Skip to contentSkip to conversation

audit: make ADUITSYSCALL optional again

PATCH In review Frederick Lawler 2 participants 4 messages Aug 8, 2025 – Aug 13, 2025 Audit

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 audit: make ADUITSYSCALL optional again

Since the introduction of commit cb74ed278f80 ("audit: always enable syscall auditing when supported and audit is enabled"), eBPF technologies are being adopted to track syscalls for auditing purposes. Those technologies add an additional overhead ontop of AUDITSYSCALL. Additionally, AUDIT infrastructure has expanded to include INTEGRITY which offers some advantages over eBPF technologies, such as early-init/boot integrity logs with. Therefore, make ADUITSYSCALL optional again, but keep it default y.

---
 init/Kconfig | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
init/Kconfig +8 -3
index af4c2f085455..2552918deb45 100644
@@ -487,16 +487,21 @@ config AUDIT
 	help
 	  Enable auditing infrastructure that can be used with another
 	  kernel subsystem, such as SELinux (which requires this for
-	  logging of avc messages output).  System call auditing is included
-	  on architectures which support it.
+	  logging of avc messages output). Does not do system-call
+	  auditing without CONFIG_AUDITSYSCALL.
 
 config HAVE_ARCH_AUDITSYSCALL
 	bool
 
 config AUDITSYSCALL
-	def_bool y
+	bool "Enable system-call auditing support"
 	depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
+	default y
 	select FSNOTIFY
+	help
+	  Enable low-overhead system-call auditing infrastructure that
+	  can be used indepdently or with another kernel subsystem,
+	  such as SELiux.
 
 source "kernel/irq/Kconfig"
 source "kernel/time/Kconfig"
Signature · 1 line
2.43.0
#1 lore raw

RFC PATCH 1/1 audit: make ADUITSYSCALL optional again

Since the introduction of commit cb74ed278f80 ("audit: always enable syscall auditing when supported and audit is enabled"), eBPF technologies are being adopted to track syscalls for auditing purposes. Those technologies add an additional overhead ontop of AUDITSYSCALL. Additionally, AUDIT infrastructure has expanded to include INTEGRITY which offers some advantages over eBPF technologies, such as early-init/boot integrity logs with. Therefore, make ADUITSYSCALL optional again, but keep it default y.

---
 init/Kconfig | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
init/Kconfig +8 -3
index af4c2f085455..2552918deb45 100644
@@ -487,16 +487,21 @@ config AUDIT
 	help
 	  Enable auditing infrastructure that can be used with another
 	  kernel subsystem, such as SELinux (which requires this for
-	  logging of avc messages output).  System call auditing is included
-	  on architectures which support it.
+	  logging of avc messages output). Does not do system-call
+	  auditing without CONFIG_AUDITSYSCALL.
 
 config HAVE_ARCH_AUDITSYSCALL
 	bool
 
 config AUDITSYSCALL
-	def_bool y
+	bool "Enable system-call auditing support"
 	depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
+	default y
 	select FSNOTIFY
+	help
+	  Enable low-overhead system-call auditing infrastructure that
+	  can be used indepdently or with another kernel subsystem,
+	  such as SELiux.
 
 source "kernel/irq/Kconfig"
 source "kernel/time/Kconfig"
Signature · 1 line
2.43.0
maintainer
#2 lore raw
In reply to Frederick Lawler
Quoted 10 lines · On Aug 8, 2025 Frederick Lawler <[email protected]> wrote:

Since the introduction of commit cb74ed278f80 ("audit: always enable syscall auditing when supported and audit is enabled"), eBPF technologies are being adopted to track syscalls for auditing purposes. Those technologies add an additional overhead ontop of AUDITSYSCALL. Additionally, AUDIT infrastructure has expanded to include INTEGRITY which offers some advantages over eBPF technologies, such as early-init/boot integrity logs with. Therefore, make ADUITSYSCALL optional again, but keep it default y.

Quoted 4 lines
---
 init/Kconfig | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

Generally speaking the less Kconfig knobs the better; it tends to complicate things and for those that rely on distro kernels, there is always at least one group that is going to be upset about the Kconfig knob being set "wrong". In my ideal world, CONFIG_AUDITSYSCALL wouldn't exist at all, but sadly not all arches have the necessary support to do that at the moment, so CONFIG_AUDITSYSCALL remains a necessary evil.

Thank you for the patch, but IMO this is not the direction we want to go with audit.

--
paul-moore.com

#3 lore raw

Hi Paul,

Quoted 23 lines · Paul Moore
Quoted 10 lines · On Aug 8, 2025 Frederick Lawler <[email protected]> wrote:

Since the introduction of commit cb74ed278f80 ("audit: always enable syscall auditing when supported and audit is enabled"), eBPF technologies are being adopted to track syscalls for auditing purposes. Those technologies add an additional overhead ontop of AUDITSYSCALL. Additionally, AUDIT infrastructure has expanded to include INTEGRITY which offers some advantages over eBPF technologies, such as early-init/boot integrity logs with. Therefore, make ADUITSYSCALL optional again, but keep it default y.

Quoted 4 lines
---
 init/Kconfig | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

Generally speaking the less Kconfig knobs the better; it tends to complicate things and for those that rely on distro kernels, there is always at least one group that is going to be upset about the Kconfig knob being set "wrong". In my ideal world, CONFIG_AUDITSYSCALL wouldn't exist at all, but sadly not all arches have the necessary support to do that at the moment, so CONFIG_AUDITSYSCALL remains a necessary evil.

Quoted 3 lines

Thank you for the patch, but IMO this is not the direction we want to go with audit.

Thanks for the response. I think setting the filters to never would be OK, but doesn't hurt to try to see if it's worth squeezing out the remaining usages.

Quoted 2 lines

--
paul-moore.com

PS. I'll be sure to use b4 next time for a
submission.

Best, Fred