My intended use case is the following:
I want to compile saned backends with different libraries (for example
glibc) than a specific application (as if it were a chroot of another
distro version). This application cannot dlopen sane
backends because of the library mismatch.
The specific application will be
built with a stripped down libsane that only has the net backend,
and therefore very few dependencies. The net backend points to saned.
saned has a full config and set of backends, and is responsible for
exporting all scanners to the application. This option allows to
reexport network scanners this way. I only have to make sure saned does
not contact itself.
Not only the Perfection1650, but other scanners need to ignore the
FSG_STATUS_CANCEL_REQ bit as well. Thanks to @matc for debugging and
providing a list. Should fix#716. Remove the hardcoded bitmask while
here.
Signed-off-by: Wolfram Sang <wsa@kernel.org>
- Stopping and releasing the device causes the ADF to purge any remaining pages,
thus only the first page gets properly scanned (observed with Samsung
SCX-4729FW).
- Instead, when sourcing from the ADF, at each page's completion it should
continue and issue "OBJECT POSITION" command to proceed to the next page.
- Once all pages have been processed, the subsequent "OBJECT POSITION" command
returns the proper "CHECK CONDITION" status (ADF is empty) and triggers the
device stop and release.
- Additionally, the status of the ADF (loaded/empty) needs to be requeried
at the `sane_start()` for proper handling of the "Auto" mode. Otherwise,
the scanning session retains the ADF state from the start of the session,
even though ADF will become empty. Thus the "Auto" mode will not properly
switch to Flatbed as intended in the client application (observed in
Simple-Scan).
> NOTE: the initial "READ" command in `sane_start()` is extraneous and
> appears to always return STATUS_BUSY (indeed, the scanner is usually
> busy loading the page). The actual READ is correctly done in
> `dev_acquire()`. It may appear benign in a single page scan mode, however,
> it becomes malfunctioning in the mutipage ADF mode. The initial READ,
> when successful, will cause the skipping of the read block from processing,
> as the actual processing is tied to content read from `dev_acquire()`.
>
> Therefore, the initial READ command in the `sane_start()` is removed.
Without this change, pixma scanning is currently broken (tested with Canon MP495). The stated reason for the workaround (libusb not supporting timeouts for interrupts) hasn't been true since 2010:
67d9ef7b68
1. Fixed scan to jpeg/png
2. Fixed bottom part of the image
3. Removed white pixels from the right part of the image
4. Adjusted physical dimensions of the scanning area