[Qemu-devel] Re: [RFC PATCH 7/6] savevm: Create a new continue flag to avoid resending block name

[Paolo Bonzini]

On 06/08/2010 11:32 PM, Alex Williamson wrote:
> On Tue, 2010-06-08 at 15:11 -0600, Alex Williamson wrote:
> > Allows us to compress the protocol a bit.
> ...
> > @@ -284,6 +290,33 @@ int ram_save_live(Monitor *mon, QEMUFile *f, int stage, 
> > void *opaque)
> >       return (stage == 2)&&  (expected_time<= migrate_max_downtime());
> >   }
> >
> > +static inline void *host_from_stream_offset(QEMUFile *f,
> > +                                            ram_addr_t offset,
> > +                                            int flags)
>
> This probably shouldn't be inline.  When sending, we'll continue from
> COMPRESS or PAGE.  We'd get out of sync on the recv if the compiler
> created separate static blocks.  Compiler folks correct me if this can't
> happen.

There is only one static variable even if the inlining happens in 
multiple places.  In fact, inlining is totally transparent.

Example:

$ cat > x.c <<\EOF
#ifndef force_inline
#define force_inline inline __attribute__((always_inline))
#endif

static force_inline int f()
{
        static int i;
        return i++;
}

int main()
{
        printf ("%d\n", f());
        printf ("%d\n", f());
        printf ("%d\n", f());
        printf ("%d\n", f());
}
EOF
$ gcc -o - -Dforce_inline= -S x.c | egrep -cw _?f
7
$ gcc -o - -S x.c | egrep -cw _?f
0
$ gcc x.c && ./a.out
0
1
2
3

Paolo