Reply 800 of 800, by superfury
A slight restart of WI=indows 95 setup (from the MS-DOS prompt again after a quick format of the hard disk) since I had a slight bug in my new 16-bit TSS cpu instruction step-based validation function call (causing an invalid recursive infinite self-call of the same function at the third line of the function instead of calling the proper wrapped function, which shares part of it's own name (just CPU_ prefixed and _protectedmode suffixed)). That improves the generic protected mode speed (not requiring TLB refills when it doesn't need to and improves support for proper nested and interruptable operation.
So far Windows 95 OSR 2.5 "C" setup is proceeding (at the file copy process so far of the first phase), I'll know if it works correctly once Internet Explorer completes it's installation after the 2 reboots after it have been done (1 reboot during setup itself and 1 for completing the installation and starting to install internet explorer (hopefully without crashing this time)). If that runs properly I can start looking into Windows 3.1's FPU issues (Pink Panther in Hocus Pokus Pink hanging the OS due to FPU emulation issues) again (and NT 4 SP6's login issues (netlogon service not running, preventing logon to the SP6 install) might have been fixed as well, but I'll need to check 95 to be working properly first).
Edit: Windows 95 setup phase 2 crashes (BSOD) before reaching the GUI, on a NULL ES register being accessed with (4GB kernel segment) with DPL=0 and Present set to 0, implying the cause is an IRET leaving the kernel segment selector in there without reloading the user-mode segment properly before the return (probably POP ES failing somehow)?
Author of the UniPCemu emulator.
UniPCemu Git repository
UniPCemu for Android, Windows, PSP, Vita and Switch on itch.io