On Mon, Mar 17, 2025 at 06:00:14PM +0530, Prasad Pandit wrote: > Hi, > > On Fri, 14 Mar 2025 at 01:40, Peter Xu <pet...@redhat.com> wrote: > >+ save_section_header(f, se, QEMU_VM_SECTION_PART); > > + ram_save_zero_page(f, se->opaque); > >I'll stop requesting a why here... > > * Earlier in this thread you mentioned 'We need a header'. I took it > as a 'RAM page' header, not save_section_header(). Section type
The question is not about the header, it's about the zero page, and why we send this zero page out of blue. IIIUC it can corrupt a page if it uses whatever cached in the previous round in the pss, and if it used to be a non-zero. > (QEMU_VM_COMMAND) was sent by qemu_savevm_command_send() as well. But it should be for generic VM operations. We have a few outliers but they're too special. For this case we'd better not add more outliers, because neither it is special, nor necessary. Thanks, -- Peter Xu
