----- Original Message ----- > From: "Alan Bateman" <alan.bate...@oracle.com> > To: "Remi Forax" <fo...@univ-mlv.fr> > Cc: "hotspot-runtime-dev" <hotspot-runtime-...@openjdk.org>, > "serviceability-dev" <serviceability-dev@openjdk.org> > Sent: Thursday, May 11, 2023 3:46:59 PM > Subject: Re: RFR: 8307478: Implementation of Prepare to Restrict The Dynamic > Loading of Agents
> On 11/05/2023 14:28, Remi Forax wrote: >> : >> I believe the flag should mention that what is traced is just the dynamic >> instrumentation, static instrumentation is not an issue, >> perhaps "-Djdk.instrument.traceDynamicUsage=true" >> > There is static, load-time, and dynamic instrumentation. The > Instrumentation API supports load-time and dynamic. If there is static > instrumentation going on then it happens before the class bytes are > loaded. Is that what you mean? > > Or maybe you mean how the agent is deployed? The opt-in tracing right > now is independent of how the agent is started so it works for > -javaagent, Launcher-Agent-Class, and agents that are dynamically loaded > via the attach mechanism. Maybe you are arguing that it should be > limited to agents that are dynamically loaded? yes ! I did not take a look to the patch, i was thinking that only dynamically loaded agents were traced. > > -Alan RĂ©mi