From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
To: Lukasz Majewski <lukma@denx.de>
Cc: GNU C Library <libc-alpha@sourceware.org>
Subject: Re: [RFC PATCH] tst: Add test for gai_suspend
Date: Tue, 23 Mar 2021 10:09:58 -0300 [thread overview]
Message-ID: <e1edeb32-6fc6-7159-e09c-c3fb7cb88e8a@linaro.org> (raw)
In-Reply-To: <20210312170934.2b29cc9d@jawa>
On 12/03/2021 13:09, Lukasz Majewski wrote:
> 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
I am not sure from where the failure come from, on my environment I am
not seeing the SEGFAULT failure on neither x86_64 nor on i686.
But I am seeing:
$ ./testrun.sh resolv/tst-gai_suspend
tst-gai_suspend.c:47: numeric comparison failure
left: 0 (0x0); from: ret
right: -3 (0xfffffffd); from: EAI_AGAIN
tst-gai_suspend.c:48: 1616504556.995728406s not after
1616504557.971222237s (difference 0.975493831s)
error: 2 test failures
The issue is tests is using the system NSS and possible DNS cache so the
expected name resolution timeout does not happen reliable. I am not
sure how we can accomplish it, maybe through running it on a container
with a custom NSS module to actually emulate the delay.
next prev parent reply other threads:[~2021-03-23 13:10 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-12 15:49 Lukasz Majewski
2021-03-12 16:09 ` Lukasz Majewski
2021-03-12 16:41 ` Andreas Schwab
2021-03-23 13:09 ` Adhemerval Zanella [this message]
2021-03-23 15:23 ` Lukasz Majewski
2021-03-26 11:30 ` Lukasz Majewski
2021-05-06 9:11 ` Lukasz Majewski
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e1edeb32-6fc6-7159-e09c-c3fb7cb88e8a@linaro.org \
--to=adhemerval.zanella@linaro.org \
--cc=libc-alpha@sourceware.org \
--cc=lukma@denx.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).