Re: [Qemu-devel] [PULL 07/22] migration: Add an ability to ignore shared RAM blocks

[Dr. David Alan Gilbert]

* Peter Maydell (address@hidden) wrote:
> On Wed, 6 Mar 2019 at 11:55, Dr. David Alan Gilbert (git)
> <address@hidden> wrote:
> >
> > From: Yury Kotov <address@hidden>
> >
> > If ignore-shared capability is set then skip shared RAMBlocks during the
> > RAM migration.
> > Also, move qemu_ram_foreach_migratable_block (and rename) to the
> > migration code, because it requires access to the migration capabilities.
> >
> 
> > --- a/migration/rdma.c
> > +++ b/migration/rdma.c
> > @@ -644,7 +644,7 @@ static int qemu_rdma_init_ram_blocks(RDMAContext *rdma)
> >
> >      assert(rdma->blockmap == NULL);
> >      memset(local, 0, sizeof *local);
> > -    qemu_ram_foreach_migratable_block(qemu_rdma_init_one_block, rdma);
> > +    foreach_not_ignored_block(qemu_rdma_init_one_block, rdma);
> >      trace_qemu_rdma_init_ram_blocks(local->nb_blocks);
> >      rdma->dest_blocks = g_new0(RDMADestBlock,
> >                                 rdma->local_ram_blocks.nb_blocks);
> 
> Hi. This change causes Coverity to gripe (CID 1399413) because
> the return value from foreach_not_ignored_block() is ignored
> here but it is checked on every other use of the function.
> 
> This is one of those Coverity errors where it's just using a
> sometimes-wrong heuristic, so we could just mark it as a false
> positive (AFAICT qemu_rdma_init_one_block() always returns 0),
> but OTOH rdma_add_block() and qemu_rdma_init_one_block()
> carefully pipe through a return value, so maybe it's worth assert()ing
> in case somebody changes rdma_add_block() to maybe fail later?
> 
> I don't think there's much in it -- let me know if you just want
> me to mark the issue as a false positive.

I'll patch it to check it, it's an easy enough check.
(Although how well a failure cleans up in practice could be fun).

Dave

> thanks
> -- PMM
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK