On Wed, Oct 12, 2016 at 02:40:08PM +0100, Wei Liu wrote: > On Wed, Oct 12, 2016 at 09:29:04AM -0400, Konrad Rzeszutek Wilk wrote: > [...] > > > > > > Wouldn't it be just as easy, and more useful, to set a "has been > > > livepatched" flag, and return errors for all gcov hypercalls if its' set? > > > > > > I would expect most users to want to build a single hypervisor that can > > > be used for both gcov testing and live patching (under different > > > circumstances). > > > > I actually would welcome livepatching and gcov to see if the test-cases I > > wrote > > cover most of the code. > > > > I don't follow. Gcov doesn't give you a call graph -- if that's what > you're after.
It gives me an idea which functions/branches have run (not the livepatch itself - just the infrastructure around adding a livepatch, doing ELF checks, etc). And more importantly - which ones haven't run and need some more test-cases. > > > Adding in checking livepatch (common/livepatch.c: prepare_payload) to > > examine > > the .gcov_info and see if it matches the hypervisor one, is fine too. > > > > This then involves a non-trivial amount of work to figure out all the > corner cases. It's better to defer that to a later stage. Sure thing. And the !LIVEPATCH is fine for now. I just need to get an idea of what this would entail so it can get done. > > Wei. > > > > > > > -George > > > _______________________________________________ Xen-devel mailing list Xenfirstname.lastname@example.org https://lists.xen.org/xen-devel