* Re: [RFC PATCH] tst: Add test for gai_suspend
2021-03-12 15:49 [RFC PATCH] tst: Add test for gai_suspend Lukasz Majewski
@ 2021-03-12 16:09 ` Lukasz Majewski
2021-03-12 16:41 ` Andreas Schwab
0 siblings, 1 reply; 3+ messages in thread
From: Lukasz Majewski @ 2021-03-12 16:09 UTC (permalink / raw)
To: Joseph Myers, Adhemerval Zanella, Florian Weimer, DJ Delorie
Cc: Paul Eggert, Alistair Francis, Arnd Bergmann, Alistair Francis,
GNU C Library, Carlos O'Donell, Florian Weimer,
Zack Weinberg, libc-help
[-- Attachment #1: Type: text/plain, Size: 6335 bytes --]
Dear Community,
> ---
> resolv/Makefile | 2 ++
> resolv/tst-gai_suspend.c | 53
> ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 55
> insertions(+) create mode 100644 resolv/tst-gai_suspend.c
>
> diff --git a/resolv/Makefile b/resolv/Makefile
> index 1047bb6ae5..e7e66fe903 100644
> --- a/resolv/Makefile
> +++ b/resolv/Makefile
> @@ -60,6 +60,7 @@ tests += \
> tst-resolv-res_init-multi \
> tst-resolv-search \
> tst-resolv-trailing \
> + tst-gai_suspend \
>
> # This test calls __res_context_send directly, which is not exported
> # from libresolv.
> @@ -209,3 +210,4 @@ $(objpfx)tst-ns_name_compress:
> $(objpfx)libresolv.so $(objpfx)tst-ns_name_pton: $(objpfx)libresolv.so
> $(objpfx)tst-res_hnok: $(objpfx)libresolv.so
> $(objpfx)tst-p_secstodate: $(objpfx)libresolv.so
> +$(objpfx)tst-gai_suspend: $(objpfx)libanl.so
> diff --git a/resolv/tst-gai_suspend.c b/resolv/tst-gai_suspend.c
> new file mode 100644
> index 0000000000..bbafd13fc4
> --- /dev/null
> +++ b/resolv/tst-gai_suspend.c
> @@ -0,0 +1,53 @@
> +/* Test for gai_suspend timeout
> + Copyright (C) 2021 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be
> useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#include <time.h>
> +#include <netdb.h>
> +#include <errno.h>
> +#include <string.h>
> +#include <support/check.h>
> +#include <support/xtime.h>
> +#include <support/timespec.h>
> +
> +static int
> +do_test (void)
> +{
> + struct timespec ts;
> + xclock_gettime (CLOCK_REALTIME, &ts);
> + struct timespec timeout = make_timespec (1, 0);
> + ts = timespec_add (ts, timeout);
> +
> + /* Ignore gaicb content - just wait for timeout. */
> + struct addrinfo result;
> +
> + struct gaicb gai = { "foo.baz", "0", NULL, &result };
> + const struct gaicb * const gai_s[] = { &gai };
> + struct gaicb *gai_p = &gai;
> +
> + int ret = getaddrinfo_a (GAI_NOWAIT, &gai_p, 1, NULL);
> + if (ret != 0)
> + FAIL_EXIT1 ("getaddrinfo_a failed: %m\n");
> +
> + ret = gai_suspend (gai_s, 1, &timeout);
> + TEST_COMPARE (ret, EAI_AGAIN);
> + TEST_TIMESPEC_NOW_OR_AFTER (CLOCK_REALTIME, ts);
> +
> + return 0;
> +}
> +
> +#include <support/test-driver.c>
I'm a bit puzzled with the above gai_suspend() test.
The idea for Y2038 test is to check if one can timeout when passed
crafted struct gaicb causes the gai_suspend() to exit with timeout.
This approach is similar to one from tst-ppoll.c
1. When I do run it by hand on the remove machine (from glibc test
suite):
--------------------------------------------------
rm -rf ./resolv/*
make PARALLELMFLAGS="-j8"
test-wrapper='/home/lukma/work/yocto/y2038/build/workspace/sources/y2038-glibc/scripts/cross-test-ssh.sh
--allow-time-setting root@192.168.7.2' subdirs=resolv check
I do see the following error:
-----------------------------
cat ./resolv/tst-gai_suspend.out
Didn't expect signal from child: got `Segmentation fault'
This is strange, as I've passed NULL to getaddrinfo_a as the last
parameter, which causes passing no signals to program running it.
What am I missing here? Has the test setup any strict policy to handle
LWPs ?
2. On the VM (armv7):
gai_suspend.c test program:
---------------------------
#define _GNU_SOURCE
#include <stdio.h>
#include <time.h>
#include <netdb.h>
int main (int argc, char **argv)
{
struct timespec ts, timeout, now;
clock_gettime (CLOCK_REALTIME, &ts);
timeout.tv_sec = ts.tv_sec + 3;
timeout.tv_nsec = ts.tv_nsec;
struct addrinfo result;
struct gaicb gai = { "foo.baz", "0", NULL, &result };
const struct gaicb * const gai_s[] = { &gai };
struct gaicb *gai_p = &gai;
int ret = getaddrinfo_a (GAI_NOWAIT, &gai_p, 1, NULL);
if (ret != 0)
printf("getaddrinfo_a error");
ret = gai_suspend (gai_s, 1, &timeout);
// According to man getaddrinfo_s I shall receive ret == -3
// when conversion from name to IP credentials is performed
if (ret != 0)
printf("gai_suspend error");
clock_gettime (CLOCK_REALTIME, &now);
printf("gai_suspend seconds diff: %llu\n", now.tv_sec -
ts.tv_sec);
return 0;
}
gcc -Wall -ggdb -D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64 -I/opt/include \
-I/opt/usr/include -L/opt/usr/lib -Wl,-rpath=/opt/lib \
-Wl,--dynamic-linker=/opt/lib/ld-linux.so.2 gai_suspend.c -o gs
-lanl \ -lpthread -static
The above line has some manual adjustments for linker, but this works
for other syscalls for some long time now - more info [1].
In this case the program quits without any issues with threads, but the
output and behavior of gai_suspend is not as expected (according at
least to the manual [2]). The return code from gai_suspend shall be
EAI_AGAIN and the result struct has some garbage in it.
The question(s):
================
1. Why I do experience thread error in glibc test suite?
2. Is the code for gai_suspend testing wrong (not much reference in
glibc), or do we have a bug in the the implementation?
Thanks in advance for any help/hints.
Links:
[1] - https://github.com/lmajewski/meta-y2038/
[2] - https://linux.die.net/man/3/gai_suspend
Best regards,
Lukasz Majewski
--
DENX Software Engineering GmbH, Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma@denx.de
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
^ permalink raw reply [flat|nested] 3+ messages in thread