warning cleanup

[sergey-safarov]

ld_fld.c:273 implicit-function-declaration

CC (gcc) [M db2_ldap.so]	db2_ldap_mod.o
CC (gcc) [M db2_ldap.so]	ld_cfg.o
CC (gcc) [M db2_ldap.so]	ld_cmd.o
CC (gcc) [M db2_ldap.so]	ld_con.o
CC (gcc) [M db2_ldap.so]	ld_fld.o
ld_fld.c: In function 'ldap_gentime2db_datetime':
ld_fld.c:273:10: warning: implicit declaration of function 'timelocal' [-Wimplicit-function-declaration]
273 | *dst = timelocal(&time);
| ^~~~~~~~~
CC (gcc) [M db2_ldap.so]	ld_res.o
CC (gcc) [M db2_ldap.so]	ld_uri.o
make[3]: 'libsrdb2.so.1.0' is up to date.

redis_table.c:586 maybe-uninitialized

CC (gcc) [M db_redis.so]	db_redis_mod.o
CC (gcc) [M db_redis.so]	redis_connection.o
CC (gcc) [M db_redis.so]	redis_dbase.o
CC (gcc) [M db_redis.so]	redis_table.o
redis_table.c: In function 'db_redis_parse_keys':
redis_table.c:586:38: warning: 'table' may be used uninitialized in this function [-Wmaybe-uninitialized]
586 | table->types = type_target = type;
| ~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~
make[3]: 'libsrdb2.so.1.0' is up to date.
make[3]: 'libsrdb1.so.1.0' is up to date.
LD (gcc) [M db_redis.so]	db_redis.so

app_python strict-aliasing

CC (gcc) [M app_python.so]	app_python_mod.o
CC (gcc) [M app_python.so]	apy_kemi.o
In file included from /usr/include/python2.7/Python.h:83,
from apy_kemi.c:25:
apy_kemi.c: In function 'sr_kemi_apy_return_true':
/usr/include/python2.7/object.h:769:6: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
769 | ((PyObject*)(op))->ob_refcnt++)
| ~^~~~~~~~~~~~~~~~
apy_kemi.c:121:2: note: in expansion of macro 'Py_INCREF'
121 | Py_INCREF(Py_True);
| ^~~~~~~~~
apy_kemi.c: In function 'sr_kemi_apy_return_false':
/usr/include/python2.7/object.h:769:6: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
769 | ((PyObject*)(op))->ob_refcnt++)
| ~^~~~~~~~~~~~~~~~
apy_kemi.c:130:2: note: in expansion of macro 'Py_INCREF'
130 | Py_INCREF(Py_False);
| ^~~~~~~~~
CC (gcc) [M app_python.so]	apy_kemi_export.o
CC (gcc) [M app_python.so]	mod_Core.o

pdb.c:348 address-of-packed-member]

LD (gcc) [M p_usrloc.so]	p_usrloc.so
CC (gcc) [M path.so]	path.o
CC (gcc) [M path.so]	path_mod.o
LD (gcc) [M path.so]	path.so
CC (gcc) [M pdb.so]	pdb.o
pdb.c: In function 'pdb_query':
pdb.c:348:60: warning: taking address of packed member of 'struct pdb_msg' may result in an unaligned pointer value [-Waddress-of-packed-member]
348 | _id = (short int *)&(msg.bdy.payload[reqlen]); /* make gcc happy */
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
LD (gcc) [M pdb.so]	pdb.so
CC (gcc) [M pdt.so]	pdt.o
CC (gcc) [M pdt.so]	pdtree.o
make[3]: 'libsrdb1.so.1.0' is up to date.

[miconda]

I pushed patches for those warnings. The one of app_python is to disable the check, on the web people say python2 is known not to handle strict aliasing and being end-of-life, I guess is no chance to fix in the libs.

For pdb: @henningw, @lbalaceanu & @smititelu - you are likely the devs familiar with the code, can you check the commit 86ebb4e and see if it is an acceptable solution? The idea was to get the short int value from the two consecutive bytes in the char buffer, to avoid reading from an unaligned memory address when buflen is not round of 4 or 8. I see there is also conversion from network to host, so it is better you have a look, not to let this change go somehow unnoticed.

[miconda]

Closing, commits were also backported to 5.3.

[Earn330]

Hi

i had to revert commit 86ebb4e because the return value wasn't correct anymore. It seems it just
returns the pointer position but not the content itself. After reverting everything worked fine again.

BR,
Bjoern

[miconda]

@Earn330 - create a new bug report and reference this issue, so it does not get forgotten and we investigate it.