}/config-${KERNELRELEASE}
+
 	# installkernel(8) says the parameters are like follows:
 	#
 	#   installkernel version zImage System.map [directory]
-- 
2.42.0

Re: [RFC PATCH 2/3] kbuild: Make sure to generate config file

[Nathan Chancellor]

Hi Tiezhu,

On Tue, Jan 20, 2026 at 08:37:29PM +0800, Tiezhu Yang wrote:
> After commit 75cffd392bfa ("LoongArch: Using generic scripts/install.sh
> in `make install`"), arch/loongarch/boot/install.sh is usually not used,
> either /root/bin/installkernel or /sbin/installkernel is used if found.
> 
> Then it can not generate the config file in most cases, just copy it to
> the installation path. Otherwise there may be an error when testing bpf,
> for example.
> 
>   $ ./test_verifier
>   gzopen /boot/config-6.19.0-rc6: No such file or directorg

This could also be resolved by enabling CONFIG_IKCONFIG_PROC.

> Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
> ---
>  scripts/install.sh | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/scripts/install.sh b/scripts/install.sh
> index 05d62ac513ee..ecf354d8f4f1 100755
> --- a/scripts/install.sh
> +++ b/scripts/install.sh
> @@ -34,6 +34,12 @@ do
>  		continue
>  	fi
>  
> +	# Install kernel config file
> +	if [ -f ${INSTALL_PATH}/config-${KERNELRELEASE} ]; then
> +	  mv ${INSTALL_PATH}/config-${KERNELRELEASE} ${INSTALL_PATH}/config-${KERNELRELEASE}.old
> +	fi
> +	cp .config ${INSTALL_PATH}/config-${KERNELRELEASE}
> +

I am a little hesistant to change the generic install logic to handle
these files, especially since the map file should be handled by
installkernel like you mention in patch 3 (but I am open to other
opinions/arguments) and the configuration can be made available via
CONFIG_IKCONFIG_PROC as I mention above.

If LoongArch wants this behavior as part of its install process, why not
just bring back the lines of 75cffd392bfa after $(cmd,install)

install:
    $(call cmd,install)
    $(Q)install -D -m 644 .config $(INSTALL_PATH)/config-$(KERNELRELEASE)
    $(Q)install -D -m 644 System.map $(INSTALL_PATH)/System.map-$(KERNELRELEASE)

>  	# installkernel(8) says the parameters are like follows:
>  	#
>  	#   installkernel version zImage System.map [directory]
> -- 
> 2.42.0
> 

Re: [RFC PATCH 2/3] kbuild: Make sure to generate config file

[Tiezhu Yang]

On 2026/1/22 上午7:47, Nathan Chancellor wrote:
> Hi Tiezhu,
> 
> On Tue, Jan 20, 2026 at 08:37:29PM +0800, Tiezhu Yang wrote:
>> After commit 75cffd392bfa ("LoongArch: Using generic scripts/install.sh
>> in `make install`"), arch/loongarch/boot/install.sh is usually not used,
>> either /root/bin/installkernel or /sbin/installkernel is used if found.
>>
>> Then it can not generate the config file in most cases, just copy it to
>> the installation path. Otherwise there may be an error when testing bpf,
>> for example.
>>
>>    $ ./test_verifier
>>    gzopen /boot/config-6.19.0-rc6: No such file or directorg
> 
> This could also be resolved by enabling CONFIG_IKCONFIG_PROC.

Currently, CONFIG_IKCONFIG_PROC is set by default on LoongArch,
the above error still exists, it checks /boot/config-* first,
here is the related code:

tools/testing/selftests/bpf/unpriv_helpers.c:
static gzFile open_config(void)
{
         struct utsname uts;
         char buf[PATH_MAX];
         gzFile config;

         if (uname(&uts)) {
                 perror("uname");
                 goto config_gz;
         }

         snprintf(buf, sizeof(buf), "/boot/config-%s", uts.release);
         config = gzopen(buf, "rb");
         if (config)
                 return config;
         fprintf(stderr, "gzopen %s: %s\n", buf, strerror(errno));

config_gz:
         config = gzopen("/proc/config.gz", "rb");
         if (!config)
                 perror("gzopen /proc/config.gz");
         return config;
}

>> Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
>> ---
>>   scripts/install.sh | 6 ++++++
>>   1 file changed, 6 insertions(+)
>>
>> diff --git a/scripts/install.sh b/scripts/install.sh
>> index 05d62ac513ee..ecf354d8f4f1 100755
>> --- a/scripts/install.sh
>> +++ b/scripts/install.sh
>> @@ -34,6 +34,12 @@ do
>>   		continue
>>   	fi
>>   
>> +	# Install kernel config file
>> +	if [ -f ${INSTALL_PATH}/config-${KERNELRELEASE} ]; then
>> +	  mv ${INSTALL_PATH}/config-${KERNELRELEASE} ${INSTALL_PATH}/config-${KERNELRELEASE}.old
>> +	fi
>> +	cp .config ${INSTALL_PATH}/config-${KERNELRELEASE}
>> +
> 
> I am a little hesistant to change the generic install logic to handle
> these files, especially since the map file should be handled by
> installkernel like you mention in patch 3 (but I am open to other
> opinions/arguments) and the configuration can be made available via
> CONFIG_IKCONFIG_PROC as I mention above.
> 
> If LoongArch wants this behavior as part of its install process, why not
> just bring back the lines of 75cffd392bfa after $(cmd,install)
> 
> install:
>      $(call cmd,install)
>      $(Q)install -D -m 644 .config $(INSTALL_PATH)/config-$(KERNELRELEASE)
>      $(Q)install -D -m 644 System.map $(INSTALL_PATH)/System.map-$(KERNELRELEASE)
> 
>>   	# installkernel(8) says the parameters are like follows:
>>   	#
>>   	#   installkernel version zImage System.map [directory]

Yes, this works well for LoongArch. If there is no need to do this for
all archs, I will only touch arch/loongarch files.

By the way, I do not know why it can not generate System.map by the
installkernel in practice, could you please check it?

Thanks,
Tiezhu

Re: [RFC PATCH 2/3] kbuild: Make sure to generate config file

[Nathan Chancellor]

On Thu, Jan 22, 2026 at 09:14:07AM +0800, Tiezhu Yang wrote:
> Currently, CONFIG_IKCONFIG_PROC is set by default on LoongArch,
> the above error still exists, it checks /boot/config-* first,
> here is the related code:
> 
> tools/testing/selftests/bpf/unpriv_helpers.c:
> static gzFile open_config(void)
> {
>         struct utsname uts;
>         char buf[PATH_MAX];
>         gzFile config;
> 
>         if (uname(&uts)) {
>                 perror("uname");
>                 goto config_gz;
>         }
> 
>         snprintf(buf, sizeof(buf), "/boot/config-%s", uts.release);
>         config = gzopen(buf, "rb");
>         if (config)
>                 return config;
>         fprintf(stderr, "gzopen %s: %s\n", buf, strerror(errno));
> 
> config_gz:
>         config = gzopen("/proc/config.gz", "rb");
>         if (!config)
>                 perror("gzopen /proc/config.gz");
>         return config;
> }

Ah but it still falls back to /proc/config.gz though, right?

> Yes, this works well for LoongArch. If there is no need to do this for
> all archs, I will only touch arch/loongarch files.

Until there is a wider insistance on including these as part of the
install target, I think it would be best to keep this in arch/loongarch,
as that is the only architecture that would notice in this case.

> By the way, I do not know why it can not generate System.map by the
> installkernel in practice, could you please check it?

Is installkernel a part of the kernel source? I am not sure where to
begin to investigate that. Could be an inaccurate comment as well if
installkernel's implementation has been updated.

Cheers,
Nathan

Re: [RFC PATCH 2/3] kbuild: Make sure to generate config file

[Tiezhu Yang]

On 2026/1/22 下午1:15, Nathan Chancellor wrote:
> On Thu, Jan 22, 2026 at 09:14:07AM +0800, Tiezhu Yang wrote:
>> Currently, CONFIG_IKCONFIG_PROC is set by default on LoongArch,
>> the above error still exists, it checks /boot/config-* first,
>> here is the related code:
>>
>> tools/testing/selftests/bpf/unpriv_helpers.c:
>> static gzFile open_config(void)
>> {
>>          struct utsname uts;
>>          char buf[PATH_MAX];
>>          gzFile config;
>>
>>          if (uname(&uts)) {
>>                  perror("uname");
>>                  goto config_gz;
>>          }
>>
>>          snprintf(buf, sizeof(buf), "/boot/config-%s", uts.release);
>>          config = gzopen(buf, "rb");
>>          if (config)
>>                  return config;
>>          fprintf(stderr, "gzopen %s: %s\n", buf, strerror(errno));
>>
>> config_gz:
>>          config = gzopen("/proc/config.gz", "rb");
>>          if (!config)
>>                  perror("gzopen /proc/config.gz");
>>          return config;
>> }
> 
> Ah but it still falls back to /proc/config.gz though, right?

Yes, so no /boot/config-* seems not a big deal.

>> Yes, this works well for LoongArch. If there is no need to do this for
>> all archs, I will only touch arch/loongarch files.
> 
> Until there is a wider insistance on including these as part of the
> install target, I think it would be best to keep this in arch/loongarch,
> as that is the only architecture that would notice in this case.
> 
>> By the way, I do not know why it can not generate System.map by the
>> installkernel in practice, could you please check it?
> 
> Is installkernel a part of the kernel source? I am not sure where to
> begin to investigate that. Could be an inaccurate comment as well if
> installkernel's implementation has been updated.

installkernel is not a part of the kernel source, IIUC it belongs to
systemd:

https://github.com/systemd/systemd/blob/main/src/kernel-install/kernel-install.c

For now, please ignore this series.

Thanks,
Tiezhu

[RFC PATCH 3/3] kbuild: Make sure to generate System.map file

[Tiezhu Yang]

After commit 75cffd392bfa ("LoongArch: Using generic scripts/install.sh
in `make install`"), arch/loongarch/boot/install.sh is usually not used,
either /root/bin/installkernel or /sbin/installkernel is used if found.

It should generate System.map when using the script installkernel on x86
and LoongArch Fedora Linux system according to the man pages but it does
not in practice, just copy it to the installation path.

Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
---
 scripts/install.sh | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/scripts/install.sh b/scripts/install.sh
index ecf354d8f4f1..93714ef65967 100755
--- a/scripts/install.sh
+++ b/scripts/install.sh
@@ -40,6 +40,12 @@ do
 	fi
 	cp .config ${INSTALL_PATH}/config-${KERNELRELEASE}
 
+	# Install system map file
+	if [ -f ${INSTALL_PATH}/System.map-${KERNELRELEASE} ]; then
+	  mv ${INSTALL_PATH}/System.map-${KERNELRELEASE} ${INSTALL_PATH}/System.map-${KERNELRELEASE}.old
+	fi
+	cp System.map ${INSTALL_PATH}/System.map-${KERNELRELEASE}
+
 	# installkernel(8) says the parameters are like follows:
 	#
 	#   installkernel version zImage System.map [directory]
-- 
2.42.0