Stephen Magladry wrote:
> I issued the
>
> export SANE_DEBUG_SANEI_THREAD=255
> only and here are the results.
>
> [blah:~] blah% scanimage >test.pnm
> [sanei_debug] Setting debug level of sanei_thread to 255.
> [sanei_debug] Setting debug level of sanei_thread to 255.
> [sanei_debug] Setting debug level of sanei_thread to 255.
> [sanei_debug] Setting debug level of sanei_thread to 255.
> [sanei_debug] Setting debug level of sanei_thread to 255.
> [sanei_debug] Setting debug level of sanei_thread to 255.
> [sanei_thread] thread started, calling func() now...
> [sanei_thread] pthread_create() created thread 25179136
> [sanei_thread] func() done - status = 0
>
The cancelling of the reader thread fails.
Since pthread_cancel is broken on MacOS X there is a workaround
implemented in sanei_thread. This workaround essentially replaces the
sanei_cancel call with a SIGUSR2 signal. This means that backends SHOULD
NEVER disable the SIGUSR2 signal on MacOS X if they want to be cancelled.
For the Plustek backend the following code:
sigfillset ( &ignore_set );
sigdelset ( &ignore_set, SIGTERM );
sigprocmask( SIG_SETMASK, &ignore_set, 0 );
should be changed to:
sigfillset ( &ignore_set );
sigdelset ( &ignore_set, SIGTERM );
#if defined (__APPLE__) && defined (__MACH__)
sigdelset ( &ignore_set, SIGUSR2 );
#endif
sigprocmask( SIG_SETMASK, &ignore_set, 0 );
This probably affects other backends as well.
Mattias
--
________________________________________________________________________
mattias.ell...@tsl.uu.se tel: +46 18 471 32 58
http://www.tsl.uu.se/~ellert/ fax: +46 18 471 35 13
________________________________________________________________________
