On Thu, 5 Mar 2009, Hans de Goede wrote:
Kyle Guinn wrote:
On Wednesday 04 March 2009 22:34:13 kilg...@banach.math.auburn.edu wrote:
On Wed, 4 Mar 2009, Kyle Guinn wrote:
On Tuesday 03 March 2009 18:12:33 kilg...@banach.math.auburn.edu wrote:
contents of file mr97310a.patch follow, for gspca/mr97310a.c
--------------------------------------------------------
--- mr97310a.c.old 2009-02-23 23:59:07.000000000 -0600
+++ mr97310a.c.new 2009-03-03 17:19:06.000000000 -0600
@@ -302,21 +302,9 @@ static void sd_pkt_scan(struct gspca_dev
data, n);
sd->header_read = 0;
gspca_frame_add(gspca_dev, FIRST_PACKET, frame, NULL, 0);
- len -= sof - data;
- data = sof;
- }
- if (sd->header_read < 7) {
- int needed;
-
- /* skip the rest of the header */
- needed = 7 - sd->header_read;
- if (len <= needed) {
- sd->header_read += len;
- return;
- }
- data += needed;
- len -= needed;
- sd->header_read = 7;
+ /* keep the header, including sof marker, for coming frame */
+ len -= n;
+ data = sof - sizeof pac_sof_marker;;
}
gspca_frame_add(gspca_dev, INTER_PACKET, frame, data, len);
A few notes:
1. There is an extra semicolon on that last added line.
Oops. My bifocals.
2. sd->header_read no longer seems necessary.
This is very likely true.
3. If the SOF marker is split over two transfers then everything falls
apart.
Are you sure about that?
Simple example: One transfer ends with FF FF 00 and the next begins with
FF 96 64. pac_find_sof() returns a pointer to 64, n is set to 0, len stays
the same, data now points at 3 bytes _before_ the transfer buffer, and we
will most likely get undefined behavior when trying to copy the data out of
the transfer buffer. Not only that, but the FF FF 00 portion of the SOF
won't get copied to the frame buffer.
Good point, since we will always pass frames to userspace which start with
the
sof, maybe we should just only pass the variable part of the header to
userspace?
That sure feels like the easiest solution to me.
Regards,
Hans
Hans, that would not solve the problem. In fact, it appears to me that
this problem was already inherent in the driver code before I proposed any
patches at all. The problem is that one must _detect_ the SOF marker. The
further problem is (and was, and nothing yet has changed that) that if the
SOF marker is split across two packets it will fail to be detected.
Two possible courses of action:
1. Leave the matter alone. It could be maintained that one of the
following is the case:
(a) this never happens
(b) even if it does happen (rarely), it will do no great harm,
because the only effect will be that the output skips a frame, having not
detected its start of frame marker. For, the frames consist of compressed
data, and if there is too much compressed data for a given frame, then
that extra compressed data will simply be tossed. The next frame which is
actually seen and used will be the next frame for which an SOF marker is
actually detected.
So, the argument for number 1 would be that, yes, this is sort of a bug,
but it is insignificant, not serious, and could never cause a problem in
practice even if it occurs, because the only result would be for a frame
to get dropped. Another argument in favor of it would be that in any event
the camera is sending isochronous packets, and there is in any event no
guarantee of the validity and accuracy of any one single packet. Instead,
the reliance is on the high rate at which the packets get sent, received,
and processed.
2. If it is deemed that, yes, it can happen that an SOF marker gets split
between two packets, and, no, if that happens it is a problem which should
not be ignored, then there is an alternative course of action:
Keep a cache consisting of the last 4 bytes of each packet, and
see if, when the next packet comes down, a SOF marker is detected in those
bytes, plus the new bytes from the beginning of the new packet. Then a SOF
marker will not be missed, even if it occurs in the situation described
above.
Alternative number two is in fact not very much more difficult, but it
would involve putting an if-then-else or two into the code and also would
require one to have a place to put those last four bytes of each packet,
keep them around until the next packet is obtained, and then parse the
result. This might slow things down, but probably not by very much. So
which way to do this would obviously depend upon some evaluation of the
danger of doing nothing.
However, the discussion of whether or not to deal with an SOF marker which
is split across two packets has nothing at all to do with the question of
whether to preserve the SOF marker in the raw frame output. The two
questions are independent.
I say that, whatever else is done, it is good to preserve the SOF marker
in the raw frame output. It is nice to see it there because then one
incontrovertibly knows that the frame is aligned with the SOF marker at
its start. I mean, if there is something wrong and one needs to do
debugging, or if there is a new compression algorithm which needs to be
studied, then if the SOF marker is absent in the raw output how in hell
does one know the output is correctly aligned? Simple answer. One does
not know. There really was a way to know that, but it has been thrown
away. It is also much easier to code in the module, because then
there is no need to code around the need to keep everything that came
down, unless it is an SOF marker and then one throws it away and is forced
to re-align the remaining data.
I wonder if anyone with wider experience could come up with an intelligent
evaluation of whether to go with item 1, or item 2? I think I have listed
the arguments for and against each of them, but to know which one of these
is actually best, is another thing.
Finally, again, the question of what happens if the SOF marker is split
between two packets in fact has nothing to do with the changes I was
proposing. It is independent. Except for one thing: Could it be that the
simplifications I was proposing have eliminated a lot of convoluted code
and thereby brought this problem to the surface?
Theodore Kilgore
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html