public inbox for ltp@lists.linux.it
 help / color / mirror / Atom feed
* [LTP] [PATCH 1/2] shell: Use $LTP_TIMEOUT_MUL also in retry functions
@ 2019-07-18  8:39 Petr Vorel
  2019-07-18  8:39 ` [LTP] [PATCH 2/2] c: " Petr Vorel
  2019-07-19  4:56 ` [LTP] [PATCH 1/2] shell: " Li Wang
  0 siblings, 2 replies; 6+ messages in thread
From: Petr Vorel @ 2019-07-18  8:39 UTC (permalink / raw)
  To: ltp

in TST_RETRY_FN_EXP_BACKOFF() and thus in TST_RETRY_FUNC() in new shell API.

This function should also address possibility of slow machine.

Because using on 2 places moved the default definition to the beginning
of the script.
+ use $((...)) instead of expr.
+ move expression using $3 after check whether we have enough parameters

+ document LTP_TIMEOUT_MUL environment variable in wiki.

Signed-off-by: Petr Vorel <pvorel@suse.cz>
---
Hi,

I can squash these 2 commits.

General question about Test Writing Guidelines in wiki [1]:
1) we have separate section for C and shell API, which is probably a good
choice, but we're trying to sync them. That leads to things like retry
function documented under chapter for shell API (2.3).

+ That page is too long and without table of contents (not easy to
navigate in it; I tried to add TOC, but without success). Maybe adding
TOC or split the page to 3: 1) general info 2) C API 3) shell API would
help. WDYT?

Kind regards,
Petr

[1] https://github.com/linux-test-project/ltp/wiki/Test-Writing-Guidelines
[2] https://github.com/linux-test-project/ltp/wiki/Test-Writing-Guidelines#retry-a-function-in-limited-time

 doc/test-writing-guidelines.txt | 2 ++
 testcases/lib/tst_test.sh       | 9 +++++----
 2 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/doc/test-writing-guidelines.txt b/doc/test-writing-guidelines.txt
index 869e6ed35..782e14f32 100644
--- a/doc/test-writing-guidelines.txt
+++ b/doc/test-writing-guidelines.txt
@@ -1881,6 +1881,8 @@ simply by setting right '$TST_NEEDS_FOO'.
                        the test (see below).
 | 'TST_NEEDS_MODULE' | Test module name needed for the test (see below).
 | 'TST_NEEDS_DRIVERS'| Checks kernel drivers support for the test.
+| 'LTP_TIMEOUT_MUL'  | Multiply timeout (> 1 is useful for slow machines to
+                       avoid unexpected timeout).
 |=============================================================================
 
 Checking for presence of commands
diff --git a/testcases/lib/tst_test.sh b/testcases/lib/tst_test.sh
index 31b3a3951..60d765424 100644
--- a/testcases/lib/tst_test.sh
+++ b/testcases/lib/tst_test.sh
@@ -17,6 +17,8 @@ export TST_ITERATIONS=1
 export TST_TMPDIR_RHOST=0
 export TST_LIB_LOADED=1
 
+export LTP_TIMEOUT_MUL=${LTP_TIMEOUT_MUL:-1}
+
 . tst_ansi_color.sh
 . tst_security.sh
 
@@ -164,12 +166,13 @@ TST_RETRY_FN_EXP_BACKOFF()
 {
 	local tst_fun="$1"
 	local tst_exp=$2
-	local tst_sec=$(expr $3 \* 1000000)
 	local tst_delay=1
+	local tst_sec
 
 	if [ $# -ne 3 ]; then
 		tst_brk TBROK "TST_RETRY_FN_EXP_BACKOFF expects 3 parameters"
 	fi
+	tst_sec=$(($3 * LTP_TIMEOUT_MUL * 1000000))
 
 	if ! tst_is_int "$tst_sec"; then
 		tst_brk TBROK "TST_RETRY_FN_EXP_BACKOFF: tst_sec must be integer ('$tst_sec')"
@@ -185,7 +188,7 @@ TST_RETRY_FN_EXP_BACKOFF()
 			tst_sleep ${tst_delay}us
 			tst_delay=$((tst_delay*2))
 		else
-			tst_brk TBROK "\"$tst_fun\" timed out"
+			tst_brk TBROK "\"$tst_fun\" timed out! If you are running on slow machine, try exporting LTP_TIMEOUT_MUL > 1"
 		fi
 	done
 
@@ -374,8 +377,6 @@ _tst_rescmp()
 
 _tst_setup_timer()
 {
-	LTP_TIMEOUT_MUL=${LTP_TIMEOUT_MUL:-1}
-
 	local sec=$((300 * LTP_TIMEOUT_MUL))
 	local h=$((sec / 3600))
 	local m=$((sec / 60 % 60))
-- 
2.22.0


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* [LTP] [PATCH 2/2] c: Use $LTP_TIMEOUT_MUL also in retry functions
  2019-07-18  8:39 [LTP] [PATCH 1/2] shell: Use $LTP_TIMEOUT_MUL also in retry functions Petr Vorel
@ 2019-07-18  8:39 ` Petr Vorel
  2019-07-19  5:02   ` Li Wang
  2019-07-19  4:56 ` [LTP] [PATCH 1/2] shell: " Li Wang
  1 sibling, 1 reply; 6+ messages in thread
From: Petr Vorel @ 2019-07-18  8:39 UTC (permalink / raw)
  To: ltp

in TST_RETRY_FN_EXP_BACKOFF() and thus in TST_RETRY_FUNC() in new C API.
This sync C API with shell API (added in previous commit).

+ mention LTP_TIMEOUT_MUL use in wiki.
+ use SPDX, add copyright

Signed-off-by: Petr Vorel <pvorel@suse.cz>
---
 doc/test-writing-guidelines.txt |  2 ++
 include/tst_common.h            | 26 +++++++++++---------------
 2 files changed, 13 insertions(+), 15 deletions(-)

diff --git a/doc/test-writing-guidelines.txt b/doc/test-writing-guidelines.txt
index 782e14f32..3548d6043 100644
--- a/doc/test-writing-guidelines.txt
+++ b/doc/test-writing-guidelines.txt
@@ -2064,6 +2064,8 @@ Sometimes LTP test needs retrying a function for many times to get success.
 This achievement makes that possible via keeping it retrying if the return
 value of the function is NOT as we expected. After exceeding a limited time,
 test will break from the retries immediately.
+NOTE: both 'TST_RETRY_FUNC()' and 'TST_RETRY_FN_EXP_BACKOFF()' use 'LTP_TIMEOUT_MUL'
+environment variable.
 
 [source,c]
 -------------------------------------------------------------------------------
diff --git a/include/tst_common.h b/include/tst_common.h
index c21505450..454a166e8 100644
--- a/include/tst_common.h
+++ b/include/tst_common.h
@@ -1,21 +1,10 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
 /*
+ * Copyright (c) Linux Test Project, 2017-2019
  * Copyright (c) 2016 Cyril Hrubis <chrubis@suse.cz>
  * Copyright (c) 2013 Stanislav Kholmanskikh <stanislav.kholmanskikh@oracle.com>
  * Copyright (c) 2010 Ngie Cooper <yaneurabeya@gmail.com>
  * Copyright (c) 2008 Mike Frysinger <vapier@gmail.com>
- *
- * This program is free software: you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation, either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
 #ifndef TST_COMMON_H__
@@ -51,15 +40,22 @@
 
 #define TST_RETRY_FN_EXP_BACKOFF(FUNC, ERET, MAX_DELAY)	\
 ({	int tst_delay_ = 1;						\
+	float m = 1;	\
+	char *mul = getenv("LTP_TIMEOUT_MUL");	\
+	if (mul) {	\
+		m = atof(mul); \
+		if (m < 1) \
+			tst_brk(TBROK, "Invalid timeout multiplier '%s'", mul); \
+	}	\
 	for (;;) {							\
 		typeof(FUNC) tst_ret_ = FUNC;				\
 		if (tst_ret_ == ERET)					\
 			break;						\
-		if (tst_delay_ < MAX_DELAY * 1000000) {			\
+		if (tst_delay_ < MAX_DELAY * m * 1000000) {			\
 			usleep(tst_delay_);				\
 			tst_delay_ *= 2;				\
 		} else {						\
-			tst_brk(TBROK, #FUNC" timed out");		\
+			tst_brk(TBROK, #FUNC" timed out! If you are running on slow machine, try exporting LTP_TIMEOUT_MUL > 1"); \
 		}							\
 	}								\
 	ERET;								\
-- 
2.22.0


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* [LTP] [PATCH 1/2] shell: Use $LTP_TIMEOUT_MUL also in retry functions
  2019-07-18  8:39 [LTP] [PATCH 1/2] shell: Use $LTP_TIMEOUT_MUL also in retry functions Petr Vorel
  2019-07-18  8:39 ` [LTP] [PATCH 2/2] c: " Petr Vorel
@ 2019-07-19  4:56 ` Li Wang
  2019-07-19  9:01   ` Petr Vorel
  1 sibling, 1 reply; 6+ messages in thread
From: Li Wang @ 2019-07-19  4:56 UTC (permalink / raw)
  To: ltp

On Thu, Jul 18, 2019 at 4:39 PM Petr Vorel <pvorel@suse.cz> wrote:

> in TST_RETRY_FN_EXP_BACKOFF() and thus in TST_RETRY_FUNC() in new shell
> API.
>
> This function should also address possibility of slow machine.
>
> Because using on 2 places moved the default definition to the beginning
> of the script.
> + use $((...)) instead of expr.
> + move expression using $3 after check whether we have enough parameters
>
> + document LTP_TIMEOUT_MUL environment variable in wiki.
>
> Signed-off-by: Petr Vorel <pvorel@suse.cz>
> ---
> Hi,
>
> I can squash these 2 commits.
>
> General question about Test Writing Guidelines in wiki [1]:
> 1) we have separate section for C and shell API, which is probably a good
> choice, but we're trying to sync them. That leads to things like retry
> function documented under chapter for shell API (2.3).
>

Thanks for point out this.

>
> + That page is too long and without table of contents (not easy to
> navigate in it; I tried to add TOC, but without success). Maybe adding
> TOC or split the page to 3: 1) general info 2) C API 3) shell API would
> help. WDYT?
>

Personally, I think adding TOC may be a better choice, since we don't have
the whole picture for the guidelines and usually hard to navigate in it.

To split the guideline that probably leads to the new LTP-user to read more
document pages. That's an annoying thing for me at least :).


>
> Kind regards,
> Petr
>
> [1] https://github.com/linux-test-project/ltp/wiki/Test-Writing-Guidelines
> [2]
> https://github.com/linux-test-project/ltp/wiki/Test-Writing-Guidelines#retry-a-function-in-limited-time
>
>  doc/test-writing-guidelines.txt | 2 ++
>  testcases/lib/tst_test.sh       | 9 +++++----
>  2 files changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/doc/test-writing-guidelines.txt
> b/doc/test-writing-guidelines.txt
> index 869e6ed35..782e14f32 100644
> --- a/doc/test-writing-guidelines.txt
> +++ b/doc/test-writing-guidelines.txt
> @@ -1881,6 +1881,8 @@ simply by setting right '$TST_NEEDS_FOO'.
>                         the test (see below).
>  | 'TST_NEEDS_MODULE' | Test module name needed for the test (see below).
>  | 'TST_NEEDS_DRIVERS'| Checks kernel drivers support for the test.
> +| 'LTP_TIMEOUT_MUL'  | Multiply timeout (> 1 is useful for slow machines
> to
> +                       avoid unexpected timeout).
>
>  |=============================================================================
>
>  Checking for presence of commands
> diff --git a/testcases/lib/tst_test.sh b/testcases/lib/tst_test.sh
> index 31b3a3951..60d765424 100644
> --- a/testcases/lib/tst_test.sh
> +++ b/testcases/lib/tst_test.sh
> @@ -17,6 +17,8 @@ export TST_ITERATIONS=1
>  export TST_TMPDIR_RHOST=0
>  export TST_LIB_LOADED=1
>
> +export LTP_TIMEOUT_MUL=${LTP_TIMEOUT_MUL:-1}
> +
>  . tst_ansi_color.sh
>  . tst_security.sh
>
> @@ -164,12 +166,13 @@ TST_RETRY_FN_EXP_BACKOFF()
>  {
>         local tst_fun="$1"
>         local tst_exp=$2
> -       local tst_sec=$(expr $3 \* 1000000)
>         local tst_delay=1
> +       local tst_sec
>
>         if [ $# -ne 3 ]; then
>                 tst_brk TBROK "TST_RETRY_FN_EXP_BACKOFF expects 3
> parameters"
>         fi
> +       tst_sec=$(($3 * LTP_TIMEOUT_MUL * 1000000))


>         if ! tst_is_int "$tst_sec"; then
>                 tst_brk TBROK "TST_RETRY_FN_EXP_BACKOFF: tst_sec must be
> integer ('$tst_sec')"
> @@ -185,7 +188,7 @@ TST_RETRY_FN_EXP_BACKOFF()
>                         tst_sleep ${tst_delay}us
>                         tst_delay=$((tst_delay*2))
>                 else
> -                       tst_brk TBROK "\"$tst_fun\" timed out"
> +                       tst_brk TBROK "\"$tst_fun\" timed out! If you are
> running on slow machine, try exporting LTP_TIMEOUT_MUL > 1"
>                 fi
>         done
>
> @@ -374,8 +377,6 @@ _tst_rescmp()
>
>  _tst_setup_timer()
>  {
> -       LTP_TIMEOUT_MUL=${LTP_TIMEOUT_MUL:-1}
> -
>         local sec=$((300 * LTP_TIMEOUT_MUL))
>

Maybe we need to check if the LTP_TIMEOUT_MUL is valid(>1) as what we do
for C API?

-- 
Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190719/ef9fca12/attachment-0001.htm>

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [LTP] [PATCH 2/2] c: Use $LTP_TIMEOUT_MUL also in retry functions
  2019-07-18  8:39 ` [LTP] [PATCH 2/2] c: " Petr Vorel
@ 2019-07-19  5:02   ` Li Wang
  2019-07-19  9:15     ` Petr Vorel
  0 siblings, 1 reply; 6+ messages in thread
From: Li Wang @ 2019-07-19  5:02 UTC (permalink / raw)
  To: ltp

On Thu, Jul 18, 2019 at 4:40 PM Petr Vorel <pvorel@suse.cz> wrote:
>...
>  #ifndef TST_COMMON_H__
> @@ -51,15 +40,22 @@
>
>  #define TST_RETRY_FN_EXP_BACKOFF(FUNC, ERET, MAX_DELAY)        \
>  ({     int tst_delay_ = 1;                                             \
> +       float m = 1;    \
> +       char *mul = getenv("LTP_TIMEOUT_MUL");  \

We also need a prefix/suffix in the variable definition to make sure
that it will
not alias with anything that has been passed to the FUNC, just like what we do
for the tst_delay_.

e.g. if the FUNC is defined as foo_func(m); the m variable will be aliased and
the function will do something very unexpected.

> +       if (mul) {      \
> +               m = atof(mul); \
> +               if (m < 1) \
> +                       tst_brk(TBROK, "Invalid timeout multiplier '%s'", mul); \

If we reverse some code order in tst_set_timeout() function, then here
we have no need to check if m < 1 again, since the LTP_TIMEOUT_MUL
valid check will be finished in setup() early phase.
(This comment is just FYI, and I also think it's OK to check twice.)

--------------------------------------
void tst_set_timeout(int timeout)
{
        float m = 1;
        char *mul = getenv("LTP_TIMEOUT_MUL");

        if (mul) {
                m = atof(mul);
                if (m < 1)
                        tst_brk(TBROK, "Invalid timeout multiplier '%s'", mul);
        }

        if (timeout == -1) {
                tst_res(TINFO, "Timeout per run is disabled");
                return;
        }

        results->timeout = timeout * m + 0.5;

        tst_res(TINFO, "Timeout per run is %uh %02um %02us",
                results->timeout/3600, (results->timeout%3600)/60,
                results->timeout % 60);

        if (getpid() == lib_pid)
                alarm(results->timeout);
        else
                heartbeat();
}


--
Regards,
Li Wang

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [LTP] [PATCH 1/2] shell: Use $LTP_TIMEOUT_MUL also in retry functions
  2019-07-19  4:56 ` [LTP] [PATCH 1/2] shell: " Li Wang
@ 2019-07-19  9:01   ` Petr Vorel
  0 siblings, 0 replies; 6+ messages in thread
From: Petr Vorel @ 2019-07-19  9:01 UTC (permalink / raw)
  To: ltp

Hi Li,

> > + That page is too long and without table of contents (not easy to
> > navigate in it; I tried to add TOC, but without success). Maybe adding
> > TOC or split the page to 3: 1) general info 2) C API 3) shell API would
> > help. WDYT?


> Personally, I think adding TOC may be a better choice, since we don't have
> the whole picture for the guidelines and usually hard to navigate in it.

> To split the guideline that probably leads to the new LTP-user to read more
> document pages. That's an annoying thing for me at least :).
Agree. I'll give TOC next try sooner or later :).

...
> >  _tst_setup_timer()
> >  {
> > -       LTP_TIMEOUT_MUL=${LTP_TIMEOUT_MUL:-1}
> > -
> >         local sec=$((300 * LTP_TIMEOUT_MUL))


> Maybe we need to check if the LTP_TIMEOUT_MUL is valid(>1) as what we do
> for C API?
Correct, sorry for omitting that => v3 needed.

Kind regards,
Petr

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [LTP] [PATCH 2/2] c: Use $LTP_TIMEOUT_MUL also in retry functions
  2019-07-19  5:02   ` Li Wang
@ 2019-07-19  9:15     ` Petr Vorel
  0 siblings, 0 replies; 6+ messages in thread
From: Petr Vorel @ 2019-07-19  9:15 UTC (permalink / raw)
  To: ltp

Hi Li,

thanks for your review!

> On Thu, Jul 18, 2019 at 4:40 PM Petr Vorel <pvorel@suse.cz> wrote:
> >...
> >  #ifndef TST_COMMON_H__
> > @@ -51,15 +40,22 @@

> >  #define TST_RETRY_FN_EXP_BACKOFF(FUNC, ERET, MAX_DELAY)        \
> >  ({     int tst_delay_ = 1;                                             \
> > +       float m = 1;    \
> > +       char *mul = getenv("LTP_TIMEOUT_MUL");  \

> We also need a prefix/suffix in the variable definition to make sure
> that it will
> not alias with anything that has been passed to the FUNC, just like what we do
> for the tst_delay_.

> e.g. if the FUNC is defined as foo_func(m); the m variable will be aliased and
> the function will do something very unexpected.
Good point, I'll fix it in v3.

> > +       if (mul) {      \
> > +               m = atof(mul); \
> > +               if (m < 1) \
> > +                       tst_brk(TBROK, "Invalid timeout multiplier '%s'", mul); \

> If we reverse some code order in tst_set_timeout() function, then here
> we have no need to check if m < 1 again, since the LTP_TIMEOUT_MUL
> valid check will be finished in setup() early phase.
> (This comment is just FYI, and I also think it's OK to check twice.)
Good point. I'd probably check twice in case the logic changes one day.

> --------------------------------------
> void tst_set_timeout(int timeout)
> {
>         float m = 1;
>         char *mul = getenv("LTP_TIMEOUT_MUL");

>         if (mul) {
>                 m = atof(mul);
>                 if (m < 1)
>                         tst_brk(TBROK, "Invalid timeout multiplier '%s'", mul);
>         }

>         if (timeout == -1) {
>                 tst_res(TINFO, "Timeout per run is disabled");
>                 return;
>         }

>         results->timeout = timeout * m + 0.5;

>         tst_res(TINFO, "Timeout per run is %uh %02um %02us",
>                 results->timeout/3600, (results->timeout%3600)/60,
>                 results->timeout % 60);

>         if (getpid() == lib_pid)
>                 alarm(results->timeout);
>         else
>                 heartbeat();
> }


Kind regards,
Petr

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2019-07-19  9:15 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2019-07-18  8:39 [LTP] [PATCH 1/2] shell: Use $LTP_TIMEOUT_MUL also in retry functions Petr Vorel
2019-07-18  8:39 ` [LTP] [PATCH 2/2] c: " Petr Vorel
2019-07-19  5:02   ` Li Wang
2019-07-19  9:15     ` Petr Vorel
2019-07-19  4:56 ` [LTP] [PATCH 1/2] shell: " Li Wang
2019-07-19  9:01   ` Petr Vorel

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox