[Libre-soc-bugs] [Bug 982] Support PowerPC ABI in ISACaller

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=982

--- Comment #38 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Dmitry Selyutin from comment #36)
> All in all, this task and its future depends on how far you want to go.

enough to run "open and read and write some files" such as /dev/ttyUSB0
or other serial console.

and enough to say "success!" and justify another grant.

> depending on which system calls you need: either intermediate copies,
> or directly shared memory, plus parts of MM, plus parts of fd tree, etc.
> etc. Basically you'll end up inventing a toy OS inside the interpreter.

indeed.

> From comment 2 I assumed the second. That's an interesting thing to try
> checking out, though I must confess, in my opinion, the budget hardly covers
> even this experiment.

i know.

> The best I can suggest is preparing an infrastructure
> which makes this possible and demonstrating some really basic bits: say
> wrapping some exit(2) variant and maybe write(2) on STDOUT_FILENO and
> perhaps stuff like sleep and OS detection via say uname(2).

you can see why cavatools just literally farmed out verbatim guest to
host, no security whatsoever, and overrode malloc and free.

can we see how far the direct-syscall-with-new-mem.py gets?

if it really doesn't work out the fallback is an emulation in python,
with a few (very few) syscalls as a start.

in theeeoryyy we can borrow the cavatools budget allocated to the
same task. see bug #981

oh. https://bugs.libre-soc.org/show_bug.cgi?id=939#c1

-- 
You are receiving this mail because:
You are on the CC list for the bug.