linux-kernel - Re: [PATCH 1/2] selftests/nolibc: add new gettimeofday test cases

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <ZHXmbNg4BdJNd4cO@1wt.eu>
Date:   Tue, 30 May 2023 14:05:00 +0200
From:   Willy Tarreau <w@....eu>
To:     Thomas Weißschuh <thomas@...ch.de>
Cc:     Zhangjin Wu <falcon@...ylab.org>, arnd@...db.de,
        linux-kernel@...r.kernel.org, linux-kselftest@...r.kernel.org,
        linux-riscv@...ts.infradead.org
Subject: Re: [PATCH 1/2] selftests/nolibc: add new gettimeofday test cases

On Tue, May 30, 2023 at 12:59:31PM +0200, Thomas Weißschuh wrote:
> On 2023-05-30 14:37:49+0800, Zhangjin Wu wrote:
> > These 3 test cases are added to cover the normal using scenes of
> > gettimeofday().
> > 
> > They have been used to trigger and fix up such issue:
> > 
> >     nolibc-test.c:(.text.gettimeofday+0x54): undefined reference to `__aeabi_ldivmod'
> > 
> > This issue happens while there is no "unsigned int" conversion in the
> > new clock_gettime / clock_gettime64 syscall path of gettimeofday():
> > 
> >     tv->tv_usec = ts.tv_nsec / 1000;
> > 
> > Signed-off-by: Zhangjin Wu <falcon@...ylab.org>
> > ---
> >  tools/testing/selftests/nolibc/nolibc-test.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> > 
> > diff --git a/tools/testing/selftests/nolibc/nolibc-test.c b/tools/testing/selftests/nolibc/nolibc-test.c
> > index 8ba8c2fc71a0..20d184da9a2b 100644
> > --- a/tools/testing/selftests/nolibc/nolibc-test.c
> > +++ b/tools/testing/selftests/nolibc/nolibc-test.c
> > @@ -533,6 +533,8 @@ static int test_stat_timestamps(void)
> >   */
> >  int run_syscall(int min, int max)
> >  {
> > +	struct timeval tv;
> > +	struct timezone tz;
> >  	struct stat stat_buf;
> >  	int euid0;
> >  	int proc;
> > @@ -588,6 +590,9 @@ int run_syscall(int min, int max)
> >  		CASE_TEST(getdents64_root);   EXPECT_SYSNE(1, test_getdents64("/"), -1); break;
> >  		CASE_TEST(getdents64_null);   EXPECT_SYSER(1, test_getdents64("/dev/null"), -1, ENOTDIR); break;
> >  		CASE_TEST(gettimeofday_null); EXPECT_SYSZR(1, gettimeofday(NULL, NULL)); break;
> > +		CASE_TEST(gettimeofday_tv);   EXPECT_SYSZR(1, gettimeofday(&tv, NULL)); break;
> > +		CASE_TEST(gettimeofday_tz);   EXPECT_SYSZR(1, gettimeofday(NULL, &tz)); break;
> 
> Calling gettimeofday(NULL, ...) will actually segfault on glibc.
> It works when calling through the VDSO, but not the logic in glibc
> itself, which is guess is allowed by POSIX.

Then that's shocking, because the man page says:

       If either tv or tz is NULL, the corresponding structure is not  set  or
       returned.   (However, compilation warnings will result if tv is NULL.)

I'd expect glibc to at least support what'd documented in the man
page :-/

> I propose to avoid doing it :-)

If you're certain that's the case, then I agree.

> Either we gate the existing test in #ifdef NOLIBC or we remove it.

Better not keep tests specific to nolibc if they aim at verifying some
compatibliity.

Willy