[LTP] [PATCH] syscalls: add rt_tgsigqueueinfo() test-case

[Sumit Garg]

Thanks Steve for the review.

On Tue, 5 Mar 2019 at 01:06, Steve Muckle <smuckle@google.com> wrote:
>
> Hi Sumit,
>
> On 02/25/2019 04:32 AM, Sumit Garg wrote:
> > This tests the rt_tgsigqueueinfo() syscall used to queue a signal and
> > data to the single thread specified by the combination of tgid, a thread
> > group ID, and tid, a thread in that thread group.
> >
> > Signed-off-by: Sumit Garg <sumit.garg@linaro.org>
> > ---
> >   runtest/syscalls                                   |  1 +
> >   .../kernel/syscalls/rt_tgsigqueueinfo/.gitignore   |  1 +
> >   .../kernel/syscalls/rt_tgsigqueueinfo/Makefile     | 10 ++++
> >   .../rt_tgsigqueueinfo/rt_tgsigqueueinfo01.c        | 53 ++++++++++++++++++++++
> >   4 files changed, 65 insertions(+)
> >   create mode 100644 testcases/kernel/syscalls/rt_tgsigqueueinfo/.gitignore
> >   create mode 100644 testcases/kernel/syscalls/rt_tgsigqueueinfo/Makefile
> >   create mode 100644 testcases/kernel/syscalls/rt_tgsigqueueinfo/rt_tgsigqueueinfo01.c
> >
> > diff --git a/runtest/syscalls b/runtest/syscalls
> > index 70d3561..861c0a6 100644
> > --- a/runtest/syscalls
> > +++ b/runtest/syscalls
> > @@ -996,6 +996,7 @@ rt_sigprocmask01 rt_sigprocmask01
> >   rt_sigprocmask02 rt_sigprocmask02
> >   rt_sigqueueinfo01 rt_sigqueueinfo01
> >   rt_sigsuspend01 rt_sigsuspend01
> > +rt_tgsigqueueinfo01 rt_tgsigqueueinfo01
> >
> >   sbrk01 sbrk01
> >   sbrk02 sbrk02
> > diff --git a/testcases/kernel/syscalls/rt_tgsigqueueinfo/.gitignore b/testcases/kernel/syscalls/rt_tgsigqueueinfo/.gitignore
> > new file mode 100644
> > index 0000000..f9ffa58
> > --- /dev/null
> > +++ b/testcases/kernel/syscalls/rt_tgsigqueueinfo/.gitignore
> > @@ -0,0 +1 @@
> > +rt_tgsigqueueinfo01
> > diff --git a/testcases/kernel/syscalls/rt_tgsigqueueinfo/Makefile b/testcases/kernel/syscalls/rt_tgsigqueueinfo/Makefile
> > new file mode 100644
> > index 0000000..035ca64
> > --- /dev/null
> > +++ b/testcases/kernel/syscalls/rt_tgsigqueueinfo/Makefile
> > @@ -0,0 +1,10 @@
> > +# Copyright (c) 2019 - Linaro Limited. All rights reserved.
> > +# SPDX-License-Identifier: GPL-2.0-or-later
> > +
> > +top_srcdir             ?= ../../../..
> > +
> > +rt_tgsigqueueinfo01: CFLAGS+=-pthread
> > +
> > +include $(top_srcdir)/include/mk/testcases.mk
> > +
> > +include $(top_srcdir)/include/mk/generic_leaf_target.mk
> > diff --git a/testcases/kernel/syscalls/rt_tgsigqueueinfo/rt_tgsigqueueinfo01.c b/testcases/kernel/syscalls/rt_tgsigqueueinfo/rt_tgsigqueueinfo01.c
> > new file mode 100644
> > index 0000000..336b968
> > --- /dev/null
> > +++ b/testcases/kernel/syscalls/rt_tgsigqueueinfo/rt_tgsigqueueinfo01.c
> > @@ -0,0 +1,53 @@
> > +// SPDX-License-Identifier: GPL-2.0-or-later
> > +/*
> > + * Copyright (c) 2019 Linaro Limited. All rights reserved.
> > + * Author: Sumit Garg <sumit.garg@linaro.org>
> > + */
> > +
> > +/*
> > + * Test rt_tgsigqueueinfo
> > + *
> > + * This tests the rt_tgsigqueueinfo() syscall. It sends the signal and data
> > + * to the single thread specified by the combination of tgid, a thread group
> > + * ID, and tid, a thread in that thread group.
> > + */
> > +
> > +#define _GNU_SOURCE
> > +#include <err.h>
> > +#include <pthread.h>
> > +#include "tst_test.h"
> > +#include "lapi/syscalls.h"
> > +
> > +void *tfunc(void *arg LTP_ATTRIBUTE_UNUSED)
> > +{
> > +     siginfo_t uinfo;
> > +
> > +     uinfo.si_errno = 0;
> > +     uinfo.si_code = SI_QUEUE;
> > +
> > +     TEST(tst_syscall(__NR_rt_tgsigqueueinfo, getpid(),
> > +                      syscall(__NR_gettid), SIGCHLD, &uinfo));
>
> It looks like the pthread here is sending the signal to itself which
> seems like a special case. Not that it isn't worth testing but I'd think
> there should be a test which sends the signal to a different thread.
> This (the self-send case) could be an additional test.
>

Hmm. So do you suggest to add following cases?

1. parent -> thread
2. thread1 -> thread2

BTW, in above cases I think we need to communicate "tid" obtained via
gettid() to parent and other thread to queue signal which could be
done via "arg" parameter in pthread_create().

> > +     if (TST_RET)
> > +             tst_res(TFAIL, "Test Failed");
> > +     else
> > +             tst_res(TPASS, "Test Succeeded");
>
> I recommend verifying the signal and its data (si_value) are delivered
> successfully prior to test completion.
>

Ok, will add this verification via signal handler.

-Sumit

> > +
> > +     return NULL;
> > +}
> > +
> > +static void verify_rt_tgsigqueueinfo(void)
> > +{
> > +     pthread_t pt;
> > +
> > +     TEST(pthread_create(&pt, NULL, tfunc, NULL));
> > +     if (TST_RET)
> > +             tst_brk(TFAIL | TTERRNO, "pthread_create failed");
> > +
> > +     TEST(pthread_join(pt, NULL));
> > +     if (TST_RET)
> > +             tst_brk(TFAIL | TTERRNO, "pthread_join failed");
> > +}
> > +
> > +static struct tst_test test = {
> > +     .test_all = verify_rt_tgsigqueueinfo,
> > +};
> >
>
> thanks,
> steve
>