Tomaž Šolc [Wed, 4 Jul 2012 14:29:13 +0000 (16:29 +0200)]
Allow setting samplerate when reading binary files
I had a binary file that I needed to decode using UART decoder. UART
decoder needs to know the sample rate for the data, but currently it's
not possible to pass parameters to input formats and so the "binary"
file format always sets the samplerate to 0.
This patch adds the possibility to append a colon-separated list of
key=value options to the -I argument, in the same way -d supports it.
Also, it makes the "binary" format support the "samplerate" option.
I included the GHashTable containing input format options directly in
the sr_input struct. I'm not sure if that's the right way to do it. I
saw that -d uses a much more elaborate system with device capabilities
and typed options, but that seemed like an overkill for input formats.
sr_session_halt() in its current state is kind of useless and even dangerous.
All it will do is mark the session as not running, but wont signal the devices
to stop to capture data. This is not so much of a problem with the blocking
sr_session_run(), but once there is support for asynchronous data acquisition by
attaching the session sources to the applications mainloop sr_session_halt()
basically becomes a no-op. sr_session_stop() already does what needs to be done,
marking the session as not running and signal the devices to stop acquisition,
so make sr_session_halt() an alias for sr_session_stop() and deprecate its
usage.
sr: session: Do not cleanup the driver state in sr_session_stop()
By cleaning up the driver state all devices will become inaccessible, which
means that is neither possible to query any information from it (like sample
rate) and it is also not possible to restart data acquisition.
sr_session_save() tries to query the sample rate from the device, as a result
calling sr_session_save() after calling sr_session_stop() - which is for example
done by sigrok-cli - will cause a segfault. This patch resolves the issue.
Both pipe channels are currently configured as blocking. We read from the pipe
in receive_data. Since the channel is configured as blocking we'll block in
receive_data until all data has been received. receive_data will be called from
the mainloop, so as consequence the mainloop will be blocked until the demo
device has finished sampling. This is not so much of a problem if we are
sampling in blocking mode (using sr_session_run()) and the demo device is the
only device in the session, but it will fail badly for all other configurations
(e.g. multiple devices or async sampling).
Free the USB transfers in the reverse order of which they were submitted. This
will avoid that while transfer 0 is cancelledd transfer 1 is started by the
host controller, and so on.
Reported-by: Peter Stuge <redacted> Signed-off-by: Lars-Peter Clausen <redacted>
The session and demo device code contain a hack to make the demo device work on
Windows. This was neccessary since polling on windows requires special handling
and we can not just pass in the raw fd to poll.
With the previous patches which added support for non-fd based event sources
this hack is no longer required. The patch moves the GIOChannels used by the
demo device to the demo device context and uses sr_session_source_add_channel
to register a source for the channels instead of using the raw pipe fds.
Currently we keep a file descriptor for each source and construct a pollfd array
from these during each loop iteration in sr_session_run(). This patch modifies
the code to keep a global pollfd array which is only modified when a source is
added or removed. On one hand this gets rid of the constant constructing and
subsequent freeing of the pollfd array in sr_session_run(), on the other hand it
will allow us to implement support for non-fd based pollfds.
Use realloc to resize the source array when adding or removing a source. This
makes the code a bit smaller. In the remove function we now check whether the fd
is valid before doing anything else and if it is not simply do nothing. If it is
valid use memove to move the elements following the source one element down in
the array. Only after that has been done the array is re-allocated.
All frontends will have to include <libsigrok/libsigrok.h> from now on.
This header includes proto.h and version.h, both installed from the
distribution into $INCLUDE/libsigrok/ as well.
The only dynamically changed header is now version.h, which has both
libsigrok and libtool compile-time versions in it.
sr: fx2lafw: Do not start a new acquisition if the old is still running
This may happen if the acquisition_start is called right after calling
acquisition_stop and not all transfers have been freed or on repeated calls to
acquisition_start. If it happens we'll enter an undefined state and all kind of
strange behavior may occur, so error out in such a case.
sr: fx2lafw: Abort pending transfers when sampling is stopped
The recent reworks of the fx2lafw made sure that the total buffer size is large
enough hold 500ms of data. This was done to improve performance and stability.
That the timeout value for a transfer was also increased to over 500ms, a side
effect of this is that when sampling is stopped there will be a additional delay
of 500ms. This is because the driver waits for all transfers to be freed
before it sends a SR_DF_END packet. Once sampling has stopped this will only
happen once a transfer times out. This patch cancels all pending transfers when
sampling is stopped, this will cause them to be freed almost immediately and the
additional delay will disappear.
Also make sure, that if we know, that we just have received the last transfer to
not resubmit this transfer again.
sr: fx2lafw: Calculate buffer size and timeout based on sample rate
Currently timeout and buffer size are hard-coded in the fx2lafw driver which is
non-optimal if we want to get good results at both high and low sample rates.
The timeout is hard-coded to 40ms, which doesn't work well when sampling at
a low sample rate. E.g. at 20kHz filling all available buffer space alone takes
6 seconds. So naturally we'll see a lot of transfers timeout in this case.
The buffer size is hard-coded to 4096 bytes, which does not work well with high
sample rates. E.g. at 24MHz these 4096 bytes are enough space for 0.17ms of
data. The total buffer size is enough for about 5ms of data. Sooner or later the
application won't be able to resubmit a transfer within this time span and the
device will abort data acquisition. Usually this happens within the first few
seconds of sampling.
This patch adds a few new helper functions which calculate the buffer size and
timeout based on the current sample rate.
The buffer size is chosen to be large enough to hold about 10ms of data and it
also must be a multiple of 512 bytes since the firmware will send us the data
in 512 byte chunks.
The timeout is set to the time it would take to fill the whole available buffer
space plus a 25% headroom to accommodate for jitter. This is more than enough,
but there is no need to make the timeout a tight deadline, since it is only
meant as a last resort in case the device stops submitting data. And in this
case data acquisition will be aborted anyway.
The patch also limits the the number of transfers so that the total buffer
space is not much more of 500ms. This will ensure that we do not have to
wait too long when aborting data acquisition.
This patch also significantly reduces the number of context switches when
sampling at a higher sample rate. On my system for example the CPU load of
sigrok-cli when sampling at 24MHz goes down from ~25% to 3-4%.
While errors are usually already implicitly caught by looking at the packet
length field there is one error status which is worth special handling. If the
device has been removed there is not really a chance to recover from this error
so data acquisition can be stopped immediately.
In receive_transfer for each completed transfer a new buffer is allocated and
the old one is freed. We can avoid this by simply reusing the buffer for the
next transfer. This is possible if we only resubmit the transfer after all
processing on the data buffer has been done. A new buffer is only allocated if
the size of the old one is not 4096 bytes.
When freeing a transfer we also have to free the transfer buffer. We also have
to keep track of the number of allocated transfers and if the freed transfer was
the last one stop acquisition. This patch introduces a helper function which
takes care of all of this.
sr: fx2lafw: Allocate header and packet struct on the stack
The header and packet struct are only used in the scope of this function and
they are freed at the end of it. Also these structs are rather small, so they
can safely be allocated on the stack. By doing so memory leaks on the error
paths are avoided.
Bert Vermeulen [Tue, 26 Jun 2012 21:26:55 +0000 (23:26 +0200)]
hantek-dso: start sending the frame only from the trigger point
Samples received before the trigger point are stored. From the
trigger point on, every chunk received from the device is sent
up the session bus. After the device has finished sending, the
stored samples are transmitted.
sr: fx2lafw: Set correct unitsize for trigger buffer
Commit 88b75eb719 ("fx2lafw: Added device caps and added support for wide
sampling") increased the size of the trigger buffer from 8 to 16 bit, but forgot
to adjust the unitsize logic packet which is used to send the contents of the
trigger buffer. This patch sets the unitsize to sizeof() of the trigger buffer.
Multistage triggers currently do no work, because there is a return statement
in the middle of the trigger detector which will be hit as soon as the first
stage in a multistage trigger matches. This patch removes the return statement
so that the trigger detector can continue to try to match the next stage. In
order for this to work we also make sure that the trigger stage is only reset
if the current sample does not match.
There are a few memory leaks in the receive_transfer transfer function. The most
serve of them is that a sample buffer is not freed if the triggered has not
matched yet, which causes a sigrok process which is waiting for a trigger to
consume several megabytes of memory within seconds. The other leaks are on the
error paths in that function.
Uwe Hermann [Thu, 21 Jun 2012 20:36:13 +0000 (22:36 +0200)]
sr: chronovu-la8: Split code into api.c and driver.c.
We should generally use api.c for API related functions and put the other
functions (mostly hardware-specific low-level code) into other C file(s)
for better readability.
Bert Vermeulen [Mon, 18 Jun 2012 21:09:37 +0000 (23:09 +0200)]
hantek-dso: capturestate packet also contains the trigger point
Not yet used, but it's the key to knowing where in the frame to
start displaying; the frame is used as a circular buffer, and what
is sent is effectively a snapshot.
Uwe Hermann [Wed, 13 Jun 2012 16:58:24 +0000 (18:58 +0200)]
sr: hantek-dso: Use portable g_ntohs() function.
The ntohs() from <arpa/inet.h> is not available on MinGW/Windows. There
are ways to work around this, but as we use glib already, using g_ntohs()
is the best option anyway.
projects / libsigrok.git / log