| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The improvements are:
- use most of the default "Makefile" for standard packaging;
- also use the default ".gitignore" with for the derived assets;
- don't impose so many $CFLAGS on the user. GCC still needs to be given
the `-ffreestanding` flag explicitly for us to get a good binary;
- stop using ad-hoc tools/* scripts, and avoid the code-generation
anti-pattern overall on the build. Some of the generated files were
checked-in, and some were removed;
- remove empty files;
- use POSIX make(1) over gmake;
- add fuzz targets;
- partial "install" and "uninstall" targets;
- complete "clean" target.
The shortcomings are:
- only working on x86_64. More platforms coming soon;
- code is still messy: way too many warnings, GNU/BSD specific
extensions, inline assembly, and all kinds of unportable code;
- still only works with GCC and GCC-like compilers, and completly fails
with tcc(1) and cproc(1);
- the `deps.mk` file is being maintained manually. As I work on the
source files I'll finish automating its generation with `mkdeps.sh`;
- still seems to be coupled with Linux;
- still is missing tests setup;
- still uses `#include <$NAME.h>` instead of the correct
`#include "$NAME.h"` form.
The generated libgrovel.a did match the previous lib/libc.a 100%.
|
|
|
this patch improves the correctness, simplicity, and size of
cancellation-related code. modulo any small errors, it should now be
completely conformant, safe, and resource-leak free.
the notion of entering and exiting cancellation-point context has been
completely eliminated and replaced with alternative syscall assembly
code for cancellable syscalls. the assembly is responsible for setting
up execution context information (stack pointer and address of the
syscall instruction) which the cancellation signal handler can use to
determine whether the interrupted code was in a cancellable state.
these changes eliminate race conditions in the previous generation of
cancellation handling code (whereby a cancellation request received
just prior to the syscall would not be processed, leaving the syscall
to block, potentially indefinitely), and remedy an issue where
non-cancellable syscalls made from signal handlers became cancellable
if the signal handler interrupted a cancellation point.
x86_64 asm is untested and may need a second try to get it right.
|
