Re: [PATCH v2] dma-buf: Split sgl into page-aligned 2G chunks

[Jason Gunthorpe]

On Tue, Jun 23, 2026 at 09:44:46AM +0100, David Laight wrote:
> On Tue, 23 Jun 2026 01:54:59 +0000
> David Hu <xuehaohu@xxxxxxxxxx> wrote:
> 
> > Currently, `fill_sg_entry()` splits the scatterlist using `UINT_MAX`.
> > This creates a non-page-aligned DMA length (`0xFFFFFFFF`) for the
> > first entry, resulting in non-page-aligned DMA addresses for all
> > subsequent entries.
> 
> There is a separate issue of whether this code is even needed at all.
> Where can transfers over 2G (never mind 4G) actually come from.

This is DMABUF land, you really can alocate DMABUFS of huge amounts of
physical memory, VFIO does this reliably and trivially for example. It
wouldn't come from the physical allocator.

So yes, these scenarios need to work in this code.

Jason