Re: [PULL 7/9] hw/guest-loader: pass random seed to fdt

[Jason A. Donenfeld]

Hi Alex,

On Thu, Jul 21, 2022 at 08:36:10PM +0100, Alex Bennée wrote:
> 
> Paolo Bonzini <pbonzini@redhat.com> writes:
> 
> > From: "Jason A. Donenfeld" <Jason@zx2c4.com>
> >
> > If the FDT contains /chosen/rng-seed, then the Linux RNG will use it to
> > initialize early. Set this using the usual guest random number
> > generation function. This FDT node is part of the DT specification.
> >
> > Cc: Alex Bennée <alex.bennee@linaro.org>
> > Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
> > Message-Id: <20220719121559.135355-1-Jason@zx2c4.com>
> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> > ---
> >  hw/core/guest-loader.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/hw/core/guest-loader.c b/hw/core/guest-loader.c
> > index 391c875a29..4f8572693c 100644
> > --- a/hw/core/guest-loader.c
> > +++ b/hw/core/guest-loader.c
> > @@ -31,6 +31,7 @@
> >  #include "hw/qdev-properties.h"
> >  #include "qapi/error.h"
> >  #include "qemu/module.h"
> > +#include "qemu/guest-random.h"
> >  #include "guest-loader.h"
> >  #include "sysemu/device_tree.h"
> >  #include "hw/boards.h"
> > @@ -46,6 +47,7 @@ static void loader_insert_platform_data(GuestLoaderState 
> > *s, int size,
> >      g_autofree char *node = g_strdup_printf("/chosen/module@0x%08" PRIx64,
> >                                              s->addr);
> >      uint64_t reg_attr[2] = {cpu_to_be64(s->addr), cpu_to_be64(size)};
> > +    uint8_t rng_seed[32];
> >  
> >      if (!fdt) {
> >          error_setg(errp, "Cannot modify FDT fields if the machine has 
> > none");
> > @@ -55,6 +57,9 @@ static void loader_insert_platform_data(GuestLoaderState 
> > *s, int size,
> >      qemu_fdt_add_subnode(fdt, node);
> >      qemu_fdt_setprop(fdt, node, "reg", &reg_attr, sizeof(reg_attr));
> >  
> > +    qemu_guest_getrandom_nofail(rng_seed, sizeof(rng_seed));
> > +    qemu_fdt_setprop(fdt, node, "rng-seed", rng_seed, sizeof(rng_seed));
> > +
> 
> Why are we inserting this here? The guest-loader is only building on
> what the machine type has already constructed which in the case of -M
> virt for riscv and ARM already has code for this.

Wish you would have replied to the list patch before Paolo queued it.

I don't know this mechanism well but I assumed it would pass a unique
seed to each chained loader. Let me know if I'm misunderstanding the
purpose; I have no qualms about dropping this patch.

Jason