In case `wl_display_roundtrip` returns an error after registering for
events, print a more user-friendly error message and exit.
Previously, if the build did not have assertions enabled, this would
likely result in a segfault. With assertions enabled, it's not user
friendly to terminate with internal implementation information.
This is the first in a series of commits to refactor the color handling
in sway. This changes parse_color to return whether it was success and
no longer uses 0xFFFFFFFF as the fallback color. This also verifies that
the string actually contains a valid hexadecimal number along with
the length checks.
In the process of altering the calls to parse_color, I also took the
opportunity to heavily refactor swaybar's ipc_parse_colors function.
This allowed for several lines of duplicated code to be removed.
Note: since strtoul() has no real error return code (both 0 and
ULONG_MAX may be returned on both success and failure), set errno=0
before calling strtoul().
If the XCURSOR_THEME and/or XCURSOR_SIZE environment variables are
set, use the theme and size they define.
If they're not set, use the same defaults as before (system default
theme, size=24).
The new upstream is https://github.com/swaywm/swaybg
This commit also refactors our use of gdk-pixbuf a bit, since the only
remaining reverse dependency is swaybar tray support.
This revamps the type configs for swaynag. All sizing attributes for
swaynag are now `ssize_t` instead of `uint32_t` to allow for a default
value of `-1`, which allows for `0` to be a valid value. Additionally,
the initialization of the type configs has been changed from a simple
calloc to use a new function `swaynag_type_new`. `swaynag_type_new`
calloc's the memory, checks for an allocation failure, sets the name,
and all sizes to -1. The layering order has also been changed to
default, general config, type config, and as highest priority command
line arguments. Finally, `swaynag_type_merge` has been modified to
handle the layering and sizing changes.
Currently on master, swaynag will retrieve a pointer instance whenever
the capabilities change and WL_SEAT_CAPBILITY_POINTER is set. The
pointer instances were never being destroyed so swaynag received events
multiple times due to having several instances of the pointer.
This fixes it so if there is already a pointer instance, swaynag does
not attempt to retrieve another. Additionally, if the pointer
capability is removed, the pointer instance is destroyed.
When swaynag is run with the -l/--detailed-message option, a crash may
occur if the detailed message read from stdin is large enough. E.g.:
swaynag -m hello -l < ~/.config/sway/config
The root cause is that the read_from_stdin() function under-allocates
memory for the destination buffer which causes that buffer to be overflowed
when copying line data to it with snprintf().
The repair is to allocate one more byte for the terminating null byte.
N.B. although getline() returns the number of bytes read excluding a
terminating null byte, the line buffer is terminated with a null byte. Thus
we have a guarantee that the line buffer will be null terminated (which is
important when copying with snprintf()).
In addition to removing unused code, two minor problems are fixed:
(1) `resize set` and `resize adjust` did not error when given
too many arguments.
(2) `orientation` was incorrectly overridden to be 'U' for
scroll events in the swaybar tray `handle_click` function.
This commit mostly duplicates the wlr_log functions, although
with a sway_* prefix. (This is very similar to PR #2009.)
However, the logging function no longer needs to be replaceable,
so sway_log_init's second argument is used to set the exit
callback for sway_abort.
wlr_log_init is still invoked in sway/main.c
This commit makes it easier to remove the wlroots dependency for
the helper programs swaymsg, swaybg, swaybar, and swaynag.
If there are no arguments or invalid arguments given, swaynag will free
`swaynag.details.button_details` under the `cleanup` label in main. It
then called `swaynag_destroy`, which would attempt to free it again.
Since `swaynag.details.button_details` is either freed on line 106 of
main (when there is no detailed message) or added to `swaynag.buttons`
on line 103 of main, there is no reason to manually free it in
`swaynag_destroy`.
Although I cannot reproduce a double free on my system, for some reason,
it should have actually resulted in a double free in all code paths.
When the cursor surface gets updated, it should be damaged.
This also bumps up `wl_compositor` to version 4 to be able to use
`wl_surface_damage_buffer`.
In `i3 4.16`, `i3-nagbar` introduces the flags `-B/--button-no-terminal`
to run the action directly instead of inside a terminal. This implements
the flags for swaynag for compatibility.
Since swaynag does not use an equivalent to `i3-sensible-terminal`, the
flags `-b/--button` only uses a terminal when the environment variable
`TERMINAL` is set, otherwise it acts the same as these new flags.
When destroying swaynag from within wl_display_dispatch, we cannot
disconnect the display as that will free the queue's event_list.
Free it after running the loop instead.
Fixes this use-after-free:
==7312==ERROR: AddressSanitizer: heap-use-after-free on address 0x612000000110 at pc 0x000000412a9f bp 0x7ffd4e811760 sp 0x7ffd4e811750
READ of size 8 at 0x612000000110 thread T0
#0 0x412a9e in wl_list_empty ../common/list.c:206
#1 0x7f5b58f0d42f in dispatch_queue src/wayland-client.c:1572
#2 0x7f5b58f0d42f in wl_display_dispatch_queue_pending src/wayland-client.c:1815
#3 0x40f465 in swaynag_run ../swaynag/swaynag.c:390
#4 0x407576 in main ../swaynag/main.c:123
#5 0x7f5b58bb9412 in __libc_start_main ../csu/libc-start.c:308
#6 0x404a3d in _start (/opt/wayland/bin/swaynag+0x404a3d)
0x612000000110 is located 208 bytes inside of 320-byte region [0x612000000040,0x612000000180)
freed by thread T0 here:
#0 0x7f5b594ab480 in free (/lib64/libasan.so.5+0xef480)
#1 0x40faff in swaynag_destroy ../swaynag/swaynag.c:454
#2 0x40cbb4 in layer_surface_closed ../swaynag/swaynag.c:82
#3 0x7f5b583e1acd in ffi_call_unix64 (/lib64/libffi.so.6+0x6acd)
previously allocated by thread T0 here:
#0 0x7f5b594aba50 in __interceptor_calloc (/lib64/libasan.so.5+0xefa50)
#1 0x7f5b58f0c902 in wl_display_connect_to_fd src/wayland-private.h:236
(you need a wayland compiled with asan, my wl_list hack, or running
with valgrind to see this trace)
It's better to use DT_RPATH dynamic section of the elf binary to store
the paths of libraries to load instead of overwriting LD_LIBRARY_PATH
for the whole environment, causing surprises. This solution is much more
transparent and perfectly suitable for running contained installations
of wayland/wlroots/sway.
The code unsetting the LD_LIBRARY_PATH/LD_PRELOAD was also deleted as
it's a placebo security at best - we should trust the execution path
that leads us to running sway, and it's way too late to care about those
variables since we already started executing our compositor, thus we
would be compromised anyway.
This does the following:
* Adds a baseline argument to get_text_size (the baseline is the
distance from the top of the texture to the baseline).
* Stores the baseline in the container when calculating the title
height.
* Takes the baseline into account when calculating the config's max font
height.
* When rendering, pads the textures according to the baseline so they
line up.
- Some platforms don't expose kill() unless _POSIX_C_SOURCE is defined.
- fork(), execl(), and setsid() need unistd.h on some platforms.
Basically, this fixes some platform-specific build errors.
