The default values for string options are set using static (i.e.
non-malloced) strings. Later, if new values are loaded from the config
file, those SANE_Strings will then point to dynamically allocated memory
which is eventually leaked.
Change the initial values for the string options to NULL, and initialize
them to the proper values within sane_init() using strdup(). This way,
whenever the value for the string is changed, we can safely free() the
previous value.
This eliminates the other main source of memory leaks in the test backend.
This patch also updates sane_init for the test backend to properly cleanup
memory if it fails.
Inside Test_Device, track whether we've already called init_options().
If we have, don't call it again when calling sane_open() again.
Add a function cleanup_options() which frees the memory used for these
options, and call it in sane_exit() if we previously called
init_options().
Change Test_Devices to be initialized with calloc instead of malloc,
so that we can assume that uninitialized fields are NULL, and can safely
call free() on them unconditionally.
This eliminates some larger memory leaks within the test backend.
Not all pthreads implementations use an integer type for pthread_t.
As a matter of fact, POSIX has explicitly withdrawn the requirement
that it must be an arithmatic type.
The musl C library uses a `struct __pthread *` which triggered the
warnings. As of this change, sanei_thread.h works around this by
providing two new macros to help keep this issue out of sight. All
backends have been changed to use these macros.
Source code was inconsistent on how they were referencing
include files in include/sane. In certain cases, sane/sane.h
and sane/saneopts.h was really including external copies (with
some preprocessor implementations anyways).
Global replace of u_int??_t with C9x standard based uintxx_t
in order to remove some tricky and overlapping portability logic
from acinclude.m4 and leave it in only m4/stdint.m4.
Fletcher Woodruff