Re: [Qemu-devel] [PATCH v7 02/26] exec: add ram_debug_ops support

[Brijesh Singh]

On 02/07/2018 10:51 AM, Paolo Bonzini wrote:
> On 07/02/2018 17:06, Brijesh Singh wrote:
> > @@ -3148,7 +3152,11 @@ MemTxResult flatview_read_continue(FlatView *fv, hwaddr 
> > addr,
> >           } else {
> >               /* RAM case */
> >               ptr = qemu_ram_ptr_length(mr->ram_block, addr1, &l, false);
> > -            memcpy(buf, ptr, l);
> > +            if (attrs.debug && mr->ram_debug_ops) {
> > +                mr->ram_debug_ops->read(buf, ptr, l, attrs);
> > +            } else {
> > +                memcpy(buf, ptr, l);
> > +            }
> >           }
> >   
> >           if (release_lock) {
>
> You also need to tweak flatview_read in include/exec/memory.h (probably
> by adding an "&& !attrs.debug", which leaves the mr->ram_debug_ops->read
> to the slow path in exec.c).


thanks, I will make the changes in flatview_read to take slow path when 
debug is enabled.


> > @@ -3218,11 +3226,13 @@ void cpu_physical_memory_rw(hwaddr addr, uint8_t *buf,
> >   
> >   enum write_rom_type {
> >       WRITE_DATA,
> > +    READ_DATA,
> >       FLUSH_CACHE,
> >   };
> >   
> > -static inline void cpu_physical_memory_write_rom_internal(AddressSpace *as,
> > -    hwaddr addr, const uint8_t *buf, int len, enum write_rom_type type)
> > +static inline void cpu_physical_memory_rw_internal(AddressSpace *as,
> > +    hwaddr addr, uint8_t *buf, int len, MemTxAttrs attrs,
> > +    enum write_rom_type type)
> >   {
> >       hwaddr l;
> >       uint8_t *ptr;
>
> I wonder if READ_DATA and WRITE_DATA still need to go down to
> cpu_physical_memory_rw_internal.  Maybe you can just call
> address_space_rw with &address_space_memory as the address space, and
> "(MemTxAttrs) { .debug = 1 }" as the attributes.


I will take a look to see if I can remove passing down the READ_DATA and 
WRITE_DATA.