[PATCH 4/5] keyctl.h: update pointer types

[PATCH 4/5] keyctl.h: update pointer types

[Ben Boeckel]

These should be `const` since the kernel should not touch them. In
addition `otherinfo` is defined to be opaque, so make it a `void *`.

Signed-off-by: Ben Boeckel <mathstuf@gmail.com>
---
 include/uapi/linux/keyctl.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/include/uapi/linux/keyctl.h b/include/uapi/linux/keyctl.h
index 1a5a7948606e..6004e05b685b 100644
--- a/include/uapi/linux/keyctl.h
+++ b/include/uapi/linux/keyctl.h
@@ -76,8 +76,8 @@ struct keyctl_dh_params {
 };
 
 struct keyctl_kdf_params {
-	char __user *hashname;
-	char __user *otherinfo;
+	const char __user *hashname;
+	const void __user *otherinfo;
 	__u32 otherinfolen;
 	__u32 __spare[8];
 };
-- 
2.17.1

Re: [PATCH 4/5] keyctl.h: update pointer types

[David Howells]

Ben Boeckel <mathstuf@gmail.com> wrote:

> These should be `const` since the kernel should not touch them. In
> addition `otherinfo` is defined to be opaque, so make it a `void *`.

You're changing the UAPI - and this no longer matches what's in keyutils.h in
userspace.

The UAPI change might potentially cause userspace to fail to compile if
someone was assigning the char * pointer from the struct to another char *
pointer.

Unless there's an actual bug to fix (see the 'private' problem), I would
strongly recommend living with this.

David

Re: [PATCH 4/5] keyctl.h: update pointer types

[Ben Boeckel]

On Thu, Sep 27, 2018 at 18:08:03 +0100, David Howells wrote:
> Ben Boeckel <mathstuf@gmail.com> wrote:
> > These should be `const` since the kernel should not touch them. In
> > addition `otherinfo` is defined to be opaque, so make it a `void *`.
> 
> You're changing the UAPI - and this no longer matches what's in keyutils.h in
> userspace.

Please see patch #5/7 of the keyutils patches on this thread for that change.

> The UAPI change might potentially cause userspace to fail to compile if
> someone was assigning the char * pointer from the struct to another char *
> pointer.

Is this in a released kernel? Isn't the KDF code being added in another
pull request of yours[1]?

--Ben

[1]https://lwn.net/Articles/764227/