lkml.org 
[lkml]   [2022]   [May]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: linux-next: Tree for May 3
On Wed, May 11, 2022 at 04:10:34PM +0200, Christoph Hellwig wrote:
> On Wed, May 11, 2022 at 05:08:52PM +0300, Mike Rapoport wrote:
> > I guess the default to use memblock_alloc_low() backfires on system with
> > physical memory living at 0x1000200000:
> >
> > [ 0.000000] Early memory node ranges
> > [ 0.000000] node 0: [mem 0x0000001000200000-0x000000103fffffff]
> >
> > The default limit for "low" memory is 0xffffffff and there is simply no
> > memory there.
>
> Is there any way to ask memblock for a specific address limit?
> swiotlb just wants <= 32-bit by default. With the little caveat
> that it should be 32-bit addressable for all devices, and we don't
> know the physical to dma address mapping at time of allocation.

There is

void *memblock_alloc_try_nid(phys_addr_t size, phys_addr_t align,
phys_addr_t min_addr, phys_addr_t max_addr,
int nid);

that lets caller to specify min and max limits

Presuming that devices see [0x1000200000-0x103fffffff] as
[0x200000-0x3fffffff] we may try something like

min = memblock_start_of_DRAM();
max = min + 0xffffffff;

if (flags & SWIOTLB_ANY)
max = MEMBLOCK_ALLOC_ACCESSIBLE;

tlb = memblock_alloc_try_nid(bytes, PAGE_SIZE, min, max, NUMA_NO_NODE);

--
Sincerely yours,
Mike.

\
 
 \ /
  Last update: 2022-05-11 16:38    [W:0.229 / U:0.060 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site