All of lore.kernel.org
 help / color / mirror / Atom feed
From: Cyril Hrubis <chrubis@suse.cz>
To: ltp@lists.linux.it
Subject: [LTP] [PATCH v4 1/2] lib: add .request_hugepages to reserve hugepage
Date: Tue, 18 Feb 2020 17:27:39 +0100	[thread overview]
Message-ID: <20200218162739.GA9985@rei.lan> (raw)
In-Reply-To: <20200208092014.24824-1-liwang@redhat.com>

Hi!
> Many of the ltp tests need to use hugepage in their testing, this allows
> the test can reserve hpages from system only via .request_hugepages = xx.
> 
> If set non-zero number of request_hugepages, test will try to reserve the
> expected number of hugepage for testing in setup phase. If system does not
> have enough hpage for using, it will try the best to reserve 80% available
> number of hpages. For a system without hugetlb supporting, variable
> 'tst_hugepage' will be set to 0.
> 
> Signed-off-by: Li Wang <liwang@redhat.com>
> ---
> 
> Notes:
>     v3 --> v4
>       * add doc to check the hpage number in setup()
>       * remove the TCONF print if hugepage is not supported
>       * check here that max_hpages * 0.8 > 1
> 
>  doc/test-writing-guidelines.txt | 58 ++++++++++++++++++++++++++++++++
>  include/tst_hugepage.h          | 29 ++++++++++++++++
>  include/tst_test.h              | 13 ++++++++
>  lib/newlib_tests/.gitignore     |  1 +
>  lib/newlib_tests/test20.c       | 45 +++++++++++++++++++++++++
>  lib/tst_hugepage.c              | 59 +++++++++++++++++++++++++++++++++
>  lib/tst_test.c                  |  6 ++--
>  7 files changed, 209 insertions(+), 2 deletions(-)
>  create mode 100644 include/tst_hugepage.h
>  create mode 100644 lib/newlib_tests/test20.c
>  create mode 100644 lib/tst_hugepage.c
> 
> diff --git a/doc/test-writing-guidelines.txt b/doc/test-writing-guidelines.txt
> index 546bb7a49..fa7cb89c1 100644
> --- a/doc/test-writing-guidelines.txt
> +++ b/doc/test-writing-guidelines.txt
> @@ -1933,6 +1933,64 @@ specified by the user plus some other heuristics.
>  
>  For full documentation see the comments in 'include/tst_fuzzy_sync.h'.
>  
> +2.2.34 Reserving hugepages
> +^^^^^^^^^^^^^^^^^^^^^^^^^^
> +
> +Many of the ltp tests need to use hugepage in their testing, this allows the
> +test can reserve hugepages from system only via .request_hugepages = xx.
> +
> +If set non-zero number of request_hugepages, test will try to reserve the
> +expected number of hugepage for testing in setup phase. If system does not
> +have enough hpage for using, it will try the best to reserve 80% available
> +number of hpages. With success test stores the reserved hugepage number in
> +'tst_hugepages. For the system without hugetlb supporting, variable
> +'tst_hugepages' will be set to 0.
> +
> +Also, we do cleanup and restore work for the hpages resetting automatically.
> +
> +[source,c]
> +-------------------------------------------------------------------------------
> +#include "tst_test.h"
> +
> +static void run(void)
> +{
> +	...
> +
> +	if (tst_hugepages == test.request_hugepages)
> +		TEST(do_hpage_test);
> +	else
> +		...
> +	...
> +}
> +
> +struct tst_test test = {
> +	.test_all = run,
> +	.request_hugepages = 2,
> +	...
> +};
> +
> +or,
> +
> +#include "tst_test.h"
> +
> +static void run(void)
> +{
> +	...
> +}
> +
> +static void setup(void)
> +{
> +?? ?? ?? ?? if (tst_hugepages != test.requested_hugepages)
> +?? ?? ?? ?? ?? ?? ?? ?? tst_brk(TCONF, "...");
> +}
> +
> +struct tst_test test = {
> +	.test_all = run,
> +	.request_hugepages = 2,
> +	...
> +};
> +-------------------------------------------------------------------------------
> +
>  2.3 Writing a testcase in shell
>  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>  
> diff --git a/include/tst_hugepage.h b/include/tst_hugepage.h
> new file mode 100644
> index 000000000..10e0eaf47
> --- /dev/null
> +++ b/include/tst_hugepage.h
> @@ -0,0 +1,29 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (c) 2019 Red Hat, Inc.
> + */
> +
> +#ifndef TST_HUGEPAGE__
> +#define TST_HUGEPAGE__
> +
> +#define PATH_HUGEPAGES	"/sys/kernel/mm/hugepages/"
> +#define PATH_NR_HPAGES	"/proc/sys/vm/nr_hugepages"
> +
> +/*
> + * Try the best to request a specified number of huge pages from system,
> + * it will store the reserved hpage number in tst_hugepages.
> + *
> + * Note: this depend on the status of system memory fragmentation.
> + */
> +int tst_request_hugepages(int hpages);
> +
> +/*
> + * This variable is used for recording the number of hugepages which system can
> + * provides. It will be equal to 'hpages' if tst_request_hugepages on success,
> + * otherwise set it to a number of hugepages that we were able to reserve.
> + *
> + * If system does not support hugetlb, then it will be set to 0.
> + */
> +extern unsigned int tst_hugepages;
> +
> +#endif /* TST_HUGEPAGE_H */
> diff --git a/include/tst_test.h b/include/tst_test.h
> index 21c7dfbdb..1026a422a 100644
> --- a/include/tst_test.h
> +++ b/include/tst_test.h
> @@ -37,6 +37,7 @@
>  #include "tst_coredump.h"
>  #include "tst_buffers.h"
>  #include "tst_capability.h"
> +#include "tst_hugepage.h"
>  
>  /*
>   * Reports testcase result.
> @@ -148,6 +149,18 @@ struct tst_test {
>  	 */
>  	int all_filesystems:1;
>  
> +	/*
> +	 * If set non-zero number of request_hugepages, test will try to reserve the
> +	 * expected number of hugepage for testing in setup phase. If system does not
> +	 * have enough hpage for using, it will try the best to reserve 80% available
> +	 * number of hpages. With success test stores the reserved hugepage number in
> +	 * 'tst_hugepages. For the system without hugetlb supporting, variable
> +	 * 'tst_hugepages' will be set to 0.
> +	 *
> +	 * Also, we do cleanup and restore work for the hpages resetting automatically.
> +	 */
> +	unsigned int request_hugepages;
> +
>  	/*
>  	 * If set non-zero denotes number of test variant, the test is executed
>  	 * variants times each time with tst_variant set to different number.
> diff --git a/lib/newlib_tests/.gitignore b/lib/newlib_tests/.gitignore
> index d4aa4935f..ab3a92c4c 100644
> --- a/lib/newlib_tests/.gitignore
> +++ b/lib/newlib_tests/.gitignore
> @@ -23,6 +23,7 @@ tst_strstatus
>  test17
>  test18
>  test19
> +test20
>  tst_expiration_timer
>  test_exec
>  test_exec_child
> diff --git a/lib/newlib_tests/test20.c b/lib/newlib_tests/test20.c
> new file mode 100644
> index 000000000..92e230976
> --- /dev/null
> +++ b/lib/newlib_tests/test20.c
> @@ -0,0 +1,45 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (c) 2019 Li Wang <liwang@redhat.com>
> + */
> +
> +/*
> + * Tests .request_hugepages + .save_restore
> + */
> +
> +#include "tst_test.h"
> +#include "tst_hugepage.h"
> +#include "tst_sys_conf.h"
> +
> +static const char * const save_restore[] = {
> +	"!/proc/sys/kernel/numa_balancing",
> +	NULL,
> +};
> +
> +static void do_test(void) {
> +
> +	int val, hpages;
> +
> +	tst_res(TINFO, "tst_hugepages = %u", tst_hugepages);
> +	SAFE_FILE_PRINTF("/proc/sys/kernel/numa_balancing", "1");
> +
> +	hpages = test.request_hugepages;
> +	SAFE_FILE_SCANF(PATH_NR_HPAGES, "%d", &val);
> +	if (val != hpages)
> +		tst_brk(TBROK, "nr_hugepages = %d, but expect %d", val, hpages);
> +	else
> +		tst_res(TPASS, "test .needs_hugepges");
> +
> +	hpages = tst_request_hugepages(3);
> +	SAFE_FILE_SCANF(PATH_NR_HPAGES, "%d", &val);
> +	if (val != hpages)
> +		tst_brk(TBROK, "nr_hugepages = %d, but expect %d", val, hpages);
> +	else
> +		tst_res(TPASS, "tst_request_hugepages");
> +}
> +
> +static struct tst_test test = {
> +	.test_all = do_test,
> +	.request_hugepages = 2,
> +	.save_restore = save_restore,
> +};
> diff --git a/lib/tst_hugepage.c b/lib/tst_hugepage.c
> new file mode 100644
> index 000000000..d37b45e3a
> --- /dev/null
> +++ b/lib/tst_hugepage.c
> @@ -0,0 +1,59 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (c) 2019 Red Hat, Inc.
> + */
> +
> +#define TST_NO_DEFAULT_MAIN
> +
> +#include "tst_test.h"
> +#include "tst_hugepage.h"
> +
> +unsigned int tst_hugepages;
> +
> +int tst_request_hugepages(int hpages)
> +{
> +	int val;
> +	long mem_avail, max_hpages;
> +
> +	if (access(PATH_HUGEPAGES, F_OK)) {
> +		tst_hugepages = 0;
> +		goto out;
> +	}
> +
> +	tst_hugepages = hpages;
> +
> +	SAFE_FILE_PRINTF("/proc/sys/vm/drop_caches", "3");
> +
> +	if (FILE_LINES_SCANF("/proc/meminfo",
> +				"MemAvailable: %ld", &mem_avail)) {
> +		/*
> +		 * Using "MemFree:" on kernel that doesn't have
> +		 * "MemAvailable:" in Meminfo
> +		 */
> +		tst_res(TINFO, "MemAvailable: not found in /proc/meminfo");
> +
> +		mem_avail = SAFE_READ_MEMINFO("MemFree:");
> +	}

I guess that we don't have to bother with MemAvailable here, since we do
drop the system caches the MemFree should be equal to MemAvailable. So I
would be for reading just MemFree unconditionally here.

If you agree with this change I can change it before applying.

> +	max_hpages = mem_avail / SAFE_READ_MEMINFO("Hugepagesize:");
> +
> +	if (hpages > max_hpages) {
> +		tst_res(TINFO, "Requested number(%d) of hugepages is too large, "
> +				"limiting to 80%% of the max hugepage count %ld",
> +				hpages, max_hpages);
> +		tst_hugepages = max_hpages * 0.8;
> +
> +		if (tst_hugepages < 1)
> +			goto out;
> +	}
> +
> +	tst_sys_conf_save("?/proc/sys/vm/nr_hugepages");
> +	SAFE_FILE_PRINTF(PATH_NR_HPAGES, "%d", tst_hugepages);
> +	SAFE_FILE_SCANF(PATH_NR_HPAGES, "%d", &val);
> +	if (val != tst_hugepages)
> +		tst_brk(TBROK, "nr_hugepages = %d, but expect %d", val, tst_hugepages);
> +
> +	tst_res(TINFO, "%d hugepage(s) reserved", tst_hugepages);
> +out:
> +	return tst_hugepages;
> +}
> diff --git a/lib/tst_test.c b/lib/tst_test.c
> index 8d7dee2cc..9a24cffc5 100644
> --- a/lib/tst_test.c
> +++ b/lib/tst_test.c
> @@ -890,6 +890,9 @@ static void do_setup(int argc, char *argv[])
>  	if (tst_test->all_filesystems)
>  		tst_test->needs_device = 1;
>  
> +	if (tst_test->request_hugepages)
> +		tst_request_hugepages(tst_test->request_hugepages);
> +
>  	setup_ipc();
>  
>  	if (tst_test->bufs)
> @@ -1006,8 +1009,7 @@ static void do_cleanup(void)
>  		tst_rmdir();
>  	}
>  
> -	if (tst_test->save_restore)
> -		tst_sys_conf_restore(0);
> +	tst_sys_conf_restore(0);
>  
>  	if (tst_test->restore_wallclock)
>  		tst_wallclock_restore();
> -- 
> 2.20.1
> 

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


-- 
Cyril Hrubis
chrubis@suse.cz

  parent reply	other threads:[~2020-02-18 16:27 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-02-08  9:20 [LTP] [PATCH v4 1/2] lib: add .request_hugepages to reserve hugepage Li Wang
2020-02-08  9:20 ` [LTP] [PATCH v4 2/2] pkey01: disable hugetlb test if failed " Li Wang
2020-02-17 16:09   ` Petr Vorel
2020-02-17 16:08 ` [LTP] [PATCH v4 1/2] lib: add .request_hugepages " Petr Vorel
2020-02-18  6:06   ` Li Wang
2020-02-18 16:27 ` Cyril Hrubis [this message]
2020-02-19  4:02   ` Li Wang
2020-02-21 12:16   ` Petr Vorel
2020-02-22  3:02     ` Li Wang
2020-02-24  8:40       ` Petr Vorel

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200218162739.GA9985@rei.lan \
    --to=chrubis@suse.cz \
    --cc=ltp@lists.linux.it \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.