[v2] acct: fix version check on big endian system
diff mbox series

Message ID 20191010081752.11334-1-pifang@redhat.com
State Accepted
Delegated to: Petr Vorel
Headers show
Series
  • [v2] acct: fix version check on big endian system
Related show

Commit Message

Ping Fang Oct. 10, 2019, 8:17 a.m. UTC
ac_version = ACCT_VERION | ACCT_BYTEORDER
which is 0x83 (131) on big endian system.

failure output:
tst_kconfig.c:62: INFO: Parsing kernel config
'/boot/config-4.18.0-147.el8.s390x'
tst_test.c:1118: INFO: Timeout per run is 0h 05m 00s
tst_kconfig.c:62: INFO: Parsing kernel config
'/boot/config-4.18.0-147.el8.s390x'
acct02.c:239: INFO: Verifying using 'struct acct_v3'
acct02.c:192: INFO: == entry 1 ==
acct02.c:147: INFO: ac_version != 3 (131)
acct02.c:192: INFO: == entry 2 ==
acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
acct02.c:133: INFO: ac_exitcode != 32768 (0)
acct02.c:141: INFO: ac_ppid != 34501 (34500)
acct02.c:147: INFO: ac_version != 3 (131)
acct02.c:182: FAIL: end of file reached

Signed-off-by: Ping Fang <pifang@redhat.com>
---
 include/lapi/acct.h                     | 5 +++++
 testcases/kernel/syscalls/acct/acct02.c | 2 +-
 2 files changed, 6 insertions(+), 1 deletion(-)

Comments

Li Wang Oct. 10, 2019, 8:33 a.m. UTC | #1
Ping Fang <pifang@redhat.com> wrote:

ac_version = ACCT_VERION | ACCT_BYTEORDER
> which is 0x83 (131) on big endian system.
>
> failure output:
> tst_kconfig.c:62: INFO: Parsing kernel config
> '/boot/config-4.18.0-147.el8.s390x'
> tst_test.c:1118: INFO: Timeout per run is 0h 05m 00s
> tst_kconfig.c:62: INFO: Parsing kernel config
> '/boot/config-4.18.0-147.el8.s390x'
> acct02.c:239: INFO: Verifying using 'struct acct_v3'
> acct02.c:192: INFO: == entry 1 ==
> acct02.c:147: INFO: ac_version != 3 (131)
> acct02.c:192: INFO: == entry 2 ==
> acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
> acct02.c:133: INFO: ac_exitcode != 32768 (0)
> acct02.c:141: INFO: ac_ppid != 34501 (34500)
> acct02.c:147: INFO: ac_version != 3 (131)
> acct02.c:182: FAIL: end of file reached
>

I'm wondering that is there any necessary to use do{}...while loop for
entry many times? Since the verify_acct() only check the dummy
program(acct02_helper) accounting struct. So in the error logs, we could
see some meaningless data(acct02) in entry 2.

    acct02.c:192: INFO: == entry 2 ==
    acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
    acct02.c:133: INFO: ac_exitcode != 32768 (0)
    acct02.c:141: INFO: ac_ppid != 34501 (34500)

This needs to be fixed in a separate patch.


> Signed-off-by: Ping Fang <pifang@redhat.com>
>

The patch itself looks good. Ack.
Li Wang Oct. 10, 2019, 9:13 a.m. UTC | #2
On Thu, Oct 10, 2019 at 4:33 PM Li Wang <liwan@redhat.com> wrote:

>
> Ping Fang <pifang@redhat.com> wrote:
>
> ac_version = ACCT_VERION | ACCT_BYTEORDER
>> which is 0x83 (131) on big endian system.
>>
>> failure output:
>> tst_kconfig.c:62: INFO: Parsing kernel config
>> '/boot/config-4.18.0-147.el8.s390x'
>> tst_test.c:1118: INFO: Timeout per run is 0h 05m 00s
>> tst_kconfig.c:62: INFO: Parsing kernel config
>> '/boot/config-4.18.0-147.el8.s390x'
>> acct02.c:239: INFO: Verifying using 'struct acct_v3'
>> acct02.c:192: INFO: == entry 1 ==
>> acct02.c:147: INFO: ac_version != 3 (131)
>> acct02.c:192: INFO: == entry 2 ==
>> acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
>> acct02.c:133: INFO: ac_exitcode != 32768 (0)
>> acct02.c:141: INFO: ac_ppid != 34501 (34500)
>> acct02.c:147: INFO: ac_version != 3 (131)
>> acct02.c:182: FAIL: end of file reached
>>
>
> I'm wondering that is there any necessary to use do{}...while loop for
> entry many times? Since the verify_acct() only check the dummy
> program(acct02_helper) accounting struct. So in the error logs, we could
> see some meaningless data(acct02) in entry 2.
>

It turns out I was wrong, if accounting is turned on, system records for
each terminating process but not only for that process, so we need to pick
up the correct acct_struct in acct_file via the while loop.

Sorry for making noise, I pull back my comments here.
Cyril Hrubis Oct. 10, 2019, 11:06 a.m. UTC | #3
Hi!
> ac_version = ACCT_VERION | ACCT_BYTEORDER
> which is 0x83 (131) on big endian system.
> 
> failure output:
> tst_kconfig.c:62: INFO: Parsing kernel config
> '/boot/config-4.18.0-147.el8.s390x'
> tst_test.c:1118: INFO: Timeout per run is 0h 05m 00s
> tst_kconfig.c:62: INFO: Parsing kernel config
> '/boot/config-4.18.0-147.el8.s390x'
> acct02.c:239: INFO: Verifying using 'struct acct_v3'
> acct02.c:192: INFO: == entry 1 ==
> acct02.c:147: INFO: ac_version != 3 (131)
> acct02.c:192: INFO: == entry 2 ==
> acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
> acct02.c:133: INFO: ac_exitcode != 32768 (0)
> acct02.c:141: INFO: ac_ppid != 34501 (34500)
> acct02.c:147: INFO: ac_version != 3 (131)
> acct02.c:182: FAIL: end of file reached

Looks good, acked.

Also please consider a patch to the man 5 acct, it says that the version
is always set to 3 for acct_v3 structure.
Petr Vorel Oct. 10, 2019, 11:26 a.m. UTC | #4
Hi Ping,

> ac_version = ACCT_VERION | ACCT_BYTEORDER
> which is 0x83 (131) on big endian system.

> failure output:
> tst_kconfig.c:62: INFO: Parsing kernel config
> '/boot/config-4.18.0-147.el8.s390x'
> tst_test.c:1118: INFO: Timeout per run is 0h 05m 00s
> tst_kconfig.c:62: INFO: Parsing kernel config
> '/boot/config-4.18.0-147.el8.s390x'
> acct02.c:239: INFO: Verifying using 'struct acct_v3'
> acct02.c:192: INFO: == entry 1 ==
> acct02.c:147: INFO: ac_version != 3 (131)
> acct02.c:192: INFO: == entry 2 ==
> acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
> acct02.c:133: INFO: ac_exitcode != 32768 (0)
> acct02.c:141: INFO: ac_ppid != 34501 (34500)
> acct02.c:147: INFO: ac_version != 3 (131)
> acct02.c:182: FAIL: end of file reached

> Signed-off-by: Ping Fang <pifang@redhat.com>
Reviewed-by: Petr Vorel <pvorel@suse.cz>

Thanks for fixing the issue.

Kind regards,
Petr
Petr Vorel Oct. 10, 2019, 11:36 a.m. UTC | #5
Hi Li, Ping, Cyril,

> On Thu, Oct 10, 2019 at 4:33 PM Li Wang <liwan@redhat.com> wrote:

> > Ping Fang <pifang@redhat.com> wrote:

> > ac_version = ACCT_VERION | ACCT_BYTEORDER
> >> which is 0x83 (131) on big endian system.

> >> failure output:
> >> tst_kconfig.c:62: INFO: Parsing kernel config
> >> '/boot/config-4.18.0-147.el8.s390x'
> >> tst_test.c:1118: INFO: Timeout per run is 0h 05m 00s
> >> tst_kconfig.c:62: INFO: Parsing kernel config
> >> '/boot/config-4.18.0-147.el8.s390x'
> >> acct02.c:239: INFO: Verifying using 'struct acct_v3'
> >> acct02.c:192: INFO: == entry 1 ==
> >> acct02.c:147: INFO: ac_version != 3 (131)
> >> acct02.c:192: INFO: == entry 2 ==
> >> acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
> >> acct02.c:133: INFO: ac_exitcode != 32768 (0)
> >> acct02.c:141: INFO: ac_ppid != 34501 (34500)
> >> acct02.c:147: INFO: ac_version != 3 (131)
> >> acct02.c:182: FAIL: end of file reached


> > I'm wondering that is there any necessary to use do{}...while loop for
> > entry many times? Since the verify_acct() only check the dummy
> > program(acct02_helper) accounting struct. So in the error logs, we could
> > see some meaningless data(acct02) in entry 2.


> It turns out I was wrong, if accounting is turned on, system records for
> each terminating process but not only for that process, so we need to pick
> up the correct acct_struct in acct_file via the while loop.

> Sorry for making noise, I pull back my comments here.

Thanks for a fix, merged.

Petr
Petr Vorel Oct. 10, 2019, 11:39 a.m. UTC | #6
Hi,

> > ac_version = ACCT_VERION | ACCT_BYTEORDER
> > which is 0x83 (131) on big endian system.

> > failure output:
> > tst_kconfig.c:62: INFO: Parsing kernel config
> > '/boot/config-4.18.0-147.el8.s390x'
> > tst_test.c:1118: INFO: Timeout per run is 0h 05m 00s
> > tst_kconfig.c:62: INFO: Parsing kernel config
> > '/boot/config-4.18.0-147.el8.s390x'
> > acct02.c:239: INFO: Verifying using 'struct acct_v3'
> > acct02.c:192: INFO: == entry 1 ==
> > acct02.c:147: INFO: ac_version != 3 (131)
> > acct02.c:192: INFO: == entry 2 ==
> > acct02.c:82: INFO: ac_comm != 'acct02_helper' ('acct02')
> > acct02.c:133: INFO: ac_exitcode != 32768 (0)
> > acct02.c:141: INFO: ac_ppid != 34501 (34500)
> > acct02.c:147: INFO: ac_version != 3 (131)
> > acct02.c:182: FAIL: end of file reached

> Looks good, acked.

> Also please consider a patch to the man 5 acct, it says that the version
> is always set to 3 for acct_v3 structure.
+1.

Kind regards,
Petr

Patch
diff mbox series

diff --git a/include/lapi/acct.h b/include/lapi/acct.h
index ebd65bbf4..c81b78b44 100644
--- a/include/lapi/acct.h
+++ b/include/lapi/acct.h
@@ -64,6 +64,11 @@  enum {
 	ACORE = 0x08,
 	AXSIG = 0x10
 };
+# if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+# define ACCT_BYTEORDER  0x80
+# elif __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+# define ACCT_BYTEORDER  0x00
+# endif
 #endif /* HAVE_STRUCT_ACCT_V3 */
 
 #endif /* LAPI_ACCT_H */
diff --git a/testcases/kernel/syscalls/acct/acct02.c b/testcases/kernel/syscalls/acct/acct02.c
index 7c2a27046..4d95aafec 100644
--- a/testcases/kernel/syscalls/acct/acct02.c
+++ b/testcases/kernel/syscalls/acct/acct02.c
@@ -142,7 +142,7 @@  static int verify_acct(void *acc, int elap_time)
 		ret = 1;
 	}
 
-	if (ACCT_MEMBER_V3(ac_version) != 3) {
+	if (ACCT_MEMBER_V3(ac_version) != (3 | ACCT_BYTEORDER)) {
 		tst_res(TINFO, "ac_version != 3 (%d)",
 			ACCT_MEMBER_V3(ac_version));
 		ret = 1;