Skip to content

GitLab

Switch branch/tag
  1. 18 Dec, 2020 10 commits
  3. 06 Dec, 2020 7 commits
  5. 08 Aug, 2020 1 commit
  7. 23 Jun, 2020 1 commit
    • Ralph Giles's avatar
      autotools: Remove AC_CANONICAL_TARGET · 475e92a1
      Ralph Giles authored 
      

We were adding the --target configure option in the mistaken
belief that it was useful for cross-compiling. In GNU terminology,
"target" refers to the architecture the programme being compiled
should itself target, as when building a cross-compiler.

To specify the platform the build itself should target one
instead uses --host, while --build refers to the system
where the build itself it taking place.

We were refering to the correct `host*` variables inside
configure.ac. The call to AC_CANONICAL_TARGET is simply
unnecessary.
Signed-off-by: Marvin Scholz's avatarMarvin Scholz <epirat07@gmail.com>
      475e92a1
  9. 08 Apr, 2020 1 commit
  11. 07 Apr, 2020 2 commits
  13. 02 Mar, 2020 6 commits
  15. 22 Jan, 2018 9 commits
  17. 21 Jan, 2018 3 commits
    • Martin Steghöfer's avatar
      Make it clear in documentation that -f needs a previous -d. · f3bf5570
      Martin Steghöfer authored and Thomas Daede's avatar Thomas Daede committed 
      Bug-Debian: https://bugs.debian.org/359948
Forwarded: https://trac.xiph.org/ticket/1679

The -f option to specify an output file name works only if a file device has previously been specified using the -d option. Some parts of the documentation don't make it very explicit that -f needs -d and that the -d option has to precede the -f option.
      f3bf5570
    • Martin Steghöfer's avatar
      Fix ogg123 freeze when interrupting at End-Of-Stream. · 98a83487
      Martin Steghöfer authored and Thomas Daede's avatar Thomas Daede committed 
      Bug-Debian: https://bugs.debian.org/307325
Forwarded: https://trac.xiph.org/ticket/1956#comment:3

When arriving at the end of the input file, the main thread waits for
the output thread to finish up the current buffer. If a cancellation
signal arrives at that stage, this signal of an empty buffer never
arrives because the output thread bails out before actually emptying
the buffer.

Fix:
1.) Make sure the output thread wakes up the main thread when bailing
out, so the main thread can go on, too.
2.) When the main thread wakes up while waiting for an empty buffer,
make sure it understands the situation (that there won't be an empty
buffer because the replay has been cancelled) and doesn't go back to
sleep.
      98a83487
    • Martin Steghöfer's avatar
      Fix synchronization issue · 6d670e7c
      Martin Steghöfer authored and Thomas Daede's avatar Thomas Daede committed 
      The following interleaving can cause problems:
The interrupt for SIGINT could set sig_request.cancel right after the consumer thread
checks it ("buf->cancel_flag || sig_request.cancel"), but before the consumer
thread enters the mutex in order to wait for more data ("COND_WAIT"). If the producer
thread reacts to the new situation immediately by exiting (which it can do because the
consumer thread hasn't entered the mutex yet), the consumer thread initiates a COND_WAIT
that might never be fulfilled because the producer thread is gone. The producer's intention
to kill the consumer using buffer_thread_kill doesn't work because it was executed after
the consumer's check for buf->cancel_flag and before the consumer's COND_WAIT.

Fix the problem by putting the cancelation flag checks in the same mutex lock zone as
the "COND_WAIT".
      6d670e7c