On Fri, Mar 28, 2025 at 5:06 AM David Laight <david.laight.linux@gmail.com> wrote:
>
> On Tue, 25 Mar 2025 08:15:41 -0400
> guoren@kernel.org wrote:
>
> > From: "Guo Ren (Alibaba DAMO Academy)" <guoren@kernel.org>
> >
> > Since 2001, the CONFIG_64BIT kernel has been built with the LP64 ABI,
> > but this patchset allows the CONFIG_64BIT kernel to use an ILP32 ABI
> > for construction to reduce cache & memory footprint (Compared to
> > kernel-lp64-abi, kernel-rv64ilp32-abi decreased the used memory by
> > about 20%, as shown in "free -h" in the following demo.)
> ...
>
> Why on earth would you want to run a 64bit application on a 32bit kernel.
> IIRC the main justification for 64bit was to get a larger address space.

Because the below chips have built the Linux ecosystem with LP64-ABI:
 - allwinner/sun20i-d1-lichee (128MB~1GB)
 - allwinner/sun20i-d1s-mangopi (64MB)
 - bouffalo/bl808 (64MB)
 - canaan/k230 (512MB)
 - canaan/k230d (128MB)
 - renesas/rzfive-smarc (256MB~1GB DDR)
 - sophgo/cv1800b (64MB DDR)
 - sophgo/cv1812h (256MB DDR)
 - sophgo/sg2002 (64/256MB DDR)
 - https://docs.banana-pi.org/en/BPI-RV2/BananaPi_BPI-RV2 (512MB DDR)
By swapping the rv64ilp32-abi Linux kernel, we could immediately reduce the 5~10% memory footprint without modifying the bootloader, firmware, or rootfs binaries.
The userspace ilp32-abi porting requires substantial effort, and customers want the porting step by step.

>
> Now you might want to compile a 32bit (ILP32) system that actually
> runs in 64bit mode (c/f x32) so that 64bit maths (long long) is
> more efficient - but that is a different issue.
We don't introduce any x32 userspace ABI here, only for the kernel self.

> (I suspect you'd need to change the process switch code to save
> all 64bits of the registers - but maybe not much else??)
Indeed, the pt_regs & context regs are the same as CONFIG_64BIT (64-bit width).

Best Regards