Some drivers check in some of their driver callbacks if the driver has been
initialized and return an error if it has not.
For the scan() callback the sigrok core checks if the driver has been
initialized and if not returns an error. So it is not possible that the
scan() callback gets called if the driver is not initialized. Without the
scan() callback succeeding it is not possible to get a reference to a
device which is associated with the driver, so it is not possible that any
of the device specific callbacks is called without the driver first being
initialized either.
In conclusion these checks are not necessary since they never evaluate to
true and can be dropped. If they should ever become necessary they should
be done in the sigrok core so all drivers and all callbacks are equally
handled.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
SR_CONF_CONTINUOUS is a capability option indicating whether a device
supports continuous capture or not. If the option exists the device
supports continuous capture and otherwise it doesn't. There is no value
associated with it and hence setting the SR_CONF_SET flag is nonsensical.
None of the drivers which set SR_CONF_SET for SR_CONF_CONTINUOUS handle it
in their config_set() callback and return an error if an application tried
to perform a config_set() operation for SR_CONF_CONTINUOUS.
Simply remove the SR_CONF_SET flag from all SR_CONF_CONTINUOUS options.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Every single hardware driver has the very same implementation of the
dev_list() callback. Put this into a helper function in the standard helper
library and use it throughout the drivers. This reduces boiler-plate code
by quite a bit.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
If a driver does not implement a dev_clear() callback the core will
automatically call std_dev_clear(di, NULL). Remove all driver dev_clear()
implementations that are identical to default. This reduces the amount of
boiler-plate code.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
std_init() allocates a drv_context struct which needs to be freed by the
driver in its cleanup struct. But the vast majority of drivers does never
does this causing memory leaks.
Instead of addressing the issue by manually adding code to free the struct
to each driver introduce a new helper function std_cleanup() that takes
care of this. In addition to freeing the drv_context struct std_cleanup()
also invokes sr_dev_clear() which takes care of freeing all devices
attached to the driver.
Combining both operations in the same helper function allows to use
std_cleanup() as the cleanup callback for all existing drivers, which
reduces the amount of boiler-plate code quite a bit.
All drivers are updated to use the new helper function.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Devices for the scpi-pps driver do have additional data attached to it that
needs to be freed when the device is freed. While the driver gets it right
for the cleanup() callback it does not for the dev_clear() callback. This
will cause memory leaks when sr_dev_clear() is called for this driver.
To fix this let the dev_clear() free the additional data.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Devices for the deree-de5000 driver do have additional data attached to it
that needs to be freed when the device is freed. While the driver gets it
right for the cleanup() callback it does not implement a dev_clear()
callback, so the default dev_clear() implementation is used which will not
free the additional data. This will cause memory leaks when sr_dev_clear()
is called for this driver.
To fix this provide a dev_clear() implementation that frees the additional
data.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Devices for the demo driver do have additional data attached to it that
needs to be freed when the device is freed. While the driver gets it right
for the cleanup() callback it does not implement a dev_clear() callback, so
the default dev_clear() implementation is used which will not free the
additional data. This will cause memory leaks when sr_dev_clear() is called
for this driver.
To fix this provide a dev_clear() implementation that frees the additional
data.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
When g_file_get_contents() encounters an error a new GError will be
allocated and passed back to the application. The application is
responsible for freeing this GError.
The baylibre-acme driver currently does not do this and as a result leaks
memory during the scan process when no device is found.
Add the missing g_error_free() invocations to fix the issue.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Make sure to free the FTDI device list and the FTDI context in scan_all()
otherwise memory leaks can be observed. Also make sure to free the FTDI
context in scan_device() on the error path.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
This makes the code shorter, simpler and more consistent, and also
ensures that the (same) debug messages are always emitted and the
packet.payload field is consistently set to NULL always, etc.
The USBee AX hardware needs a sampling clock that is lower than
the 30MHz or 48MHz that the FX2 has to offer. This flag will enable
clocking via the CTL2 pin that is an even divisor of the main clock.
The driver did not look at the mq_flags provided with the
SR_CONF_MEASURED_QUANTITY key, and it defaulted to DC measurements.
Use the second member of the tuple provided by the config key, which
represents the flags for the measurement, and set the instrument's
measurement mode accordingly.
On the high-end bench multimeters, resistance can be measured with a
kelvin connection as well as the more common two wire method. Provide
a flag which can indicate if four-wire mode is used.
Due to a PEBKAC error, after the last sample was sent, a new
measurement was triggered, but its value was never sent down the
session bus. This is easily fixed by incrementing devc->num_samples
right after a measurement is sent instead of when a measurement is
retriggered.
g_variant_print() allocates memory during call. Save the pointer
so that it can be free'd afterwards.
==10048== 16 bytes in 1 blocks are definitely lost in loss record 17 of 37
==10048== at 0x4C2DEAE: realloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==10048== by 0x536C85D: g_realloc (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x53877C6: ??? (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x5388B60: g_string_append_vprintf (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x5388D83: g_string_append_printf (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x539D034: g_variant_print_string (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x539C92C: g_variant_print (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x4E5F713: log_key (hwdriver.c:597)
==10048== by 0x4E5FCD5: sr_config_set (hwdriver.c:752)
==10048== by 0x408C1A: run_session (session.c:661)
==10048== by 0x404FC5: main (main.c:267)
==10048==
==10048== 16 bytes in 1 blocks are definitely lost in loss record 18 of 37
==10048== at 0x4C2DEAE: realloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==10048== by 0x536C85D: g_realloc (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x53877C6: ??? (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x5388B60: g_string_append_vprintf (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x5388D83: g_string_append_printf (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x539D034: g_variant_print_string (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x539C92C: g_variant_print (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4400.1)
==10048== by 0x4E5F713: log_key (hwdriver.c:597)
==10048== by 0x4E5FBDD: sr_config_get (hwdriver.c:709)
==10048== by 0x4080D7: datafeed_in (session.c:196)
==10048== by 0x4E5D47E: sr_session_send (session.c:1192)
==10048== by 0x4E62682: std_session_send_df_header (std.c:101)
Using non power of 2 sizes causes the driver to not work on OS X.
Depending on the usb transfer mode the buffer sizes will map
perfectly to the underlying transport protocol.
scpi_serial generate an POLLIN event after the requested data is returned
by the instrument. For USBTMC it is necessary to
1. send an REQUEST_DEV_DEP_MSG_IN request
2. submit an USB bulk read transfer asynchronously.
Using the synchronous libusb_bulk_read() does not generate an POLLIN event.
Solving this properly needs major surgery in spci_usbtmc_libusb.
While some transports add a terminating (carriagereturn+)linefeed
unconditionally, the USBTMC transport does not. At least the R&S HMO1002
requires the linefeed and locks up otherwise. Fixes bug #784.
This changes the TCP and VXI transport from CR+LF to LF only.
Also fixes a possible memory leak for VXI, where the temporary command
buffer was not freed in case of a write error.
According to USBTMC usb488 subclass spec, wValue hast to be 0 for both
LOCAL_LOCKOUT and GO_TO_LOCAL. At least required for R&S HMO1002, the
bad request results in a STALL. Fixes bug #783.
After acquisition start, DSLogic stores samples in memory, and when done it
sends a USB packet with the trigger position.
This initial fillup can take some time. If the user requests a session stop
in between, the USB transfer is cancelled and the session hangs because it
is not closed properly.
This commit manages this case and closes the session properly when
acquisition is stopped by the user.
Signed-off-by: Diego F. Asanza <f.asanza@gmail.com>
According to the infos I have, VCD files should be plain ASCII, but we
got report of a version adding a UTF8 BOM at the beginning of the file,
so we need to skip it.
This fixes bug #755.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
For some reason, uploading the FPGA binary into DSLogic in small chunks
does not work. In this commit, the whole binary image is loaded into memory
and transfer is done in one chunk.
Furthermore, the FPGA configuration structure was not initialized
properly. This was changed with the initialization values taken from the
original DSLogic software.
Signed-off-by: Diego Asanza <f.asanza@gmail.com>
Certain output modules do not understand double-precision data.
Although we need double precision to represent the full resolution
of 7.5 digit readings, temporarily convert data to single precision
so that the output modules understand it. While the reasing might be
off by a few counts, we ensure the output modules do not display
completely bogus data.
For details, see bug #779.
KNOWN ISSUES:
- When sampling with 100 NPLC, the poll function will timeout a few
times before the first sample is acquired. Increasing the timeout
passed to sr_scpi_source_add() will cause all the other commands to
be processed slowly, producing a sampling rate of about one sample
every ten seconds.
- Support for plug-in cards (44491A and 44492A) is not implemented.
- Support for AC, AC+DC and four-wire resistance measurements is not
implemented.
- Support for configuring the frequency measurement source is not
implemented.
High precision multimeters have a special setting, called "number of
powerline cycles" (NPLC) which determines the integration time of the
ADC in terms of the power line period. Some devices need their NPLC
adjusted from the default value before they can measure at their full
rated precision.
Devices connected on a real GPIB bus are placed in remote mode when
opening them. libgpib does not automatically place devices back in
local mode when closing the handle. It is thus possible to lock out a
GPIB device by probing it with libsigrok.
This happens on the HP 3457A meter, which does not have a "LOCAL"
command, and must be put back in local mode via GPIB handshake.
ibloc() takes care of this, and it does it on a per-device basis,
such that other devices on the GPIB bus are not affected.
libgpib has an error_string which formats a numeric error code into a
human-readable description. Use that instead of printing the numeric
code, as it makes debugging easier.
- Don't set capturefile if no logic channels are saved
- Don't set total probes if no logic channels are saved
- Save analog channels without index gaps (e.g. probe1/probe4)
Note: This commit is based on the initial implementation by
Christer Ekholm (but stashed into one commit), with some adaptations
(forward porting, coding style and consistency fixes) by Uwe Hermann.
We want the size of the struct, not of a pointer to the struct. And to
be absolutely future proof, dereference the pointer we are assigning the
memory to (not the one we are copying the data from). Found by Coverity,
CID 50858.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
The firmware has "on\n" and "off\n" commands since 1.10, so use them.
Apparently you can only set the state (on/off) of the load, but it's
not possible to query the current state.
Also note that the __ namespace is reserved by POSIX for its private
usage, so user land software should never rely on any kind of API
with a __ prefix.
For session sources without a file descriptor to poll a negative number
should be passed for the fd parameter. The hung-chang-dso-2100 driver
currently passes 0 instead, which is the stdin stream. Fix the issue by
passing -1 for the fd parameter.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Accept (and set to default) 0 for numchannels which means 'unlimited'.
I think it is convenient to read all channels of a vcd file by default.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Benefits:
* only channels really used in vcd will be added
* we can give them the proper name found in the vcd file
* less code :)
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Current code moves the identifier string one byte to the front to
overwrite the bit value, so 'tokens[i]' is a string to compare against
the desired value. This copying is unnecessary, just pass a properly
setup pointer.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
No need to bail out on vectors. As long as there are tokens left, we can
try to parse the rest. Also, print a message so the user knows what's
going on. Here is a testcase vcd:
$timescale 1 ns $end
$var wire 1 n0 addr_0 $end
$var wire 1 n1 addr_1 $end
$enddefinitions $end
#0
0n0
b1 n1
#1
1n0
b0 n1
#2
0n0
b1 n1
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
If we hit the missing identifier case, then we reached the end of the
token list. So, we should break out of the loop, and not continue.
Otherwise we will go past the end of the array as this minimal testcase
shows:
$timescale 1 ns $end
$var wire 1 n0 addr_0 $end
$enddefinitions $end
1
gives:
$ ./sigrok-cli -I vcd -i no_mod.vcd -O vcd -o /tmp/o.vcd
Segmentation fault
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
A $var block can have an optional index item which looks like '[<sth>]'.
Parse it, too, and append it to the channel name.
This fixes bug #322. A first version was posted by Simon Richter. This
version is rebased and simplified.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
As we're downsampling, several record time stamps can match the specified
trigger time. For this reason, it's possible that several trigger packets
are sent when a file is loaded. This prevents the issue and sends a
trigger packet only on the first matching record.
src/hardware/lecroy-logicstudio/protocol.c: In function 'handle_fetch_samples_done':
src/hardware/lecroy-logicstudio/protocol.c:261:3: warning: passing argument 6 of 'libusb_fill_bulk_transfer' from incompatible pointer type
recv_bulk_transfer, (void *)sdi, USB_TIMEOUT_MS);
^
In file included from ./src/libsigrok-internal.h:31:0,
from src/hardware/lecroy-logicstudio/protocol.h:26,
from src/hardware/lecroy-logicstudio/protocol.c:23:
/home/uwe/sr_mingw/include/libusb-1.0/libusb.h:1546:20: note: expected 'libusb_transfer_cb_fn' but argument is of type 'void (*)(struct libusb_transfer *)'
static inline void libusb_fill_bulk_transfer(struct libusb_transfer *transfer,
^
src/hardware/lecroy-logicstudio/protocol.c: In function 'fetch_samples_async':
src/hardware/lecroy-logicstudio/protocol.c:314:4: warning: passing argument 6 of 'write_registers_async' from incompatible pointer type
handle_fetch_samples_done);
^
src/hardware/lecroy-logicstudio/protocol.c:200:12: note: expected 'libusb_transfer_cb_fn' but argument is of type 'void (*)(struct libusb_transfer *)'
static int write_registers_async(const struct sr_dev_inst *sdi,
^
src/hardware/lecroy-logicstudio/protocol.c: In function 'lls_start_acquisition':
src/hardware/lecroy-logicstudio/protocol.c:1122:3: warning: passing argument 6 of 'libusb_fill_interrupt_transfer' from incompatible pointer type
recv_intr_transfer, (void *) sdi, USB_TIMEOUT_MS);
^
In file included from ./src/libsigrok-internal.h:31:0,
from src/hardware/lecroy-logicstudio/protocol.h:26,
from src/hardware/lecroy-logicstudio/protocol.c:23:
/home/uwe/sr_mingw/include/libusb-1.0/libusb.h:1602:20: note: expected 'libusb_transfer_cb_fn' but argument is of type 'void (*)(struct libusb_transfer *)'
static inline void libusb_fill_interrupt_transfer(
^
There is no status bit for RMS. We know about RMS if certain modes are
active. So, drop this superfluous variable.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
The primary display is the power factor, the secondary is the frequency.
This got mixed up, so change the order. We also need to fix the
conversion factor.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
is_power_apparent_power is index 0 of function 0x39, so it is better to
process it first and the later indices after that (we need to add
another one with a different patch later).
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Testing showed that AC current needs to be handled different from DC.
Note that ACA is still untested because of limited testing equipment.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
It was confusing to see the display value (5 digits) printed in debug
output as a float. Print it the same way as shown on the real device,
without comma, of course.
This also allows to simplify the code a little.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
This g_close(), the only one in the whole code-base, would unnecessarily
raise the minimum glib version to 2.36.
Thanks to Daniel Glöckner for the report.
This fixes bug #724.
This function replaces the pattern of calling config_list() with
SR_CONF_DEVICE_OPTIONS to obtain a list of device options. Note
that this does not include the SR_CONF_{GET,SET,LIST} bitmask,
which is now retrieved for a specific key by calling
sr_dev_config_capabilties().
Refactor handling the size of modbus_devs, so it doesn't produce a build
warning and still allows the compiler to remove unused code.
This fixes bug #637. It could be reverted once modbus_devs
unconditionally has a member in the struct.
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
In some situations, the reply to the *IDN? command contains an
additional trailing 0x01 byte for unknown reasons.
This issue seems to be reproducible by changing the voltage using the knobs
on the device, then turning on the output and turning it off again.
The next korad-kaxxxxp scan() operation would contain the trailing 0x01
byte, which would lead to the detection of the device in libsigrok no
longer working until the next power-cycle.
Work around this issue by treating both the ID string with and without
the trailing 0x01 byte as valid.
This is a desperate measure to improve the success rate of device
initialization even after it got into a bad state. Combine this
with a reduced USB timeout (1 second) so that if it fails, it fails
quickly. Also ignore USB errors from the initial dummy read of the
device test ID.
Detect whether the FX2 firmware of the LWLA device exhibits the
short transfer bug. If so, work around the problem by limiting
reads to at most 64 bytes at a time. This slows down the memory
read after acquisition quite noticably, but makes the device
usable even in adverse conditions.
Reduce the number of long registers read in bulk during status
polling from 10 to 5. The LWLA1034 driver used to do that already
in an earlier iteration, which was then changed to be more like
the original vendor software. The reason for bringing it back now
is that it reduces the response size to 40 bytes, which works
around the spurious 64 byte limit bug in the FX2 firmware of the
LWLA devices.
The sr_input_dev_inst_get API documentation guarantees an input is fully
initialized as soon as the device instance is returned. An sdi
implementation should not set sdi_ready any earlier.
This fixes parts of bug #387.
The sixth character from ISET? is read and discarded. If the device is
turned off and on again, this won't be there and causes 10 ms delay in
every ISET? Luckily, this value isn't queried that often. To get the
sixth byte, the *IDN? command has to be issued before ISET?.
==18779== 800,000 bytes in 196 blocks are definitely lost in loss record 29 of 29
==18779== at 0x4C29110: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==18779== by 0x4E635C3: receive (analog.c:319)
==18779== by 0x40870B: datafeed_in (session.c:316)
==18779== by 0x4E59D4E: sr_session_send (session.c:1201)
==18779== by 0x4E59F8B: sr_session_send (session.c:1159)
==18779== by 0x4E62595: send_chunk (wav.c:234)
==18779== by 0x4E62A06: process_buffer (wav.c:290)
==18779== by 0x40954A: load_input_file_module (input.c:123)
==18779== by 0x4097AB: load_input_file (input.c:157)
==18779== by 0x40531E: main (main.c:288)
==17549== 32 (16 direct, 16 indirect) bytes in 1 blocks are definitely lost in loss record 22 of 39
==17549== at 0x4C29110: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==17549== by 0x5359200: g_malloc (in /usr/lib64/libglib-2.0.so.0.4200.2)
==17549== by 0x536EE2D: g_slice_alloc (in /usr/lib64/libglib-2.0.so.0.4200.2)
==17549== by 0x5370165: g_slist_append (in /usr/lib64/libglib-2.0.so.0.4200.2)
==17549== by 0x4E595C3: sr_session_datafeed_callback_add (session.c:512)
==17549== by 0x409527: load_input_file_module (input.c:111)
==17549== by 0x4097AB: load_input_file (input.c:157)
==17549== by 0x40531E: main (main.c:288)
==7478== Invalid write of size 8
==7478== at 0x4E59182: sr_session_dev_remove_all (session.c:302)
==7478== by 0x4E591CD: sr_session_destroy (session.c:265)
==7478== by 0x4095D9: load_input_file_module (input.c:143)
==7478== by 0x4097AB: load_input_file (input.c:157)
==7478== by 0x40531E: main (main.c:288)
==7478== Address 0x7877eb8 is 88 bytes inside a block of size 96 free'd
==7478== at 0x4C2A37C: free (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==7478== by 0x4E5F454: sr_input_free (input.c:573)
==7478== by 0x4095C3: load_input_file_module (input.c:140)
==7478== by 0x4097AB: load_input_file (input.c:157)
==7478== by 0x40531E: main (main.c:288)
Do not use size_t for values whose width is defined by the device,
not the host. Also don't use size_t for simple indices with known
small range, unless type compatibility considerations apply.
Refactor the sysclk-lwla driver to separate the generic logic from
the model-specific implementation. Based on this, implement support
for the SysClk LWLA1016 device.
On some systems it can happen that the USB 'bus' number is a lot larger
than 64, but sr_usb_find() currently errors out if it is > 64.
Example:
Bus 250 Device 006: ID 1ab1:04ce 1ab1 DS1000Z Series[...]
Increase that limit so that the code will work everywhere. This bus number
is queried via libusb_get_bus_number() which returns an uint8_t, so we're
limiting to 255 here.
Thanks to 'ssi' on IRC for reporting the issue.
Apparently (some versions of) Mac OS X have the same problem with
usb_get_port_path() as FreeBSD does. Work around this in the same way.
Thanks to hanyazou@gmail.com for the patch!
This fixes bug #673.
The driver should now be able to cope with e.g. multiple ChronoVu LA8
and/or ChronoVu LA16 devices being connected to the same PC.
It now also provides the serial number and connection ID, which can be
used by frontends to differentiate multiple devices.
Also improve the scanopts / drvopts / devopts lists handling.
This fixes bug #504.
With the current driver API and the corresponding session event
handling, it is not possible to destroy and then re-create an
event source with the same key within the same main loop iteration.
The next generation driver API will fix this problem. But for now,
just change the driver to make do without that sort of thing. Also
increase the I/O timeout to 100 ms to be safer in the event of all
kind of delays the OS environment may induce.
This fixes bug #678.
Evaluate all 64 bit of the duration field in the capture status
record. Although unlikely in practical use, due to compression
it is possible for the duration in ms to exceed 32 bit.
zip_discard() was introduced in libzip 0.11, which some systems
do not have yet. Provide a fallback replacement for zip_discard(),
and reduce the requirement to libzip 0.10 again.
This fixes bug #674.
During initialization of the LWLA1034, read the 64-bit test word
twice and verify the result of the second read only. This better
matches what the original vendor software does.
Apparently, these four registers form an interface for indirect
access to another internal 64 bit wide memory. This is likely the
same memory as that accessed by the bulk transfer commands 7 and 8.
Added support for SR_CONF_REGULATION which returns value for CH1
Also VELLEMAN LABPS3005D (only device currently supported) sends single
'M' character in beginning of return value, which is specially discarded.
Trying to configure an invalid capture ratio would reset the
previously configured value. Instead, we should just reject the
new value and keep the original one.
Introduce a new API function sr_session_stopped_callback_set()
which can be used to receive notification when a session stops
running. This allows applications to integrate libsigrok event
processing with their own main loop, instead of blocking in
sr_session_run().
With this driver it is possible to set voltage target and current
limit. Also enabling and disabling the output is possible.
Analog output sends read back values from output. If output is
disabled analog outputs 0.00.
In protocol.c there is a g_usleep() call. This gives almost
every time enough time for PSU to parse and process input.
Multichannel devices aren't supported.
Instead of using a non-standard packed attribute, read the contents of the
probe EEPROM into a buffer and then process it using the R* macros.
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Standard integer types may differ in size on different targets. Use fixed-size
types instead.
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
SR_CONF_NUM_LOGIC_CHANNELS is defined as SR_T_INT32. Create the
GVariant with the correct type to avoid a type mismatch error in
sr_variant_type_check().
The resource API provides a generic means for accessing resources
that are bundled with sigrok, such as device firmware files. Since
the manner of resource bundling is platform-dependent, users of
libsigrok may override the functions used to open, close and read
a resource. The default implementation accesses resources as files
located in one of the XDG data directories or a directory defined
at compile time.
Use in-memory buffers instead of temporary files. This avoids
the need for low-level I/O on the FD returned by g_mkstemp().
Refactor the code accordingly. Also plug a number of leaks and
tighten the error checking.
Completely remove the old session save code that has been
superseded by the srzip output module. Also refactor a bit,
plug a number of leaks and tighten the error checking.
Introduce the sr_file_get_size() utility function to retrieve the
size of an open FILE stream. This is based on fseeko() followed by
ftello(), which are POSIX functions but quite portable in practice.
Since these calls operate on FILE streams instead of filenames, the
issue of filename encoding no longer arises.
Do not use Unix low-level I/O for reading a regular input file.
Read in the file header once and re-use the buffer for all input
modules participating in the scan. Also re-use a prefilled metadata
table instead of creating it anew for each input module tried.
This fixes parts of bug #423.
The list of fixed warnings:
src/output/srzip.c:285:3: warning: Value stored to 'ret' is never read
ret = zip_append(o, logic->data, logic->unitsize, logic->length);
^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/scpi/scpi.c:610:2: warning: Value stored to 'ret' is never read
ret = SR_OK;
^ ~~~~~
src/scpi/scpi.c:667:2: warning: Value stored to 'ret' is never read
ret = SR_OK;
^ ~~~~~
src/dmm/vc870.c:410:2: warning: Value stored to 'info_local' is never read
info_local = (struct vc870_info *)info;
^ ~~~~~~~~~~~~~~~~~~~~~~~~~
src/hardware/conrad-digi-35-cpu/api.c:130:2: warning: Value stored to 'ret' is never read
ret = SR_OK;
^ ~~~~~
src/hardware/fx2lafw/api.c:658:2: warning: Value stored to 'timeout' is never read
timeout = fx2lafw_get_timeout(devc);
^ ~~~~~~~~~~~~~~~~~~~~~~~~~
src/hardware/gmc-mh-1x-2x/protocol.c:941:3: warning: Value stored to 'retc' is never read
retc = SR_ERR_ARG;
^ ~~~~~~~~~~
src/hardware/gmc-mh-1x-2x/api.c:168:2: warning: Value stored to 'model' is never read
model = METRAHIT_NONE;
^ ~~~~~~~~~~~~~
src/hardware/ikalogic-scanalogic2/api.c:325:2: warning: Value stored to 'ret' is never read
ret = SR_OK;
^ ~~~~~
src/hardware/openbench-logic-sniffer/api.c:185:3: warning: Value stored to 'devc' is never read
devc = sdi->priv;
^ ~~~~~~~~~
src/hardware/rigol-ds/api.c:813:3: warning: Value stored to 'devc' is never read
devc = sdi->priv;
^ ~~~~~~~~~
src/hardware/scpi-pps/api.c:405:2: warning: Value stored to 'ret' is never read
ret = SR_OK;
^ ~~~~~
src/hardware/yokogawa-dlm/api.c:239:2: warning: Value stored to 'ret' is never read
ret = SR_ERR_NA;
^ ~~~~~~~~~
Timer intervals shorter than about 100 ms are unnecessarily taxing
on system resources. Also, on systems like Windows the smallest
resolvable time unit without using high precision timers is about
15 ms. Regular timer intervals should be well above that value to
avoid being dominated by noise and round-off.
Firmware versions starting with 00.02.04 apparently cause the in and out
bulk endpoints to end up in a HALT state. This is likely related to the
larger transfer size quirk implemented in the Linux kernel for the Rigol
DS1000: this USBTMC implementation does not have that workaround.
Instead, if the firmware version is >= 00.02.04, both endpoints have the HALT
condition cleared on device close.
This fixes bug #354.
The wrong byte was being used to test for the nano indicator.
This resulted in reported resistance and capacitance readings being off
by orders of magnitude.
This fixes bug #657.
The check for p == q is basically checking whether p/q == 1. We should
be normalising the rational before it gets here though, so in this case
we should have p == q == 1 here.
scpi-pps at line 212 assumes that an SR_OK return means that the gvar
is valid, which leads to the following error:
** GLib:ERROR:/build/glib2.0-2.45.8/./glib/gvarianttypeinfo.c:184:g_variant_type_info_check: assertion failed: (0 <= index && index < 24)
GPIO direction should be set once right after exporting. There's no need
to reset it again - in fact it's a bug which causes the probe to be reset
every time the value is read/set and gives incorrect results when reading
the GPIO values with direction == 'in'.
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Revision B of ACME hardware introduces probes with on-board at24cs02
EEPROM. Extend the ACME driver to support reading the contents of
the EEPROM via linux' sysfs interface.
Also: make the driver be able to tell the difference between revisions,
add new GPIO layout and set the shunt resistance for revB at probe
registration.
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Only perform a single check at initialization time to see if the probe is
equipped with a power-switch. This is done in preparation for revision B
support which has this kind of information encoded in EEPROM.
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
After opening the USB device, set the device configuration to 1.
Actually, do it twice, just as the vendor driver seems to do. This
is supposed to trigger a lightweight reset of the device.
Originally, I omitted this reset sequence from the sigrok driver
because it simply did not work at all for me. However, it does seem
to work now, so that may have been a problem in libusb or the kernel
which is now fixed.
With some luck, this change may finally fix#327.
Use states SR_ST_ACTIVE and SR_ST_INACTIVE to indicate that the
device is open or closed, respectively. Do not use any of the
other state values. Improve the robustness of the open and close
methods in face of errors. Introduce a separate flag to indicate
that a running acquisition should be canceled.
Prepare the trigger masks at config_commit() time, so that the
trigger setup can be validated before starting an acquisition.
Accordingly, do actually report validation errors back to the
caller.
It turns out that g_stat() breaks apart when using 64 bit stat on
32-bit systems. Since the actual type of GStatBuf is decided when
glib/gstdio.h is included, it is thus possible for GLib itself to
be compiled with a different type than user code.
Ouch. Unfortunately going back to plain stat() also means that we
lose Unicode filename support on Windows.
Since Autoconf places some important feature flags only into the
configuration header, it is necessary to include it globally to
guarantee a consistent build.
Disallow polling for input/error and output-ready events at the
same time, and ensure only a single FD event source is installed.
Also, do not leak if the FD event source is removed by means
other than calling serial_source_remove().
On MinGW, two implementations of printf() are available: either
the Microsoft native one or a standard-conforming replacement from
gnulib. Since we build in C99 mode, headers such as <inttypes.h>
already select the standard-conforming variant. However, MinGW's
GCC does not seem to know about this and assumes MS-style format
syntax by default, which triggers a lot of wrong warnings.
Thus, on MinGW, explicitly decorate sr_log() with the gnu_printf
format flavor attribute. Also use GLib's printf replacements in
the logging implementation to make sure we link to a conforming
printf on any platform, independently of the compiler flags.
This gets rid of the mistaken -Wformat warnings for sr_log(), but
does not cover functions such as g_strdup_printf() which do not
explicitly specify the gnu_printf flavor in the format attribute.
This can be overcome by adding "-D__printf__=__gnu_printf__" to
CPPFLAGS, but it would be inappropriate for libsigrok to define
this on its own.
The confusingly named sr_log_logdomain_set() simply set a global
string prefixed to the log message by the default log callback.
This is pretty much useless, misleadingly named, and not used by
either sigrok-cli or PulseView.
A few of these were pretty serious, like missing arguments,
passing integers where a string was expected, and so on.
In some places, change the types used by the code rather than
just the format strings.
SR_LOG_DBG and above are targeted at developers, so it makes sense
to extend timestamp output to that. Also sanitize the calculation
of the timestamp components a bit.
Get rid of the specicialized sr_err(), sr_warn(), etc. functions.
Instead, define the logging helper macros in terms of sr_log(),
and remove the sr_log() helper macro so that no function is hidden
by a macro anymore.
Decorate sr_log() with G_GNUC_PRINTF to detect varargs errors. This
unearthed a gazillion warnings all over the place which will have
to be fixed.
Also convert the helper macros to ISO C99 __VA_ARGS__ style instead
of relying on a GNU C extension. Paste the log prefix directly into
the format string to make this work.
Some drivers, such as zeroplus-logic-cube, run everything they do
right away in dev_acquisition_start(), never installing any event
sources. Handle that evilness by returning from sr_session_run()
immediately if there are no sources.
Replace the custom session main loop with the GLib main loop.
This is phase one of the port, which leaves the session and
driver APIs unchanged while replacing the internals.
The Hung-Chang DSO-2100 is a parallel port PC oscilloscope sold back
in 1999 under brand names like Protek and Voltcraft.
This inital version of the driver has the following limitations:
- Hardcoded calibration values. All parameters are set to 50%.
- No support for auto triggering
- No support for TV sync trigger modes
- No support for the "scroll acquisition" mode
In scroll acquisition mode the device behaves more like a multimeter
and reports the current voltage of a probe on request. While in this
mode the sample rate is limited by the parallel port interface, it is
the only way to capture both channels at the same time (well, sort of).
Calibration would need auto triggering. The calibration values are very
temperature dependent and the device takes literally hours to reach its
final temperature. Every vdiv setting needs its own set of calibration
values. Without hardware modifications, the calibration settings wear
of in less than a second while waiting for a trigger because the
capacitors storing those values are not recharged in state 0x21.
This is intended to make people notice when libusb is too old
for the new Windows code. However, this is not foolproof, since
the libusb version may be different at runtime.
Introduce new internal session API for changing the set of polled
file descriptors for an already installed event source. Use the
new API to apply changes to the USB poll FDs when requested to do
so by libusb. Doing so is necessary to make the generic USB code
work on Windows.
There was a problem in scpi_serial.c in the scpi_serial_read_data()
function. Incoming data was written at the read position in the buffer,
although it should be written at the count position in the buffer.
This is another attempt at getting the mess that is libusb event
handling on Windows under control. Until libusb makes its HANDLEs
available for polling, we have no choice but to block while waiting
for libusb events. Since we do not want to force drivers to deal
with multi-threading issues, that means we have to block in the
session main loop.
Fortunately, it turns out that our drivers aren't using multiple
event sources, so it is actually possible to block the main loop
without disrupting too much. This also gets rid of the USB thread
on Windows. Thankfully, libusb does not seem to care that we are
now calling libusb_handle_events_timeout_completed() twice per
iteration: first a blocking call (with timeout) in the callback
wrapper, followed by the non-blocking call in the driver-supplied
callback.
Turns out that having one event source per libusb poll FD is
a bad idea. There is only a single callback for all poll FDs,
and libusb expects to be called only once per poll iteration,
no matter how many FDs triggered.
Also, they should all share the same timeout, which should get
reset on events from any polled FD. The new timeout handling made
this problem apparent, as it caused the callback to be invoked
multiple times on timeouts, once for each separate event source.
In order to fix this, change the implementation to allow for an
arbitrary number of poll FDs per event source. This number is
zero for timer FDs, one for normal I/O sources, and one or more
for libusb sources (Unix only).
Also, on Windows, do not get an additional timeout from libusb
in the event loop. This is only appropriate when polling the
libusb FDs directly, which we aren't doing on Windows.
Lars-Peter Clausen
Bert Vermeulen
Uwe Hermann
Benjamin Larsson
Joel Holdsworth
Alexandru Gagniuc
Sergey Alirzaev
Markus Siegert
Stefan Brüns
Diego F. Asanza
Wolfram Sang
Soeren Apel
Martin Ling
Chriter
Aurelien Jacobs
Daniel Elstner
Martin Ling
Tilman Sauerbeck
Hannu Vuolasaho
Petteri Aimonen
Matthieu Gaillet
jry
Karl Palsson
Bartosz Golaszewski
Martin Lederhilger
Scott Allen
Daniel Glöckner