Cross-Hyperthread Spectre V2 Mitigation Ready For Linux With STIBP
On the Spectre front for the recently-started Linux 4.20~5.0 kernel is STIBP support for cross-hyperthread Spectre Variant Two mitigation.
Going back to the end of the summer was the patch work for this cross-hyperthread Spectre V2 mitigation with STIBP while now it's being merged to mainline.
The Single Thread Indirect Branch Predictors (STIBP) allows for preventing cross-hyperthread control of decisions that are made by indirect branch predictors (IBP). Obviously this is only relevant to the Intel CPU models that have Hyper Threading and where it is enabled... If going for "full" mitigation for L1TF/Foreshadow it will already disable SMT/HT support and OpenBSD for example now ships with Hyper Threading disabled over security concerns.
STIBP also requires the functionality be supported by the Intel microcode in use. This cross-hyperthread Spectre V2 STIBP mitigation is landing as part of the x86/pti changes queued for the Linux 4.20~5.0 kernel. Also included in that pull request are changes to make the IBPB (Indirect Branch Predictor Barrier) more strict as well as some other minor optimizations.
Going back to the end of the summer was the patch work for this cross-hyperthread Spectre V2 mitigation with STIBP while now it's being merged to mainline.
The Single Thread Indirect Branch Predictors (STIBP) allows for preventing cross-hyperthread control of decisions that are made by indirect branch predictors (IBP). Obviously this is only relevant to the Intel CPU models that have Hyper Threading and where it is enabled... If going for "full" mitigation for L1TF/Foreshadow it will already disable SMT/HT support and OpenBSD for example now ships with Hyper Threading disabled over security concerns.
STIBP also requires the functionality be supported by the Intel microcode in use. This cross-hyperthread Spectre V2 STIBP mitigation is landing as part of the x86/pti changes queued for the Linux 4.20~5.0 kernel. Also included in that pull request are changes to make the IBPB (Indirect Branch Predictor Barrier) more strict as well as some other minor optimizations.
5 Comments