linux-input.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] Input: gtco - fix potential out-of-bound access
@ 2017-10-24  5:28 Dmitry Torokhov
  2017-10-24 11:04 ` Andrey Konovalov
  0 siblings, 1 reply; 5+ messages in thread
From: Dmitry Torokhov @ 2017-10-24  5:28 UTC (permalink / raw)
  To: linux-input; +Cc: Andrey Konovalov, linux-kernel

parse_hid_report_descriptor() has a while (i < length) loop, which
only guarantees that there's at least 1 byte in the buffer, but the
loop body can read multiple bytes which causes out-of-bounds access.

Reported-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
 drivers/input/tablet/gtco.c | 24 +++++++++++++++++-------
 1 file changed, 17 insertions(+), 7 deletions(-)

diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
index b796e891e2ee..0351203b8c24 100644
--- a/drivers/input/tablet/gtco.c
+++ b/drivers/input/tablet/gtco.c
@@ -230,13 +230,24 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
 
 	/* Walk  this report and pull out the info we need */
 	while (i < length) {
-		prefix = report[i];
-
-		/* Skip over prefix */
-		i++;
+		prefix = report[i++];
 
 		/* Determine data size and save the data in the proper variable */
-		size = PREF_SIZE(prefix);
+		if (PREF_SIZE(prefix) < 1) {
+			dev_err(ddev,
+				"Invalid size %d in element at offset %d\n",
+				PREF_SIZE(prefix), i);
+			break;
+		}
+
+		size = 1U << (PREF_SIZE(prefix) - 1);
+		if (i + size >= length) {
+			dev_err(ddev,
+				"Not enough data (need %d, have %d)\n",
+				i + size, length);
+			break;
+		}
+
 		switch (size) {
 		case 1:
 			data = report[i];
@@ -244,8 +255,7 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
 		case 2:
 			data16 = get_unaligned_le16(&report[i]);
 			break;
-		case 3:
-			size = 4;
+		case 4:
 			data32 = get_unaligned_le32(&report[i]);
 			break;
 		}
-- 
2.15.0.rc0.271.g36b669edcc-goog


-- 
Dmitry

^ permalink raw reply related	[flat|nested] 5+ messages in thread

* Re: [PATCH] Input: gtco - fix potential out-of-bound access
  2017-10-24  5:28 [PATCH] Input: gtco - fix potential out-of-bound access Dmitry Torokhov
@ 2017-10-24 11:04 ` Andrey Konovalov
  2017-10-24 16:39   ` Dmitry Torokhov
  0 siblings, 1 reply; 5+ messages in thread
From: Andrey Konovalov @ 2017-10-24 11:04 UTC (permalink / raw)
  To: Dmitry Torokhov; +Cc: linux-input, LKML

On Tue, Oct 24, 2017 at 7:28 AM, Dmitry Torokhov
<dmitry.torokhov@gmail.com> wrote:
> parse_hid_report_descriptor() has a while (i < length) loop, which
> only guarantees that there's at least 1 byte in the buffer, but the
> loop body can read multiple bytes which causes out-of-bounds access.
>
> Reported-by: Andrey Konovalov <andreyknvl@google.com>
> Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> ---
>  drivers/input/tablet/gtco.c | 24 +++++++++++++++++-------
>  1 file changed, 17 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
> index b796e891e2ee..0351203b8c24 100644
> --- a/drivers/input/tablet/gtco.c
> +++ b/drivers/input/tablet/gtco.c
> @@ -230,13 +230,24 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
>
>         /* Walk  this report and pull out the info we need */
>         while (i < length) {
> -               prefix = report[i];
> -
> -               /* Skip over prefix */
> -               i++;
> +               prefix = report[i++];
>
>                 /* Determine data size and save the data in the proper variable */
> -               size = PREF_SIZE(prefix);
> +               if (PREF_SIZE(prefix) < 1) {

AFAIU PREF_SIZE(prefix) == 0 is a perfectly valid item data size.

> +                       dev_err(ddev,
> +                               "Invalid size %d in element at offset %d\n",
> +                               PREF_SIZE(prefix), i);
> +                       break;
> +               }
> +
> +               size = 1U << (PREF_SIZE(prefix) - 1);

This formula doesn't work with PREF_SIZE(prefix) == 0.

> +               if (i + size >= length) {
> +                       dev_err(ddev,
> +                               "Not enough data (need %d, have %d)\n",
> +                               i + size, length);
> +                       break;
> +               }
> +
>                 switch (size) {
>                 case 1:
>                         data = report[i];
> @@ -244,8 +255,7 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
>                 case 2:
>                         data16 = get_unaligned_le16(&report[i]);
>                         break;
> -               case 3:
> -                       size = 4;
> +               case 4:
>                         data32 = get_unaligned_le32(&report[i]);
>                         break;
>                 }
> --
> 2.15.0.rc0.271.g36b669edcc-goog
>
>
> --
> Dmitry

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] Input: gtco - fix potential out-of-bound access
  2017-10-24 11:04 ` Andrey Konovalov
@ 2017-10-24 16:39   ` Dmitry Torokhov
  2017-10-25 10:33     ` Andrey Konovalov
  0 siblings, 1 reply; 5+ messages in thread
From: Dmitry Torokhov @ 2017-10-24 16:39 UTC (permalink / raw)
  To: Andrey Konovalov; +Cc: linux-input, LKML

On Tue, Oct 24, 2017 at 01:04:03PM +0200, Andrey Konovalov wrote:
> On Tue, Oct 24, 2017 at 7:28 AM, Dmitry Torokhov
> <dmitry.torokhov@gmail.com> wrote:
> > parse_hid_report_descriptor() has a while (i < length) loop, which
> > only guarantees that there's at least 1 byte in the buffer, but the
> > loop body can read multiple bytes which causes out-of-bounds access.
> >
> > Reported-by: Andrey Konovalov <andreyknvl@google.com>
> > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> > ---
> >  drivers/input/tablet/gtco.c | 24 +++++++++++++++++-------
> >  1 file changed, 17 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
> > index b796e891e2ee..0351203b8c24 100644
> > --- a/drivers/input/tablet/gtco.c
> > +++ b/drivers/input/tablet/gtco.c
> > @@ -230,13 +230,24 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
> >
> >         /* Walk  this report and pull out the info we need */
> >         while (i < length) {
> > -               prefix = report[i];
> > -
> > -               /* Skip over prefix */
> > -               i++;
> > +               prefix = report[i++];
> >
> >                 /* Determine data size and save the data in the proper variable */
> > -               size = PREF_SIZE(prefix);
> > +               if (PREF_SIZE(prefix) < 1) {
> 
> AFAIU PREF_SIZE(prefix) == 0 is a perfectly valid item data size.

Fair enough. How about the below instead then?

-- 
Dmitry


Input: gtco - fix potential out-of-bound access

From: Dmitry Torokhov <dmitry.torokhov@gmail.com>

parse_hid_report_descriptor() has a while (i < length) loop, which
only guarantees that there's at least 1 byte in the buffer, but the
loop body can read multiple bytes which causes out-of-bounds access.

Reported-by: Andrey Konovalov <andreyknvl@google.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
---
 drivers/input/tablet/gtco.c |   17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
index b796e891e2ee..7d8e9fb831c4 100644
--- a/drivers/input/tablet/gtco.c
+++ b/drivers/input/tablet/gtco.c
@@ -230,13 +230,17 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
 
 	/* Walk  this report and pull out the info we need */
 	while (i < length) {
-		prefix = report[i];
-
-		/* Skip over prefix */
-		i++;
+		prefix = report[i++];
 
 		/* Determine data size and save the data in the proper variable */
-		size = PREF_SIZE(prefix);
+		size = (1U << PREF_SIZE(prefix)) >> 1;
+		if (size && i + size >= length) {
+			dev_err(ddev,
+				"Not enough data (need %d, have %d)\n",
+				i + size, length);
+			break;
+		}
+
 		switch (size) {
 		case 1:
 			data = report[i];
@@ -244,8 +248,7 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
 		case 2:
 			data16 = get_unaligned_le16(&report[i]);
 			break;
-		case 3:
-			size = 4;
+		case 4:
 			data32 = get_unaligned_le32(&report[i]);
 			break;
 		}

^ permalink raw reply related	[flat|nested] 5+ messages in thread

* Re: [PATCH] Input: gtco - fix potential out-of-bound access
  2017-10-24 16:39   ` Dmitry Torokhov
@ 2017-10-25 10:33     ` Andrey Konovalov
  2017-10-25 16:10       ` Dmitry Torokhov
  0 siblings, 1 reply; 5+ messages in thread
From: Andrey Konovalov @ 2017-10-25 10:33 UTC (permalink / raw)
  To: Dmitry Torokhov; +Cc: linux-input, LKML

On Tue, Oct 24, 2017 at 6:39 PM, Dmitry Torokhov
<dmitry.torokhov@gmail.com> wrote:
> On Tue, Oct 24, 2017 at 01:04:03PM +0200, Andrey Konovalov wrote:
>> On Tue, Oct 24, 2017 at 7:28 AM, Dmitry Torokhov
>> <dmitry.torokhov@gmail.com> wrote:
>> > parse_hid_report_descriptor() has a while (i < length) loop, which
>> > only guarantees that there's at least 1 byte in the buffer, but the
>> > loop body can read multiple bytes which causes out-of-bounds access.
>> >
>> > Reported-by: Andrey Konovalov <andreyknvl@google.com>
>> > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
>> > ---
>> >  drivers/input/tablet/gtco.c | 24 +++++++++++++++++-------
>> >  1 file changed, 17 insertions(+), 7 deletions(-)
>> >
>> > diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
>> > index b796e891e2ee..0351203b8c24 100644
>> > --- a/drivers/input/tablet/gtco.c
>> > +++ b/drivers/input/tablet/gtco.c
>> > @@ -230,13 +230,24 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
>> >
>> >         /* Walk  this report and pull out the info we need */
>> >         while (i < length) {
>> > -               prefix = report[i];
>> > -
>> > -               /* Skip over prefix */
>> > -               i++;
>> > +               prefix = report[i++];
>> >
>> >                 /* Determine data size and save the data in the proper variable */
>> > -               size = PREF_SIZE(prefix);
>> > +               if (PREF_SIZE(prefix) < 1) {
>>
>> AFAIU PREF_SIZE(prefix) == 0 is a perfectly valid item data size.
>
> Fair enough. How about the below instead then?
>
> --
> Dmitry
>
>
> Input: gtco - fix potential out-of-bound access
>
> From: Dmitry Torokhov <dmitry.torokhov@gmail.com>
>
> parse_hid_report_descriptor() has a while (i < length) loop, which
> only guarantees that there's at least 1 byte in the buffer, but the
> loop body can read multiple bytes which causes out-of-bounds access.
>
> Reported-by: Andrey Konovalov <andreyknvl@google.com>
> Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> ---
>  drivers/input/tablet/gtco.c |   17 ++++++++++-------
>  1 file changed, 10 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
> index b796e891e2ee..7d8e9fb831c4 100644
> --- a/drivers/input/tablet/gtco.c
> +++ b/drivers/input/tablet/gtco.c
> @@ -230,13 +230,17 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
>
>         /* Walk  this report and pull out the info we need */
>         while (i < length) {
> -               prefix = report[i];
> -
> -               /* Skip over prefix */
> -               i++;
> +               prefix = report[i++];
>
>                 /* Determine data size and save the data in the proper variable */
> -               size = PREF_SIZE(prefix);
> +               size = (1U << PREF_SIZE(prefix)) >> 1;
> +               if (size && i + size >= length) {

I think this should be >, not >=, as when i + size == length, item
data fits into the remaining space precisely. Also I don't see much
point in checking that size is not 0, but I don't mind it.

> +                       dev_err(ddev,
> +                               "Not enough data (need %d, have %d)\n",
> +                               i + size, length);
> +                       break;
> +               }
> +
>                 switch (size) {
>                 case 1:
>                         data = report[i];
> @@ -244,8 +248,7 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
>                 case 2:
>                         data16 = get_unaligned_le16(&report[i]);
>                         break;
> -               case 3:
> -                       size = 4;
> +               case 4:
>                         data32 = get_unaligned_le32(&report[i]);
>                         break;
>                 }
>
>

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] Input: gtco - fix potential out-of-bound access
  2017-10-25 10:33     ` Andrey Konovalov
@ 2017-10-25 16:10       ` Dmitry Torokhov
  0 siblings, 0 replies; 5+ messages in thread
From: Dmitry Torokhov @ 2017-10-25 16:10 UTC (permalink / raw)
  To: Andrey Konovalov; +Cc: linux-input, LKML

On Wed, Oct 25, 2017 at 12:33:38PM +0200, Andrey Konovalov wrote:
> On Tue, Oct 24, 2017 at 6:39 PM, Dmitry Torokhov
> <dmitry.torokhov@gmail.com> wrote:
> > On Tue, Oct 24, 2017 at 01:04:03PM +0200, Andrey Konovalov wrote:
> >> On Tue, Oct 24, 2017 at 7:28 AM, Dmitry Torokhov
> >> <dmitry.torokhov@gmail.com> wrote:
> >> > parse_hid_report_descriptor() has a while (i < length) loop, which
> >> > only guarantees that there's at least 1 byte in the buffer, but the
> >> > loop body can read multiple bytes which causes out-of-bounds access.
> >> >
> >> > Reported-by: Andrey Konovalov <andreyknvl@google.com>
> >> > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> >> > ---
> >> >  drivers/input/tablet/gtco.c | 24 +++++++++++++++++-------
> >> >  1 file changed, 17 insertions(+), 7 deletions(-)
> >> >
> >> > diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
> >> > index b796e891e2ee..0351203b8c24 100644
> >> > --- a/drivers/input/tablet/gtco.c
> >> > +++ b/drivers/input/tablet/gtco.c
> >> > @@ -230,13 +230,24 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
> >> >
> >> >         /* Walk  this report and pull out the info we need */
> >> >         while (i < length) {
> >> > -               prefix = report[i];
> >> > -
> >> > -               /* Skip over prefix */
> >> > -               i++;
> >> > +               prefix = report[i++];
> >> >
> >> >                 /* Determine data size and save the data in the proper variable */
> >> > -               size = PREF_SIZE(prefix);
> >> > +               if (PREF_SIZE(prefix) < 1) {
> >>
> >> AFAIU PREF_SIZE(prefix) == 0 is a perfectly valid item data size.
> >
> > Fair enough. How about the below instead then?
> >
> > --
> > Dmitry
> >
> >
> > Input: gtco - fix potential out-of-bound access
> >
> > From: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> >
> > parse_hid_report_descriptor() has a while (i < length) loop, which
> > only guarantees that there's at least 1 byte in the buffer, but the
> > loop body can read multiple bytes which causes out-of-bounds access.
> >
> > Reported-by: Andrey Konovalov <andreyknvl@google.com>
> > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> > ---
> >  drivers/input/tablet/gtco.c |   17 ++++++++++-------
> >  1 file changed, 10 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/input/tablet/gtco.c b/drivers/input/tablet/gtco.c
> > index b796e891e2ee..7d8e9fb831c4 100644
> > --- a/drivers/input/tablet/gtco.c
> > +++ b/drivers/input/tablet/gtco.c
> > @@ -230,13 +230,17 @@ static void parse_hid_report_descriptor(struct gtco *device, char * report,
> >
> >         /* Walk  this report and pull out the info we need */
> >         while (i < length) {
> > -               prefix = report[i];
> > -
> > -               /* Skip over prefix */
> > -               i++;
> > +               prefix = report[i++];
> >
> >                 /* Determine data size and save the data in the proper variable */
> > -               size = PREF_SIZE(prefix);
> > +               size = (1U << PREF_SIZE(prefix)) >> 1;
> > +               if (size && i + size >= length) {
> 
> I think this should be >, not >=, as when i + size == length, item
> data fits into the remaining space precisely. Also I don't see much
> point in checking that size is not 0, but I don't mind it.

Yes, you are right, I ma not sure why I thought that condition should be
greater or equal, not simply greater. I'll adjust this and drop the size
!= 0 check as it is definitely not needed with the updated condition.

I'll put you down as reviewed-by.

Thanks!

-- 
Dmitry

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2017-10-25 16:10 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2017-10-24  5:28 [PATCH] Input: gtco - fix potential out-of-bound access Dmitry Torokhov
2017-10-24 11:04 ` Andrey Konovalov
2017-10-24 16:39   ` Dmitry Torokhov
2017-10-25 10:33     ` Andrey Konovalov
2017-10-25 16:10       ` Dmitry Torokhov

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).