Re: radeon, amdgpu improvements for aarch64

[Taylor R Campbell <campbell+netbsd-tech-kern%mumble.net@localhost>]

> Date: Thu, 28 Dec 2023 01:02:08 +0100
> From: Tobias Nygren <tnn%NetBSD.org@localhost>
> 
> I've spent some time testing GPUs on aarch64 and fixing bugs.
> Attached are some patches to make things more usable.

Cool, thanks!  Some notes below.  Everything else seems fine.

> [2] /libdata/firmware/amdgpu is not shipped.
>     (Can we include these in the sets?)

amdgpu firmware should be included in the gpufw set, I think.

> --- sys/arch/evbarm/fdt/fdt_machdep.c	4 Aug 2023 09:06:33 -0000	1.106
> +++ sys/arch/evbarm/fdt/fdt_machdep.c	27 Dec 2023 22:07:38 -0000
> @@ -194,6 +194,11 @@ fdt_add_boot_physmem(const struct fdt_me
>  	bp->bp_start = atop(saddr);
>  	bp->bp_pages = atop(eaddr) - bp->bp_start;
>  	bp->bp_freelist = VM_FREELIST_DEFAULT;
> +#ifdef _LP64
> +	if (eaddr < (1UL<<40)) {

Tiny nit: I'd spell this as:

	if (eaddr < ((paddr_t)1 << 40)) {

or as:

	if (eaddr <= BITS(0,39)) {

Just to confirm: eaddr is inclusive here, right?  As in, if the range
were [0x10000, 0x20000), we would have saddr=0x10000 eaddr=0x1ffff (or
maybe eaddr=0x1f000), right?

> --- sys/external/bsd/drm2/include/drm/bus_dma_hacks.h	19 Jul 2022 23:19:44 -0000	1.25
> +++ sys/external/bsd/drm2/include/drm/bus_dma_hacks.h	27 Dec 2023 22:07:38 -0000
> @@ -99,6 +99,8 @@ bus_dmamem_pgfl(bus_dma_tag_t tag)
>  {
>  #if defined(__i386__) || defined(__x86_64__)
>  	return x86_select_freelist(tag->_bounce_alloc_hi - 1);
> +#elif defined(__aarch64__)
> +	return VM_FREELIST_FIRST1T;

This should look through the tag->_ranges to choose
VM_FREELIST_FIRST1T if it has been restricted with
bus_dmatag_subregion to lie in that range, and VM_FREELIST_DEFAULT if
not.

> --- sys/uvm/uvm_pglist.c	21 Dec 2021 08:27:49 -0000	1.90
> +++ sys/uvm/uvm_pglist.c	27 Dec 2023 22:06:10 -0000
> @@ -112,8 +112,9 @@ static int
>  uvm_pglistalloc_c_ps(uvm_physseg_t psi, int num, paddr_t low, paddr_t high,
>      paddr_t alignment, paddr_t boundary, struct pglist *rlist)
>  {
> -	signed int candidate, limit, candidateidx, end, idx, skip;
> -	int pagemask;
> +	long candidate, limit, candidateidx, end, idx;
> +	int skip;
> +	long pagemask;

I don't really have an issue with this but I think we may need to
switch int to pfn_t for page frame numbers much more systematically.
Curious how changing only start_hint from int to long helps?  (Commit
message doesn't explain.)