Re: [Xen-devel] [PATCH 2/2] docs/hypervisor-guide: Code Coverage

[Julien Grall <julien.grall@xxxxxxx>]

Hi,

On 04/04/2019 13:46, Andrew Cooper wrote:
> On 04/04/2019 13:29, Julien Grall wrote:
> > On 04/04/2019 13:07, Artem Mygaiev wrote:
> > > Hello Andrew
> > Hi,
> >
> > > Minor note below.
> > >
> > > On Wed, 2019-04-03 at 16:00 +0100, Andrew Cooper wrote:
> > >
> > > [snip]
> > >
> > > > +  For ARM builds, while Xen will compile with ``CONFIG_COVERAGE``
> > > > enabled, the
> > > > +  resulting binary will not successfully boot if it exceeds 2MB in
> > > > size.
> > > > +  Xen's early memory management code needs adjusting to resolve this
> > > > issue.
> > > Julien mentioned that on staging the binary is 1.5MB only so we must be
> > > able to run it without changes to Xen mem management. I will ask
> > > someone to try it today or tomorrow and get back with results
> > I tried with a different .config and Xen will use 1.8MB. I have
> > computed the size using (_start - _end).
> >
> > This is dangerously close to 2MB and it might be possible be
> > possible that we reach the limit with another .config.
> >
> > I think Andrew's statement is correct. It does not say the binary
> > will always be above 2MB but it might exceeds it. It only warns
> > that it may happen and fail to boot.
> >
> > The boot failure may not be obvious for the user. I wrote the
> > patch below to fail when linking instead. We can consider it
> > for backporting as well:
> >
> > commit 256528c97b4c265911d960073facd4fdffcc650a
> > Author: Julien Grall <julien.grall@xxxxxxx>
> > Date:   Thu Apr 4 13:22:46 2019 +0100
> >
> >      xen/arm: Check Xen size when linking
> >      
> >      The linker will happily link Xen if it is bigger than what we can handle
> >      (e.g 2MB). This will result to unexpected failure after boot.
> >      
> >      This unexpected failure can be prevented by forbidding linking if Xen is
> >      bigger than 2MB.
> >      
> >      Signed-off-by: Julien Grall <julien.grall@xxxxxxx>
> > diff --git a/xen/arch/arm/xen.lds.S b/xen/arch/arm/xen.lds.S
> > index 1e72906477..28dc2a6070 100644
> > --- a/xen/arch/arm/xen.lds.S
> > +++ b/xen/arch/arm/xen.lds.S
> > @@ -242,3 +242,4 @@ ASSERT( _end_boot - start <= PAGE_SIZE, "Boot code is larger 
> > than 4K")
> >    */
> >   ASSERT(IS_ALIGNED(__init_begin,     4), "__init_begin is misaligned")
> >   ASSERT(IS_ALIGNED(__init_end,       4), "__init_end is misaligned")
> > +ASSERT(_end - start < XEN_SLOT_SIZE, "Xen is larger than 2M")
> I think you probably want an opencoded MB(2) here, or the comment will
> become stale as soon as the other issues get fixed.  That way, someone
> will be forced to adjust this comment when the limit is lifted.
I would prefer to make the message agnostic. This could potentially allow us to 
make XEN_SLOT_SIZE configurable.
> Given a patch like this, I'll drop the warning from the docs.
>
> As an aside, do you happen to know which versions of GCC are expected to
> work.  Alternatively, shall I just say that ARM isn't supported, by
> virtue of it never having actually been tested.
I have no idea. I haven't tried more than building GCOV.

It would be best if it is marked as unsupported until someone actually test it.

Cheers,

--
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel