When a new connection came in we would close the old connection. This in
turn would trigger a receive error causing the *new* connection to be
closed from the error handler.
Fix this by simply cancelling any pending transfers when a new
connection comes in. Also change the error handling code to catch issues
like partial writes correctly.
This fixes an issue for the fprintd test where some tests were flaky.
While the image device has its own finger status tracking, we use a simpler
version as public data information, so let's just report the finger-on/off
and when a finger is expected to the parent class.
Verify that this happens as expected using the virtual-image class
When porting the driver to the new libfprint 1.90.0 a mistake was made
where the device was not passed through user_data anymore but it was
still read from there. Stop using user_data in the callback to fix this.
See: #320
The elan driver always "deactivates" the device after a cpature run. We
can simplify the while internal state into a single "active" state and
rely on the image device driving this state machine correctly.
i.e.:
* We start a callibrate/capture/deactivate when we go into the
AWAIT_FINGER_ON state
* We only store the fact that we are active and want to deactivate
* We rely on the image device never going into the AWAIT_FINGER_ON
state without first waiting for the finger to be off (which implies
deactivation).
This adds a number of new internal states to better capture what is
going on. Also added are checks that all transitions we make are in the
set of expected and valid transitions.
Only three drivers use the state_change notification. These drivers are
updated accordingly.
The boolean is just used to emit a warning for unexpected state
transitions. It is sufficient to pass it to the deactivate function
directly for that purpose.
We might redo image transfers, but we only ever had one reference that
was implicitly removed after the transfer completed. Add a new reference
each time it is submitted and only free the last reference in the stop
handler.
For some reason we re-submitted the interrupt transfer with a timeout of
1s while the original submission was without a timeout. This looks like
typo, remove the timeout.
Annotate the USB transfer creation functions with the correct buffer
access attribute. Note that we only annotate them as read_only as the
functions may be used for sending and receiving.
Hopefully this will catch buffer overflows in drivers in the future.
Doing this avoids race conditions in testing code where the code might
try to submit errors/images before the device has been activated. This
would then (sometimes) trigger assertions in the image driver code.
The earlier image device code tried to hide deactivation from the
surrounding library. However, this does not make any sense anymore with
the early reporting feature present.
This changes the operation to complete only once the device is
deactivated. Also changed that in most cases (except for cancellation)
we wait for the finger to be removed before deactivating the device.
The driver seems to have relied on the device to be fully
deactivated and activated again for each capture. This is not the case
during enroll, and as such in can cause issues.
Try fixing this by splitting out the last few commands send to the
device and assuming that this starts the capture.
Fixes: #306
We have plenty of code paths where a transfer may be cancelled before it
is submitted. Unfortunately, libgusb up to and including version 0.3.6
are not handling that case correctly (due to libusb ignoring
cancellation on transfers that are not yet submitted).
Work around this, but do so in a somewhat lazy fashion that is not
entirely race free.
Closes: #306
Delay the open/close callbacks by 100ms so that we have a window of
opportunity for race conditions. This is needed to test certain
conditiosn in fprintd.
Adding a trailing \n to g_message, g_debug, g_warning and g_error is not
neccessary, as a newline will be added automatically by the logging
infrastructure.
The cancellable needs to be free'ed at deactivation. Also free it if we
run into a fatal error, which then in turn indicates that the device is
deactivated already.
The change_state function is called synchronously from the
image_captured callback. This means that deactivation of the device
happens during the img_cb function, causing the USB transfer to be
re-registered even though the device is already deactivating.
There are various ways to fix this, but it makes sense to directly bind
the cancellation to the deactivation. So create a cancellable that we
cancel at deactivation time, and make sure we always deactivate by going
through cancellation.
closes: #306
Image devices are simply deactivated suddenly. As such, the cancellation
logic of FpDevice is not really useful there, but the documentation was
apparently accidentally copied unmodified.
This is unlikely to happen in a real world scenario and currently breaks
running the CI test (not the umockdev based ones) while building the
flatpak. Lower the severity to avoid aborting because there is a
warning.
This is based on the patch and observation from Bastien that some
URU4000B devices do not use encryption by default (it is a configuration
stored within the firmware). As such, it makes sense to always detect
whether encryption is in use by inspecting the image.
The encryption option would disable flipping of the image for the
URU400B device. Retain this behaviour for backward compatibility.
The code would just read 4096 bytes from the packet, without checking
the size and neither setting short_is_error. It is not clear whether
packets from the device are always 4096 bytes or not. But the code
assume we always get a full line, so enforce that and use the actual
packet size otherwise.
This was not used. The old driver used this if creating a USB transfer
failed, however, we delay any such failures (which cannot really happen)
into the callback today, where the error is handled differently.
The GPtrArray needs to be created at some point. Also, reference
counting was wrong as submitting the transfer sinks the ref, but we rely
on it surviving.
Note that we really should change this to only have one in-flight
transfer and starting a new one after it finishes.
Co-authored-by: Vasily Khoruzhick <anarsoul@gmail.com>
When sending static data, it would not be copied. The function that
sends it assumed that it should be free'ed though.
Fix this by simply always making a copy.
The driver would warn about the fact that a state change is queued, but
still queue it a second time. This would result in deactivation to run
twice.
See: #216
The image driver may still be deactivating when a new activation request
comes in. This is because of a hack to do early reporting, which is
technically not needed anymore.
Fix the immediate issue by properly reporting the retry case. The proper
fix is to only finish the previous operation after the device has been
deactivated.
The driver might forget to set the type of the print. Catch that error a
bit earlier rather than failing when the API user tries to load it from
disk again.
The type of the print (RAW or NBIS) needs to be filled in by the driver.
For most drivers the image devices does this (NBIS), but the
corresponding call was missing in the upekts driver, rendering the
enrolled print unusable.
__handle_incoming_msg would copy the payload of the message into a newly
created buffer just to destroy it again immediately after calling the
callback. Just reference the correct address inside the original package
instead.
Also, in one case the extra buffer was leaked afterwards.
The driver would correctly calculate the amount of extra space needed to
receive the whole packet. It would also request the correct number of
bytes for this transfer.
However, the reallocated buffer to hold this data was directly derived
from the expected payload size and did not include the overhead.
Make the code more explicit and get rid of the confusing
MAX_DATA_IN_READ_BUF define that hides details on buffer allocation
calculation from the code.
The device is already beeing de-initialised from the verify/enroll
commands. Trying it again will result in a timeout error as it is not
responding properly at that point.
And activate perimeter points removal if this flag is set
This flag should be set for aes1610, aesx660, aes2501, aes2550
and upektc_img since these sensors may produce incomplete image.
Fixes: #142
This step helps to drop false minutiae for short sensors and
it was accidentally dropped during NBIS update. Re-add this step
and add a patch to update script to ensure that it's not dropped
during next update.
Fixes: 9fb789dc78 ("nbis: Update to NBIS 5.0.0")
If a transfer errors out then actual_length is negative. The only error
that is not caught is a timeout error, which should also result in the
SSM to move to the next state.
For some reason static checkers report that the auto-free is not run if
the goto exists the loop. It seems to me like that should work fine, but
we can simply make the analysers happy by moving it.
We need more than 1 line for assembling, but in general, we should
have a reasonable amount of lines. So use the width in the hope we'll
get an image that is about square at least.
Closes: #135
When no driver is found for an USB device a debug message is
printed. However, it has PID and VID in wrong order - usually it
is Vendor ID which goes first. This is how 'lsusb' prints it.
Matching the order helps debugging.
Signed-off-by: Michal Privoznik <michal@privoznik.com>
In the fpi_print_fill_from_user_id() GDate is defined using
g_autoptr(). However, this requires new enough GLib. For older
versions there's a definition provided locally in fpi-compact.h.
Include the file to fix build with older version of GLib.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
If we fail when setting the scanned image to a print, we'll have a fatal
error, in such case we can terminate the current action and deactivate the
device.
In some cases nbim's get_minutiae returns no minutiae without providing an
error code, and if this happens we would crash in the task callback function
as we would try to deference the data->minutiae pointer.
Related to: #251
If the image height is less than the sensor horizontal resolution, then
return a retry error rather than trying to submit the image for further
processing.
Related to: #251
The minutiae detection might fail and we must not copy any data at that
point. So check g_task_had_error to ensure that we only do so when the
task was successful.
Fixes: #251
libfprint/fp-print.c: In function ‘fp_print_equal’:
libfprint/fp-print.c:596:21: warning: comparison of integer expressions of different signedness: ‘gint’ {aka ‘int’} and ‘guint’ {aka ‘unsigned int’} [-Wsign-compare]
596 | for (i = 0; i < self->prints->len; i++)
| ^
libfprint/fp-print.c: In function ‘fp_print_serialize’:
libfprint/fp-print.c:667:21: warning: comparison of integer expressions of different signedness: ‘gint’ {aka ‘int’} and ‘guint’ {aka ‘unsigned int’} [-Wsign-compare]
667 | for (i = 0; i < print->prints->len; i++)
| ^
libfprint/fp-print.c: In function ‘fp_print_deserialize’:
libfprint/fp-print.c:823:21: warning: comparison of integer expressions of different signedness: ‘gint’ {aka ‘int’} and ‘gsize’ {aka ‘long unsigned int’} [-Wsign-compare]
823 | for (i = 0; i < g_variant_n_children (prints); i++)
| ^
The public API uses gio and gobject header, ensure that these are in the
list of Required pkg-config modules, otherwise they are added to
Required.private which is not OK.
It appears the order of linking is relevant in this case, change it to
fix some linking issues.
It may be that there are better solutions to this problem.
Devices with no storage don't allow listing prints, and if we try to do
that, we'd end up in trying to call a NULL function pointer, causing a crash
So always check if the device has storage before calling the list vfunc, and
if we fail, return an error.
Include an unit-test to verify this situation
Without this we get warnings like the following:
cc1: warning: .../_build/libfprint/nbis/libfprint-include: No such file or directory [-Wmissing-include-dirs]
We are already using a number of defines and autoptrs from newer GLib
releases. Add the appropriate compatibility defines rather than removing
the corresponding code.
Closes: #222
To avoid conflicts with previous libfprint version and make sure that the
target version is the correct one (plus to allow parallel install in some
distros), let's use a versioned naming for the library keeping the abi
version in sync.
Fixes#223
The callback function would continue processing even after having failed
the SSM already. This causes further invalid operations on the SSM.
This error was found using a coverity scan.
When quickly scanning a finger with the synaptic driver, it may wait forever
for finger removal even if this has already happened.
In fact we don't take care of the finger status when reporting the
verification.
To avoid this, add a function that delays the completion of the verification
until the finger removal if the finger is on sensor, otherwise it just
performs it.
Fixes#228
Remove the never-used cmd_complete_data value and the repetition of
cmd_complete_error.
In fact now as per the fpi_device_verify_report() usage, we already pass
the match information to the device, such as the error and it will be
they will be used on completion if needed.
This allows to simplify cmd_ssm_done() as well.
In some cases we want to complete the verification after that the finger has
been removed, but we still need to promptly report the match state otherwise
fpi-device will complain about, and will eventually cause a match error
instead that reporting a non-match:
synaptics: Finger is now on the sensor
synaptics: Received message with 0 sequence number 0x91, ignoring!
synaptics: interrupt transfer done
synaptics: delaying match failure until after finger removal!
synaptics: interrupt transfer done
device: Driver reported successful verify complete but did not report the
result earlier. Reporting error instead
libfprint: Failed to verify print: An unspecified error occured!
Fixes#227
This was added as alias to the error check, but given we're passing to the
callback both the error and the match itself, we can just avoid adding an
extra parameter that could be confusing (as may imply that the matching
happened).
Also clarify the documentation and ensure that the match value is properly
set in tests.
It is a good idea to report match results early, to e.g. log in a user
immediately even if more device interaction is needed. Add new _full
variants for the verify/identify functions, with a corresponding
callback. Also move driver result reporting into new
fpi_device_{identify,verify}_report functions and remove the reporting
from the fpi_device_{identify,verify}_complete calls.
Basic updates to code is done in places. Only the upekts driver is
actually modified from a behaviour point of view. The image driver code
should be restructured quite a bit to split the reporting and only
report completion after device deactivation. This should simplifiy the
code quite a bit again.
Some things were odd with regard to the ownership of passed objects. Try
to make things sane overall, in particular with the possible floating
FpPrint reference.
The progress report user data free func was not assigned and therefore
never called. Add the missing assign, potentially fixing memory leaks
(mostly relevant for bindings).
Unfortunately, the timeout handling cannot be simulated properly. This
also adds a workaround in the driver to not consider it a protocol error
if this happens.
Continuing an enroll was broken in case of a retry error. Explicitly add
code to wait for the finger to go OFF after a retry error, and ensure
that the enroll will continue once that has happened.
We cannot copy information from the class in the _init routine, instead,
this needs to be done in _constructed. Move the relevant code into a new
_constructed function to fix importing the bz3_threshold override from
drivers.
Fixes: #206
The offset stored for a frame was not always relative to the previous
frame. This was the case for reverse movement, but for forwrad movement
the offset was the one to the next frame.
Make the offset handling consistent and alwasy store the offset to the
previous frame. Also update the frame assembling code to add the offset
before blitting the frame (i.e. making it relative to the previous frame
there too).
Note that fpi_assemble_lines already made the assumption that this was
the case as it forced the offset for the first frame to be zero. As
such, the code was inconsistent.
This could affect the AES drivers slightly as they use hardware reported
values which might not adhere to these assumptions.
We might want to iterate through the supported fingers values, to do that we
were hardcoding FP_FINGER_LEFT_THUMB and FP_FINGER_RIGHT_LITTLE as first and
last fingers.
Not that we'd ever get more fingers (unless some weird radiation would do
the job), but it's logically nicer to read than random hardcoded values.
We prefixed them with fp- which is not as obvious as fpi-. Also,
explicitly mark them as private and to be skipped in the GObject
Introspection annotatinos.
Warning: FPrint: (Signal)fp-image-device-state-changed: argument object: Unresolved type: 'FpiImageDeviceState'
Don't make headers inclusions dependent on each others, given that FpiSsm
depends on FpiUsbTransfer and vice-versa, so fix the dependency cycle by
using forwarded declarations.
The private library needs to indirectly include fp-enum.h. This
dependency was not listed anyway, resulting in a race condition during
the build process.
The elan driver converts frames into a different format. These frames
are only needed to assemable the image and should be free'ed afterwards.
Fixes: #213
While timeout was already cleared for parent, we didn't properly delete the
cancellable.
Although we'd warn anyways when starting the SSM, is still better to clear
any delayed action also for the sub-SSM
In order to be able to test the private device code (used by drivers) we
need to have that split a part in a different .c file so that we can compile
it alone and link with it both the shared library and the test executables.
Redefine fp_image_device_get_instance_private for private usage, not to move
the private struct as part of FpDevice.
In order to be able to test the private device code (used by drivers) we
need to have that split a part in a different .c file so that we can compile
it alone and link with it both the shared library and the test executables.
Redefine fp_device_get_instance_private for private usage, not to move
the private struct as part of FpDevice.
A Fp-device use an union to track the handle to the lower-level device, and
the value depends on the object type.
So in case of using a virtual device, the priv->usb_device location matches
the priv->virtual_env string location, and thus we'd end up unreffing a
string location as it was a GObject, while we'd leak the string.
To avoid such errors, instead of just checking the device type when we
finalize the device, let's just use different pointers to avoid other
possible clashes.
This fixes the the problem that the sensor becomes unresponsive after
pressing the wrong fingerprint. We fix the problem by making sure that
the non-match report is delayed until the finger is removed. With this
we cannot run into the situation that the next match request fails
immediately as the finger is still present.
Fixes: #208
The FPI_MATCH_ERROR constant was set to 0, however it is propagated to
the task completion using g_task_propagate_int. As g_task_propagate_int
will always return -1 on error, we either need to add an explicit -1
check or we just need to match the semantics.
Change the constant to -1, also rearange FP_MATCH_SUCCESS so that it
does not end up being 0.
Add commands to disable automatic finger reporting for images and to
send a specific finger report. This is useful to test more code paths
inside the image-device code.
Add fpi_ssm_new_full() that allows to pass a name to the state-machine
constructor, not to change all the drivers, provide a fpi_ssm_new() macro
that stringifies the nr_parameters value (usually an enum value) as the name
so that we can use it for better debugging.
The IRQ handler will re-register itself automatically. However, if this
happens after the callback is called, then the check whether the IRQ
handler is running fails.
Re-start the IRQ handler before calling the callback. This way the state
changes happening from the callback will see the correct IRQ handler
registration state.
See: #205
At the end of enroll, the image device would put the driver into the
AWAIT_FINGER_ON state and then deactivate the device afterwards. Doing
this adds additional complexity to drivers which would need to handle
both cancellation and normal deactivation from that state.
Only put the device into the AWAIT_FINGER_ON state when we know that
another enroll stage is needed. This avoids the critical state
transition simplifying the driver state machine.
Fixes: #203
Fixes: 689aff0232
During calibration, the internal state was stored as INACTIVE. This is
not true though, the device is actively running state machines.
Move the state update to the start of the operation, therefore ensuring
we don't deactivate without completing the SSM.
Note that this will prevent a crash, but the driver still does not
behave quite correctly when such a state change does happen. However,
this is just a safety measure as the state change should not happen in
the first place.
See: #203
These were probably added in previous iterations, but they are not uneeded
anymore as the GSource embeds already a callback function.
So make just this clearer in the dispatch function.
Now that nbis is a static library it should be possible to compile it
without any fprint-built dependency, although since it included fp_internal
there was a compile-time dependency on the fp-enums that can be generated at
later times.
So:
- Move nbis-helpers to nbis includes (and remove inclusion in fp_internal)
- Move the Minutiae definitions inside a standalone fpi-minutiae header
- Include fpi-minutiae.h in fp_internal.h
- Include nbis-hepers.h and fpi-minutiae.h in nbis' lfs.h
- Adapt missing definitions in libfprint
The nbis headers are full of redundant declarations, we can't fix them all
now, so in the mean time let's use an header using pragma to ignore such
errors.
Add the error to nbis private include folder that should be used only by
headers of libfprint-nbis.
As nbis is an external source bundle, it does not necessarily make sense
to enable/fix all warnings to the extend we do for our own library code.
As such, separate the build process into multiple stages.
We were passing around the common cflags and setting them for each library
or executable, but this is just a repetition given we can just use
add_project_arguments for this.
Properly follow function signature using a temporary gsize variable address
to make the function use the same pointer type and avoid troubles at
deferencing it, while use automatic-casting to switch to signed one if
transfer succeeded.
Add a GCancellable parameter to fpi_ssm_nex_state_delayed and
fpi_ssm_jump_to_state_delayed() so that it's possible to cancel an action
from the caller and in case the driver wants to cancel a delayed operation
when a device action has been cancelled.
This allows to have an automatic cleanup of the timeout source when the
the callback is reached and to avoid to do further state changes in the
middle.
Since GSource data can be automatically cleaned up on source destruction, we
can mimic this for the devices timeout easily as well.
Add an extra parameter, and let's use this cocci file to adapt all the
drivers like magic:
@@
expression e1, e2, e3, e4;
@@
fpi_device_add_timeout (e1, e2, e3, e4
+ , NULL
)
When using fpi_usb_transfer_submit_sync we still need to unref the transfer
once done with it, so let's use an auto pointer so we free it also on
errors and early returns without having to handle this manually.
In rare occasions it could happen that the driver was reading
insufficient data. Fix this by using g_input_stream_read_all_async
which will ensure that incomplete data will not be misinterpreted.
This fixes rare test failures seen in fprintd.
Seems like the older uncrustify versions did not find these indentation
issues. Fix them.
Old versions of uncrustify will leave things as is, so this is not a
problem if developers are using an old version of uncrustify.
Using floating point causes architecture dependent results due to
accuracy/rounding differences. It is not hard to switch to fixed point,
and while this does cause quite different rounding errors, the
difference is small.
Fixes: #200
g_variant_new_from_data() allows to destroy some other user_data passed as
parameter that might be different from the aligned_data itself.
But since in this case they match, pass it to be set as g_free parameter
or it won't be free'd.
When we notify the enroll progress with a print, this needs to be unreffed
once we're done, but this only was happening in case of error.
Since it's not up to the callback function to free it, let's do it at the
end of the function.
As per this, clarify the docs for FpEnrollProgress marking it as transfer
none.
Allow to pass a double-pointer to be nullified as the transfer data in order
to mark it as NULL when the transfer is done.
This is useful if we're keeping the transfer around in order to check that
no one is currently running.
When a transfer is completed, we automatically unref it since we can't
consider it valid anymore since this point.
Update the drivers not to free the transfer after submitting anymore.
When a machine is completed, we automatically free it since we can't
consider it valid anymore since this point.
Update the drivers not to free the SSM on completion callback anymore.
As per commit 201b5a961 we use g_date_copy() to copy the date, however the
GLib implementation is done assuming that the GDate getters are always used
as the copy function doesn't preserve the original format of the date
(whether is using julian days or dmy), and the synaptics driver access to
the dmy values directly, without using the getter that would recompute the
proper values.
Causing a read error of unset values.
So, to avoid this, just use the g_date_get_* getters to retrieve the day
month and year for for defining the print enroll id.
This is the same logic we apply to fp-device by default: any completed
action should trigger the subsequent one when it is finished.
So in case we want reactivate after a deactivation, let's do it in an idle,
after removing the current pending timeout.
Meson files are normally using 4-spaces to indent and functions use first
parameter on the same line while others at next indentation level, not
following the parenthesis indentation.
So adapt libfprint to follow the meson standard.
