media: uvcvideo: Silence memcpy() run-time false positive warnings
Subsystems (2)
PATCH 1/1 media: uvcvideo: Silence memcpy() run-time false positive warnings
The memcpy() in uvc_video_decode_meta() intentionally copies across the length and flags members and into the trailing buf flexible array. Split the copy so that the compiler can better reason about (the lack of) buffer overflows here. Avoid the run-time false positive warning:
memcpy: detected field-spanning write (size 12) of single field "&meta->length" at drivers/media/usb/uvc/uvc_video.c:1355 (size 1)
Additionally fix a typo in the documentation for struct uvc_meta_buf.
--- drivers/media/usb/uvc/uvc_video.c | 4 +++- include/uapi/linux/uvcvideo.h | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-)
drivers/media/usb/uvc/uvc_video.c +3 -1
@@ -1352,7 +1352,9 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,if (has_scr)memcpy (stream->clock.last_scr, scr,6 ); -memcpy (&meta->length, mem, length); + meta->length = mem[0 ]; + meta->flags = mem[1 ]; +memcpy (meta->buf, &mem[2 ], length -2 ); meta_buf->bytesused += length +sizeof (meta->ns) +sizeof (meta->sof); uvc_dbg(stream->dev, FRAME,
include/uapi/linux/uvcvideo.h +1 -1
@@ -86,7 +86,7 @@ struct uvc_xu_control_query {* struct. The first two fields are added by the driver, they can be used for * clock synchronisation. The rest is an exact copy of a UVC payload header. * Only complete objects with complete buffers are included. Therefore it's -* always sizeof(meta->ts) + sizeof(meta->sof) + meta->length bytes large. +* always sizeof(meta->ns) + sizeof(meta->sof) + meta->length bytes large. */ struct uvc_meta_buf {__u64 ns;
Signature · 1 line
2.34.1
Hi Kees
Thanks for the patch
Would it make more sense to replace *mem with a structure/union. Something like:
https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/
Regards!
Quoted 52 lines · Kees Cook
The
memcpy()inuvc_video_decode_meta()intentionally copies across the length and flags members and into the trailing buf flexible array. Split the copy so that the compiler can better reason about (the lack of) buffer overflows here. Avoid the run-time false positive warning:memcpy: detected field-spanning write (size 12) of single field "
&meta->length" atdrivers/media/usb/uvc/uvc_video.c:1355(size 1)Additionally fix a typo in the documentation for struct
uvc_meta_buf.Reported-by [email protected]Cc Laurent Pinchart <[email protected]>Cc Mauro Carvalho Chehab <[email protected]>Signed-off-by Kees Cook <[email protected]>--- drivers/media/usb/uvc/uvc_video.c | 4 +++- include/uapi/linux/uvcvideo.h | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-)drivers/media/usb/uvc/uvc_video.c +3
-1@@ -1352,7 +1352,9 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,if (has_scr)memcpy (stream->clock.last_scr, scr,6 ); -memcpy (&meta->length, mem, length); + meta->length = mem[0 ]; + meta->flags = mem[1 ]; +memcpy (meta->buf, &mem[2 ], length -2 ); meta_buf->bytesused += length +sizeof (meta->ns) +sizeof (meta->sof); uvc_dbg(stream->dev, FRAME,include/uapi/linux/uvcvideo.h +1
-2@@ -86,7 +86,7 @@ struct uvc_xu_control_query {* struct. The first two fields are added by the driver, they can be used for * clock synchronisation. The rest is an exact copy of a UVC payload header. * Only complete objects with complete buffers are included. Therefore it's -* always sizeof(meta->ts) + sizeof(meta->sof) + meta->length bytes large. +* always sizeof(meta->ns) + sizeof(meta->sof) + meta->length bytes large. */ struct uvc_meta_buf {__u64 ns; --2.34.1
Signature · 1 line
Ricardo Ribalda
Quoted 52 lines · Ricardo Ribalda
Quoted 49 lines · Kees Cook
The
memcpy()inuvc_video_decode_meta()intentionally copies across the length and flags members and into the trailing buf flexible array. Split the copy so that the compiler can better reason about (the lack of) buffer overflows here. Avoid the run-time false positive warning:memcpy: detected field-spanning write (size 12) of single field "
&meta->length" atdrivers/media/usb/uvc/uvc_video.c:1355(size 1)Additionally fix a typo in the documentation for struct
uvc_meta_buf.Reported-by [email protected]Cc Laurent Pinchart <[email protected]>Cc Mauro Carvalho Chehab <[email protected]>Signed-off-by Kees Cook <[email protected]>--- drivers/media/usb/uvc/uvc_video.c | 4 +++- include/uapi/linux/uvcvideo.h | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-)drivers/media/usb/uvc/uvc_video.c +3
-1@@ -1352,7 +1352,9 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,if (has_scr)memcpy (stream->clock.last_scr, scr,6 ); -memcpy (&meta->length, mem, length); + meta->length = mem[0 ]; + meta->flags = mem[1 ]; +memcpy (meta->buf, &mem[2 ], length -2 ); meta_buf->bytesused += length +sizeof (meta->ns) +sizeof (meta->sof); uvc_dbg(stream->dev, FRAME,include/uapi/linux/uvcvideo.h +1
-1@@ -86,7 +86,7 @@ struct uvc_xu_control_query {* struct. The first two fields are added by the driver, they can be used for * clock synchronisation. The rest is an exact copy of a UVC payload header. * Only complete objects with complete buffers are included. Therefore it's -* always sizeof(meta->ts) + sizeof(meta->sof) + meta->length bytes large. +* always sizeof(meta->ns) + sizeof(meta->sof) + meta->length bytes large. */ struct uvc_meta_buf {__u64 ns;[...]
Quoted 2 lines
Would it make more sense to replace *mem with a structure/union. Something like:
https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/
I wasn't sure -- it seemed like this routine was doing the serializing into a struct already and an additional struct overlay wasn't going to improve readability. But I can certainly do that if it's preferred!
-Kees
Signature · 1 line
Kees Cook
Hello,
Quoted 53 lines · Kees Cook
Quoted 52 lines · Ricardo Ribalda
Quoted 49 lines · Kees Cook
The
memcpy()inuvc_video_decode_meta()intentionally copies across the length and flags members and into the trailing buf flexible array. Split the copy so that the compiler can better reason about (the lack of) buffer overflows here. Avoid the run-time false positive warning:memcpy: detected field-spanning write (size 12) of single field "
&meta->length" atdrivers/media/usb/uvc/uvc_video.c:1355(size 1)Additionally fix a typo in the documentation for struct
uvc_meta_buf.Reported-by [email protected]Cc Laurent Pinchart <[email protected]>Cc Mauro Carvalho Chehab <[email protected]>Signed-off-by Kees Cook <[email protected]>--- drivers/media/usb/uvc/uvc_video.c | 4 +++- include/uapi/linux/uvcvideo.h | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-)drivers/media/usb/uvc/uvc_video.c +3
-1@@ -1352,7 +1352,9 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,if (has_scr)memcpy (stream->clock.last_scr, scr,6 ); -memcpy (&meta->length, mem, length); + meta->length = mem[0 ]; + meta->flags = mem[1 ]; +memcpy (meta->buf, &mem[2 ], length -2 ); meta_buf->bytesused += length +sizeof (meta->ns) +sizeof (meta->sof); uvc_dbg(stream->dev, FRAME,include/uapi/linux/uvcvideo.h +1
-1@@ -86,7 +86,7 @@ struct uvc_xu_control_query {* struct. The first two fields are added by the driver, they can be used for * clock synchronisation. The rest is an exact copy of a UVC payload header. * Only complete objects with complete buffers are included. Therefore it's -* always sizeof(meta->ts) + sizeof(meta->sof) + meta->length bytes large. +* always sizeof(meta->ns) + sizeof(meta->sof) + meta->length bytes large. */ struct uvc_meta_buf {__u64 ns;[...]
Quoted 6 lines
Quoted 2 lines
Would it make more sense to replace *mem with a structure/union. Something like:
https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/I wasn't sure -- it seemed like this routine was doing the serializing into a struct already and an additional struct overlay wasn't going to improve readability. But I can certainly do that if it's preferred!
I'm not sure to see how using an additional struct or union would help. We can't use struct assignment as the data may be unaligned, so memcpy() is required. The issue isn't with the source operand of the memcpy() but with the destination operand. Ricardo, if I'm missing something, please submit an alternative patch to explain what you meant.
Signature · 3 lines
Regards, Laurent Pinchart
Hi Laurent
I was thinking about something on the line of the attached patch,
uvc_frame_header->data could also be replaced with a union.
Warning, not tested ;)
Quoted 76 lines · Laurent Pinchart
Hello,
Quoted 53 lines · Kees Cook
Quoted 52 lines · Ricardo Ribalda
Quoted 49 lines · Kees Cook
The
memcpy()inuvc_video_decode_meta()intentionally copies across the length and flags members and into the trailing buf flexible array. Split the copy so that the compiler can better reason about (the lack of) buffer overflows here. Avoid the run-time false positive warning:memcpy: detected field-spanning write (size 12) of single field "
&meta->length" atdrivers/media/usb/uvc/uvc_video.c:1355(size 1)Additionally fix a typo in the documentation for struct
uvc_meta_buf.Reported-by [email protected]Cc Laurent Pinchart <[email protected]>Cc Mauro Carvalho Chehab <[email protected]>Signed-off-by Kees Cook <[email protected]>--- drivers/media/usb/uvc/uvc_video.c | 4 +++- include/uapi/linux/uvcvideo.h | 2 +- 2 files changed, 4 insertions(+), 2 deletions(-)drivers/media/usb/uvc/uvc_video.c +3
-1@@ -1352,7 +1352,9 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,if (has_scr)memcpy (stream->clock.last_scr, scr,6 ); -memcpy (&meta->length, mem, length); + meta->length = mem[0 ]; + meta->flags = mem[1 ]; +memcpy (meta->buf, &mem[2 ], length -2 ); meta_buf->bytesused += length +sizeof (meta->ns) +sizeof (meta->sof); uvc_dbg(stream->dev, FRAME,include/uapi/linux/uvcvideo.h +1
-1@@ -86,7 +86,7 @@ struct uvc_xu_control_query {* struct. The first two fields are added by the driver, they can be used for * clock synchronisation. The rest is an exact copy of a UVC payload header. * Only complete objects with complete buffers are included. Therefore it's -* always sizeof(meta->ts) + sizeof(meta->sof) + meta->length bytes large. +* always sizeof(meta->ns) + sizeof(meta->sof) + meta->length bytes large. */ struct uvc_meta_buf {__u64 ns;[...]
Quoted 6 lines
Quoted 2 lines
Would it make more sense to replace *mem with a structure/union. Something like:
https://patchwork.linuxtv.org/project/linux-media/patch/[email protected]/I wasn't sure -- it seemed like this routine was doing the serializing into a struct already and an additional struct overlay wasn't going to improve readability. But I can certainly do that if it's preferred!
I'm not sure to see how using an additional struct or union would help. We can't use struct assignment as the data may be unaligned, so
memcpy()is required. The issue isn't with the source operand of thememcpy()but with the destination operand. Ricardo, if I'm missing something, please submit an alternative patch to explain what you meant.Reviewed-by Laurent Pinchart <[email protected]>--
Regards,Laurent Pinchart
Signature · 1 line
Ricardo Ribalda
From dce72fe7f003ba40ba2d534e5f84bafc73b37314 Mon Sep 17 00:00:00 2001
From: Ricardo Ribalda <[email protected]>
Date: Mon, 9 Jan 2023 11:42:21 +0100
Subject: [PATCH] media: uvcvideo: Refactor uvc_video_decode_meta
NOT TESTED!
--- drivers/media/usb/uvc/uvc_video.c | 58 ++++++++++++++++++++----------- 1 file changed, 37 insertions(+), 21 deletions(-)
drivers/media/usb/uvc/uvc_video.c +37 -21
@@ -1330,6 +1330,17 @@ static int uvc_video_encode_data(struct uvc_streaming *stream,* Metadata */ +struct uvc_frame_header { +u8 length; +u8 flags; +u8 data[]; +} __packed; + +struct uvc_scr { +u32 scr; +u16 frame_sof; +} __packed; +/* * Additionally to the payload headers we also want to provide the user with USB * Frame Numbers and system time values. The resulting buffer is thus composed
@@ -1343,7 +1354,8 @@ static int uvc_video_encode_data(struct uvc_streaming *stream,*/ static void uvc_video_decode_meta (struct uvc_streaming *stream,struct uvc_buffer *meta_buf, -const u8 *mem,unsigned int length) +const struct uvc_frame_header *header, +unsigned int length) {struct uvc_meta_buf *meta;size_t len_std =2 ;
@@ -1351,7 +1363,8 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,unsigned long flags;unsigned int sof;ktime_t time; -const u8 *scr; +const struct uvc_scr *scr; +const u32 *pts;if (!meta_buf || length ==2 )return ;
@@ -1362,28 +1375,31 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,return ; } - has_pts = mem[1 ] &UVC_STREAM_PTS ; - has_scr = mem[1 ] &UVC_STREAM_SCR ; + has_pts = header->flags &UVC_STREAM_PTS ; + has_scr = header->flags &UVC_STREAM_SCR ;if (has_pts) { - len_std +=4 ; - scr = mem +6 ; - }else { - scr = mem +2 ; + pts = (u32 *) header->data; + len_std +=sizeof (*pts); } -if (has_scr) - len_std +=6 ; +if (has_scr) { +u8 offset; + + offset = has_pts ?sizeof (*pts) :0 ; + scr = (struct uvc_scr *) header->data + offset; + len_std +=sizeof (struct uvc_scr ); + }if (stream->meta.format ==V4L2_META_FMT_UVC ) length = len_std;if (length == len_std && (!has_scr || - !memcmp (scr, stream->clock.last_scr,6 ))) + !memcmp (scr, stream->clock.last_scr,sizeof (struct uvc_scr ))))return ; meta = (struct uvc_meta_buf *)((u8 *)meta_buf->mem + meta_buf->bytesused); -local_irq_save (flags); +local_irq_save (flags);//do we need this? time =uvc_video_get_time (); sof =usb_get_current_frame_number (stream->dev->udev);local_irq_restore (flags);
@@ -1391,20 +1407,20 @@ static void uvc_video_decode_meta(struct uvc_streaming *stream,put_unaligned (sof, &meta->sof);if (has_scr) -memcpy (stream->clock.last_scr, scr,6 ); +memcpy (stream->clock.last_scr, scr,sizeof (struct uvc_scr )); - meta->length = mem[0 ]; - meta->flags = mem[1 ]; -memcpy (meta->buf, &mem[2 ], length -2 ); + meta->length = header->length; + meta->flags = header->flags; +memcpy (meta->buf, header->data, length - offsetof(struct uvc_frame_header , data)); meta_buf->bytesused += length +sizeof (meta->ns) +sizeof (meta->sof);uvc_dbg (stream->dev,FRAME ,"%s(): t-sys %lluns, SOF %u, len %u, flags 0x%x, PTS %u, STC %u frame SOF %u\n" , __func__,ktime_to_ns (time), meta->sof, meta->length, meta->flags, - has_pts ? *(u32 *)meta->buf :0 , - has_scr ? *(u32 *)scr :0 , - has_scr ? *(u32 *)(scr +4 ) &0x7ff :0 ); + has_pts ? *pts :0 , + has_scr ? scr->scr :0 , + has_scr ? scr->frame_sof &0x7ff :0 ); }/* ------------------------------------------------------------------------
@@ -1479,7 +1495,7 @@ static void uvc_video_decode_isoc(struct uvc_urb *uvc_urb,if (ret <0 )continue ; -uvc_video_decode_meta (stream, meta_buf, mem, ret); +uvc_video_decode_meta (stream, meta_buf, (struct uvc_frame_header *) mem, ret);/* Decode the payload data. */ uvc_video_decode_data(uvc_urb, buf, mem + ret,
@@ -1531,7 +1547,7 @@ static void uvc_video_decode_bulk(struct uvc_urb *uvc_urb,memcpy (stream->bulk.header, mem, ret); stream->bulk.header_size = ret; -uvc_video_decode_meta (stream, meta_buf, mem, ret); +uvc_video_decode_meta (stream, meta_buf, (struct uvc_frame_header *)mem, ret); mem += ret; len -= ret;
Signature · 1 line
2.39.0.314.g84b9a713c41-goog
Quoted 7 lines · Ricardo Ribalda
Hi Laurent
I was thinking about something on the line of the attached patch,
uvc_frame_header->datacould also be replaced with a union.Warning, not tested ;)
...
drivers/media/usb/uvc/uvc_video.c static int uvc_video_encode_data(struct uvc_streaming *stream, · line 1337
Quoted 5 lines
drivers/media/usb/uvc/uvc_video.c +5
-0+struct uvc_frame_header { +u8 length; +u8 flags; +u8 data[]; +} __packed;
__packed! (See below)
...
drivers/media/usb/uvc/uvc_video.c static void uvc_video_decode_meta(struct uvc_streaming *stream, · line 1382
Quoted 1 line
drivers/media/usb/uvc/uvc_video.c +1
-0+ pts = (u32 *) header->data;
Ai-ai-ai.
Here is just a yellow flag...
...
drivers/media/usb/uvc/uvc_video.c static void uvc_video_decode_meta(struct uvc_streaming *stream, · line 1421
Quoted 5 lines
drivers/media/usb/uvc/uvc_video.c +1
-0uvc_dbg(stream->dev, FRAME, "%s(): t-sys %lluns, SOF %u, len %u, flags 0x%x, PTS %u, STC %u frame SOF %u\n", __func__, ktime_to_ns(time), meta->sof, meta->length, meta->flags, + has_pts ? *pts : 0,
...and here is a red flag. What you need to have is
void *pts; u32 pts_val; pts_val = get_unaligned_be32(); // or le32 ...use pts_val...
Quoted 2 lines
drivers/media/usb/uvc/uvc_video.c +2
-0+ has_scr ? scr->scr :0 , + has_scr ? scr->frame_sof & 0x7ff : 0);
Signature · 2 lines
With Best Regards, Andy Shevchenko
Hi Andy
Quoted 29 lines · Andy Shevchenko
Quoted 7 lines · Ricardo Ribalda
Hi Laurent
I was thinking about something on the line of the attached patch,
uvc_frame_header->datacould also be replaced with a union.Warning, not tested ;)
...
drivers/media/usb/uvc/uvc_video.c static int uvc_video_encode_data(struct uvc_streaming *stream, · line 1337
Quoted 5 lines
drivers/media/usb/uvc/uvc_video.c +5
-0+struct uvc_frame_header { +u8 length; +u8 flags; +u8 data[]; +} __packed;
__packed! (See below)...
Quoted 1 line
+ pts = (u32 *)
header->data;Ai-ai-ai.
Here is just a yellow flag......
Quoted 7 lines
Quoted 5 lines
uvc_dbg(stream->dev, FRAME, "%s(): t-sys %lluns, SOF %u, len %u, flags 0x%x, PTS %u, STC %u frame SOF %u\n", __func__, ktime_to_ns(time), meta->sof, meta->length, meta->flags,+
has_pts? *pts : 0,...and here is a red flag. What you need to have is
Thanks! you are absolutely right :)
Luckily Laurent merged the original patch not mine.
Regards!
Quoted 17 lines
void *pts; u32 pts_val; pts_val = get_unaligned_be32(); // or le32 ...use pts_val...drivers/media/usb/uvc/uvc_video.c static void uvc_video_decode_meta(struct uvc_streaming *stream, · line 1423
Quoted 2 lines
drivers/media/usb/uvc/uvc_video.c +2
-0+ has_scr ? scr->scr :0 , + has_scr ? scr->frame_sof & 0x7ff : 0);--
With Best Regards,
Andy Shevchenko
Signature · 1 line
Ricardo Ribalda