| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
| |
tv_sub() had a bug: if 'a' represented the same time as 'b' then it
added extra 1000000 usecs.
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Update documentation with configuration file details.
Update help output with missing help for '-N' option.
|
|
|
|
| |
using proper %zu for size_t argument in printf
|
| |
|
| |
|
|
|
|
|
| |
Add ARGS_STRING_POINTER_LENGTH option. It allows to display
pointer value and string length for "char *" arguments.
|
|
|
|
|
|
| |
Added automated tests for most of the LIBS*/SYM* options.
Missing tests for options:
LIBS_TO, LIBS_FROM, SYM_BELOW, SYM_NOEXIT
|
|
|
|
|
|
|
|
|
| |
Added SYM_NOEXIT option to be able to run only pltentry
callback for specified symbols.
This option is needed for setjmp symbol, otherwise the
tracee segfaults. The "_setjmp" symbol is added as default
value to the global config file.
|
|
|
|
|
|
|
|
| |
Added following config file options:
LIBS, LIBS_TO, LIBS_FROM, SYM, SYM_OMIT, SYM_BELOW
They match their command line equivalents.
|
|
|
|
|
| |
Moving test-args/test-common into scripts, to have common
way of running and cheking tests results.
|
| |
|
|
|
|
|
|
| |
fix global_symbols tree management
- add only if we care about the symbol
- make bind call dependent on the global_symbols variable
|
| |
|
|
|
|
|
| |
it makes more sense to read the -N config file immediatelly within
the arguments processing, since more confing files can be added
|
|
|
|
|
| |
if the tracee died we break the loop leaving the tracee
socket(s) full of data, caused by recent code refactoring
|
|
|
|
|
| |
- moved specific cleanup into src/Makefile
- added OBJS_DEPS_OMIT variable to ease DEPS_OBJS generation
|
|
|
|
|
| |
moving tty fd into the config struct, since it's better fit
and it's needed for error simulation, which is comming in shortly
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As per David Gilbert's email:
On library functions like strcpy() where there is a destination
string, latrace -A ends up trying to print out the
destination buffer which may be completely invalid.
Maybe it's worth doing the same trick that ltrace does, and just
declare the destination parameters as void*
until we handle bad characters somehow, replacing
destination char pointers with void pointers
|
| |
|
|
|
|
| |
+ refactoring lt_run to check the latrace got killed
|
|
|
|
| |
contributed by Dr. David Alan Gilbert <david.gilbert@linaro.org>
|
|
|
|
|
|
|
| |
- separated 32/64 into single objects that will get included
by each arch
- this way we can choose the appropriate test version for
each architecture
|
|
|
|
|
|
|
| |
- removed the complex math from test_[short|int]_* functions
and return simple numbers, it's easier to debug
- kept and simplified math for test_long
|
|
|
|
|
|
|
| |
+ force test.sh to use bash
+ fix test_char tests
contributed by Dr. David Alan Gilbert <david.gilbert@linaro.org>
|
|
|
|
|
|
|
|
| |
cppcheck discovered some memory leaks
(http://cppcheck.sourceforge.net/)
It'd be worth adding Makefile target to run/check
cppcheck output automatically.
|
|
|
|
|
|
| |
recognize arm properly
contributed by Dr. David Alan Gilbert <david.gilbert@linaro.org>
|
|
|
|
|
|
|
|
|
|
| |
- added OUTPUT_TTY config file option
it is possible to catch traced program tty output and
storing it to the configured file (0/1/2 descriptors
are monitored)
- added automated test for the option
- refactoring process method a bit
- disabling connection between -R and -q options
|
|
|
|
|
|
|
| |
So far for x86 and x86_64 others are disabled.
From this time on, I'll try to force addition of automated
test for each fix/feature.. ;)
|
| |
|
|
|
|
| |
so the notification is not affected by other files
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- separating bison/flex functions for args and config
- the "include file support" unified among new conf and C header
parsing
- support for following options:
HEADERS
INDENT_SYM
PIPE
TIMESTAMP
FRAMESIZE
FRAMESIZE_CHECK
HIDE_TID
FOLLOW_FORK
FOLLOW_EXEC
DEMANGLE
BRACES
ENABLE_ARGS
DETAIL_ARGS
|
| |
|
| |
|
| |
|
|
|
|
| |
fix LDFLAGS/CFLAGS usage
|
|
|
|
|
|
| |
one global tree to rule them all
- only one tree is searched during the plt entry/exit
- symbols are added during the bind audit callback
|
|
|
|
| |
Artur Skawina <art.08.09@gmail.com>
|
| |
|
|
|
|
| |
add '-n' option, allowing to omit tracing certain symbols
|
|
|
|
| |
more sophisticated parsing of /proc/self/maps
|
| |
|
|
|
|
|
|
| |
based on Akos' comments:
- dynamic rlimit stack check
- recognize [stack] map arrea and handle it properly
|
| |
|