mirrors/security-misc
1
0
Fork 0
mirror of https://github.com/Kicksecure/security-misc.git synced 2024-12-22 16:13:35 +07:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #287 from raja-grewal/patch

Browse Source 
Refactor and add two CPU mitigations
This commit is contained in:
Patrick Schleizer 2024-12-19 00:31:25 -05:00 committed by GitHub
commit c7f7196471
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 99 additions and 53 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
README.md
View File

@ -37,11 +37,11 @@ Kernel space:
- Entirely disable the SysRq key so that the Secure Attention Key (SAK)
can no longer be utilized. See [documentation](https://www.kicksecure.com/wiki/SysRq).
- Optional - Disable all use of user namespaces.
- Optional - Restrict user namespaces to `CAP_SYS_ADMIN` as they can lead to substantial
privilege escalation.
- Optional - Disable all use of user namespaces.
- Restrict kernel profiling and the performance events system to `CAP_PERFMON`.
- Force the kernel to panic on both "oopses", which can potentially indicate and thwart

92
etc/default/grub.d/40_cpu_mitigations.cfg
View File

@ -8,6 +8,7 @@
## If there is no explicit KSPP compliance notice, the setting is not mentioned by the KSPP.
## Enable known mitigations for CPU vulnerabilities.
## Note, the mitigations for SSB and Retbleed are not currently mentioned in the first link.
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/index.html
## https://www.kernel.org/doc/html/latest/admin-guide/kernel-parameters.html
## https://forums.whonix.org/t/should-all-kernel-patches-for-cpu-bugs-be-unconditionally-enabled-vs-performance-vs-applicability/7647
@ -17,7 +18,7 @@
## https://www.intel.com/content/www/us/en/developer/topic-technology/software-security-guidance/advisory-guidance.html
## https://www.intel.com/content/www/us/en/developer/topic-technology/software-security-guidance/disclosure-documentation.html
## Enable a subset of known mitigations for CPU vulnerabilities and disable SMT.
## Enable a subset of known mitigations for some CPU vulnerabilities and disable SMT.
##
## KSPP=yes
## KSPP sets the kernel parameters.
@ -27,7 +28,7 @@ GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX mitigations=auto,nosmt"
## Disable SMT as it has been the cause of and amplified numerous CPU exploits.
## The only full mitigation of cross-HT attacks is to disable SMT.
## Disabling will significantly decrease system performance on multi-threaded tasks.
## To enable SMT, remove this line and all other occurrences of "nosmt" in this file.
## Note, this setting will prevent re-enabling SMT via the sysfs interface.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/core-scheduling.html
## https://forums.whonix.org/t/should-all-kernel-patches-for-cpu-bugs-be-unconditionally-enabled-vs-performance-vs-applicability/7647/17
@ -36,95 +37,136 @@ GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX mitigations=auto,nosmt"
## KSPP=yes
## KSPP sets the kernel parameter.
##
## To re-enable SMT:
## - Remove "nosmt=force".
## - Remove all occurrences of ",nosmt" in this file (note the comma ",").
## - Downgrade "l1tf=full,force" protection to "l1tf=flush".
## - Regenerate the dracut initramfs and then reboot system.
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX nosmt=force"
## Enable mitigations for both Spectre Variant 2 (indirect branch speculation)
## and Intel branch history injection (BHI) vulnerabilities.
## Spectre Side Channels (BTI and BHI):
## Unconditionally enable mitigation for Spectre Variant 2 (branch target injection).
## Enable mitigation for the Intel branch history injection vulnerability.
## Currently affects both AMD and Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/spectre.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX spectre_v2=on"
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX spectre_bhi=on"
## Disable Speculative Store Bypass (Spectre Variant 4).
## Unconditionally enable mitigation for both kernel and userspace.
## Speculative Store Bypass (SSB):
## Mitigate Spectre Variant 4 by disabling speculative store bypass system-wide.
## Unconditionally enable the mitigation for both kernel and userspace.
## Currently affects both AMD and Intel CPUs.
##
## https://en.wikipedia.org/wiki/Speculative_Store_Bypass
## https://www.suse.com/support/kb/doc/?id=000019189
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX spec_store_bypass_disable=on"
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX ssbd=force-on"
## Enable mitigations for the L1TF vulnerability through disabling SMT
## and L1D flush runtime control.
## L1 Terminal Fault (L1TF):
## Mitigate the vulnerability by disabling L1D flush runtime control and SMT.
## If L1D flushing is conditional, mitigate the vulnerability for certain KVM hypervisor configurations.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/l1tf.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX l1tf=full,force"
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX kvm-intel.vmentry_l1d_flush=always"
## Enable mitigations for the MDS vulnerability through clearing buffer cache
## and disabling SMT.
## Microarchitectural Data Sampling (MDS):
## Mitigate the vulnerability by clearing the CPU buffer cache and disabling SMT.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/mds.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX mds=full,nosmt"
## Patches the TAA vulnerability by disabling TSX and enables mitigations using
## TSX Async Abort along with disabling SMT.
## TSX Asynchronous Abort (TAA):
## Mitigate the vulnerability by disabling TSX.
## If TSX is enabled, clear CPU buffer rings on transitions and disable SMT.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/tsx_async_abort.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX tsx=off"
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX tsx_async_abort=full,nosmt"
## Mark all huge pages in the EPT as non-executable to mitigate iTLB multihit.
## iTLB Multihit:
## Mitigate the vulnerability by marking all huge pages in the EPT as non-executable.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/multihit.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX kvm.nx_huge_pages=force"
## Mitigations for SRBDS to prevent MDS attacks on RDRAND and RDSEED instructions
## are only possible through microcode updates from Intel.
## Special Register Buffer Data Sampling (SRBDS):
## Mitigation of the vulnerability is only possible via microcode update from Intel.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/special-register-buffer-data-sampling.html
## https://access.redhat.com/solutions/5142691
## Enable the prctl() interface to prevent leaks from L1D on context switches.
## L1D Flushing:
## Mitigate leaks from the L1D cache on context switches by enabling the prctl() interface.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/l1d_flush.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX l1d_flush=on"
## Mitigate numerous MMIO Stale Data vulnerabilities and disable SMT.
## Processor MMIO Stale Data:
## Mitigate the vulnerabilities by appropriately clearing the CPU buffer and disabling SMT.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/processor_mmio_stale_data.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX mmio_stale_data=full,nosmt"
## Enable mitigations for RETBleed (Arbitrary Speculative Code Execution with
## Return Instructions) vulnerability and disable SMT.
## Arbitrary Speculative Code Execution with Return Instructions (Retbleed):
## Mitigate the vulnerability through CPU-dependent implementation and disable SMT.
## Currently affects both AMD Zen 1-2 and Intel CPUs.
##
## https://en.wikipedia.org/wiki/Retbleed
## https://comsec.ethz.ch/research/microarch/retbleed/
## https://www.suse.com/support/kb/doc/?id=000020693
## https://access.redhat.com/solutions/retbleed
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX retbleed=auto,nosmt"
## Control RAS overflow mitigation on AMD Zen CPUs.
## Cross-Thread Return Address Predictions:
## Mitigate the vulnerability for certain KVM hypervisor configurations.
## Currently affects AMD Zen 1-2 CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/cross-thread-rsb.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX kvm.mitigate_smt_rsb=1"
## Speculative Return Stack Overflow (SRSO):
## Mitigate the vulnerability by ensuring all RET instructions speculate to a controlled location.
## Currently affects AMD Zen 1-4 CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/srso.html
##
## The default kernel setting will be utilized until provided sufficient evidence to modify.
## Using "spec_rstack_overflow=ipbp" may provide stronger security at a greater performance impact.
##
#GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX spec_rstack_overflow=safe-ret"
## Enable Gather Data Sampling (GDS) mitigation.
## Note for systems that have not received a suitable microcode update this will
## entirely disable use of the AVX instructions set.
## Gather Data Sampling (GDS):
## Mitigate the vulnerability either via microcode update or by disabling AVX.
## Note, without a suitable microcode update, this will entirely disable use of the AVX instructions set.
## Currently affects Intel CPUs.
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/gather_data_sampling.html
##
GRUB_CMDLINE_LINUX="$GRUB_CMDLINE_LINUX gather_data_sampling=force"
## Enable Register File Data Sampling (RFDS) mitigation on Intel Atom CPUs which
## encompasses E-cores on hybrid architectures.
## Register File Data Sampling (RFDS):
## Mitigate the vulnerability by appropriately clearing the CPU buffer.
## Currently affects Intel Atom CPUs (which encompasses E-cores on hybrid architectures).
##
## https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/reg-file-data-sampling.html
##

11
etc/default/grub.d/41_recovery_restrict.cfg
View File

@ -7,10 +7,15 @@
## KSPP=no: not (currently) compliant with recommendations by the KSPP
## If there is no explicit KSPP compliance notice, the setting is not mentioned by the KSPP.
## Disable access to single-user mode (i.e. recovery mode).
## Disable access to single-user (recovery) mode.
##
## https://forums.kicksecure.com/t/remove-linux-recovery-mode-boot-option-from-default-grub-boot-menu/727
GRUB_DISABLE_RECOVERY='true'
##
GRUB_DISABLE_RECOVERY="true"
## Disable access to Dracut's recovery console.
##
## https://forums.kicksecure.com/t/harden-dracut-initramfs-generator-by-disabling-recovery-console/724
GRUB_CMDLINE_LINUX_DEFAULT="$GRUB_CMDLINE_LINUX_DEFAULT rd.shell=0 rd.emergency=halt"
##
GRUB_CMDLINE_LINUX_DEFAULT="$GRUB_CMDLINE_LINUX_DEFAULT rd.emergency=halt"
GRUB_CMDLINE_LINUX_DEFAULT="$GRUB_CMDLINE_LINUX_DEFAULT rd.shell=0"

45
usr/lib/sysctl.d/990-security-misc.conf
View File

@ -117,42 +117,41 @@ vm.unprivileged_userfaultfd=0
##
kernel.sysrq=0
## Restrict user namespaces to users with CAP_SYS_ADMIN.
## Disable user namespaces entirely.
## User namespaces aim to improve sandboxing and accessibility for unprivileged users.
## Unprivileged user namespaces pose substantial privilege escalation risks.
## Restricting may lead to breakages in numerous software packages.
##
## Flatpak requires unprivileged users to create new user namespaces for sandboxing.
## https://github.com/flatpak/flatpak/wiki/User-namespace-requirements
## https://salsa.debian.org/debian/bubblewrap/-/blob/debian/latest/debian/README.Debian
## https://forums.kicksecure.com/t/can-not-run-flatpak-apps-after-kicksecure-update/592
##
## Disabling entirely will reduce compatibility with some AppArmor profiles.
## Disabling entirely is known to break the UPower systemd service.
##
## Also breaks (some?) AppImages.
## https://forums.kicksecure.com/t/cannot-run-some-appimage-apps-after-kicksecure-upate/594
##
## Might also break evolution (e-mail client):
## https://forums.kicksecure.com/t/impossible-to-start-evolution-app-since-the-last-update/601
## Not recommended due to well-known breakages across numerous software packages.
##
## https://lwn.net/Articles/673597/
## https://madaidans-insecurities.github.io/linux.html#kernel
## https://gitlab.com/apparmor/apparmor/-/wikis/unprivileged_userns_restriction
## https://github.com/a13xp0p0v/kernel-hardening-checker#questions-and-answers
## https://github.com/NixOS/nixpkgs/pull/84522#issuecomment-614640601
## https://github.com/flatpak/flatpak/wiki/User-namespace-requirements
## https://github.com/Kicksecure/security-misc/pull/263
## https://github.com/Kicksecure/security-misc/issues/274
##
## KSPP=no
## KSPP sets user.max_user_namespaces=0 sysctl, a Linux mainline, stricter setting.
## KSPP sets the sysctl.
##
## kernel.unprivileged_userns_clone is a Debian specific kernel feature. Not Linux mainline.
#kernel.unprivileged_userns_clone=0
## Uncomment the following sysctl to entirely disable user namespaces.
#user.max_user_namespaces=0
## Restrict user namespaces to users with CAP_SYS_ADMIN.
## See the user.max_user_namespaces setting for more details.
## This is a Debian-specific kernel feature, not a Linux mainline setting.
## Unprivileged user namespaces pose substantial privilege escalation risks.
## Flatpak requires unprivileged users to create new user namespaces for sandboxing.
## Restricting is known to cause breakages in some AppImages and the Evolution Email Client.
## Not recommended due to widespread breakages across many software packages.
##
## https://salsa.debian.org/debian/bubblewrap/-/blob/debian/latest/debian/README.Debian
## https://gitlab.com/apparmor/apparmor/-/wikis/unprivileged_userns_restriction
## https://github.com/flatpak/flatpak/wiki/User-namespace-requirements
## https://forums.kicksecure.com/t/can-not-run-flatpak-apps-after-kicksecure-update/592
## https://forums.kicksecure.com/t/cannot-run-some-appimage-apps-after-kicksecure-upate/594
## https://forums.kicksecure.com/t/impossible-to-start-evolution-app-since-the-last-update/601
## https://github.com/Kicksecure/security-misc/issues/274
##
#kernel.unprivileged_userns_clone=0
## Restricts kernel profiling to users with CAP_PERFMON.
## The performance events system should not be accessible by unprivileged users.
## Other distributions such as Ubuntu and Fedora may permit further restricting.
@ -281,7 +280,7 @@ fs.protected_fifos=2
fs.protected_regular=2
## Enable ASLR for mmap base, stack, VDSO pages, and heap.
## Forces shared libraries to be loaded to random addresses
## Forces shared libraries to be loaded to random addresses.
## Start location of PIE-linked binaries is randomized.
## Heap randomization can lead to breakages with legacy applications.
##