JJS wrote:Alright, something like this should work. It is still pretty rough, things that could be optimized are
- the hbl load address could be higher
- the p5 stubs don't have to be parsed
- function hooks are mostly unnecessary, but have to be controlled by a different flag than g->syscalls_known (because setting it has sideeffects)
- the launcher has more imports than are needed because I just used prxtool to build import stubs from 3.x kernel modules and then removed those without a known name
There are no changes to the HBL files except for a new exploit directory and I removed the -Werror flag from the compiler because otherwise it complains because of the empty arrays in exploit_config.h.
I couldn't get it to work on OFW 6.37.
I have to dig deeper, but the symptoms are the following:
The Sony gameboot starts, then black screen, no error message, no memstick activity.
Therefore this doesn't look like a problem on the HBL side, but more the signed Eboot itself.
It's very possible that I incorrectly signed the thing (I have so many versions of the various tools right now...), care to share your signed launcher EBOOT?
Edit: my bad, after adding some debugging, what's not working for me seems to be located in h.bin.
Even the first debug statement is not generated, so I'm assuming something very bad is happening at compile time.