This allows compiling a full 128-bit product of two 64-bit integers as
a mul/mulh pair, instead of a nasty long sequence of 20+ instructions.

Signed-off-by: Xi Ruoyao <xry111@xry111.site>
---
 arch/loongarch/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
index a5f300ec6f28..5a769bb92d7c 100644
--- a/arch/loongarch/Kconfig
+++ b/arch/loongarch/Kconfig
@@ -56,6 +56,7 @@ config LOONGARCH
 	select ARCH_SUPPORTS_ACPI
 	select ARCH_SUPPORTS_ATOMIC_RMW
 	select ARCH_SUPPORTS_HUGETLBFS
+	select ARCH_SUPPORTS_INT128 if CC_HAS_INT128
 	select ARCH_SUPPORTS_LTO_CLANG
 	select ARCH_SUPPORTS_LTO_CLANG_THIN
 	select ARCH_SUPPORTS_NUMA_BALANCING
-- 
2.44.0

Hi, Ruoyao,

If enable CONFIG_DEBUG_INFO_BTF, we get:

loongarch64-unknown-linux-gnu-ld: kernel/sched/fair.o: in function
`mul_u64_u32_shr':
/home/chenhuacai/linux-test.git/./include/linux/math64.h:161:(.text+0x5e4):
undefined reference to `__lshrti3'

So, could you please implement _ashlti3(), __ashrti3(), and
__lshrti3() as commit fbac266f095de92b93e7 ("s390: select
ARCH_SUPPORTS_INT128") does?

Huacai

On Thu, Mar 28, 2024 at 1:18 AM Xi Ruoyao <xry111@xry111.site> wrote:
>
> This allows compiling a full 128-bit product of two 64-bit integers as
> a mul/mulh pair, instead of a nasty long sequence of 20+ instructions.
>
> Signed-off-by: Xi Ruoyao <xry111@xry111.site>
> ---
>  arch/loongarch/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
> index a5f300ec6f28..5a769bb92d7c 100644
> --- a/arch/loongarch/Kconfig
> +++ b/arch/loongarch/Kconfig
> @@ -56,6 +56,7 @@ config LOONGARCH
>         select ARCH_SUPPORTS_ACPI
>         select ARCH_SUPPORTS_ATOMIC_RMW
>         select ARCH_SUPPORTS_HUGETLBFS
> +       select ARCH_SUPPORTS_INT128 if CC_HAS_INT128
>         select ARCH_SUPPORTS_LTO_CLANG
>         select ARCH_SUPPORTS_LTO_CLANG_THIN
>         select ARCH_SUPPORTS_NUMA_BALANCING
> --
> 2.44.0
>
>

On Fri, 2024-04-26 at 12:00 +0800, Huacai Chen wrote:
> Hi, Ruoyao,
> 
> If enable CONFIG_DEBUG_INFO_BTF, we get:
> 
> loongarch64-unknown-linux-gnu-ld: kernel/sched/fair.o: in function
> `mul_u64_u32_shr':
> /home/chenhuacai/linux-test.git/./include/linux/math64.h:161:(.text+0x5e4):
> undefined reference to `__lshrti3'
> 
> So, could you please implement _ashlti3(), __ashrti3(), and
> __lshrti3() as commit fbac266f095de92b93e7 ("s390: select
> ARCH_SUPPORTS_INT128") does?

It seems not related to DEBUG_INFO_BTF, but related to
CC_OPTIMIZE_FOR_SIZE.

I'm making a patch.


-- 
Xi Ruoyao <xry111@xry111.site>
School of Aerospace Science and Technology, Xidian University

Queued for loongarch-next, thanks.

Huacai

On Thu, Mar 28, 2024 at 1:18 AM Xi Ruoyao <xry111@xry111.site> wrote:
>
> This allows compiling a full 128-bit product of two 64-bit integers as
> a mul/mulh pair, instead of a nasty long sequence of 20+ instructions.
>
> Signed-off-by: Xi Ruoyao <xry111@xry111.site>
> ---
>  arch/loongarch/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/arch/loongarch/Kconfig b/arch/loongarch/Kconfig
> index a5f300ec6f28..5a769bb92d7c 100644
> --- a/arch/loongarch/Kconfig
> +++ b/arch/loongarch/Kconfig
> @@ -56,6 +56,7 @@ config LOONGARCH
>         select ARCH_SUPPORTS_ACPI
>         select ARCH_SUPPORTS_ATOMIC_RMW
>         select ARCH_SUPPORTS_HUGETLBFS
> +       select ARCH_SUPPORTS_INT128 if CC_HAS_INT128
>         select ARCH_SUPPORTS_LTO_CLANG
>         select ARCH_SUPPORTS_LTO_CLANG_THIN
>         select ARCH_SUPPORTS_NUMA_BALANCING
> --
> 2.44.0
>
>