cmd: add support for building with address and undefined behavior sanitizers #15059
Conversation
|
Tue Feb 18 12:09:09 UTC 2025 Spread tests skipped |
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #15059 +/- ##
==========================================
+ Coverage 78.07% 78.13% +0.06%
==========================================
Files 1182 1172 -10
Lines 157743 157622 -121
==========================================
+ Hits 123154 123156 +2
+ Misses 26943 26823 -120
+ Partials 7646 7643 -3
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
bboozzoo
force-pushed
the
bboozzoo/asan
branch
from
60354fc to
d5010ec
Compare
bboozzoo
force-pushed
the
bboozzoo/asan
branch
from
d5010ec to
a1b04d7
Compare
| @@ -319,6 +319,20 @@ AS_IF([test "x$with_unit_tests" = "xyes"], [ | |||
| AX_APPEND_COMPILE_FLAGS([-Werror], [CHECK_CFLAGS]) | |||
| ]) | |||
|
|
|||
| AC_ARG_ENABLE([sanitize], | |||
There was a problem hiding this comment.
Since this is only using asan at the moment, would it be better to just call it asan rather than sanitize?
Do you think it makes sense to add ubsan too, even if for unit tests only? |
Sure, the more the merrier particularly if its just the unit tests. |
bboozzoo
force-pushed
the
bboozzoo/asan
branch
from
06d8368 to
12c286f
Compare
bboozzoo
changed the title
bboozzoo
force-pushed
the
bboozzoo/asan
branch
from
12c286f to
1b967cc
Compare
| endif | ||
|
|
||
| ## | ||
| ## snap-gdb-shim | ||
| ## | ||
|
|
||
| if !ENABLE_SANITIZE |
There was a problem hiding this comment.
I suppose you do not add SANITIZE_* to snap_gdb_shim_snap_gdb_shim_*. So that would still build, right? Is it just to disable what we are not testing here?
There was a problem hiding this comment.
If it is so, we should maybe document in the AS_HELP_STRING of --enable-sanitize that it gives a partial build.
There was a problem hiding this comment.
Bit of both. We don't test it and it won't build. Even with -static-libasan dlopen() is retained for some reason. I guess it just makes libasan parts to be linked statically, but still relies on hijacking libc's malloc dynamically.
There was a problem hiding this comment.
I've tweaked the help message.
Add support for building C code with address sanitizer (-fsanitize=address). Signed-off-by: Maciej Borzecki <[email protected]>
Signed-off-by: Maciej Borzecki <[email protected]>
Signed-off-by: Maciej Borzecki <[email protected]>
Signed-off-by: Maciej Borzecki <[email protected]>
Signed-off-by: Maciej Borzecki <[email protected]>
Signed-off-by: Maciej Borzecki <[email protected]>
bboozzoo
force-pushed
the
bboozzoo/asan
branch
from
1b967cc to
76dfa04
Compare
The branch builds on #15045 and adds support for building with asan and ubsan. This is only useful for testing local installation or running unit tests, as otherwise we'd have to ship libasan/libubsan in the snapd snap to make it universally useful in all scenarios.
Note, some binaries which are linked statically,
snap-gdb*-shimspecifically, had to be skipped, since libasan relies on dlopen().