Re: [Qemu-devel] [PATCH v3 08/10] migration: handle the error condition properly

[Peter Xu <peterx@redhat.com>]

On Wed, Aug 08, 2018 at 03:23:22PM +0800, Xiao Guangrong wrote:
> 
> 
> On 08/08/2018 02:56 PM, Peter Xu wrote:
> > On Wed, Aug 08, 2018 at 02:29:52PM +0800, Xiao Guangrong wrote:
> > > 
> > > 
> > > On 08/08/2018 01:08 PM, Peter Xu wrote:
> > > > On Tue, Aug 07, 2018 at 05:12:07PM +0800, guangrong.xiao@gmail.com wrote:
> > > > > From: Xiao Guangrong <xiaoguangrong@tencent.com>
> > > > > 
> > > > > ram_find_and_save_block() can return negative if any error hanppens,
> > > > > however, it is completely ignored in current code
> > > > 
> > > > Could you hint me where we'll return an error?
> > > > 
> > > 
> > > I think control_save_page() may return a error condition but i am not
> > > good at it ... Other places look safe _currently_. These functions were
> > > designed to have error returned anyway.
> > 
> > Ah, the RDMA codes...
> > 
> > Then I feel like this patch would be more suitable to be put into some
> > of the RDMA series - at least we'd better be clear about what errors
> > we're going to capture.  For non-RDMA, it seems a bit helpless after
> > all - AFAIU we're depending on the few qemu_file_get_error() calls to
> > detect output errors.
> 
> So, are you talking about to modify the semantic of these functions,
> ram_save_host_page(), ram_save_target_page(), etc, and make them
> be:
> "Returns the number of pages written where zero means no dirty pages,
> error conditions are indicated in the QEMUFile @rs->file if it
> happened."
> 
> If it's what you want, i will update the comments and make the
> implementation more clear to reflect this fact for these
> functions

Not really; I am just unclear about how this patch could help current
code, however I have no objection on the content.  Let's see whether
Dave or Juan would like it.

Thanks,

-- 
Peter Xu