[PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[Nathan Lynch via B4 Relay]

From: Nathan Lynch <nathanl@linux.ibm.com>

Smatch warns:

  arch/powerpc/kernel/rtas.c:1932 __do_sys_rtas() warn: potential
  spectre issue 'args.args' [r] (local cap)

The 'nargs' and 'nret' locals come directly from a user-supplied
buffer and are used as indexes into a small stack-based array and as
inputs to copy_to_user() after they are subject to bounds checks.

Use array_index_nospec() after the bounds checks to clamp these values
for speculative execution.

Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
Reported-by: Breno Leitao <leitao@debian.org>
---
Based on a change originally submitted by Breno Leitao in 2018:

https://lore.kernel.org/linuxppc-dev/1534876926-21849-1-git-send-email-leitao@debian.org/

I've used a Reported-by: tag to credit Breno, let me know if you would
prefer a different tag (perhaps Co-developed-by?)
---
 arch/powerpc/kernel/rtas.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/arch/powerpc/kernel/rtas.c b/arch/powerpc/kernel/rtas.c
index 8064d9c3de86..f7e86e09c49f 100644
--- a/arch/powerpc/kernel/rtas.c
+++ b/arch/powerpc/kernel/rtas.c
@@ -19,6 +19,7 @@
 #include <linux/lockdep.h>
 #include <linux/memblock.h>
 #include <linux/mutex.h>
+#include <linux/nospec.h>
 #include <linux/of.h>
 #include <linux/of_fdt.h>
 #include <linux/reboot.h>
@@ -1916,6 +1917,9 @@ SYSCALL_DEFINE1(rtas, struct rtas_args __user *, uargs)
 	    || nargs + nret > ARRAY_SIZE(args.args))
 		return -EINVAL;
 
+	nargs = array_index_nospec(nargs, ARRAY_SIZE(args.args));
+	nret = array_index_nospec(nret, ARRAY_SIZE(args.args) - nargs);
+
 	/* Copy in args. */
 	if (copy_from_user(args.args, uargs->args,
 			   nargs * sizeof(rtas_arg_t)) != 0)

---
base-commit: be2fc65d66e0406cc9d39d40becaecdf4ee765f3
change-id: 20240530-sys_rtas-nargs-nret-a188eaf5a500

Best regards,
-- 
Nathan Lynch <nathanl@linux.ibm.com>

Re: [PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[Breno Leitao]

On Thu, May 30, 2024 at 07:44:12PM -0500, Nathan Lynch via B4 Relay wrote:
> From: Nathan Lynch <nathanl@linux.ibm.com>
> 
> Smatch warns:
> 
>   arch/powerpc/kernel/rtas.c:1932 __do_sys_rtas() warn: potential
>   spectre issue 'args.args' [r] (local cap)
> 
> The 'nargs' and 'nret' locals come directly from a user-supplied
> buffer and are used as indexes into a small stack-based array and as
> inputs to copy_to_user() after they are subject to bounds checks.
> 
> Use array_index_nospec() after the bounds checks to clamp these values
> for speculative execution.
> 
> Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
> Reported-by: Breno Leitao <leitao@debian.org>

Thanks for working on it. 

Reviewed-by: Breno Leitao <leitao@debian.org>

> +	nargs = array_index_nospec(nargs, ARRAY_SIZE(args.args));
> +	nret = array_index_nospec(nret, ARRAY_SIZE(args.args) - nargs);

On an unrelated note, can nargs and nret are integers and could be
eventually negative. Is this a valid use case?

Thanks!

Re: [PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[Nathan Lynch]

Breno Leitao <leitao@debian.org> writes:

> On Thu, May 30, 2024 at 07:44:12PM -0500, Nathan Lynch via B4 Relay wrote:
>> From: Nathan Lynch <nathanl@linux.ibm.com>
>> 
>> Smatch warns:
>> 
>>   arch/powerpc/kernel/rtas.c:1932 __do_sys_rtas() warn: potential
>>   spectre issue 'args.args' [r] (local cap)
>> 
>> The 'nargs' and 'nret' locals come directly from a user-supplied
>> buffer and are used as indexes into a small stack-based array and as
>> inputs to copy_to_user() after they are subject to bounds checks.
>> 
>> Use array_index_nospec() after the bounds checks to clamp these values
>> for speculative execution.
>> 
>> Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
>> Reported-by: Breno Leitao <leitao@debian.org>
>
> Thanks for working on it. 
>
> Reviewed-by: Breno Leitao <leitao@debian.org>

Thanks!

>
>> +	nargs = array_index_nospec(nargs, ARRAY_SIZE(args.args));
>> +	nret = array_index_nospec(nret, ARRAY_SIZE(args.args) - nargs);
>
> On an unrelated note, can nargs and nret are integers and could be
> eventually negative. Is this a valid use case?

No, it's not valid for a caller to provide negative nargs or nret. I
convinced myself that this bounds check:

	nargs = be32_to_cpu(args.nargs);
	nret  = be32_to_cpu(args.nret);

	if (nargs >= ARRAY_SIZE(args.args)
	    || nret > ARRAY_SIZE(args.args)
	    || nargs + nret > ARRAY_SIZE(args.args))
		return -EINVAL;

rejects negative values of nargs or nret due to C's "usual arithmetic
conversions", where nargs and nret are implicitly converted to size_t
for the comparisons.

However I don't see any value in keeping them as signed int. I have some
changes in progress in this area and I'll plan on making these unsigned.

Re: [PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[Breno Leitao]

On Fri, May 31, 2024 at 11:45:48AM -0500, Nathan Lynch wrote:
> Breno Leitao <leitao@debian.org> writes:
> 
> > On Thu, May 30, 2024 at 07:44:12PM -0500, Nathan Lynch via B4 Relay wrote:
> >> From: Nathan Lynch <nathanl@linux.ibm.com>

> >> +	nargs = array_index_nospec(nargs, ARRAY_SIZE(args.args));
> >> +	nret = array_index_nospec(nret, ARRAY_SIZE(args.args) - nargs);
> >
> > On an unrelated note, can nargs and nret are integers and could be
> > eventually negative. Is this a valid use case?
> 
> No, it's not valid for a caller to provide negative nargs or nret. I
> convinced myself that this bounds check:
> 
> 	nargs = be32_to_cpu(args.nargs);
> 	nret  = be32_to_cpu(args.nret);
> 
> 	if (nargs >= ARRAY_SIZE(args.args)
> 	    || nret > ARRAY_SIZE(args.args)
> 	    || nargs + nret > ARRAY_SIZE(args.args))
> 		return -EINVAL;
> 
> rejects negative values of nargs or nret due to C's "usual arithmetic
> conversions", where nargs and nret are implicitly converted to size_t
> for the comparisons.
> 
> However I don't see any value in keeping them as signed int. I have some
> changes in progress in this area and I'll plan on making these unsigned.

yea, I think it will help to make this code easier to read/review.

Thanks again for fixing it.

Re: [PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[Michael Ellerman]

On Thu, 30 May 2024 19:44:12 -0500, Nathan Lynch wrote:
> Smatch warns:
> 
>   arch/powerpc/kernel/rtas.c:1932 __do_sys_rtas() warn: potential
>   spectre issue 'args.args' [r] (local cap)
> 
> The 'nargs' and 'nret' locals come directly from a user-supplied
> buffer and are used as indexes into a small stack-based array and as
> inputs to copy_to_user() after they are subject to bounds checks.
> 
> [...]

Applied to powerpc/next.

[1/1] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()
      https://git.kernel.org/powerpc/c/0974d03eb479384466d828d65637814bee6b26d7

cheers

[PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[haixiao.yan.cn]

From: Nathan Lynch <nathanl@linux.ibm.com>

[ Upstream commit 0974d03eb479384466d828d65637814bee6b26d7 ]

Smatch warns:

  arch/powerpc/kernel/rtas.c:1932 __do_sys_rtas() warn: potential
  spectre issue 'args.args' [r] (local cap)

The 'nargs' and 'nret' locals come directly from a user-supplied
buffer and are used as indexes into a small stack-based array and as
inputs to copy_to_user() after they are subject to bounds checks.

Use array_index_nospec() after the bounds checks to clamp these values
for speculative execution.

Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
Reported-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Breno Leitao <leitao@debian.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240530-sys_rtas-nargs-nret-v1-1-129acddd4d89@linux.ibm.com
Signed-off-by: Haixiao Yan <haixiao.yan.cn@windriver.com>
---
This commit is backporting 0974d03eb479 to the branch linux-5.15.y to
solve the CVE-2024-46774.

 arch/powerpc/kernel/rtas.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/arch/powerpc/kernel/rtas.c b/arch/powerpc/kernel/rtas.c
index d01a0ad57e38..f2378f51cbed 100644
--- a/arch/powerpc/kernel/rtas.c
+++ b/arch/powerpc/kernel/rtas.c
@@ -16,6 +16,7 @@
 #include <linux/capability.h>
 #include <linux/delay.h>
 #include <linux/cpu.h>
+#include <linux/nospec.h>
 #include <linux/sched.h>
 #include <linux/smp.h>
 #include <linux/completion.h>
@@ -1076,6 +1077,9 @@ SYSCALL_DEFINE1(rtas, struct rtas_args __user *, uargs)
 	    || nargs + nret > ARRAY_SIZE(args.args))
 		return -EINVAL;
 
+	nargs = array_index_nospec(nargs, ARRAY_SIZE(args.args));
+	nret = array_index_nospec(nret, ARRAY_SIZE(args.args) - nargs);
+
 	/* Copy in args. */
 	if (copy_from_user(args.args, uargs->args,
 			   nargs * sizeof(rtas_arg_t)) != 0)
-- 
2.35.5

Re: [PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[Greg KH]

On Fri, Dec 13, 2024 at 11:44:22AM +0800, haixiao.yan.cn@eng.windriver.com wrote:
> From: Nathan Lynch <nathanl@linux.ibm.com>
> 
> [ Upstream commit 0974d03eb479384466d828d65637814bee6b26d7 ]
> 
> Smatch warns:
> 
>   arch/powerpc/kernel/rtas.c:1932 __do_sys_rtas() warn: potential
>   spectre issue 'args.args' [r] (local cap)
> 
> The 'nargs' and 'nret' locals come directly from a user-supplied
> buffer and are used as indexes into a small stack-based array and as
> inputs to copy_to_user() after they are subject to bounds checks.
> 
> Use array_index_nospec() after the bounds checks to clamp these values
> for speculative execution.
> 
> Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
> Reported-by: Breno Leitao <leitao@debian.org>
> Reviewed-by: Breno Leitao <leitao@debian.org>
> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
> Link: https://msgid.link/20240530-sys_rtas-nargs-nret-v1-1-129acddd4d89@linux.ibm.com
> Signed-off-by: Haixiao Yan <haixiao.yan.cn@windriver.com>
> ---
> This commit is backporting 0974d03eb479 to the branch linux-5.15.y to
> solve the CVE-2024-46774.

Now deleted, please see:
	https://lore.kernel.org/r/2024121322-conjuror-gap-b542@gregkh
for what you all need to do, TOGETHER, to get this fixed and so that I
can accept patches from your company in the future.

thanks,

greg k-h

Re: [PATCH] powerpc/rtas: Prevent Spectre v1 gadget construction in sys_rtas()

[Greg KH]

On Fri, Dec 13, 2024 at 11:44:22AM +0800, haixiao.yan.cn@eng.windriver.com wrote:
> From: Nathan Lynch <nathanl@linux.ibm.com>
> 
> [ Upstream commit 0974d03eb479384466d828d65637814bee6b26d7 ]
> 


Now deleted, please see:
        https://lore.kernel.org/r/2024121322-conjuror-gap-b542@gregkh
for what you all need to do, TOGETHER, to get this fixed and so that I
can accept patches from your company in the future.

thanks,

greg k-h