Linux Being Patched For Buggy MWAIT Behavior On Intel Ice Lake Servers
([Intel] 5 Hours Ago
Buggy MWAIT)
- Reference: 0001541964
- News link: https://www.phoronix.com/news/Linux-Buggy-MWAIT-Ice-Lake
- Source link:
The Linux kernel has seen safeguards for select prior Intel CPU cores due to bugs around the MONITOR/MWAIT implementation with the processors. MWAIT/MONITOR bugs was found to be the cause of [1]annoying issues at boot for Lunar Lake laptops and also previously plagued Goldmont Atom cores. It also turns out that Ice Lake servers can be subject to similar MWAIT/MONITOR behavior.
Surprisingly it's taken until now for this issue to creep up and be fixed soon in the mainline Linux kernel. After all, 3rd Gen Xeon Scalable "Ice Lake" [2]launched back in 2021 .
Intel via a specification document update sent out a notice in 2022 that their 3rd Gen Xeon Scalable "Ice Lake" server processors indeed had a known issue where MONITOR may not be triggered that was armed via the MWAIT instruction. In turn this can lead to a processor hang as noted in the ICX143 spec update.
Finally in 2025 this is being patched by the Linux kernel in the same manner as the Lunar Lake and Goldmont to avoid the MWAIT issue. This CPU bug has been hitting server folks now at least under some Xen virtualization usage. [3][PATCH] Handle Ice Lake MONITOR erratum was posted yesterday and explained:
"Andrew Cooper reported some boot issues on Ice Lake servers when running Xen that he tracked down to MWAIT not waking up. Do the safe thing and consider them buggy since there's a published erratum. Note: I've seen no reports of this occurring on Linux.
Add Ice Lake servers to the list of shaky MONITOR implementations with no workaround available."
Hopefully the future Intel CPU cores will have less worries around their MONITOR/MWAIT handling but at least for this Xeon Ice Lake issue doesn't seem to be too common unlike the Lunar Lake issues that came up shortly after launch.
This Ice Lake patch is currently for review on the Linux kernel mailing list but will likely be submitted in the coming days or so as part of fixes to Linux 6.15 and then back-ported.
[1] https://www.phoronix.com/news/Intel-Lunar-Lake-Monitor-Bug
[2] https://www.phoronix.com/review/intel-xeon-8380-linux
[3] https://lore.kernel.org/lkml/20250421192205.7CC1A7D9@davehans-spike.ostc.intel.com/
Surprisingly it's taken until now for this issue to creep up and be fixed soon in the mainline Linux kernel. After all, 3rd Gen Xeon Scalable "Ice Lake" [2]launched back in 2021 .
Intel via a specification document update sent out a notice in 2022 that their 3rd Gen Xeon Scalable "Ice Lake" server processors indeed had a known issue where MONITOR may not be triggered that was armed via the MWAIT instruction. In turn this can lead to a processor hang as noted in the ICX143 spec update.
Finally in 2025 this is being patched by the Linux kernel in the same manner as the Lunar Lake and Goldmont to avoid the MWAIT issue. This CPU bug has been hitting server folks now at least under some Xen virtualization usage. [3][PATCH] Handle Ice Lake MONITOR erratum was posted yesterday and explained:
"Andrew Cooper reported some boot issues on Ice Lake servers when running Xen that he tracked down to MWAIT not waking up. Do the safe thing and consider them buggy since there's a published erratum. Note: I've seen no reports of this occurring on Linux.
Add Ice Lake servers to the list of shaky MONITOR implementations with no workaround available."
Hopefully the future Intel CPU cores will have less worries around their MONITOR/MWAIT handling but at least for this Xeon Ice Lake issue doesn't seem to be too common unlike the Lunar Lake issues that came up shortly after launch.
This Ice Lake patch is currently for review on the Linux kernel mailing list but will likely be submitted in the coming days or so as part of fixes to Linux 6.15 and then back-ported.
[1] https://www.phoronix.com/news/Intel-Lunar-Lake-Monitor-Bug
[2] https://www.phoronix.com/review/intel-xeon-8380-linux
[3] https://lore.kernel.org/lkml/20250421192205.7CC1A7D9@davehans-spike.ostc.intel.com/
reba