On Thu, Oct 26, 2017 at 4:36 PM, Peter Maydell <peter.mayd...@linaro.org> wrote: > This has *never* been guaranteed or indeed true. It's just happened > to be true for the x86 PC machine and perhaps for some others that > you've been using. For instance, the MIPS boston board only has > one uart. highbank has one. integratorcp has 2. The microblaze > boards only have one uart. And so on. In cases where we're > modelling real hardware, if the real board only has one UART > then we're only going to provide one UART, so in some cases you > just have to deal with that and have some way in your communication > protocol to multiplex across a single communications channel.
Interesting. Malta provides quite a few, which is what I'm using on MIPS, so I guess I've benefited from the uniformity. Since the ARM virt machine is virtual, it seems like it'd be worthwhile to make it broadly useful, and adding a few more serial outputs is basically free. Right now on build.wireguard.com, I'm just using this horrendous patch that hopefully hurts to look at: https://א.cc/iKBdlvI3/patch (since I only need 2). The goal would be for everything else to remain the same in the deployment, but to not have to use that patch. >> Could we just do the simple patch of adding four pl011s in there? > > I'm still thinking about this. (You definitely can't have four, > though, because we already have 2 in the trustzone-enabled > config, so we can only add another 2.) The trustzone one could be the 5th? Since it's only useful for trustzone.