Il 28/08/22 15:19, Samuel Thibault ha scritto:
Have you gone through the whole list of the syscalls? I guess we have to
do that to be somehow sure we have not missed anything (rather than
relying on noticing odd bugs in userland)
I went over all syscalls and I found I didn't cover all, but these are
still missing:
* syscall_device_writev_request
* syscall_device_write_request
* syscall_thread_depress_abort
I'm not sure whether they are used, but I'll convert them.
Luca Dariz, le mar. 28 juin 2022 12:10:50 +0200, a ecrit:
+mach_port_name_t mach_reply_port (void);
Ideally we'd propagate the type fix into ipc_port_alloc's name
parameter, etc. That'd bring less confusion to source readers.
That is another cleanup I started to do, but not completed yet. It also
requires to fix other cases at the same time, for example some functions
seem to accept both a port name and a kernel port, and use the value as
kernel port if the port name lookup failed. I think this will be a big
patch with some smaller fixes like this.
Luca
