http://bugs.winehq.org/show_bug.cgi?id=58146
Bernhard Übelacker bernhardu@mailbox.org changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |bernhardu@mailbox.org
--- Comment #7 from Bernhard Übelacker bernhardu@mailbox.org --- A git bisect led me to this commit: commit 0692d07f85f30ee5191a781b8b6f51006f2ae407 ntdll: Look at CurrentMajor/MinorVersionNumber registry values before CurrentVersion. (released in wine-8.2)
Before winedevice.exe was already crashing, but HWiNFO was still showing its user interface.
So I am unsure how the current regression commit got determined.
Before 0692d07f8 HWiNFO did load a driver file "HWiNFO_x64_W7_204.sys", after this commit it loads "HWiNFO_x64_204.sys".
This newer file makes winedevice "crash" already a few instructions in DriverInit. There is a comparison with 0x2b992ddfa232 which seems related to "Buffer Security Check" and the default value of the "security cookie". This cookie needs to be changed already at this point, if not a "int $0x29"/fastfail" gets executed.
Further searching seems to show this should probably be initialized by the loader in update_load_config. But this does not happen because the offset in the LoadConfig directory shows a value as if the image was not relocated.
I tried to attach a file with more details, and a workaround to enter the relocation path for this driver file, which seems to make DriverInit to continue, and HWiNFO shows its user interface, like with wine-8.1. However also winedevice.exe crashes like with wine-8.1.
Unfortunately bugzilla shows me "Sorry, you have been blocked" How to find out what makes coudflare upset?