| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
| |
to detect issues such as #958
|
| |
|
|
| |
use `cmake --build` instead of `make` directly
|
| |
|
|
|
|
|
| |
on top of providing a central definition place, which eases maintenance,
it might also help for #1021.
Also : updated doc
|
| |
|
|
|
|
|
|
|
|
|
|
| |
To support older compiler which doesn't have explicit dialect option
for C90 (gcc-4.4), this change set split "c_standards" into multiple
part.
Original "make c_standards" still works as intended. But this change
gives extra freedom of choice for external program For example, CI
can choose test for standards which is supported by specimen compiler.
With this separation, we can also introduce C17 smoothly.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The following command doesn't work as intended
```
cd
git clone https://github.com/lz4/lz4.git
cd lz4
make usan MOREFLAGS='-Wcomma -Werror'
```
```
Cleaning completed
cc: error: unrecognized command line option ‘-Wcomma’; did you mean ‘-Wcomment’?
make[3]: *** [<builtin>: ../lib/lz4.o] Error 1
make[2]: *** [Makefile:65: lz4] Error 2
make[1]: *** [Makefile:133: test] Error 2
make: *** [Makefile:158: usan] Error 2
```
Because the following part of the `Makefile` doesn't propagate `CC`, `CFLAGS` and `LDFLAGS` to child `$(MAKE)` properly.
```
.PHONY: usan
usan: CC = clang
usan: CFLAGS = -O3 -g -fsanitize=undefined -fno-sanitize-recover=undefined -fsanitize-recover=pointer-overflow
usan: LDFLAGS = $(CFLAGS)
usan: clean
$(MAKE) test FUZZER_TIME="-T30s" NB_LOOPS=-i1
```
We need explicit propagation for child process
```
- $(MAKE) test FUZZER_TIME="-T30s" NB_LOOPS=-i1
+ CC=$(CC) CFLAGS=$(CFLAGS) LDFLAGS=$(LDFLAGS) $(MAKE) test FUZZER_TIME="-T30s" NB_LOOPS=-i1
```
After that, `make usan` works and it shows expected runtime error.
```
$ make usan MOREFLAGS='-Wcomma -Werror'
Cleaning completed
../lib/lz4frame.c:907:25: runtime error: applying non-zero offset 65536 to null pointer
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../lib/lz4frame.c:907:25 in
../lib/lz4frame.c:907:58: runtime error: applying zero offset to null pointer
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../lib/lz4frame.c:907:58 in
...
```
Please note that `make usan` is working at travis-ci. Because `.travis.yml` has the following [explicit `compiler:` setting](https://github.com/lz4/lz4/blob/7a966c1511816b53ac93aa2f2a2ff97e036a4a60/.travis.yml#L66).
```
- name: (Trusty) USan test
dist: trusty
compiler: clang
script:
- make usan MOREFLAGS=-Wcomma -Werror
```
|
| | |
|
| |
|
|
| |
so now, `%p` _requires_ a `void*` pointer ?
|
| |
|
|
| |
seems to require explicit environment variables
|
| | |
|
| |
|
|
|
| |
remove usage of include Makefile.inc in too Makefile
as it seems to somehow unexport CFLAGS ...
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
the way `base` is used must allow 2-complement address overflow.
`base` is effectively a virtual pointer, never dereferenced directly.
What matters is `base + index`.
This is the address that will be dereference and must be valid (it is properly validated).
|
| |
|
|
|
|
| |
which now fails with a clear error as intended
(not just print a status and move on).
should be reproduced on travisCI
|
| |
|
|
| |
Fixes #852.
|
| | |
|
| | |
|
| |
|
|
|
| |
- only play listTest with `make test`, not `make all` which is limited to build
- update `clangtest`, so that it's possible to disable O3 optimization, for faster processing
|
| | |
|
| |
|
|
| |
"install" in one place to try to isolate it.
|
| | |
|
| |
|
|
|
|
|
|
| |
- promoted LZ4_resetStreamHC_fast() to stable
- moved LZ4_resetStreamHC() to deprecated (but do not generate a warning yet)
- Updated doc, to highlight difference between init and reset
- switched all invocations of LZ4_resetStreamHC() onto LZ4_initStreamHC()
- misc: ensure `make all` also builds /tests
|
| | |
|
| | |
|
| |
|
|
|
|
| |
as Makefile target and Travis CI test.
Fixed last cppcheck warnings in tests and examples
|
| |
|
|
|
|
|
| |
as detected in #485 by @JoachimSchneider.
Refactored the c_standards tests
so that these issues get automatically detected in CI tests.
|
| | |
|
| | |
|
| |
|
|
| |
lz4 1.8.2 works fine on Haiku and passes all tests.
|
| | |
|
| |
|
|
| |
to catch `//` comments
|
| |
|
|
|
|
|
|
|
| |
`make all` will trigger several sub-directory makefiles.
several of them need `liblz4.a`.
When built with `-j#`, there are several concurrent liblz4.a built
Make liblz4.a a dependency, which is built once,
before moving to sub-directory Makefiles
|
| |
|
|
|
|
|
|
|
|
| |
according to GNU Makefile conventions,
the Makefile should feature a make check target
to self-test the generated program:
https://www.gnu.org/prep/standards/html_node/Standard-Targets.html .
this is much less thorough and less taxing than `make test`,
and can be run on any target in a reasonable timeframe (several seconds).
|
| |
|
|
| |
static library *.a must come after source files *.c on linux
|
| |\ |
|
| | | |
|
| |/
|
|
| |
to better reflect LZ4F API usage.
|
| | |
|
| |
|
|
| |
since it's not needed.
|
| |
|
|
|
|
|
|
|
| |
instead of selected at compilation time depending on a macro.
This design makes it possible to have a single binary which supports both modes.
An advantageous side effect is that when doing `make; make install`
no additional binary is created during `make install`
(it used to create `lz4c`, because `make` would only build `lz4`)
|
| |
|
|
|
|
| |
supports lowercase directory variables
add an "Installation" section in README.md
added an INSTALL file
|
| |
|
|
| |
and added entry "make list"
|
| |
|
|
|
| |
to automatically build manual files with
make all
|
| |\
| |
| | |
Dev
|
| | | |
|
| |/ |
|
| | |
|
| | |
|
