lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <8734c9wwvv.fsf@mpe.ellerman.id.au>
Date: Mon, 09 Jun 2025 11:01:56 +1000
From: Michael Ellerman <mpe@...erman.id.au>
To: Masahiro Yamada <masahiroy@...nel.org>, Christophe Leroy
 <christophe.leroy@...roup.eu>
Cc: Madhavan Srinivasan <maddy@...ux.ibm.com>, Nicholas Piggin
 <npiggin@...il.com>, Naveen N Rao <naveen@...nel.org>,
 linuxppc-dev@...ts.ozlabs.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] powerpc: use always-y instead of extra-y in Makefiles

Masahiro Yamada <masahiroy@...nel.org> writes:
> On Tue, Jun 3, 2025 at 3:50 PM Christophe Leroy
> <christophe.leroy@...roup.eu> wrote:
>> Le 02/06/2025 à 18:32, Masahiro Yamada a écrit :
>> > The extra-y syntax is planned for deprecation because it is similar
>> > to always-y.
>> >
>> > When building the boot wrapper, always-y and extra-y are equivalent.
>> > Use always-y instead.
>> >
>> > In arch/powerpc/kernel/Makefile, I added ifdef KBUILD_BUILTIN to
>> > keep the current behavior: prom_init_check is skipped when building
>> > only modular objects.
>>
>> I don't understand what you mean.
>>
>> CONFIG_PPC_OF_BOOT_TRAMPOLINE is a bool, it cannot be a module.
>>
>> prom_init_check is only to check the content of prom_init.o which is
>> never a module.
>>
>> Is always-y to run _after_ prom_init.o is built ?
>
> The intent of "make ARCH=powerpc modules"
> is to compile objects that are necessary for modules,
> that is, all built-in objects are skipped.
>
> However,
> always-$(CONFIG_PPC_OF_BOOT_TRAMPOLINE) += prom_init_check
> would generate prom_init_check regardless,
> and its prerequisite, prom_init.o as well.
>
> With CONFIG_MODULES=y and
> CONFIG_MODVERSIONS=n,
> and without ifdef KBUILD_BUILTIN,
>
> $ make ARCH=powerpc CROSS_COMPILE=powerpc64-linux-gnu-   modules
>
> would result in this:
>
>
>   CC [M]  arch/powerpc/kvm/book3s_xive_native.o
>   CC [M]  arch/powerpc/kvm/book3s_64_vio.o
>   LD [M]  arch/powerpc/kvm/kvm.o
>   CC [M]  arch/powerpc/kvm/book3s_hv.o
>   AS [M]  arch/powerpc/kvm/book3s_hv_interrupts.o
>   CC [M]  arch/powerpc/kvm/book3s_64_mmu_hv.o
>   CC [M]  arch/powerpc/kvm/book3s_64_mmu_radix.o
>   CC [M]  arch/powerpc/kvm/book3s_hv_nested.o
>   CC [M]  arch/powerpc/kvm/book3s_hv_tm.o
>   LD [M]  arch/powerpc/kvm/kvm-hv.o
>   CC [M]  arch/powerpc/kernel/rtas_flash.o
>   CC      arch/powerpc/kernel/prom_init.o
>   PROMCHK arch/powerpc/kernel/prom_init_check
>   CC [M]  kernel/locking/locktorture.o
>   CC [M]  kernel/time/test_udelay.o
>   CC [M]  kernel/time/time_test.o
>   CC [M]  kernel/backtracetest.o
>   CC [M]  kernel/torture.o
>   CC [M]  kernel/resource_kunit.o
>   CC [M]  kernel/sysctl-test.o
>   CC [M]  fs/ext4/inode-test.o
>   LD [M]  fs/ext4/ext4-inode-test.o
>   CC [M]  fs/fat/namei_vfat.o
>   LD [M]  fs/fat/vfat.o
>   CC [M]  fs/fat/fat_test.o
>   CC [M]  fs/nls/nls_ucs2_utils.o
>   CC [M]  fs/netfs/buffered_read.o
>   CC [M]  fs/netfs/buffered_write.o
> ...
>
>
>
> You can see these two lines:
>
>   CC      arch/powerpc/kernel/prom_init.o
>   PROMCHK arch/powerpc/kernel/prom_init_check
>
> are supposed to be skipped when "make modules",
> but actually compiled without ifdef.
>
> So, I added ifdef KBUILD_BUILTIN to preserve
> the current behavior.

OK, that makes sense.

I don't really ever build just modules, so I wouldn't notice, but some
folks probably do.

Acked-by: Michael Ellerman <mpe@...erman.id.au> (powerpc)

cheers

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ