crash-context-0.6.0

Changed

• PR#70 removed the winapi dependency in favor of embedded bindings to avoid dependencies.
• PR#70 removed the asm implementations for Windows CPU context retrieval in favor of using RtlCaptureContext. This means that floating state is not captured, but is otherwise and improvement.

Added

• PR#68 added capture context support for x86 Windows, but this change was supplanted in PR#70 to use RtlCaptureContext instead.

Fixed

• PR#71 fixed the definition of mcontext_t for i686-unknow-linux. Thanks @afranchuk!

minidumper-0.7.0

Fixed

• PR#63 replaced windows-sys with winapi, resolving #61. Eventually winapi will be removed as well.

crash-handler-0.5.1

Fixed

• PR#64 fixed compilation for aarch64-linux-android. Additional targets were added to CI so they get caught before release.

crash-handler-0.5.0

Added

• PR#62 and PR#63 added support for handling SIGABRT on Windows.

Fixed

• PR#62 changed from using RtlCaptureContext on Windows to using crash_context::capture_context. This implementation fixes a crash issue due to winapi and windows-sys having improper bindings.

Changed

• PR#62 changed from using RtlCaptureContext on Windows to using crash_context::capture_context. This implementation additionally captures floating point and vector state on x86_64 unlike RtlCaptureContext.

crash-context-0.5.1

Fixed

• PR#66 (apparently) resolved #65 by...changing from AT&T to Intel syntax. This shouldn't have changed anything, but it did, and I'm too tired and have other things to work on, so here we are.

crash-context-0.5.0

Added

• PR#62 added a replacement implementation of RtlCaptureContext, and defines its own CONTEXT structure. This was needed due to both winapi and windows-sys incorrectly defining the CONTEXT and related structures on both x86_64 and aarch64 to not be correctly aligned, as well as RtlCaptureContext not actually capturing the floating point and vector state.

minidumper-0.6.0

Changed

• PR#60 updated minidump-writer -> 0.5 and windows-sys -> 0.42.

crash-handler-0.4.0

Added

• PR#60 resolved #59 by adding support for PR_SET_PTRACER before invoking the user callback, ensuring that an external process has permissions to perform ptrace operations on the crashing process, even if /proc/sys/kernel/yama/ptrace_scope is set to restricted (1), as this is the default for most newer distributions.

Changed

• PR#60 bumped windows-sys to 0.42.

sadness-generator-0.4.2

Changed

• PR#52 changed the address used to produce a sigsegv to be a canonical rather than a non-canonical address on 64-bit architectures. We will eventually add back support for non-canonical addresses.

minidumper-0.5.1

Fixed

• PR#58 fixed an issue on Linux where the socket poll would erroneously return an error if it was interrupted. This is now handled gracefully. Thanks @MarcusGrass!