This BSOD does only happen with GCC Release builds, the debug builds are not affected.
It's RosBE2.1.6 Win GCC4.7.2.
The test was done on VBox 4.3.12. No USB controllers enabled. 10GB HDD. no VT-x. 1024MB RAM spent.
- configure -DCMAKE_BUILD_TYPE=Release
- ninja bootcd
- mount the VS2010Express1_UDFS_ISO.PNG at some other CDROM drive. The bug does not happen otherwise, I double-checked. Not sure whether it fails only with that specific UDFS iso, or whether every UDFS iso would crash it. https://archive.org/download/vs-2010-express-1/VS2010Express1.iso
- now install ros and see what it does at 2nd stage
It BSODs after the drivers loading in 2nd stage.
I can not attach any logs for a release build obviously.
I tried the same with 0.4.7-release-91-g54e7bcf and got the exact same result at the exact same point in time, so we do know that this regression is quite old.
But I am sure that I could run those kind of builds to the 3rd stage desktop and further in older revisions of ros without issues, maybe it was around 0.4.0 or 0.4.2.
Should be bisected at some point, when that did break in master.
Regression test (all builds RosBE2.1.6 GCC4.7.2 rls cfg)
0.4.14-RC-67-gef623b1 BSOD IRQL_NOT_LESS_OR_EQUAL
0.4.7-release-91-g54e7bcf BSOD IRQL_NOT_LESS_OR_EQUAL
r76032 BSOD IRQL_NOT_LESS_OR_EQUAL
r75200 BSOD IRQL_NOT_LESS_OR_EQUAL (2compile exclude rdbsslib, rxce, nfs)
r75025 BSOD IRQL_NOT_LESS_OR_EQUAL
r74940 BSOD IRQL_NOT_LESS_OR_EQUAL
r74917 BSOD IRQL_NOT_LESS_OR_EQUAL
r74906 BSOD IRQL_NOT_LESS_OR_EQUAL
r74901 BSOD IRQL_NOT_LESS_OR_EQUAL <- guilty
r74900_ok.png no changes in src needed
r74895_ok.png no changes in src needed
r74850_ok.png no changes in src needed
r74500_ok.png no changes in src needed
r73000_ok.png (2compile change SAL in psdk/ks.h a bit _in -> _In)