RE: [PATCH] net: micro-optimize skb_datagram_iter

[David Laight <David.Laight@ACULAB.COM>]

From: Matthew Wilcox
> Sent: 16 June 2024 22:53
> 
> On Sun, Jun 16, 2024 at 09:51:05PM +0000, David Laight wrote:
> > From: Sagi Grimberg
> > > Sent: 16 June 2024 10:24
> > ...
> > > > [ 13.498663][ T189] EIP: usercopy_abort (mm/usercopy.c:102 (discriminator 12))
> > > > [   13.499424][  T194] usercopy: Kernel memory exposure attempt detected from kmap (offset 0,
> size
> > > 8192)!
> > >
> > > Hmm, not sure I understand exactly why changing kmap() to
> > > kmap_local_page() expose this,
> > > but it looks like mm/usercopy does not like size=8192 when copying for
> > > the skb frag.
> >
> > Can't a usercopy fault and have to read the page from swap?
> > So the process can sleep and then be rescheduled on a different cpu?
> > So you can't use kmap_local_page() here at all.
> 
> I don't think you understand how kmap_local_page() works.

Quite likely :-)

But I thought it was a cheap way of temporarily mapping a physical memory
page into the current cpu's page tables without having to do any IPI to
tell other cpu about the insert or removal?
Which would require that the process not be migrated, which pretty much
implies that pre-emption be disabled.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)