On (13/01/15 11:14), Jakub Hrozek wrote:
On Fri, Jan 09, 2015 at 05:32:41PM +0100, Lukas Slebodnik wrote:
> On (09/01/15 15:18), Pavel Březina wrote:
> >On 01/09/2015 03:17 PM, Pavel Březina wrote:
> >>If OR is used in assert macro we never know what part of the condition
> >>was not met from a failure message directly.
> >>
> >>I'm seeing a failure in server-tests, because pid number of sssd process
> >>is greater than 0xffff, it is bigger than 100k. Where does this constant
> >>come from?
> cat /proc/sys/kernel/pid_max
>
> On my system i is 0x8000
> echo "obase=16; `cat /proc/sys/kernel/pid_max`" | bc
>
> >>
> >>[ RUN ] test_run_as_root_daemon
> >>tmp > 0xFFFF
> >>/home/pbrezina/workspace/sssd/src/tests/cwrap/test_server.c:79: error:
> >>Failure
> >
> >And now with the patch.
> >
>
> >From 7d5912f3206205dc24537f3640e9a8701995ceaf Mon Sep 17 00:00:00 2001
> >From: =?UTF-8?q?Pavel=20B=C5=99ezina?= <pbrezina(a)redhat.com>
> >Date: Fri, 9 Jan 2015 15:12:31 +0100
> >Subject: [PATCH] server-tests: do not use complex condition in assert
> >
> >If OR is used in assert macro we never know what part of the
> >condition was not met from a failure message directly.
> >---
> > src/tests/cwrap/test_server.c | 4 +++-
> > 1 file changed, 3 insertions(+), 1 deletion(-)
> >
> >diff --git a/src/tests/cwrap/test_server.c b/src/tests/cwrap/test_server.c
> >index
d0aeac47d0b067fdbc3399037c0a74f150337a23..560d73330f8caa5ba9706d13bae296594bb2841f 100644
> >--- a/src/tests/cwrap/test_server.c
> >+++ b/src/tests/cwrap/test_server.c
> >@@ -75,7 +75,9 @@ static void wait_for_bg_server(const char *pidfile)
> > buf[sizeof(buf) - 1] = '\0';
> >
> > tmp = strtol(buf, NULL, 10);
> >- assert_false(tmp == 0 || tmp > 0xFFFF || errno == ERANGE);
> >+ assert_false(tmp == 0);
> >+ assert_false(tmp > 0xFFFF);
> ^^^^^^
> I think we should not rely on such constant
> because you can set the value higher
> (up to 2^22 on 32-bit machines: 4,194,304) with:
> echo 4194303 > /proc/sys/kernel/pid_max
>
> LS
git blame points to me and IIRC I took part of these tests from
uid_wrapper code without too much thinking about them.
I think we can just remove the 0xFFFF part completely.
I agree.
or we can use our function strtouint32 to convert number instead of strtol.
There is assumption sizeof(pid_t) == sizeof(uint32_t)
LS