Re: [PATCH for-4.16] xen/arm: don't assign domU static-mem to dom0 as reserved-memory

[Stefano Stabellini <sstabellini@xxxxxxxxxx>]

On Tue, 9 Nov 2021, Julien Grall wrote:
> On 09/11/2021 00:48, Stefano Stabellini wrote:
> > From: Stefano Stabellini <stefano.stabellini@xxxxxxxxxx>
> > 
> > DomUs static-mem ranges are added to the reserved_mem array for
> > accounting, but they shouldn't be assigned to dom0 as the other regular
> > reserved-memory ranges in device tree.
> > 
> > In make_memory_nodes, fix the error by skipping banks with xen_domain
> > set to true in the reserved-memory array. Also make sure to use the
> > first valid (!xen_domain) start address for the memory node name.
> > 
> 
> This is a bug fix. So please add a Fixes tag. In this case, I think it should
> be:
> 
> Fixes: 41c031ff437b ("xen/arm: introduce domain on Static Allocation")

Thanks, will add



> > Signed-off-by: Stefano Stabellini <stefano.stabellini@xxxxxxxxxx>
> > ---
> > 
> > This patch should be considered for 4.16 as it fixes an incorrect
> > behavior.
> > 
> > The risk is low for two reasons:
> > - the change is simple
> > - make_memory_node is easily tested because it gets called at every
> >    boot, e.g. gitlab-ci and OSSTest exercise this path
> > 
> > I tested this patch successfully with and without xen,static-mem.
> > 
> > ---
> >   xen/arch/arm/domain_build.c | 13 +++++++++++--
> >   1 file changed, 11 insertions(+), 2 deletions(-)
> > 
> > diff --git a/xen/arch/arm/domain_build.c b/xen/arch/arm/domain_build.c
> > index 1fbafeaea8..56d3ff9d08 100644
> > --- a/xen/arch/arm/domain_build.c
> > +++ b/xen/arch/arm/domain_build.c
> > @@ -874,11 +874,17 @@ static int __init make_memory_node(const struct domain
> > *d,
> >       if ( mem->nr_banks == 0 )
> >           return -ENOENT;
> >   +    for ( i = 0; i < mem->nr_banks && mem->bank[i].xen_domain; i++ )
> > +        ;
> > +    /* No reserved-memory ranges to expose to Dom0 */
> I find this comment a bit misleading because make_memory_node() will also be
> called to create normal memory region. I would drop "reserved-memory" and add
> a comment on top of the loop explaining what the loop does.

Yeah, I agree, I moved it and changed it


> > +    if ( i == mem->nr_banks )
> > +        return 0;
> > +
> >       dt_dprintk("Create memory node (reg size %d, nr cells %d)\n",
> >                  reg_size, nr_cells);
> 
> I think you need to adjust nr_cells otherwise we would write garbagge in the
> DT if we need to exclude some regions.

Good point! Fixed in the next version


> >         /* ePAPR 3.4 */
> > -    snprintf(buf, sizeof(buf), "memory@%"PRIx64, mem->bank[0].start);
> > +    snprintf(buf, sizeof(buf), "memory@%"PRIx64, mem->bank[i].start);
> >       res = fdt_begin_node(fdt, buf);
> >       if ( res )
> >           return res;
> > @@ -888,11 +894,14 @@ static int __init make_memory_node(const struct domain
> > *d,
> >           return res;
> >         cells = &reg[0];
> > -    for ( i = 0 ; i < mem->nr_banks; i++ )
> > +    for ( ; i < mem->nr_banks; i++ )
> >       {
> >           u64 start = mem->bank[i].start;
> >           u64 size = mem->bank[i].size;
> >   +        if ( mem->bank[i].xen_domain )
> > +            continue;
> > +
> >           dt_dprintk("  Bank %d: %#"PRIx64"->%#"PRIx64"\n",
> >                      i, start, start + size);