On 5/27/20 10:55 AM, Bruce Dubbs wrote:
On 5/27/20 8:11 AM, Karel Zak wrote:
On Tue, May 26, 2020 at 05:34:13PM -0500, Bruce Dubbs wrote:
I'm getting three test failures and hope I can get someone to help.
The test failures are ipcs/mk-rm-msg, ipcs/mk-rm-sem, and
ipcs/mk-rm-shm.
Concentrating on mk-rm-msg, I found on the system:
./tests/output/ipcs/mk-rm-msg
./tests/ts/ipcs/mk-rm-msg
./tests/expected/ipcs/mk-rm-msg
The contents of the output and the expected files are identical:
Message queue id: <was_number>
1
Message queue id: <was_number>
I am running the tests as a non-privileged user.
If I run the commands:
$ ipcmk -Q (returns 24)
$ ipcs -q -i 24
$ ipcrm -q 24
$ echo $?
0
Everything looks OK. Does anyone have an idea why this test (these
tests)
are failing?
I had no time to play with it yet, but according to feedback from
https://github.com/karelzak/util-linux/issues/1042 there is some kernel
regression:
The ipcs tests are failing due to a regression in kernels 5.6.8
through 5.6.13. Or mainline commits between
89163f93c6f969da5811af5377cc10173583123b and
5e698222c70257d13ae0816720dde57c56f81e15 (apparently including your
5.7.0-rc0).
Thanks. I'll try 5.6.14 and report back.
Indeed updating to kernel 5.6.14 fixed the ipcs test issues. Thanks for
the help.
Additionally I'd like to report one minor warning when building with gcc10:
login-utils/logindefs.c: In function ‘get_hushlogin_status’:
login-utils/logindefs.c:370:20: warning: ‘%s’ directive writing likely
15 or more bytes into a region of size between 9 and 8191
[-Wformat-overflow=]
370 | sprintf(buf, "%s/%s", pwd->pw_dir, file);
| ^~
login-utils/logindefs.c:370:16: note: assuming directive output of 15 bytes
370 | sprintf(buf, "%s/%s", pwd->pw_dir, file);
| ^~~~~~~
login-utils/logindefs.c:370:3: note: ‘sprintf’ output 2 or more bytes
(assuming 8199) into a destination of size 8192
370 | sprintf(buf, "%s/%s", pwd->pw_dir, file);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-- Bruce
