Re: [PATCH v3 0/9] bulk: Replace CONFIG_SOFTMMU by !CONFIG_USER_ONLY/CONFIG_SYSTEM_ONLY

[Nicholas Piggin]

On Tue Jun 13, 2023 at 11:33 PM AEST, Philippe Mathieu-Daudé wrote:
> Missing review: 1, 7, 8
>
> Since v2:
> - Rebased
> - Added R-b tags
> - Rework i386_tr_init_disas_context() patch (Richard)
> - Dropped RFC prefix
>
> This series aims to clarify the CONFIG_[USER|SYSTEM] vs CONFIG_SOFTMMU
> confusion [*] by using explicit definitions, removing mentions of
> CONFIG_SOFTMMU in non-TCG code.
>
> We replace CONFIG_SOFTMMU by !CONFIG_USER_ONLY in C code and
> by CONFIG_SYSTEM_ONLY in meson config files.

I like the change in general, SOFTMMU does not read well (and is
not exactly correct for system code as pointed out).

Sorry for chiming in late and if I missed it, but was there
a reason not to define a complementary CONFIG_SYSTEM so system
code does not have to test !CONFIG_USER_ONLY and invert a bunch
of the tests?

Actually I thought you would have CONFIG_SYSTEM and CONFIG_USER
options and the _ONLY variants could be derivative for convenience,
but I'm probably missing some detail.

Thanks,
Nick