Re: [Qemu-devel] [PATCH] null-machine: Add support for the "-kernel" parameter

[Eduardo Habkost]

On Mon, Feb 27, 2017 at 12:43:23PM +0100, Thomas Huth wrote:
> On 25.01.2017 09:40, Thomas Huth wrote:
> > We can have basic support for the "-kernel" parameter quite easily
> > by using the generic loader device. This should be enough for most
> > boards which do not need special machine-specific magic for loading
> > a kernel (and for those that need special magic, the generic "none"
> > machine is likely not suitable for using it as an instruction set
> > simulator board anyway).
> > 
> > Signed-off-by: Thomas Huth <address@hidden>
> > ---
> >  PS: If we can't agree on using the generic loader here, I can also
> >      prepare a patch instead that simply prints out an error message
> >      if the user tried to use the "-kernel" parameter.
> > 
> >  hw/core/null-machine.c | 14 ++++++++++++++
> >  1 file changed, 14 insertions(+)
> > 
> > diff --git a/hw/core/null-machine.c b/hw/core/null-machine.c
> > index 27c8369..866e699 100644
> > --- a/hw/core/null-machine.c
> > +++ b/hw/core/null-machine.c
> > @@ -5,6 +5,7 @@
> >   *
> >   * Authors:
> >   *  Anthony Liguori   <address@hidden>
> > + *  Thomas Huth       <address@hidden>
> >   *
> >   * This work is licensed under the terms of the GNU GPL, version 2 or 
> > later.
> >   * See the COPYING file in the top-level directory.
> > @@ -16,6 +17,7 @@
> >  #include "qemu/error-report.h"
> >  #include "hw/hw.h"
> >  #include "hw/boards.h"
> > +#include "hw/core/generic-loader.h"
> >  #include "sysemu/sysemu.h"
> >  #include "exec/address-spaces.h"
> >  #include "cpu.h"
> > @@ -40,6 +42,18 @@ static void machine_none_init(MachineState *mch)
> >          memory_region_allocate_system_memory(ram, NULL, "ram", 
> > mch->ram_size);
> >          memory_region_add_subregion(get_system_memory(), 0, ram);
> >      }
> > +
> > +    /* Load kernel */
> > +    if (mch->kernel_filename) {
> > +        DeviceState *loader;
> > +
> > +        loader = qdev_create(sysbus_get_default(), TYPE_GENERIC_LOADER);
> > +        qdev_prop_set_string(loader, "file", mch->kernel_filename);
> > +        if (cpu) {
> > +            qdev_prop_set_uint32(loader, "cpu-num", cpu->cpu_index);
> > +        }
> > +        qdev_init_nofail(loader);
> > +    }
> >  }
> >  
> >  static void machine_none_machine_init(MachineClass *mc)
> 
> *ping*
> 
> Apparently the discussion has ceased ... can we get a consensus whether
> we want to support the "-kernel" parameter for the "none" machine or not?

I think Peter's point is still valid:

] If you just want "load a blob and start it" then we already
] have -device loader. Making -kernel have yet another set of
] semantics that this time depends on the machine being selected
] seems like a bad idea. If -kernel doesn't do what it does
] for the other machines of the same architecture then we should
] just not accept it.

If you add a mechanism to ensure "-machine none -kernel" has the
same behavior as the other machines in the same QEMU binary, then
I believe it will be OK.

I believe we don't need to make this work on all architectures at
the same time, though. We can do this using an optional
per-architecture kernel-loading hook.

-- 
Eduardo