| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
|
|
Github action by using builtin GITHUB_TOKEN
|
|
* Linux/MacOSX: during build, fallback to /tmp if wxWidgets sources not found in parent directory
* Linux: Add Github workflow to build .deb files and test generated binaries
* Linux: fix wrong wxWidgets extract path in yaml
* fix: fix wrong wxWidgets extract path in yaml
* Linux: harmonize content of test volumes and update Github workflow for this
* Fix corrupted test container
* Linux: optimize Github workflow by caching wxBuildConsole and wxBuildGUI folders
We also modify build script to detect the presence of wxBuildConsole and wxBuildGUI folders and reuse them
* Remove files mistakenly added
* Linux: Fix wrong test for Github workflow
* Linux: use static build of wxWidgets for GUI in Github workflow
* Linux: upload also .deb SHA256SUM file if it is generated in Github workflow
* Linux: fix detection logic of parent folder in build scripts
* Github workflow: add build_cmake_deb.sh as trigger to build-linux.yml
* Github workflow: add wxWidgets source code to the cache of workflow
|
|
for generic GUI installer
|
|
Reported on https://sourceforge.net/p/veracrypt/discussion/technical/thread/90f967e642
reference: https://grsecurity.net/toolchain_necromancy_past_mistakes_haunting_aslr
script to check: https://github.com/opensrcsec/paxtest/blob/master/contrib/check_align.sh
|
|
Full Xcode application which can take up-to 40GB of disk space is not
necessary for building VeraCrypt, rather the CommandLineTools for XCode
are sufficient which only take few gigabytes. Instead of hardcoding the SDK
location, use xcrun --show-sdk-path to support also the CommandLineTools
instance.
The reason for switching the logic for the XCode version is because the
xcodebuild will not report a correct XCode version for the CommandLineTools,
thus it can't be relied for whether to use the -Wl,-ld_classic flags.
Instead, we can just check the ld version in use in the active developer
directory, and see whether we are using dyld (the new) or ld64 (the old).
|
|
|
|
|
|
|
|
config since it detected automatically
We also update copyrights date
|
|
GTK2 specific gui installer.
|
|
makefile argument.
|
|
Introducing data into the standard out in the core service will
cause deserialization problems, as the pipes are used for interprocess
communication.
|
|
Update friend declaration in FuseService.h and refactor GetCharWidth to ComputeCharWidth in WaitDialog.h to avoif hiding GetCharWidth inherited from wxWindow
|
|
|
|
FUSE-T volumes.
|
|
|
|
official VeraCrypt dmg
|
|
minimum supported macOS version
|
|
instead of MacFUSE
|
|
|
|
|
|
The build script build_veracrypt_macosx.h now accepts the argument -f to enable fuse-t support.
It is also possible to set the environment variable VC_OSX_FUSET to 1 for FUSE-T support.
A change was done in CoreUnix::GetMountedVolumes to add a waiting loop for control file to be accessible because when using FUSE-T there always a delay before control file can be serialized.
|
|
release
|
|
This is a regression caused by macOS changes. And endif statement was misplaced.
|
|
Replaced an incorrect 'end' statement with 'endif' in the
MacOSX-specific section of the Main.make file. This resolves the make
error caused by an unclosed conditional block.
|
|
|
|
|
|
"r" getopts option.
|
|
|
|
|
|
|
|
* Add missing macOS requirement for 'make package'
We need packages for the last build step on macOS, update docs
to reflect the requirement.
* Add build instructions using homebrew
On macOS, we can use a package manager to easily install
dependencies. This simplifies onboarding and building Veracrypt.
* Add flag to use homebrew packages
When building, we can use prebuilt wxwidgets from homebrew to
simplify and speed up local building. We also put the package
behind a flag as it's optional during development.
* Skip signing for local builds
When building with homebrew, skip signing. This can be put behind
a flag to enable, if needed.
* Use system yasm on macOS if available
The binary in the repo is not universal (x86_64) and therefore
building fails on arm architecture if Rosetta is not installed.
Use local yasm if available.
* Build local arch only in development
When building via homebrew and locally, build only the local arch
which skips ASM for arm(Mx) for MacOS. This removes the need to
have rosetta installed for building.
* Fix compilation issue when COMPILE_ASM is undefined
Use a conditional check for COMPILE_ASM not being false instead of true.
This avoids passing the variable to other parts of the build script.
* Set SDK 12 as the minimum requirement and target
Align the requirement to SDK 12 in both the makefile and script,
and update the comment to remove confusion.
I chose to leave this on 12 to be on the side of err and support
as many building platforms as possible, when we can support.
The local script now also sets the target using the local sdk
version. This should improve the local development experience.
* Fix wrong architecture for macOS in x86 builds
We now build only the current arch for local development builds
in macOS. This change also fixes the x86 builds failing.
* Add instructions brew backed macOS local builds
Flags to build a local build using homebrew packages are not
default and require parameter -b to build. We also don't build
packages directly, which requires -p.
* Fix wxwidgets not linking in local x86 macOS development builds
* Clarify build location in the document
|
|
|
|
wxDir::GetAllFiles will throw an assert when opening the preferences dialog
if for some reason the user has not installed VeraCrypt properly and is
missing the intended folder. This patch adds a check to ensure the folder
first exists before querying its files.
|
|
filesystems (#1350)
|
|
The behaviour of `wc -l` is different on FreeBSD, in which the stdout
result is padded by spaces in the beginning, which causes that the
result[0] is not actually the value we care about. This patch adds
a translate removing all whitespace from the output.
|
|
|
|
packages were renamed and we linked statically against wxWidgets in GUI mode because of a bug in wxWidegts that was solved in 3.2.5 (https://github.com/wxWidgets/wxWidgets/issues/24081)
|
|
* Move from deprecated wxScopedPtr to std::unique_ptr
wxScopedPtr was included previously through some header hierarchy which as of 3.3 is
no longer the case causing it to break. But instead of including a header for a deprecated
function explicitly, just move to std::unique_ptr as recommended by upstream.
* Convert to explicit conversions from wxString
As of https://github.com/wxWidgets/wxWidgets/pull/23449/commits/35c35c235e9c29b40002131602e050dca8d65b8c
wxWidgets defaults to STL classes, which has a side-effect that
some implicit conversions break. This patch converts those conversions
to explicit in anticipation of wxWidgets 3.3 release.
|
|
Currently the maximum hidden volume size for file containers is
limited by available free space on the device the file container
resides on, which we do not care about. This commit changes so
that only Normal volumes get their `maxVolumeSize` limited by
`AvailableDiskSpace`. Also the --size=max parameter is restricted
from hidden volume creation as there is no way to determine a good
size as we do not mount the outer volume through the CLI process
flow to determine available free space on the outer volume.
|
|
from granting permissions to itself.
|
|
Signed-off-by: RoboSchmied <github@roboschmie.de>
|
|
|
|
Fixes the order of the language strings to match the correct
exceptions.
|
|
We don't need to reset PIM in PasswordException as it is immediately
fell back to if PIM is specified in text mode, but password is not.
This causes an exception that resets the PIM when it shouldn't.
|
|
Currently we check whether the system has DISPLAY environment variable
set, which is the case in a system that uses X11 natively or XWayland.
This variable is not set in a system with only Wayland, so we need to
also check whether WAYLAND_DISPLAY is set.
Fixes: #184
|
|
|
|
(#1274)
* macOS: Fix issue where PIM box has no width in Wizard
VolumePimTextCtrl has a problem with width on macOS which
we can fix by adjusting the proportions of the elements
inside the PimSizer, which seems like a better solution than
using a forced minimum size in pixels.
Adjacent, simplifies the validator logic for digits in PIM field.
Fixes #1219
|
|
|
