On Wed, 28 Oct 2020 08:44:59 +0100 Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote: > On 28/10/20 08:11, AlexChen wrote: > > The current 'DEBUG_KVM' macro is defined in many files, and turning on > > the debug switch requires code modification, which is very inconvenient, > > so this series add an option to configure to support the definition of > > the 'DEBUG_KVM' macro. > > In addition, patches 3 and 4 also make printf always compile in debug output > > which will prevent bitrot of the format strings by referring to the > > commit(08564ecd: s390x/kvm: make printf always compile in debug output). > > Mostly we should use tracepoints, but the usefulness of these printf > statements is often limited (except for s390 that maybe could make them > unconditional error_reports). I would leave this as is, maintainers can > decide which tracepoints they like to have. Looking at the s390 statements, they look more like something to put into trace events (the "unimplemented instruction" cases are handled gracefully, but the information might be interesting when developing.)
