>> Oh, and one more reason why the proposal in this patch is inconsistent: >> >> When migrating resizable memory regions (RAM_SAVE_FLAG_MEM_SIZE) we >> store the block->used_length (ram_save_setup()) and use that value to >> resize the region on the target (ram_load_precopy() -> qemu_ram_resize()). >> >> This will be the value the callback will be called with. Page aligned. >> > > Sorry, I didn’t quite get that point and not sure how "req_length" approach > will affect the migration.
The issue is that on migration, you will lose the sub-page size either way. So your callback will be called - on the migration source with a sub-page size (via memory_region_ram_resize() from e.g., hw/i386/acpi-build.c) - on the migration target with a page-aligned size (via qemu_ram_resize() from migration/ram.c) So this is inconsistent, especially when migrating. Is there a way to get access to the sub-page size without passing it through the callback? Like in fw_cfg_modify_file() do some fancy lookup and use the sub-page size instead of the passed size? (might have to be stored somewhere and refetched - and migrated) -- Thanks, David / dhildenb
