Am 07.01.2012 04:14, schrieb Peter Maydell: > On 6 January 2012 20:11, Andreas Färber <afaer...@suse.de> wrote: >> Not sure how hardcoding the cpu_model would work with CPU features, >> would they be still included or stripped out before. Peter? > > Interesting question. It's certainly more likely to work to have > a board where the only tweak you made to the CPU was to disable > Neon, say, but I'm not sure "likely to work" is a very firm criterion.
What I meant was, if sometime we allow, e.g., '-cpu cortex-m4,+FPU', will cpu_model technically contain "cortex-m4" or "cortex-m4,+FPU"? If the latter, then we should not hardcode cpu_model anywhere. I'm sure we can always find a scenario that doesn't work, but the emulated instructions are rather unlikely to conflict with a machine-specific memory layout and device instantiation, given that the guest matches the emulated CPU features. Andreas -- SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg
