diff mbox series

[2/2] powerpc/rtas: block error injection when locked down

Message ID 20220922193817.106041-3-nathanl@linux.ibm.com (mailing list archive)
State Superseded
Delegated to: Paul Moore
Headers show
Series powerpc/pseries: restrict error injection and DT changes when locked down | expand

Commit Message

Nathan Lynch Sept. 22, 2022, 7:38 p.m. UTC
The error injection facility on pseries VMs allows corruption of
arbitrary guest memory, potentially enabling a sufficiently privileged
user to disable lockdown or perform other modifications of the running
kernel via the rtas syscall.

Block the PAPR error injection facility from being opened or called
when locked down.

Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
---
 arch/powerpc/kernel/rtas.c | 25 ++++++++++++++++++++++++-
 include/linux/security.h   |  1 +
 security/security.c        |  1 +
 3 files changed, 26 insertions(+), 1 deletion(-)

Comments

Paul Moore Sept. 23, 2022, 1:28 a.m. UTC | #1
On Thu, Sep 22, 2022 at 3:38 PM Nathan Lynch <nathanl@linux.ibm.com> wrote:
>
> The error injection facility on pseries VMs allows corruption of
> arbitrary guest memory, potentially enabling a sufficiently privileged
> user to disable lockdown or perform other modifications of the running
> kernel via the rtas syscall.
>
> Block the PAPR error injection facility from being opened or called
> when locked down.
>
> Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
> ---
>  arch/powerpc/kernel/rtas.c | 25 ++++++++++++++++++++++++-
>  include/linux/security.h   |  1 +
>  security/security.c        |  1 +
>  3 files changed, 26 insertions(+), 1 deletion(-)

...

> diff --git a/include/linux/security.h b/include/linux/security.h
> index 1ca8dbacd3cc..b5d5138ae66a 100644
> --- a/include/linux/security.h
> +++ b/include/linux/security.h
> @@ -123,6 +123,7 @@ enum lockdown_reason {
>         LOCKDOWN_BPF_WRITE_USER,
>         LOCKDOWN_DBG_WRITE_KERNEL,
>         LOCKDOWN_DEVICE_TREE,
> +       LOCKDOWN_RTAS_ERROR_INJECTION,

With the understanding that I've never heard of RTAS until now, are
there any other RTAS events that would require a lockdown reason?  As
a follow up, is it important to distinguish between different RTAS
lockdown reasons?

I'm trying to determine if we can just call it LOCKDOWN_RTAS.

>         LOCKDOWN_INTEGRITY_MAX,
>         LOCKDOWN_KCORE,
>         LOCKDOWN_KPROBES,
> diff --git a/security/security.c b/security/security.c
> index 2863fc31eec6..6518b239ada2 100644
> --- a/security/security.c
> +++ b/security/security.c
> @@ -61,6 +61,7 @@ const char *const lockdown_reasons[LOCKDOWN_CONFIDENTIALITY_MAX+1] = {
>         [LOCKDOWN_BPF_WRITE_USER] = "use of bpf to write user RAM",
>         [LOCKDOWN_DBG_WRITE_KERNEL] = "use of kgdb/kdb to write kernel RAM",
>         [LOCKDOWN_DEVICE_TREE] = "modifying device tree contents",
> +       [LOCKDOWN_RTAS_ERROR_INJECTION] = "RTAS error injection",

See above.

>         [LOCKDOWN_INTEGRITY_MAX] = "integrity",
>         [LOCKDOWN_KCORE] = "/proc/kcore access",
>         [LOCKDOWN_KPROBES] = "use of kprobes",
> --
> 2.37.3
Michael Ellerman Sept. 23, 2022, 7:12 a.m. UTC | #2
Paul Moore <paul@paul-moore.com> writes:
> On Thu, Sep 22, 2022 at 3:38 PM Nathan Lynch <nathanl@linux.ibm.com> wrote:
>>
>> The error injection facility on pseries VMs allows corruption of
>> arbitrary guest memory, potentially enabling a sufficiently privileged
>> user to disable lockdown or perform other modifications of the running
>> kernel via the rtas syscall.
>>
>> Block the PAPR error injection facility from being opened or called
>> when locked down.
>>
>> Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
>> ---
>>  arch/powerpc/kernel/rtas.c | 25 ++++++++++++++++++++++++-
>>  include/linux/security.h   |  1 +
>>  security/security.c        |  1 +
>>  3 files changed, 26 insertions(+), 1 deletion(-)
>
> ...
>
>> diff --git a/include/linux/security.h b/include/linux/security.h
>> index 1ca8dbacd3cc..b5d5138ae66a 100644
>> --- a/include/linux/security.h
>> +++ b/include/linux/security.h
>> @@ -123,6 +123,7 @@ enum lockdown_reason {
>>         LOCKDOWN_BPF_WRITE_USER,
>>         LOCKDOWN_DBG_WRITE_KERNEL,
>>         LOCKDOWN_DEVICE_TREE,
>> +       LOCKDOWN_RTAS_ERROR_INJECTION,
>
> With the understanding that I've never heard of RTAS until now, are
> there any other RTAS events that would require a lockdown reason?  As
> a follow up, is it important to distinguish between different RTAS
> lockdown reasons?
>
> I'm trying to determine if we can just call it LOCKDOWN_RTAS.

Yes I think we should.

Currently it only locks down the error injection calls, not all of RTAS.

But firmware can/will add new RTAS calls in future, so it's always
possible something will need to be added to the list of things that need
to be blocked during lockdown.

So I think calling it LOCKDOWN_RTAS would be more general and future
proof, and can be read to mean "lockdown the parts of RTAS that need
to be locked down".

Similarly we have LOCKDOWN_ACPI_TABLES which locks down modification to
ACPI data, but doesn't disable ACPI use entirely AIUI.

cheers
Nathan Lynch Sept. 23, 2022, 3:39 p.m. UTC | #3
Michael Ellerman <mpe@ellerman.id.au> writes:
> Paul Moore <paul@paul-moore.com> writes:
>> On Thu, Sep 22, 2022 at 3:38 PM Nathan Lynch <nathanl@linux.ibm.com> wrote:
>>>
>>> The error injection facility on pseries VMs allows corruption of
>>> arbitrary guest memory, potentially enabling a sufficiently privileged
>>> user to disable lockdown or perform other modifications of the running
>>> kernel via the rtas syscall.
>>>
>>> Block the PAPR error injection facility from being opened or called
>>> when locked down.
>>>
>>> Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
>>> ---
>>>  arch/powerpc/kernel/rtas.c | 25 ++++++++++++++++++++++++-
>>>  include/linux/security.h   |  1 +
>>>  security/security.c        |  1 +
>>>  3 files changed, 26 insertions(+), 1 deletion(-)
>>
>> ...
>>
>>> diff --git a/include/linux/security.h b/include/linux/security.h
>>> index 1ca8dbacd3cc..b5d5138ae66a 100644
>>> --- a/include/linux/security.h
>>> +++ b/include/linux/security.h
>>> @@ -123,6 +123,7 @@ enum lockdown_reason {
>>>         LOCKDOWN_BPF_WRITE_USER,
>>>         LOCKDOWN_DBG_WRITE_KERNEL,
>>>         LOCKDOWN_DEVICE_TREE,
>>> +       LOCKDOWN_RTAS_ERROR_INJECTION,
>>
>> With the understanding that I've never heard of RTAS until now, are
>> there any other RTAS events that would require a lockdown reason?  As
>> a follow up, is it important to distinguish between different RTAS
>> lockdown reasons?

1. Not to my current knowledge.
2. Yes, I think so, see below.

>>
>> I'm trying to determine if we can just call it LOCKDOWN_RTAS.
>
> Yes I think we should.
>
> Currently it only locks down the error injection calls, not all of RTAS.
>
> But firmware can/will add new RTAS calls in future, so it's always
> possible something will need to be added to the list of things that need
> to be blocked during lockdown.
>
> So I think calling it LOCKDOWN_RTAS would be more general and future
> proof, and can be read to mean "lockdown the parts of RTAS that need
> to be locked down".

RTAS provides callable interfaces for a broad variety of functions,
including device configuration, halt/reboot/suspend, CPU online/offline,
NVRAM access, firmware upgrade, platform diagnostic data retrieval, and
others.

Currently I don't know of other RTAS-provided functions that should be
restricted. But if we were to add more, then -- in answer to Paul -- yes
I think it would be important to have distinct reasons and
messages. Taking the point of view of someone diagnosing lockdown denial
messages and relating them to kernel code and user space activity, I
would rather we err toward specificity.

Also: LOCKDOWN_RTAS_ERROR_INJECTION is proposed for lockdown's integrity
mode. But consider that future RTAS-related additions could be proposed
for confidentiality mode, which is more restrictive. (A plausible
example could be platform dump retrieval.) We would need at least two
RTAS reasons and one wouldn't suffice.

So a single RTAS catch-all lockdown reason doesn't appeal to me, but
lockdown reasons and messages aren't ABI (right?) and we could
eventually change whatever decision we reach here. So if you both still
prefer a single LOCKDOWN_RTAS reason, I can do that for v2.

That said, I could see a case for instead adding
LOCKDOWN_HW_ERROR_INJECTION (without the RTAS), to indicate restriction
of hardware- or platform-level error injection. I was a little surprised
to find that an error injection reason doesn't already exist for the
ACPI-based facility (drivers/acpi/apei/einj.c), but since the user
interface is debugfs-based I suppose it's already restricted in practice
by LOCKDOWN_DEBUGFS.
Paul Moore Sept. 23, 2022, 5:42 p.m. UTC | #4
On Fri, Sep 23, 2022 at 11:40 AM Nathan Lynch <nathanl@linux.ibm.com> wrote:
> Michael Ellerman <mpe@ellerman.id.au> writes:
> > Paul Moore <paul@paul-moore.com> writes:
> >> On Thu, Sep 22, 2022 at 3:38 PM Nathan Lynch <nathanl@linux.ibm.com> wrote:
> >>>
> >>> The error injection facility on pseries VMs allows corruption of
> >>> arbitrary guest memory, potentially enabling a sufficiently privileged
> >>> user to disable lockdown or perform other modifications of the running
> >>> kernel via the rtas syscall.
> >>>
> >>> Block the PAPR error injection facility from being opened or called
> >>> when locked down.
> >>>
> >>> Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
> >>> ---
> >>>  arch/powerpc/kernel/rtas.c | 25 ++++++++++++++++++++++++-
> >>>  include/linux/security.h   |  1 +
> >>>  security/security.c        |  1 +
> >>>  3 files changed, 26 insertions(+), 1 deletion(-)
> >>
> >> ...
> >>
> >>> diff --git a/include/linux/security.h b/include/linux/security.h
> >>> index 1ca8dbacd3cc..b5d5138ae66a 100644
> >>> --- a/include/linux/security.h
> >>> +++ b/include/linux/security.h
> >>> @@ -123,6 +123,7 @@ enum lockdown_reason {
> >>>         LOCKDOWN_BPF_WRITE_USER,
> >>>         LOCKDOWN_DBG_WRITE_KERNEL,
> >>>         LOCKDOWN_DEVICE_TREE,
> >>> +       LOCKDOWN_RTAS_ERROR_INJECTION,
> >>
> >> With the understanding that I've never heard of RTAS until now, are
> >> there any other RTAS events that would require a lockdown reason?  As
> >> a follow up, is it important to distinguish between different RTAS
> >> lockdown reasons?
>
> 1. Not to my current knowledge.
> 2. Yes, I think so, see below.
>
> >>
> >> I'm trying to determine if we can just call it LOCKDOWN_RTAS.
> >
> > Yes I think we should.
> >
> > Currently it only locks down the error injection calls, not all of RTAS.
> >
> > But firmware can/will add new RTAS calls in future, so it's always
> > possible something will need to be added to the list of things that need
> > to be blocked during lockdown.
> >
> > So I think calling it LOCKDOWN_RTAS would be more general and future
> > proof, and can be read to mean "lockdown the parts of RTAS that need
> > to be locked down".
>
> RTAS provides callable interfaces for a broad variety of functions,
> including device configuration, halt/reboot/suspend, CPU online/offline,
> NVRAM access, firmware upgrade, platform diagnostic data retrieval, and
> others.
>
> Currently I don't know of other RTAS-provided functions that should be
> restricted. But if we were to add more, then -- in answer to Paul -- yes
> I think it would be important to have distinct reasons and
> messages. Taking the point of view of someone diagnosing lockdown denial
> messages and relating them to kernel code and user space activity, I
> would rather we err toward specificity.

As I said before, RTAS is a great mystery to me, if it can be extended
in the future then having a targeted lockdown name makes perfect
sense.

> So a single RTAS catch-all lockdown reason doesn't appeal to me, but
> lockdown reasons and messages aren't ABI (right?) ...

Correct.  Or at least that is my understanding, but there have been
some odd rulings on lockdown in the past so my advice would be to make
*very* sure you get this right the first time.  From what you and
Michael have said, it seems like a function specific name is the way
to go here, and based on your explanations of the situation it seems
like putting this in the integrity bin is the right way to go.
diff mbox series

Patch

diff --git a/arch/powerpc/kernel/rtas.c b/arch/powerpc/kernel/rtas.c
index 693133972294..c2540d393f1c 100644
--- a/arch/powerpc/kernel/rtas.c
+++ b/arch/powerpc/kernel/rtas.c
@@ -23,6 +23,7 @@ 
 #include <linux/memblock.h>
 #include <linux/slab.h>
 #include <linux/reboot.h>
+#include <linux/security.h>
 #include <linux/syscalls.h>
 #include <linux/of.h>
 #include <linux/of_fdt.h>
@@ -464,6 +465,9 @@  void rtas_call_unlocked(struct rtas_args *args, int token, int nargs, int nret,
 	va_end(list);
 }
 
+static int ibm_open_errinjct_token;
+static int ibm_errinjct_token;
+
 int rtas_call(int token, int nargs, int nret, int *outputs, ...)
 {
 	va_list list;
@@ -476,6 +480,16 @@  int rtas_call(int token, int nargs, int nret, int *outputs, ...)
 	if (!rtas.entry || token == RTAS_UNKNOWN_SERVICE)
 		return -1;
 
+	if (token == ibm_open_errinjct_token || token == ibm_errinjct_token) {
+		/*
+		 * It would be nicer to not discard the error value
+		 * from security_locked_down(), but callers expect an
+		 * RTAS status, not an errno.
+		 */
+		if (security_locked_down(LOCKDOWN_RTAS_ERROR_INJECTION))
+			return -1;
+	}
+
 	if ((mfmsr() & (MSR_IR|MSR_DR)) != (MSR_IR|MSR_DR)) {
 		WARN_ON_ONCE(1);
 		return -1;
@@ -1227,6 +1241,14 @@  SYSCALL_DEFINE1(rtas, struct rtas_args __user *, uargs)
 	if (block_rtas_call(token, nargs, &args))
 		return -EINVAL;
 
+	if (token == ibm_open_errinjct_token || token == ibm_errinjct_token) {
+		int err;
+
+		err = security_locked_down(LOCKDOWN_RTAS_ERROR_INJECTION);
+		if (err)
+			return err;
+	}
+
 	/* Need to handle ibm,suspend_me call specially */
 	if (token == rtas_token("ibm,suspend-me")) {
 
@@ -1325,7 +1347,8 @@  void __init rtas_initialize(void)
 #ifdef CONFIG_RTAS_ERROR_LOGGING
 	rtas_last_error_token = rtas_token("rtas-last-error");
 #endif
-
+	ibm_open_errinjct_token = rtas_token("ibm,open-errinjct");
+	ibm_errinjct_token = rtas_token("ibm,errinjct");
 	rtas_syscall_filter_init();
 }
 
diff --git a/include/linux/security.h b/include/linux/security.h
index 1ca8dbacd3cc..b5d5138ae66a 100644
--- a/include/linux/security.h
+++ b/include/linux/security.h
@@ -123,6 +123,7 @@  enum lockdown_reason {
 	LOCKDOWN_BPF_WRITE_USER,
 	LOCKDOWN_DBG_WRITE_KERNEL,
 	LOCKDOWN_DEVICE_TREE,
+	LOCKDOWN_RTAS_ERROR_INJECTION,
 	LOCKDOWN_INTEGRITY_MAX,
 	LOCKDOWN_KCORE,
 	LOCKDOWN_KPROBES,
diff --git a/security/security.c b/security/security.c
index 2863fc31eec6..6518b239ada2 100644
--- a/security/security.c
+++ b/security/security.c
@@ -61,6 +61,7 @@  const char *const lockdown_reasons[LOCKDOWN_CONFIDENTIALITY_MAX+1] = {
 	[LOCKDOWN_BPF_WRITE_USER] = "use of bpf to write user RAM",
 	[LOCKDOWN_DBG_WRITE_KERNEL] = "use of kgdb/kdb to write kernel RAM",
 	[LOCKDOWN_DEVICE_TREE] = "modifying device tree contents",
+	[LOCKDOWN_RTAS_ERROR_INJECTION] = "RTAS error injection",
 	[LOCKDOWN_INTEGRITY_MAX] = "integrity",
 	[LOCKDOWN_KCORE] = "/proc/kcore access",
 	[LOCKDOWN_KPROBES] = "use of kprobes",