[LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Cyril Hrubis]

Calling tst_reinit() when library is already initialized or calling it
for a second time is always mistake. With that we silenty drop the old
IPC region, that may have contained test results and there are likely
other problems possibly caused by initializing the test library twice.

To fix that this patch adds tst_brk() that abort the test if we call
tst_reinit() with already initialized test library.

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
---
 lib/tst_test.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/lib/tst_test.c b/lib/tst_test.c
index 26f6510a0..fa61e1be4 100644
--- a/lib/tst_test.c
+++ b/lib/tst_test.c
@@ -196,6 +196,9 @@ void tst_reinit(void)
 	size_t size = getpagesize();
 	int fd;
 
+	if (ipc)
+		tst_brk(TBROK, "Test library already initialized!");
+
 	if (!path)
 		tst_brk(TBROK, IPC_ENV_VAR" is not defined");
 
-- 
2.53.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Andrea Cervesato via ltp]

Reviewed-by: Andrea Cervesato <andrea.cervesato@suse.com>

--
Andrea Cervesato
SUSE QE Automation Engineer Linux
andrea.cervesato@suse.com

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Li Wang]

On Thu, Apr 30, 2026 at 11:19:56AM +0200, Cyril Hrubis wrote:
> Calling tst_reinit() when library is already initialized or calling it
> for a second time is always mistake. With that we silenty drop the old
> IPC region, that may have contained test results and there are likely
> other problems possibly caused by initializing the test library twice.
> 
> To fix that this patch adds tst_brk() that abort the test if we call
> tst_reinit() with already initialized test library.
> 
> Signed-off-by: Cyril Hrubis <chrubis@suse.cz>

Reviewed-by: Li Wang <li.wang@linux.dev>

The patch itself looks good, I'm just wondering how you found
this issue, was it called twice somewhere/in a test?

-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] lib: tst_test: tst_reinit() abort when already initialized

[linuxtestproject.agent]

Hi Cyril,

On 2026-04-30, Cyril Hrubis wrote:
> lib: tst_test: tst_reinit() abort when already initialized

> +       if (ipc)
> +               tst_brk(TBROK, "Test library already initialized!");

Looks good.

Reviewed-by: LTP AI Reviewer <ltp-ai@noreply.github.com>

---
Note:

Our agent completed the review of the patch. The full review can be
found at: https://github.com/linux-test-project/ltp-agent/actions/runs/25163012944

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Cyril Hrubis]

Hi!
> > Calling tst_reinit() when library is already initialized or calling it
> > for a second time is always mistake. With that we silenty drop the old
> > IPC region, that may have contained test results and there are likely
> > other problems possibly caused by initializing the test library twice.
> > 
> > To fix that this patch adds tst_brk() that abort the test if we call
> > tst_reinit() with already initialized test library.
> > 
> > Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
> 
> Reviewed-by: Li Wang <li.wang@linux.dev>
> 
> The patch itself looks good, I'm just wondering how you found
> this issue, was it called twice somewhere/in a test?

https://lists.linux.it/pipermail/ltp/2026-April/048084.html

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Li Wang]

Patch merged, thanks!

-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Cyril Hrubis]

Hi!
I guess that we also need to revert:

https://lists.linux.it/pipermail/ltp/2026-April/048007.html

Since the IMA test is broken after this patch:

$./ima_mmap -f foo
tst_test.c:2042: TINFO: LTP version: 20260130-210-gcd9899ce4
tst_test.c:2045: TINFO: Tested kernel: 6.16.0 #1 SMP PREEMPT_DYNAMIC Tue Aug 12 10:49:53 CEST 2025 x86_64
tst_kconfig.c:90: TINFO: Parsing kernel config '/lib/modules/6.16.0/build/.config'
tst_test.c:1870: TINFO: Overall timeout per run is 0h 00m 30s
tst_test.c:199: TBROK: Test library already initialized!

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Li Wang]

Cyril Hrubis wrote:

> I guess that we also need to revert:
> 
> https://lists.linux.it/pipermail/ltp/2026-April/048007.html
> 
> Since the IMA test is broken after this patch:

You're right, tst_reinit() shouldn't be called in an LTP test with
already initialized the IPC.

Or, we should convert ima_mmap.c into raw main() format proceduer.

> $./ima_mmap -f foo
> tst_test.c:2042: TINFO: LTP version: 20260130-210-gcd9899ce4
> tst_test.c:2045: TINFO: Tested kernel: 6.16.0 #1 SMP PREEMPT_DYNAMIC Tue Aug 12 10:49:53 CEST 2025 x86_64
> tst_kconfig.c:90: TINFO: Parsing kernel config '/lib/modules/6.16.0/build/.config'
> tst_test.c:1870: TINFO: Overall timeout per run is 0h 00m 30s
> tst_test.c:199: TBROK: Test library already initialized!

-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Li Wang]

> > I guess that we also need to revert:

I look into Petr's patch, seems we can go another way to replace
the checkpoint method.

How about this:

--- a/testcases/kernel/security/integrity/ima/src/ima_mmap.c
+++ b/testcases/kernel/security/integrity/ima/src/ima_mmap.c
@@ -34,10 +34,7 @@ static void run(void)
        file = SAFE_MMAP(NULL, MMAPSIZE, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
        SAFE_CLOSE(fd);

-       tst_reinit();
-       TST_CHECKPOINT_WAIT(0);
-       /* keep running until ima_violations.sh open and close file */
-       TST_CHECKPOINT_WAKE_AND_WAIT(0);
+       raise(SIGSTOP);

        tst_res(TPASS, "test completed");
 }
diff --git a/testcases/kernel/security/integrity/ima/tests/ima_violations.sh b/testcases/kernel/security/integrity/ima/tests/ima_violations.sh
index d7dcd077b..e772cf53b 100755
--- a/testcases/kernel/security/integrity/ima/tests/ima_violations.sh
+++ b/testcases/kernel/security/integrity/ima/tests/ima_violations.sh
@@ -156,6 +156,14 @@ test2()
        validate $num_violations $count $search
 }

+wait_process_stopped()
+{
+       while true; do
+               [ "$(awk '{print $3}' /proc/$1/stat 2>/dev/null)" = "T" ] && break
+               usleep 100000
+       done
+}
+
 test3()
 {
        local pid
@@ -173,13 +181,13 @@ test3()
        ima_mmap -f $FILE &
        pid=$!

-       # wait for violations appear in logs
-       TST_CHECKPOINT_WAKE_AND_WAIT 0
+       wait_process_stopped $pid

        open_file_read
        close_file_read

-       TST_CHECKPOINT_WAKE 0
+       kill -CONT $pid
+       wait $pid

        validate $num_violations $count $search


-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Cyril Hrubis]

Hi!
> I look into Petr's patch, seems we can go another way to replace
> the checkpoint method.

I would rather change the binary to use TST_NO_DEFAULT_MAIN. It's not a
test, just a helper binary, the tst_test structure shouldn't have been
used to begin with.

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH] lib: tst_test: tst_reinit() abort when already initialized

[Li Wang]

Cyril Hrubis wrote:

> Hi!
> > I look into Petr's patch, seems we can go another way to replace
> > the checkpoint method.
> 
> I would rather change the binary to use TST_NO_DEFAULT_MAIN. It's not a
> test, just a helper binary, the tst_test structure shouldn't have been
> used to begin with.

That sounds reasonable! To avoid mixing with real LTP tests,
we should keep the auxiliary binary clean.

-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp