Okay, I need to classify this bug report about QEMU's increased memory usage when using TCG. Let me break it down step by step. First, the problem is that instances with low memory allocations (like 256MB) are using much more RSS ( Resident Set Size) when running with accel=tcg. This wasn't an issue with older versions of QEMU or when using KVM acceleration instead. The user provided steps to reproduce: deploying devstack, starting a Cirros instance with a 256MB flavor, and observing the RSS jump to around 512MB. They expected it to stay under 256MB as it did before. Looking at the possible categories: - Instruction: Faulty instructions in code. - Syscall: Errors related to system calls. - Runtime: Other runtime errors. This issue is about memory usage behavior during runtime, specifically with TCG acceleration. It's not pointing to a specific instruction error or a syscall problem. Instead, it's how the memory is being managed while the instance runs, which falls under runtime issues. runtime