ACPI: Drop the custom_method debugfs interface
PATCH v3 1/1 ACPI: Drop the custom_method debugfs interface
From: Rafael J. Wysocki <[email protected]>
The ACPI custom_method debugfs interface is security-sensitive and concurrent access to it is broken [1].
Moreover, the recipe for preparing a customized version of a given control method has changed at one point due to ACPICA changes, which has not been reflected in its documentation, so whoever used it before has had to adapt and it had gone unnoticed for a long time.
This interface was a bad idea to start with and its implementation is fragile at the design level. It's been always conceptually questionable, problematic from the security standpoint and implemented poorly.
Patches fixing its most apparent functional issues (for example, [2]) don't actually address much of the above.
Granted, at the time it was introduced, there was no alternative, but there is the AML debugger in the kernel now and there is the configfs interface allowing custom ACPI tables to be loaded. The former can be used for extensive AML debugging and the latter can be use for testing new AML. [3]
Accordingly, drop custom_method along with its (outdated anyway) documentation.
https://lore.kernel.org/linux-acpi/[email protected]/ [2]
https://stackoverflow.com/questions/62849113/how-to-unload-an-overlay-loaded-using-acpi-config-sysfs # [3]
---
v2 -> v3:
* Update changelog.
* Add R-by from Rui.
v1 -> v2: Update index.rst too.
--- Documentation/firmware-guide/acpi/index.rst | 1 Documentation/firmware-guide/acpi/method-customizing.rst | 89 ------------ drivers/acpi/Kconfig | 14 -- drivers/acpi/Makefile | 1 drivers/acpi/custom_method.c | 103 --------------- 5 files changed, 208 deletions(-)
drivers/acpi/Kconfig +0 -14
@@ -449,20 +449,6 @@ config ACPI_HED which is used to report some hardware errors notified via SCI, mainly the corrected errors. -config ACPI_CUSTOM_METHOD - tristate "Allow ACPI methods to be inserted/replaced at run time" - depends on DEBUG_FS - help - This debug facility allows ACPI AML methods to be inserted and/or - replaced without rebooting the system. For details refer to: - Documentation/firmware-guide/acpi/method-customizing.rst. - - NOTE: This option is security sensitive, because it allows arbitrary - kernel memory to be written to by root (uid=0) users, allowing them - to bypass certain security measures (e.g. if root is not allowed to - load additional kernel modules after boot, this feature may be used - to override that restriction). - config ACPI_BGRT bool "Boottime Graphics Resource Table support" depends on EFI && (X86 || ARM64)
drivers/acpi/Makefile +0 -1
@@ -101,7 +101,6 @@ obj-$(CONFIG_ACPI_SBS) += sbshc.o obj-$(CONFIG_ACPI_SBS) += sbs.o obj-$(CONFIG_ACPI_HED) += hed.o obj-$(CONFIG_ACPI_EC_DEBUGFS) += ec_sys.o -obj-$(CONFIG_ACPI_CUSTOM_METHOD)+= custom_method.o obj-$(CONFIG_ACPI_BGRT) += bgrt.o obj-$(CONFIG_ACPI_CPPC_LIB) += cppc_acpi.o obj-$(CONFIG_ACPI_SPCR_TABLE) += spcr.o
drivers/acpi/custom_method.c deleted +0 -103
103 lines not on this page · raw message
Documentation/firmware-guide/acpi/method-customizing.rst deleted +0 -89
89 lines not on this page · raw message
Documentation/firmware-guide/acpi/index.rst +0 -1
@@ -14,7 +14,6 @@ ACPI Support dsd/phy enumeration osi - method-customizing method-tracing DSD-properties-rules debug
Quoted 23 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
The ACPI
custom_methoddebugfs interface is security-sensitive and concurrent access to it is broken [1].Moreover, the recipe for preparing a customized version of a given control method has changed at one point due to ACPICA changes, which has not been reflected in its documentation, so whoever used it before has had to adapt and it had gone unnoticed for a long time.
This interface was a bad idea to start with and its implementation is fragile at the design level. It's been always conceptually questionable, problematic from the security standpoint and implemented poorly.
Patches fixing its most apparent functional issues (for example, [2]) don't actually address much of the above.
Granted, at the time it was introduced, there was no alternative, but there is the AML debugger in the kernel now and there is the configfs interface allowing custom ACPI tables to be loaded. The former can be used for extensive AML debugging and the latter can be use for testing new AML. [3]
Quoted 6 lines
Accordingly, drop
custom_methodalong with its (outdated anyway) documentation.https://lore.kernel.org/linux-acpi/[email protected]/ [2]
https://stackoverflow.com/questions/62849113/how-to-unload-an-overlay-loaded-using-acpi-config-sysfs # [3]
I believe you missed Link: tags for 2 and 3.
Quoted 3 lines
Reported-by Hang Zhang <[email protected]>Signed-off-by Rafael J. Wysocki <[email protected]>Reviewed-by Zhang Rui <[email protected]>
Signature · 2 lines
With Best Regards, Andy Shevchenko
Rafael J. Wysocki
maintainerQuoted 25 lines · Andy Shevchenko
Quoted 23 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
The ACPI
custom_methoddebugfs interface is security-sensitive and concurrent access to it is broken [1].Moreover, the recipe for preparing a customized version of a given control method has changed at one point due to ACPICA changes, which has not been reflected in its documentation, so whoever used it before has had to adapt and it had gone unnoticed for a long time.
This interface was a bad idea to start with and its implementation is fragile at the design level. It's been always conceptually questionable, problematic from the security standpoint and implemented poorly.
Patches fixing its most apparent functional issues (for example, [2]) don't actually address much of the above.
Granted, at the time it was introduced, there was no alternative, but there is the AML debugger in the kernel now and there is the configfs interface allowing custom ACPI tables to be loaded. The former can be used for extensive AML debugging and the latter can be use for testing new AML. [3]
Quoted 8 lines
Quoted 6 lines
Accordingly, drop
custom_methodalong with its (outdated anyway) documentation.https://lore.kernel.org/linux-acpi/[email protected]/ [2]
https://stackoverflow.com/questions/62849113/how-to-unload-an-overlay-loaded-using-acpi-config-sysfs # [3]I believe you missed Link: tags for 2 and 3.
Obviously, you are right.
That's easy to fix, though, and I'd rather not confuse people by sending a v4 just because of this.
Quoted 3 lines
Quoted 3 lines
Reported-by Hang Zhang <[email protected]>Signed-off-by Rafael J. Wysocki <[email protected]>Reviewed-by Zhang Rui <[email protected]>
Quoted 23 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
The ACPI
custom_methoddebugfs interface is security-sensitive and concurrent access to it is broken [1].Moreover, the recipe for preparing a customized version of a given control method has changed at one point due to ACPICA changes, which has not been reflected in its documentation, so whoever used it before has had to adapt and it had gone unnoticed for a long time.
This interface was a bad idea to start with and its implementation is fragile at the design level. It's been always conceptually questionable, problematic from the security standpoint and implemented poorly.
Patches fixing its most apparent functional issues (for example, [2]) don't actually address much of the above.
Granted, at the time it was introduced, there was no alternative, but there is the AML debugger in the kernel now and there is the configfs interface allowing custom ACPI tables to be loaded. The former can be used for extensive AML debugging and the latter can be use for testing new AML. [3]
Quoted 8 lines
Accordingly, drop
custom_methodalong with its (outdated anyway) documentation.https://lore.kernel.org/linux-acpi/[email protected]/ [2]
https://stackoverflow.com/questions/62849113/how-to-unload-an-overlay-loaded-using-acpi-config-sysfs # [3]Reported-by Hang Zhang <[email protected]>Signed-off-by Rafael J. Wysocki <[email protected]>
Yay! Happy to see this removed. :)
Signature · 1 line
Kees Cook