14 Mar
2025
14 Mar
'25
12:02 p.m.
On Fri Mar 14 11:46:11 2025 +0000, Nikolay Sivov wrote:
Correct, it has to be changed too. It's certainly easier to do the way you did it, and maybe it's acceptable, I don't know. On Windows it's structured differently.
objdump -p .../windows/system32/advapi32.dll says that yes, advapi32 should forward a bunch of SystemFunctions to cryptsp and cryptbase, only implementing 017 and 019 itself. So yes, our cryptsp is inaccurate.
But if it works, it works; I vote accept, and leave it like that until something actually relies on things being in the right DLL. But it's not my call.
I do, however, question that kernelbase patch. Looks like it belongs in a separate MR.