diff mbox series

[HID,v3,15/16] HID: bpf: rework hid_bpf_ops_btf_struct_access

Message ID 20240608-hid_bpf_struct_ops-v3-15-6ac6ade58329@kernel.org (mailing list archive)
State Not Applicable
Headers show
Series HID: convert HID-BPF into using bpf_struct_ops | expand

Checks

Context Check Description
netdev/tree_selection success Not a local patch, async

Commit Message

Benjamin Tissoires June 8, 2024, 9:01 a.m. UTC
The idea is to provide a list of stucts and their editable fields.

Currently no functional changes are introduced here, we will add some
more writeable fields in the next patch.

Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>

---

changes in v3:
- rewrote WRITE_RANGE macro to not deal with offset from the caller side

new in v2
---
 drivers/hid/bpf/hid_bpf_struct_ops.c | 91 +++++++++++++++++++++++++++---------
 1 file changed, 69 insertions(+), 22 deletions(-)

Comments

Alexei Starovoitov June 10, 2024, 6:39 p.m. UTC | #1
On Sat, Jun 8, 2024 at 2:01 AM Benjamin Tissoires <bentiss@kernel.org> wrote:
>
> The idea is to provide a list of stucts and their editable fields.
>
> Currently no functional changes are introduced here, we will add some
> more writeable fields in the next patch.
>
> Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>
>
> ---
>
> changes in v3:
> - rewrote WRITE_RANGE macro to not deal with offset from the caller side
>
> new in v2
> ---
>  drivers/hid/bpf/hid_bpf_struct_ops.c | 91 +++++++++++++++++++++++++++---------
>  1 file changed, 69 insertions(+), 22 deletions(-)
>
> diff --git a/drivers/hid/bpf/hid_bpf_struct_ops.c b/drivers/hid/bpf/hid_bpf_struct_ops.c
> index 056d05d96962..b14eccb121e0 100644
> --- a/drivers/hid/bpf/hid_bpf_struct_ops.c
> +++ b/drivers/hid/bpf/hid_bpf_struct_ops.c
> @@ -16,6 +16,7 @@
>  #include <linux/hid_bpf.h>
>  #include <linux/init.h>
>  #include <linux/module.h>
> +#include <linux/stddef.h>
>  #include <linux/workqueue.h>
>  #include "hid_bpf_dispatch.h"
>
> @@ -52,40 +53,86 @@ static int hid_bpf_ops_check_member(const struct btf_type *t,
>         return 0;
>  }
>
> +struct hid_bpf_offset_write_range {
> +       const char *struct_name;
> +       u32 struct_length;
> +       u32 start;
> +       u32 end;
> +};
> +
>  static int hid_bpf_ops_btf_struct_access(struct bpf_verifier_log *log,
>                                            const struct bpf_reg_state *reg,
>                                            int off, int size)
>  {
> -       const struct btf_type *state;
> -       const struct btf_type *t;
> -       s32 type_id;
> +#define WRITE_RANGE(_name, _field, _is_string)                                 \
> +       {                                                                       \
> +               .struct_name = #_name,                                          \
> +               .struct_length = sizeof(struct _name),                          \
> +               .start = offsetof(struct _name, _field),                        \
> +               .end = offsetofend(struct _name, _field) - !!(_is_string),

so it works because char name[128]; had last byte as zero
before prog writes into it (in addition to potentially having
earlier 0 bytes), so the string is guaranteed
to be null-terminated regardless of what prog writes into it.
Right?

Overall:
Acked-by: Alexei Starovoitov <ast@kernel.org>
Benjamin Tissoires June 11, 2024, 4:57 p.m. UTC | #2
On Jun 10 2024, Alexei Starovoitov wrote:
> On Sat, Jun 8, 2024 at 2:01 AM Benjamin Tissoires <bentiss@kernel.org> wrote:
> >
> > The idea is to provide a list of stucts and their editable fields.
> >
> > Currently no functional changes are introduced here, we will add some
> > more writeable fields in the next patch.
> >
> > Signed-off-by: Benjamin Tissoires <bentiss@kernel.org>
> >
> > ---
> >
> > changes in v3:
> > - rewrote WRITE_RANGE macro to not deal with offset from the caller side
> >
> > new in v2
> > ---
> >  drivers/hid/bpf/hid_bpf_struct_ops.c | 91 +++++++++++++++++++++++++++---------
> >  1 file changed, 69 insertions(+), 22 deletions(-)
> >
> > diff --git a/drivers/hid/bpf/hid_bpf_struct_ops.c b/drivers/hid/bpf/hid_bpf_struct_ops.c
> > index 056d05d96962..b14eccb121e0 100644
> > --- a/drivers/hid/bpf/hid_bpf_struct_ops.c
> > +++ b/drivers/hid/bpf/hid_bpf_struct_ops.c
> > @@ -16,6 +16,7 @@
> >  #include <linux/hid_bpf.h>
> >  #include <linux/init.h>
> >  #include <linux/module.h>
> > +#include <linux/stddef.h>
> >  #include <linux/workqueue.h>
> >  #include "hid_bpf_dispatch.h"
> >
> > @@ -52,40 +53,86 @@ static int hid_bpf_ops_check_member(const struct btf_type *t,
> >         return 0;
> >  }
> >
> > +struct hid_bpf_offset_write_range {
> > +       const char *struct_name;
> > +       u32 struct_length;
> > +       u32 start;
> > +       u32 end;
> > +};
> > +
> >  static int hid_bpf_ops_btf_struct_access(struct bpf_verifier_log *log,
> >                                            const struct bpf_reg_state *reg,
> >                                            int off, int size)
> >  {
> > -       const struct btf_type *state;
> > -       const struct btf_type *t;
> > -       s32 type_id;
> > +#define WRITE_RANGE(_name, _field, _is_string)                                 \
> > +       {                                                                       \
> > +               .struct_name = #_name,                                          \
> > +               .struct_length = sizeof(struct _name),                          \
> > +               .start = offsetof(struct _name, _field),                        \
> > +               .end = offsetofend(struct _name, _field) - !!(_is_string),
> 
> so it works because char name[128]; had last byte as zero
> before prog writes into it (in addition to potentially having
> earlier 0 bytes), so the string is guaranteed
> to be null-terminated regardless of what prog writes into it.
> Right?

Yeah, struct hid_device is created through hid_allocate_device(), which
does a kzalloc. Then all operations are supposedly safe in the current
transport layers, so the last byte should always be \0.

> 
> Overall:
> Acked-by: Alexei Starovoitov <ast@kernel.org>

Thanks a lot.

I might send a v4 or not depending if I get other reviews, but I'll make
sure to take your nitpick in 3/16 into account (cast kdata/udate in the
beginning of the function to make the lines shorter and less verbose).

Cheers,
Benjamin
diff mbox series

Patch

diff --git a/drivers/hid/bpf/hid_bpf_struct_ops.c b/drivers/hid/bpf/hid_bpf_struct_ops.c
index 056d05d96962..b14eccb121e0 100644
--- a/drivers/hid/bpf/hid_bpf_struct_ops.c
+++ b/drivers/hid/bpf/hid_bpf_struct_ops.c
@@ -16,6 +16,7 @@ 
 #include <linux/hid_bpf.h>
 #include <linux/init.h>
 #include <linux/module.h>
+#include <linux/stddef.h>
 #include <linux/workqueue.h>
 #include "hid_bpf_dispatch.h"
 
@@ -52,40 +53,86 @@  static int hid_bpf_ops_check_member(const struct btf_type *t,
 	return 0;
 }
 
+struct hid_bpf_offset_write_range {
+	const char *struct_name;
+	u32 struct_length;
+	u32 start;
+	u32 end;
+};
+
 static int hid_bpf_ops_btf_struct_access(struct bpf_verifier_log *log,
 					   const struct bpf_reg_state *reg,
 					   int off, int size)
 {
-	const struct btf_type *state;
-	const struct btf_type *t;
-	s32 type_id;
+#define WRITE_RANGE(_name, _field, _is_string)					\
+	{									\
+		.struct_name = #_name,						\
+		.struct_length = sizeof(struct _name),				\
+		.start = offsetof(struct _name, _field),			\
+		.end = offsetofend(struct _name, _field) - !!(_is_string),	\
+	}
 
-	type_id = btf_find_by_name_kind(reg->btf, "hid_bpf_ctx",
-					BTF_KIND_STRUCT);
-	if (type_id < 0)
-		return -EINVAL;
+	const struct hid_bpf_offset_write_range write_ranges[] = {
+		WRITE_RANGE(hid_bpf_ctx, retval, false),
+	};
+#undef WRITE_RANGE
+	const struct btf_type *state = NULL;
+	const struct btf_type *t;
+	const char *cur = NULL;
+	int i;
 
 	t = btf_type_by_id(reg->btf, reg->btf_id);
-	state = btf_type_by_id(reg->btf, type_id);
-	if (t != state) {
-		bpf_log(log, "only access to hid_bpf_ctx is supported\n");
-		return -EACCES;
-	}
 
-	/* out-of-bound access in hid_bpf_ctx */
-	if (off + size > sizeof(struct hid_bpf_ctx)) {
-		bpf_log(log, "write access at off %d with size %d\n", off, size);
-		return -EACCES;
+	for (i = 0; i < ARRAY_SIZE(write_ranges); i++) {
+		const struct hid_bpf_offset_write_range *write_range = &write_ranges[i];
+		s32 type_id;
+
+		/* we already found a writeable struct, but there is a
+		 * new one, let's break the loop.
+		 */
+		if (t == state && write_range->struct_name != cur)
+			break;
+
+		/* new struct to look for */
+		if (write_range->struct_name != cur) {
+			type_id = btf_find_by_name_kind(reg->btf, write_range->struct_name,
+							BTF_KIND_STRUCT);
+			if (type_id < 0)
+				return -EINVAL;
+
+			state = btf_type_by_id(reg->btf, type_id);
+		}
+
+		/* this is not the struct we are looking for */
+		if (t != state) {
+			cur = write_range->struct_name;
+			continue;
+		}
+
+		/* first time we see this struct, check for out of bounds */
+		if (cur != write_range->struct_name &&
+		    off + size > write_range->struct_length) {
+			bpf_log(log, "write access for struct %s at off %d with size %d\n",
+				write_range->struct_name, off, size);
+			return -EACCES;
+		}
+
+		/* now check if we are in our boundaries */
+		if (off >= write_range->start && off + size <= write_range->end)
+			return NOT_INIT;
+
+		cur = write_range->struct_name;
 	}
 
-	if (off < offsetof(struct hid_bpf_ctx, retval)) {
+
+	if (t != state)
+		bpf_log(log, "write access to this struct is not supported\n");
+	else
 		bpf_log(log,
-			"write access at off %d with size %d on read-only part of hid_bpf_ctx\n",
-			off, size);
-		return -EACCES;
-	}
+			"write access at off %d with size %d on read-only part of %s\n",
+			off, size, cur);
 
-	return NOT_INIT;
+	return -EACCES;
 }
 
 static const struct bpf_verifier_ops hid_bpf_verifier_ops = {