The configure script doesn't test for presence of TPM device or support. It activates TPM support if not explicitly disabled, and disables TPM support if explicitly enabled on Windows. With TPM support compiled in, it causes an assertion failure on launch of qemu at util/async.c:669 qemu_set_current_aio_context() !my_aiocontext. I haven't yet figured out why though, but disabling TPM might be best, and there's precedent for it as it's disabled if compiling for Windows.
În dum., 4 iul. 2021 la 14:29, Richard Zak <richard.j....@gmail.com> a scris: > > În dum., 4 iul. 2021 la 13:11, Peter Maydell <peter.mayd...@linaro.org> a > scris: > >> On Sun, 4 Jul 2021 at 17:44, Richard Zak <richard.j....@gmail.com> wrote: >> > >> > Use system capstone, for which a port is maintained by Haiku. Disable >> TPM which isn't supported. >> > >> > Signed-off-by: Richard Zak <richard.j....@gmail.com> >> > --- >> > configure | 3 +++ >> > 1 file changed, 3 insertions(+) >> > >> > diff --git a/configure b/configure >> > index e799d908a3..c928071f69 100755 >> > --- a/configure >> > +++ b/configure >> > @@ -358,6 +358,7 @@ oss_lib="" >> > bsd="no" >> > linux="no" >> > solaris="no" >> > +haiku="no" >> > profiler="no" >> > cocoa="auto" >> > softmmu="yes" >> > @@ -769,6 +770,8 @@ SunOS) >> > ;; >> > Haiku) >> > haiku="yes" >> > + tpm="no" >> >> If the autodetect for tpm doesn't get this right, we should fix >> the autodetect. >> >> As a general principle we prefer to avoid "do this specific thing >> for this specific host OS" whenever we can, in favour of "test >> whether we have whatever feature/function/library is required". >> >> thanks >> -- PMM >> > > Totally makes sense, and I'll be mindful of that. In this case, the > configure script is enabling TPM support on Haiku, but I don't think it > breaks anything, but I haven't tested it yet. > > -- > Regards, > > Richard J. Zak > Professional Genius > PGP Key: https://keybase.io/rjzak/key.asc > -- Regards, Richard J. Zak Professional Genius PGP Key: https://keybase.io/rjzak/key.asc
