[PATCH] test: Increase syscall entry timeout

[PATCH] test: Increase syscall entry timeout

[eugene.loh]

From: Eugene Loh <eugene.loh@oracle.com>

The run time for this test seems twice as long for aarch64 as for
x86_64.  Further, the run time seems to have jumped significantly from
kernel 5.15 to 6.12 and then again to kernel 6.18.  E.g.,

                  x86_64     aarch64
    5.15          7 secs     18 secs
    6.12         12 secs     33 secs
    6.18         22 secs     54 secs

Looking at a run on the 6.18 aarch64 system, the time is basically
spent in the dtrace_close() call to dt_probe_detach_all(), which does:

    for (prp = ...)
        prp->prov->impl->detach(dtp, prp);

Then, dt_tp_probe_detach() calls dt_tp_detach(), which does:

    close(tpp->fd);

This close() averages over 0.1 secs.  For hundreds of syscall probes,
we get nearly a minute of run time, exceeding the default test timeout.

For the time being, increase the timeout on this test.

Signed-off-by: Eugene Loh <eugene.loh@oracle.com>
---
 test/unittest/syscall/tst.entry.d | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/test/unittest/syscall/tst.entry.d b/test/unittest/syscall/tst.entry.d
index f2d56150d..dfd7304e5 100644
--- a/test/unittest/syscall/tst.entry.d
+++ b/test/unittest/syscall/tst.entry.d
@@ -1,6 +1,6 @@
 /*
  * Oracle Linux DTrace.
- * Copyright (c) 2020, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2020, 2026, Oracle and/or its affiliates. All rights reserved.
  * Licensed under the Universal Permissive License v 1.0 as shown at
  * http://oss.oracle.com/licenses/upl.
  */
@@ -8,6 +8,7 @@
 /* @@trigger: syscall-tst-args */
 /* @@trigger-timing: before */
 /* @@runtest-opts: $_pid */
+/* @@timeout: 80 */
 
 /* check that syscall:::entry picks up mmap:entry */
 /* (mmap is called repeatedly by the trigger) */
-- 
2.47.3

Re: [DTrace-devel] [PATCH] test: Increase syscall entry timeout

[Nick Alcock]

On 22 Jan 2026, eugene loh outgrape:

> From: Eugene Loh <eugene.loh@oracle.com>
>
> The run time for this test seems twice as long for aarch64 as for
> x86_64.  Further, the run time seems to have jumped significantly from
> kernel 5.15 to 6.12 and then again to kernel 6.18.  E.g.,

Ew.

>                   x86_64     aarch64
>     5.15          7 secs     18 secs
>     6.12         12 secs     33 secs
>     6.18         22 secs     54 secs
>
> Looking at a run on the 6.18 aarch64 system, the time is basically
> spent in the dtrace_close() call to dt_probe_detach_all(), which does:
>
>     for (prp = ...)
>         prp->prov->impl->detach(dtp, prp);
>
> Then, dt_tp_probe_detach() calls dt_tp_detach(), which does:
>
>     close(tpp->fd);
>
> This close() averages over 0.1 secs.  For hundreds of syscall probes,
> we get nearly a minute of run time, exceeding the default test timeout.

Sounds like another O(n^2) crept back in somewhere in the kernel :(

> For the time being, increase the timeout on this test.
>
> Signed-off-by: Eugene Loh <eugene.loh@oracle.com>

Reviewed-by: Nick Alcock <nick.alcock@oracle.com>