https://bugs.winehq.org/show_bug.cgi?id=50002
Bug ID: 50002 Summary: BlackBerry (Research in Motion) USB and Modem Drivers installs files in the wrong place and fails to install kernel drivers Product: Wine Version: 5.19 Hardware: x86-64 URL: https://swdownloads.blackberry.com/Downloads/contactFo rmPreload.do?code=A8BAA56554F96369AB93E4F3BB068C22&dl= F5937D3FABCEC4C05D26AB6F3FEEC891 OS: Linux Status: NEW Keywords: download, Installer Severity: normal Priority: P2 Component: msi Assignee: wine-bugs@winehq.org Reporter: z.figura12@gmail.com Distribution: ---
Created attachment 68409 --> https://bugs.winehq.org/attachment.cgi?id=68409 hack: set CommonFiles64Folder to the 32-bit common files dir
The installer is more than a bit of a mess. It seems to be trying to install 64-bit files into the 64-bit "Common Files" folder, and 32-bit files into the 32-bit folder. It has this <trimmed> in its Directory table:
Directory Directory_Parent DefaultDir ---------------------------------------------------------- TARGETDIR SourceDir RESEARCH_IN_MOTION ProgramFilesFolder RESEAR~1|Research In Motion BLACKBERRY RESEARCH_IN_MOTION BLACKB~1|BlackBerry INSTALLDIR BLACKBERRY . CommonFiles64Folder TARGETDIR .:Common64 CommonFilesFolder TARGETDIR .:Common CommonFilesRIM.191... CommonFilesFolder RESEAR~1|Research In Motion CommonFilesRIM.07D... CommonFiles64Folder RESEAR~1|Research In Motion
On Wine this behaves essentially as expected, and files get installed into 64-bit "Common Files". On Windows, however, for some reason, everything gets installed into the 32-bit "Common Files" directory. This problem is compounded when the program subsequently tries to install PnP drivers (from a 64-bit custom action) using files from the 32-bit "Common Files" folder—working around their own bug, I guess—and on Wine nothing gets installed.
I'm more than a little baffled as to how this could possibly break on Windows. It'll probably take some more involved debugging.
Anyway, the attached patch works around the bug. The installer successfully creates a couple of root PnP devices, but the driver subsequently fails to load due to missing WDF. (In fact it needs quite a substantial chunk of WDF, which I have partially implemented in my local tree).