Re: [PATCH v2 1/2] nvme-pci: Use size_t for length fields to handle larger sizes

[Leon Romanovsky <leon@kernel.org>]

On Wed, Nov 19, 2025 at 01:36:15PM +0000, David Laight wrote:
> On Wed, 19 Nov 2025 11:55:16 +0200
> Leon Romanovsky <leon@kernel.org> wrote:
> 
> > On Tue, Nov 18, 2025 at 06:03:11AM +0100, Christoph Hellwig wrote:
> > > On Mon, Nov 17, 2025 at 09:22:43PM +0200, Leon Romanovsky wrote:  
> > > > From: Leon Romanovsky <leonro@nvidia.com>
> > > > 
> > > > This patch changes the length variables from unsigned int to size_t.
> > > > Using size_t ensures that we can handle larger sizes, as size_t is
> > > > always equal to or larger than the previously used u32 type.
> > > > 
> > > > Originally, u32 was used because blk-mq-dma code evolved from
> > > > scatter-gather implementation, which uses unsigned int to describe length.
> > > > This change will also allow us to reuse the existing struct phys_vec in places
> > > > that don't need scatter-gather.
> > > > 
> > > > Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
> > > > Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
> > > > ---
> > > >  block/blk-mq-dma.c      | 8 ++++++--
> > > >  drivers/nvme/host/pci.c | 4 ++--
> > > >  2 files changed, 8 insertions(+), 4 deletions(-)
> > > > 
> > > > diff --git a/block/blk-mq-dma.c b/block/blk-mq-dma.c
> > > > index e9108ccaf4b0..e7d9b54c3eed 100644
> > > > --- a/block/blk-mq-dma.c
> > > > +++ b/block/blk-mq-dma.c
> > > > @@ -8,7 +8,7 @@
> > > >  
> > > >  struct phys_vec {
> > > >  	phys_addr_t	paddr;
> > > > -	u32		len;
> > > > +	size_t		len;
> > > >  };  
> > > 
> > > So we're now going to increase memory usage by 50% again after just
> > > reducing it by removing the scatterlist?  
> > 
> > It is slightly less.
> > 
> > Before this change: 96 bits
> 
> Did you actually look?

No, I simply performed sizeof(phys_addr_t) + sizeof(size_t).

> There will normally be 4 bytes of padding at the end of the structure.
> 
> About the only place where it will be 12 bytes is a 32bit system with
> 64bit phyaddr that aligns 64bit items on 32bit boundaries - so x86.

So does it mean that Christoph's comment about size increase is not correct?

Thanks

> 
> 	David
> 
> > After this change (on 64bits system): 128 bits.
> > 
> > It is 33% increase per-structure.
> > 
> > So what is the resolution? Should I drop this patch or not?
> > 
> > Thanks 
> > 
>