On 07/25/2012 06:31 AM, Liu Ping Fan wrote:
> From: Liu Ping Fan <[email protected]>
>
> acquire device's refcnt with qemu_device_tree_mutex rwlock, so we
> can safely handle it when mmio dispatch.
>
> If in radix-tree, leaf is subpage, then move further step to acquire
> opaque which is the type --DeiveState.
>
>
> +static MemoryRegionSection *subpage_get_backend(subpage_t *mmio,
> + target_phys_addr_t addr)
> +{
> + MemoryRegionSection *section;
> + unsigned int idx = SUBPAGE_IDX(addr);
> +
> + section = &phys_sections[mmio->sub_section[idx]];
> + return section;
> +}
> +
> +void *get_backend(MemoryRegion* mr, target_phys_addr_t addr)
> +{
> + MemoryRegionSection *p;
> + Object *ret;
> +
> + if (mr->subpage) {
> + p = subpage_get_backend(mr->opaque, addr);
> + ret = OBJECT(p->mr->opaque);
> + } else {
> + ret = OBJECT(mr->opaque);
> + }
> + return ret;
> +}
> +
You don't enforce that mr->opaque is an object.
The name 'backend' is inappropriate here (actually I don't like it
anywhere). If we can s/opaque/object/ (and change the type too, we can
call it get_object() (and return an Object *).
> static const MemoryRegionOps subpage_ops = {
> .read = subpage_read,
> .write = subpage_write,
> @@ -3396,13 +3420,25 @@ void cpu_physical_memory_rw(target_phys_addr_t addr,
> uint8_t *buf,
> uint32_t val;
> target_phys_addr_t page;
> MemoryRegionSection *section;
> + Object *bk;
>
> while (len > 0) {
> page = addr & TARGET_PAGE_MASK;
> l = (page + TARGET_PAGE_SIZE) - addr;
> if (l > len)
> l = len;
> +
> + qemu_rwlock_rdlock_devtree();
> section = phys_page_find(page >> TARGET_PAGE_BITS);
Does the devtree lock also protect the data structures accessed by
phys_page_find()? Seems wrong.
> + if (!(memory_region_is_ram(section->mr) ||
> + memory_region_is_romd(section->mr)) && !is_write) {
> + bk = get_backend(section->mr, addr);
> + object_ref(bk);
> + } else if (!memory_region_is_ram(section->mr) && is_write) {
> + bk = get_backend(section->mr, addr);
> + object_ref(bk);
> + }
Best push the ugliness that computes bk into a small helper, and do just
the object_ref() here.
> + qemu_rwlock_unlock_devtree();
>
> if (is_write) {
> if (!memory_region_is_ram(section->mr)) {
> @@ -3426,6 +3462,7 @@ void cpu_physical_memory_rw(target_phys_addr_t addr,
> uint8_t *buf,
> io_mem_write(section->mr, addr1, val, 1);
> l = 1;
> }
> + object_unref(bk);
> } else if (!section->readonly) {
> ram_addr_t addr1;
> addr1 = memory_region_get_ram_addr(section->mr)
> @@ -3464,6 +3501,7 @@ void cpu_physical_memory_rw(target_phys_addr_t addr,
> uint8_t *buf,
> stb_p(buf, val);
> l = 1;
> }
> + object_unref(bk);
> } else {
> /* RAM case */
> ptr = qemu_get_ram_ptr(section->mr->ram_addr
> diff --git a/memory.h b/memory.h
> index 740c48e..e5a86dc 100644
> --- a/memory.h
> +++ b/memory.h
> @@ -748,6 +748,8 @@ void memory_global_dirty_log_stop(void);
>
> void mtree_info(fprintf_function mon_printf, void *f);
>
> +void *get_backend(MemoryRegion* mr, target_phys_addr_t addr);
> +
This is a private interface, shouldn't be in memory.h.
--
error compiling committee.c: too many arguments to function
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
