On Fri, Jun 27, 2025 at 02:38:19PM +0200, Jocelyn Falempe wrote:
> On 32bits ARM, u64 divided by a constant is not optimized to a
> multiply by inverse by the compiler [1].
> So do the multiply by inverse explicitly for this architecture.
>
> Link: https://github.com/llvm/llvm-project/issues/37280 [1]
> Reported-by: Andrei Lalaev <andrey.lal...@gmail.com>
> Closes:
> https://lore.kernel.org/dri-devel/c0a2771c-f3f5-4d4c-aa82-d673b3c5c...@gmail.com/
> Fixes: 675008f196ca ("drm/panic: Use a decimal fifo to avoid u64 by u64
> divide")
> Signed-off-by: Jocelyn Falempe <jfale...@redhat.com>
> ---
> drivers/gpu/drm/drm_panic_qr.rs | 22 +++++++++++++++++++++-
> 1 file changed, 21 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/drm_panic_qr.rs b/drivers/gpu/drm/drm_panic_qr.rs
> index dd55b1cb764d..774a17de4f2f 100644
> --- a/drivers/gpu/drm/drm_panic_qr.rs
> +++ b/drivers/gpu/drm/drm_panic_qr.rs
> @@ -381,6 +381,26 @@ struct DecFifo {
> len: usize,
> }
>
> +// On arm32 architecture, dividing an `u64` by a constant will generate a
> call
> +// to `__aeabi_uldivmod` which is not present in the kernel.
> +// So use the multiply by inverse method for this architecture.
I think the problem here is the u64 by u64 division. u64 by u32 should work.
Unfortunately Rust doesn't seem to have a way to perform a mixed-type division.
We already have optimized C/ASM helpers for u64 divisions. For example
div_u64() does such an u64 by u32 division.
While it may be slower than the inverse multiplication, it is less code, easier
to understand and will work for all architectures automatically.
> +fn div10(val: u64) -> u64 {
> + if cfg!(target_arch = "arm") {
> + let val_h = val >> 32;
> + let val_l = val & 0xFFFFFFFF;
> + let b_h: u64 = 0x66666666;
> + let b_l: u64 = 0x66666667;
> +
> + let tmp1 = val_h * b_l + ((val_l * b_l) >> 32);
> + let tmp2 = val_l * b_h + (tmp1 & 0xffffffff);
> + let tmp3 = val_h * b_h + (tmp1 >> 32) + (tmp2 >> 32);
> +
> + tmp3 >> 2
> + } else {
> + val / 10
> + }
> +}
> +
> impl DecFifo {
> fn push(&mut self, data: u64, len: usize) {
> let mut chunk = data;
> @@ -389,7 +409,7 @@ fn push(&mut self, data: u64, len: usize) {
> }
> for i in 0..len {
> self.decimals[i] = (chunk % 10) as u8;
> - chunk /= 10;
> + chunk = div10(chunk);
> }
> self.len += len;
> }
>
> base-commit: 3529cb5ab16b4f1f8bbc31dc39a1076a94bd1e38
> --
> 2.49.0
>
