| Commit message (Collapse) | Author | Age | Lines | |
|---|---|---|---|---|
| * | Merge branch 'async-ptmx' | Daniel Eklöf | 2019-11-03 | -88/+210 |
| |\ | ||||
| | * | terminal: cleanup asynchronous ptmx output handling | Daniel Eklöf | 2019-11-03 | -37/+41 |
| | | | | | | | | | | | Break out actual writing to a separate function, and call this function both from the synchronous and the asynchronous code paths. | |||
| | * | Merge branch 'master' into async-ptmx | Daniel Eklöf | 2019-11-03 | -1/+10 |
| | |\ | |/ |/| | ||||
| * | | slave: turn on FD_CLOEXEC after slave has been spawned | Daniel Eklöf | 2019-11-03 | -0/+9 |
| | | | | | | | | | | | This ensures that our ptmx FD isn't carried over to other terminals spawned later. | |||
| * | | server: FD_CLOEXEC is a file descriptor flag, not a file status flag | Daniel Eklöf | 2019-11-03 | -1/+1 |
| | | | ||||
| | * | term: asynchronous writes to slave | Daniel Eklöf | 2019-11-03 | -21/+106 |
| | | | | | | | | | | | | | | | | | | | | | Make ptmx non-blocking. Then, when writing data to the slave would have blocked, use the FDM to asynchronously write the remaining data. This is done by enabling EPOLLOUT on ptmx, and enqueueing all outgoing data. The FDM handler will go through the enqueued data, and once all of it has been written, we turn off EPOLLOUT again (thus switching back to synchronous writes) | |||
| | * | fdm: invert check to get rid of one level of indentation | Daniel Eklöf | 2019-11-02 | -13/+14 |
| | | | ||||
| | * | Rename: vt_to_slave() -> term_to_slave() | Daniel Eklöf | 2019-11-02 | -46/+46 |
| | | | ||||
| | * | misc: fdm already logs failures | Daniel Eklöf | 2019-11-02 | -30/+8 |
| | | | ||||
| | * | fdm: add fdm_event_add() and fdm_event_del() | Daniel Eklöf | 2019-11-02 | -1/+55 |
| |/ | | | | | These functions allow users to modify already registered FDs, to add or remove events they are interested in. | |||
| * | fdm: bug: 'ret' was renamed to 'r', and replaced with another 'ret' | Daniel Eklöf | 2019-11-02 | -1/+1 |
| | | ||||
| * | fdm: rename struct fd -> struct handler | Daniel Eklöf | 2019-11-02 | -11/+11 |
| | | ||||
| * | fdm: fdm_destroy(): assert deferred delete list is empty | Daniel Eklöf | 2019-11-02 | -0/+2 |
| | | ||||
| * | fdm: purge deferred delete list also when handler fails | Daniel Eklöf | 2019-11-02 | -5/+8 |
| | | ||||
| * | server: server_destroy(): unlink UNIX socket | Daniel Eklöf | 2019-11-02 | -0/+2 |
| | | ||||
| * | server: use a non-blocking socket when trying to connect | Daniel Eklöf | 2019-11-02 | -4/+23 |
| | | | | | | | | | | | | | | | When checking of a 'foot --server' instance is already running, we try to connect to the UNIX socket we're planning on listening on. In most cases, this will fail hard and fast. But under certain circumstances, we can get stuck in connect() waiting for a connection timeout. Since it should be possible to establish a connection immediately *if* there's someone actually listening on the socket, rely on the fact that connect() will fail with a EINPROGRESS if a connection can *not* be established immediately (and thus no one is listening on it). | |||
| * | main: pre-allocate fonts | Daniel Eklöf | 2019-11-02 | -1/+28 |
| | | | | | | | | | Since fonts are cached, this adds no additional memory. However, it makes the first terminal window in --server mode start much faster, since the (primary) fonts have already been loaded. Fallback fonts are still loaded on-demand. | |||
| * | fdm: fdm_poll(): disallow nested calls | Daniel Eklöf | 2019-11-02 | -0/+6 |
| | | ||||
| * | server: don't get stuck in read() if TERM length is 0 | Daniel Eklöf | 2019-11-02 | -2/+4 |
| | | ||||
| * | server: don't get stuck in read() if length of argv argument is 0 | Daniel Eklöf | 2019-11-02 | -0/+3 |
| | | ||||
| * | terminal: don't get stuck waiting for misbehaving slaves to terminate | Daniel Eklöf | 2019-11-02 | -4/+64 |
| | | | | | | | | | | | | | | | While unusual, it *is* possible for a client *not* to terminate when we close ptmx. We need to handle this *somehow*. Since it is so unusual, we'll go with a fairly easy, but synchronous method: * Register a signal handler for SIGALRM, and setup a 2 second alarm * Wait for slave to die * If it didn't die, sent SIGTERM, then re-set the alarm for another 2 seconds. * If it still hasn't died, send SIGKILL (this time without an alarm). | |||
| * | terminal: 'child_ret' variables isn't needed | Daniel Eklöf | 2019-11-02 | -5/+3 |
| | | ||||
| * | timerfd: read() returns -1 with errno == EAGAIN, not 0 | Daniel Eklöf | 2019-11-02 | -20/+12 |
| | | | | | | | When there hasn't been a timeout (or in our case, there was a timeout, but we reset the timer before we got to the read()), read() returns, not 0, but -1 with errno == EAGAIN. | |||
| * | shm: purge buffers *before* we (maybe) return one from the cache | Daniel Eklöf | 2019-11-02 | -16/+16 |
| | | ||||
| * | Revert "render: last_buf may point to a free:d buffer" | Daniel Eklöf | 2019-11-02 | -23/+9 |
| | | | | | This reverts commit 4d3251a93b4ba309d573aca6ff397df44c4ff760. | |||
| * | shm: automatic buffer purging is now delayed one cycle | Daniel Eklöf | 2019-11-02 | -5/+22 |
| | | | | | | This ensures we don't purge a buffer that a user is holding a reference to. | |||
| * | timerfd: use non-blocking mode, fixes dead lock | Daniel Eklöf | 2019-11-02 | -5/+25 |
| | | | | | | | | | | | | Since we cancel the timers every now and then, there's a (small) chance that one handler cancels a timer that has triggered in the same epoll() iteration. When this happens, read() blocks. Fix by making the timer FDs non-blocking, and simply returning when we read 0 bytes. | |||
| * | server: purge shm buffers when terminal is destroyed | Daniel Eklöf | 2019-11-01 | -0/+6 |
| | | ||||
| * | shm: add shm_purge() | Daniel Eklöf | 2019-11-01 | -0/+19 |
| | | | | | Destroys *all* buffers associated with the specified cookie. | |||
| * | render: last_buf may point to a free:d buffer | Daniel Eklöf | 2019-11-01 | -9/+23 |
| | | | | | So, store last buf's width/height separately | |||
| * | shm: purge unused buffers | Daniel Eklöf | 2019-11-01 | -7/+28 |
| | | | | | | | When we need to create a new buffer (because the cache doesn't have any buffers of correct size, or because they're all busy), purge buffers with a size mismatch. | |||
| * | shm: associate a 'cookie' with each buffer | Daniel Eklöf | 2019-11-01 | -8/+21 |
| | | | | | | | | When re-using a buffer from cache, only re-use ones with a matching cookie. This prevents contention between multiple terminal windows. | |||
| * | shm: get_buffer(): remove 'copies' argument; it's not used | Daniel Eklöf | 2019-11-01 | -7/+4 |
| | | ||||
| * | shm: add debug logging of how large the allocated memory buffer is | Daniel Eklöf | 2019-11-01 | -0/+3 |
| | | ||||
| * | shm: fix name of memfd | Daniel Eklöf | 2019-11-01 | -1/+1 |
| | | ||||
| * | client: bug: verify argv was sent correctly | Daniel Eklöf | 2019-11-01 | -1/+1 |
| | | | | | We checked we had written 2 bytes (sizeof(len)), not <len> bytes. | |||
| * | client: add debug logging of argc/argv | Daniel Eklöf | 2019-11-01 | -0/+4 |
| | | ||||
| * | server: add debug logging of client argc/argv | Daniel Eklöf | 2019-11-01 | -0/+4 |
| | | ||||
| * | Merge branch 'multi-window' | Daniel Eklöf | 2019-11-01 | -171/+923 |
| |\ | ||||
| | * | client: register signal handler for SIGTERM too | Daniel Eklöf | 2019-11-01 | -1/+2 |
| | | | ||||
| | * | client/server: add -t,--term to footclient | Daniel Eklöf | 2019-11-01 | -3/+40 |
| | | | ||||
| | * | term_init(): initialize slave TERM from term_init() argument | Daniel Eklöf | 2019-11-01 | -6/+6 |
| | | | ||||
| | * | slave: set TERM environment variable in slave process | Daniel Eklöf | 2019-11-01 | -4/+6 |
| | | | ||||
| | * | PKGBUILD: need to execute footclient, to generate profiling data | Daniel Eklöf | 2019-11-01 | -0/+2 |
| | | | ||||
| | * | client: add long options for --verbose and --help | Daniel Eklöf | 2019-11-01 | -1/+7 |
| | | | ||||
| | * | wayland: wayl_win_destroy(): looks like we need another roundtrip | Daniel Eklöf | 2019-11-01 | -0/+2 |
| | | | ||||
| | * | font: cache loaded fonts globally | Daniel Eklöf | 2019-11-01 | -0/+52 |
| | | | | | | | | | | | | | | | | | | | Each font instance has a ref-counter. Whenever we want to instantiate a font that has already been loaded, we instead return the already-loaded instance, and bump the ref counter. When the last font instance is destroyed, it is also removed from the cache. | |||
| | * | completions: zsh: add --server and --term to completions | Daniel Eklöf | 2019-11-01 | -4/+10 |
| | | | ||||
| | * | client: a standalone binary that connects to a foot --server | Daniel Eklöf | 2019-11-01 | -0/+143 |
| | | | ||||
| | * | server: implement a --server mode | Daniel Eklöf | 2019-11-01 | -2/+334 |
| | | | | | | | | | | | | | | | | | | | | | | | In this mode, foot listens on a UNIX socket and creates terminal windows when clients connect. A connecting client sends argc/argv to the server, and the server instantiates a new terminal window. When the terminal window is closed, the exit code is sent back to the client. | |||
| | * | main: register signal handlers for SIGINT and SIGTERM | Daniel Eklöf | 2019-11-01 | -0/+15 |
| | | | ||||
| | * | terminal: call user-defined callback when destroying terminal | Daniel Eklöf | 2019-11-01 | -13/+38 |
| | | | | | | | | | main uses this to get the exit code of the terminal. | |||
| | * | term_shutdown(): cleanup | Daniel Eklöf | 2019-11-01 | -49/+35 |
| | | | | | | | | | | | * Close FDs, then delay destruction to next epoll() call * Destroy window before destroying terminal | |||
| | * | terminal: fdm_del() closes the FD | Daniel Eklöf | 2019-11-01 | -20/+6 |
| | | | ||||
| | * | terminal: delayed rendering: fdm_delayed_render() may be called with is_armged==false | Daniel Eklöf | 2019-11-01 | -1/+2 |
| | | | | | | | | | This would happen if *both* timers triggered in the same epoll() call. | |||
| | * | terminal: no need to check for EAGAIN - we don't use non-blocking | Daniel Eklöf | 2019-11-01 | -18/+10 |
| | | | ||||
| | * | render: limit length of title in call to xdg_toplevel_set_title() | Daniel Eklöf | 2019-11-01 | -1/+13 |
| | | | | | | | | | | | | | Trying to pass a too long title (not sure what "too long" is though...) will trigger a call to abort() inside the wayland-client library. | |||
| | * | wayland: wayl_destroy(): destroy any remaining terminals | Daniel Eklöf | 2019-11-01 | -0/+9 |
| | | | ||||
| | * | Don't use non-blocking FDs | Daniel Eklöf | 2019-11-01 | -33/+6 |
| | | | | | | | | | | | We use epoll() to determine when we can read/write FDs so there's absolutely no need for non-blocking. | |||
| | * | wayland: wayl_win_destroy(): unmap windows before destroying | Daniel Eklöf | 2019-11-01 | -0/+23 |
| | | | | | | | | | | | | | | | | | | | This will trigger e.d. keyboard_leave() and wl_pointer_leave() events, which ensures there aren't any references to the destroyed window from the global wayland struct. Call wl_display_roundtrip() to trigger those events *before* we destroy the window. | |||
| | * | wayland: wayl_init(): call wl_display_roundtrip() when done | Daniel Eklöf | 2019-11-01 | -2/+3 |
| | | | ||||
| | * | render: call wl_display_flush() after rendering | Daniel Eklöf | 2019-11-01 | -1/+2 |
| | | | | | | | | | This allows us to remove that call from the main event loop. | |||
| | * | wayland: fdm_del() now closes the FD | Daniel Eklöf | 2019-11-01 | -4/+3 |
| | | | ||||
| | * | fdm: don't free FD data that may be referenced by epoll return data | Daniel Eklöf | 2019-11-01 | -9/+60 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It is perfectly possible, and legal, for a FDM handler to delete another handler. The problem however is when the epoll returned array of FD events contain the removed FD handler *after* the handler that removed it. That is, if the epoll returned array is: [FD=13, FD=37] and the handler for FD=13 removes FD=37, then given the current implementation, the FD user data (our handler callback etc) will point to a free:d address. Add support for this situation by deferring FD handler removal *when called from another handler*. This is done by "locking" the FDM struct before looping the handlers for FDs with events (and unlocking afterwards). In fdm_del(), we check if the FDM has been locked, in which case the FD is marked as deleted, and put on a deferred list. But not *actually* deleted. Meanwhile, in the FDM poll function, we skip calling handlers for marked-for-deletion FDs. Then, when all FDs have been processed, we loop the deferred list and finally deletes the FDs for real. | |||
| | * | term: fdm_shutdown(): remove debug output | Daniel Eklöf | 2019-10-30 | -1/+0 |
| | | | ||||
| | * | main: get exit value from wayland struct | Daniel Eklöf | 2019-10-30 | -2/+3 |
| | | | ||||
| | * | term: term_shutdown(): set exit value in wayland | Daniel Eklöf | 2019-10-30 | -5/+7 |
| | | | ||||
| | * | wayland: temporary: track last terminal's exit value | Daniel Eklöf | 2019-10-30 | -0/+1 |
| | | | ||||
| | * | wayland: wayl_win_destroy(): return early if win == NULL | Daniel Eklöf | 2019-10-30 | -0/+3 |
| | | | ||||
| | * | term: assign client PID to term->slave | Daniel Eklöf | 2019-10-30 | -1/+1 |
| | | | ||||
| | * | term: term_init(): create wayland window *last* | Daniel Eklöf | 2019-10-30 | -48/+57 |
| | | | | | | | | | | | * Configure all FDs *completely* before moving on * Start client before creating the wayland window | |||
| | * | slave: log child's errno, not parents | Daniel Eklöf | 2019-10-30 | -2/+2 |
| | | | ||||
| | * | slave: fix debug logging (no 'term' variable) | Daniel Eklöf | 2019-10-30 | -2/+2 |
| | | | ||||
| | * | wayland: xdg_toplevel_close(): call term_shutdown() | Daniel Eklöf | 2019-10-30 | -3/+1 |
| | | | ||||
| | * | term: add term_shutdown() | Daniel Eklöf | 2019-10-30 | -32/+103 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This function unmaps the terminal window, removes itself from the wayland list of terminals and then finally destroys itself. We ensure we don't get any callbacks/events referring to a free:d terminal struct, we close all terminal related FDs and unmap the wayland window. Then, to be really really sure there aren't any (by the FDM) queued up events, we delay the self-destruction to the next FDM poll iteration, by opening an event FD and adding it to the FDM. The callback for the event FD removes the FD from the FDM again, and closes it. And then proceeds to destroy the terminal. | |||
| | * | wayland: track multiple terminals | Daniel Eklöf | 2019-10-30 | -15/+26 |
| |/ | | | | | The wayland 'term' member is gone and replaced by a list, 'terms'. This list contains all currently running terminal (windows). | |||
| * | Merge branch 'term-init-cleanup' | Daniel Eklöf | 2019-10-30 | -137/+196 |
| |\ | ||||
| | * | main: free 'config' on command line parse error | Daniel Eklöf | 2019-10-30 | -0/+4 |
| | | | ||||
| | * | term: move client startup to a new function, slave_spawn() | Daniel Eklöf | 2019-10-30 | -61/+79 |
| | | | ||||
| | * | term: break out font initialization to a separate function | Daniel Eklöf | 2019-10-30 | -32/+41 |
| | | | ||||
| | * | term: break out render worker threads initialization to a separate function | Daniel Eklöf | 2019-10-30 | -21/+50 |
| | | | ||||
| | * | term: break out color cube initialization to a separate function | Daniel Eklöf | 2019-10-30 | -24/+23 |
| | | | ||||
| | * | term: require all font variants to load | Daniel Eklöf | 2019-10-30 | -5/+5 |
| |/ | ||||
| * | Merge branch 'fdm' | Daniel Eklöf | 2019-10-30 | -1483/+1945 |
| |\ | ||||
| | * | slave: rename slave_spawn() -> slave_exec() | Daniel Eklöf | 2019-10-30 | -3/+3 |
| | | | ||||
| | * | term: no need to stack-allocate an array of worker contexts | Daniel Eklöf | 2019-10-28 | -6/+9 |
| | | | ||||
| | * | render worker context: allocate, and let worker threads free | Daniel Eklöf | 2019-10-28 | -4/+6 |
| | | | | | | | | | | | | | | | | | | | | | Since we now initialize the worker threads from term_init(), which returns before the threads terminate, we can no longer use stack-allocated worker contexts. We _could_ put them in the terminal struct. But a simpler solution is to allocate them in term_init(), and let the threads free them when they don't need them anymore. | |||
| | * | term: open FDs before allocating terminal struct | Daniel Eklöf | 2019-10-28 | -42/+69 |
| | | | ||||
| | * | term: integrate directly with FDM | Daniel Eklöf | 2019-10-28 | -222/+223 |
| | | | ||||
| | * | terminal: implement term_init() and term_destroy() | Daniel Eklöf | 2019-10-28 | -328/+375 |
| | | | ||||
| | * | wayland: move reload_cursor_theme() and update_cursor_surface() to wayland | Daniel Eklöf | 2019-10-27 | -61/+64 |
| | | | ||||
| | * | main: remove a couple of includes | Daniel Eklöf | 2019-10-27 | -3/+1 |
| | | | ||||
| | * | main: kbd repeat FD is handled by the wayland backend | Daniel Eklöf | 2019-10-27 | -1/+1 |
| | | | ||||
| | * | wayland: backend now configures FD as non-blocking, not main | Daniel Eklöf | 2019-10-27 | -16/+17 |
| | | | ||||
| | * | main: cleanup | Daniel Eklöf | 2019-10-27 | -7/+0 |
| | | | ||||
| | * | wayland: free wayland/wl_window instances | Daniel Eklöf | 2019-10-27 | -0/+3 |
| | | | ||||
| | * | wayl_destroy: early return on NULL | Daniel Eklöf | 2019-10-27 | -0/+3 |
| | | | ||||
| | * | wayland: move keyboard repeat handling into wayland | Daniel Eklöf | 2019-10-27 | -31/+39 |
| | | | ||||
| | * | wayland: integrate directly with the FDM | Daniel Eklöf | 2019-10-27 | -17/+23 |
| | | | ||||
| | * | wayland: implement wayl_init() | Daniel Eklöf | 2019-10-27 | -526/+569 |
| | | | | | | | | | Wayland instantiation is now done by the wayland backend, not in main. | |||
