make-scripts breaks on more recent kernels
n0toose opened this issue · comments
So, as far as I am aware of, the way make-scripts works is by applying certain patches to the header sources before compiling them.
This seems to result in unexpected behavior in more recent kernel versions (e.g. Linux 5.6.2)
~$ sudo make-scripts
Patching and running make scripts on the kernel headers.
The next patch would create the file tools/include/tools/be_byteshift.h,
which already exists! Assume -R? [n]
Apply anyway? [n]
Skipping patch.
1 out of 1 hunk ignored
The next patch would create the file tools/include/tools/le_byteshift.h,
which already exists! Assume -R? [n]
Apply anyway? [n]
Skipping patch.
1 out of 1 hunk ignored
YACC scripts/kconfig/parser.tab.[ch]
HOSTCC scripts/kconfig/conf.o
HOSTCC scripts/kconfig/confdata.o
HOSTCC scripts/kconfig/expr.o
/bin/sh: 1: scripts/basic/fixdep: Exec format error
make[2]: *** [scripts/Makefile.host:124: scripts/kconfig/conf.o] Error 2
make[2]: *** Deleting file 'scripts/kconfig/conf.o'
make[2]: *** Waiting for unfinished jobs....
/bin/sh: 1: scripts/basic/fixdep: Exec format error
make[2]: *** [scripts/Makefile.host:124: scripts/kconfig/expr.o] Error 2
make[2]: *** Deleting file 'scripts/kconfig/expr.o'
/bin/sh: 1: scripts/basic/fixdep: Exec format error
make[2]: *** [scripts/Makefile.host:124: scripts/kconfig/confdata.o] Error 2
make[2]: *** Deleting file 'scripts/kconfig/confdata.o'
make[1]: *** [Makefile:568: syncconfig] Error 2
make: *** [Makefile:678: include/config/auto.conf.cmd] Error 2
Done.
Interesting. It seems as though the patch has already been applied and no longer needed in 5.6.y. Although it doesn't seem like running make scripts actually works? So something isn't right.
I know on mainline compiling against the headers is also broken at the moment.
Interesting. It seems as though the patch has already been applied and no longer needed in 5.6.y. Although it doesn't seem like running make scripts actually works? So something isn't right.
I know on mainline compiling against the headers is also broken at the moment.
Yup, I discovered this after dealing with difficulties with the wireguard-dkms package.
I'm assuming ur using a Pi4? Try this kernel. It isn't cross-compiled, it's native.
About to run dkms install steps...
Creating symlink /var/lib/dkms/rtl8812au/5.6.4.2/source ->
/usr/src/rtl8812au-5.6.4.2
DKMS: add completed.
Kernel preparation unnecessary for this kernel. Skipping...
Building module:
cleaning build area...
'make' -j4 KVER=5.6.2-v8 KSRC=/lib/modules/5.6.2-v8/build....................................................
cleaning build area...
DKMS: build completed.
88XXau.ko:
Running module version sanity check.
- Original module
- No original module exists within this kernel
- Installation
- Installing to /lib/modules/5.6.2-v8/updates/dkms/
depmod...
DKMS: install completed.
Finished running dkms install steps.
Make sure you purge the old kernel, before installing.
I'm assuming ur using a Pi4? Try this kernel. It isn't cross-compiled, it's native.
Should a script be provided additionally for the time being?
A script to compile native? You can already do that with the current kernel script. Would just need to compile it on the Pi4 or another SBC running aarch64.
You can actually run the whole build script from the Pi4 if you like. Grab the aarch64 IMG from the hub, git clone and ur off to the races.
Cons: Compiling the kernel on the Pi4 will take about 52 minutes or longer.
Pros: The rest of the process (rootfs and image) is actually quicker.
Unless ur using a beast of an X86_64 machine of course. ;-)
Seems like 5.4.32 is also having problems on this front. Patching and compiling make scripts works as it should, but compiling against the headers results in a exec error.
Only fool proof way of getting around this, as far as I can tell, is native compiling.
update:
Before going to bed last night I decided to run make allv7 using the rpi-4.19.y branch and check it over in the morning.
This kernel seems to be playing as it should.
Building modules, stage 2.
MODPOST 1 modules
CC /home/patrick/rtl8812au/88XXau.mod.o
LD [M] /home/patrick/rtl8812au/88XXau.ko
make[1]: Leaving directory '/usr/src/linux-headers-4.19.115-v7l'
Whatever bug is present in the 5.4.y and 5.6.y branch, hopefully gets sorted out soon.
Fixed in edge branch
MODPOST 1 modules
CC [M] /home/patrick/rtl8812au/88XXau.mod.o
LD [M] /home/patrick/rtl8812au/88XXau.ko
make[1]: Leaving directory '/usr/src/linux-headers-5.6.7-v8'