Daniel P. Berrangé <[email protected]> writes: > On Mon, Sep 28, 2026 at 01:41:58PM +0200, Markus Armbruster wrote: >> Yonggang Luo <[email protected]> writes: >> >> > Extract common_ss, system_ss, and family-common objects into one >> > qemu-system binary. Only arm, aarch64, microblaze-softmmu are >> > listed; a new target must have arch_srcs equal to target_info_def >> > only (target-info-def.c). Config headers stay in arch_genh. >> > >> > - Union non-TARGET_* keys with y taking precedence >> > - Reject leftover arch_srcs on participating targets >> > - Reject TARGET_* conditions on common_ss and system_ss >> > - Link SDL, install the binary, and apply Darwin entitlements >> > >> > Adding a target that still has per-target sources fails meson, for >> > example: >> > >> > combined qemu-system leftover for : >> > expected: >> > target-info-def.c >> > arch_srcs: >> > target/microblaze/mmu.c >> > target/microblaze/machine.c >> > target-info-def.c >> > >> > Signed-off-by: Yonggang Luo <[email protected]> >> >> Watch query-target lie: >> >> $ qemu-system -S -nodefaults -display none -qmp stdio >> {"QMP": {"version": {"qemu": {"micro": 50, "minor": 1, "major": 11}, >> "package": "v11.1.0-1902-g88f617913a"}, "capabilities": ["oob"]}} >> qemu-system: warning: nic xlnx.xps-ethernetlite.0 has no peer >> {"execute": "qmp_capabilities", "arguments": {"enable": ["oob"]}} >> {"return": {}} >> {'execute': 'query-target'} >> --> {"return": {"arch": "aarch64"}} >> >> Perhaps you can fix this by reordering patches. If you can't, >> deficiencies like this one need a prominent mention in the commit >> message. > > What do you think the behaviour of 'query-target' ought to be in > this scenario ? > > The 'none' machine has no associated target. Should we add a > 'none' or 'unspecified' target to SysEmuTarget, or should we > just declare that "query-target" will report an error if no > target is activate.
Both solutions could work. To have a preference, I'd have to think some more. Thinking costs extra! ;) For what it's worth, the next patch adds member "none" to SysEmuTarget. > I'd probably lean towards the latter. Latter in this series > is a new 'query-targets' command that can enumerate all > available targets in 'qemu-system', so 'query-target' merely > tells us which is active (if any). Perhaps 'query-targets' > should in fact tell us that straightaway, such that we can > declare 'query-target' to be obsolete (and even deprecate > it) > > With regards, > Daniel
