From: David Marchand <david.marchand@redhat.com>
To: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>
Cc: Amit Gupta <agupta3@marvell.com>,
"Wang, Yipeng1" <yipeng1.wang@intel.com>,
"Gobriel, Sameh" <sameh.gobriel@intel.com>,
"thomas@monjalon.net" <thomas@monjalon.net>, dev <dev@dpdk.org>,
nd <nd@arm.com>
Subject: Re: [dpdk-dev] [PATCH v2 3/5] test/hash: add lock free reader writer functional tests
Date: Wed, 5 Feb 2020 17:41:38 +0100 [thread overview]
Message-ID: <CAJFAV8wXe0mjT-9eT_wQ5Qh_qODbDSW2K77yM2e_=BeAGEa3sg@mail.gmail.com> (raw)
In-Reply-To: <VE1PR08MB5149905796CA08D184501E8398020@VE1PR08MB5149.eurprd08.prod.outlook.com>
On Wed, Feb 5, 2020 at 5:22 PM Honnappa Nagarahalli
<Honnappa.Nagarahalli@arm.com> wrote:
>
> >
> > On Mon, Feb 3, 2020 at 8:49 PM Honnappa Nagarahalli
> > <honnappa.nagarahalli@arm.com> wrote:
> > >
> > > Add lock-free reader writer concurrency functional tests.
> > > These tests will provide the same coverage that non lock-free APIs
> > > have.
> > >
> > > Signed-off-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
> > > ---
> > > app/test/test_hash_readwrite.c | 58
> > > +++++++++++++++++++++-------------
> > > 1 file changed, 36 insertions(+), 22 deletions(-)
> > >
> > > diff --git a/app/test/test_hash_readwrite.c
> > > b/app/test/test_hash_readwrite.c index 635ed5a9f..a9429091c 100644
> > > --- a/app/test/test_hash_readwrite.c
> > > +++ b/app/test/test_hash_readwrite.c
> > > @@ -121,7 +121,7 @@
> > test_hash_readwrite_worker(__attribute__((unused))
> > > void *arg) }
> > >
> > > static int
> > > -init_params(int use_ext, int use_htm, int use_jhash)
> > > +init_params(int use_ext, int use_htm, int rw_lf, int use_jhash)
> > > {
> > > unsigned int i;
> > >
> > > @@ -140,15 +140,16 @@ init_params(int use_ext, int use_htm, int
> > use_jhash)
> > > else
> > > hash_params.hash_func = rte_hash_crc;
> > >
> > > + hash_params.extra_flag =
> > > + RTE_HASH_EXTRA_FLAGS_MULTI_WRITER_ADD;
> > > if (use_htm)
> > > - hash_params.extra_flag =
> > > - RTE_HASH_EXTRA_FLAGS_TRANS_MEM_SUPPORT |
> > > - RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY |
> > > - RTE_HASH_EXTRA_FLAGS_MULTI_WRITER_ADD;
> > > + hash_params.extra_flag |=
> > > + RTE_HASH_EXTRA_FLAGS_TRANS_MEM_SUPPORT;
> > > + if (rw_lf)
> > > + hash_params.extra_flag |=
> > > + RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY_LF;
> > > else
> > > - hash_params.extra_flag =
> > > - RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY |
> > > - RTE_HASH_EXTRA_FLAGS_MULTI_WRITER_ADD;
> > > + hash_params.extra_flag |=
> > > + RTE_HASH_EXTRA_FLAGS_RW_CONCURRENCY;
> > >
> > > if (use_ext)
> > > hash_params.extra_flag |= @@ -195,7 +196,7 @@
> > > init_params(int use_ext, int use_htm, int use_jhash) }
> > >
> > > static int
> > > -test_hash_readwrite_functional(int use_ext, int use_htm)
> > > +test_hash_readwrite_functional(int use_htm, int use_rw_lf, int
> > > +use_ext)
> >
> > This is a bit hard to read, please keep the same order than init_params.
> It looks like it is better to change the init_params. Otherwise, the code in test_hash_rw_func_main becomes hard to read. See the comment below.
>
> >
> >
> > > {
> > > unsigned int i;
> > > const void *next_key;
> > > @@ -214,7 +215,7 @@ test_hash_readwrite_functional(int use_ext, int
> > use_htm)
> > > rte_atomic64_init(&ginsertions);
> > > rte_atomic64_clear(&ginsertions);
> > >
> > > - if (init_params(use_ext, use_htm, use_jhash) != 0)
> > > + if (init_params(use_ext, use_htm, use_rw_lf, use_jhash) != 0)
> > > goto err;
> > >
> > > if (use_ext)
> > > @@ -229,6 +230,8 @@ test_hash_readwrite_functional(int use_ext, int
> > use_htm)
> > > tbl_rw_test_param.num_insert
> > > * slave_cnt;
> > >
> > > + printf("\nHTM = %d, RW-LF = %d, EXT-Table = %d\n",
> > > + use_htm, use_rw_lf, use_ext);
> > > printf("++++++++Start function tests:+++++++++\n");
> > >
> > > /* Fire all threads. */
> > > @@ -379,7 +382,7 @@ test_hash_readwrite_perf(struct perf *perf_results,
> > int use_htm,
> > > rte_atomic64_init(&gwrite_cycles);
> > > rte_atomic64_clear(&gwrite_cycles);
> > >
> > > - if (init_params(0, use_htm, use_jhash) != 0)
> > > + if (init_params(0, use_htm, 0, use_jhash) != 0)
> > > goto err;
> > >
> > > /*
> > > @@ -700,7 +703,6 @@ test_hash_rw_func_main(void)
> > > * than writer threads. This is to timing either reader threads or
> > > * writer threads for performance numbers.
> > > */
> > > - int use_htm, use_ext;
> >
> > The comments block just before is out of sync.
> >
> >
> > > unsigned int i = 0, core_id = 0;
> > >
> > > if (rte_lcore_count() < 3) {
> > > @@ -721,29 +723,41 @@ test_hash_rw_func_main(void)
> > >
> > > printf("Test read-write with Hardware transactional
> > > memory\n");
> > >
> > > - use_htm = 1;
> > > - use_ext = 0;
> > > + /* htm = 1, rw_lf = 0, ext = 0 */
> >
> > I didn't like those local variables.
> > But comments tend to get out of sync fairly easily, please remove too.
> >
> >
> > > + if (test_hash_readwrite_functional(1, 0, 0) < 0)
> > > + return -1;
> > >
> > > - if (test_hash_readwrite_functional(use_ext, use_htm) < 0)
> > > + /* htm = 1, rw_lf = 1, ext = 0 */
> > > + if (test_hash_readwrite_functional(1, 1, 0) < 0)
> > > return -1;
> > >
> > > - use_ext = 1;
> > > - if (test_hash_readwrite_functional(use_ext, use_htm) < 0)
> > > + /* htm = 1, rw_lf = 0, ext = 1 */
> > > + if (test_hash_readwrite_functional(1, 0, 1) < 0)
> > > return -1;
> > >
> > > + /* htm = 1, rw_lf = 1, ext = 1 */
> > > + if (test_hash_readwrite_functional(1, 1, 1) < 0)
> > > + return -1;
> > > } else {
> > > printf("Hardware transactional memory (lock elision) "
> > > "is NOT supported\n");
> > > }
> > >
> > > printf("Test read-write without Hardware transactional memory\n");
> > > - use_htm = 0;
> > > - use_ext = 0;
> > > - if (test_hash_readwrite_functional(use_ext, use_htm) < 0)
> > > + /* htm = 0, rw_lf = 0, ext = 0 */
> > > + if (test_hash_readwrite_functional(0, 0, 0) < 0)
> > > + return -1;
> > > +
> > > + /* htm = 0, rw_lf = 1, ext = 0 */
> > > + if (test_hash_readwrite_functional(0, 1, 0) < 0)
> > > + return -1;
> > > +
> > > + /* htm = 0, rw_lf = 0, ext = 1 */
> > > + if (test_hash_readwrite_functional(0, 0, 1) < 0)
> > > return -1;
> > >
> > > - use_ext = 1;
> > > - if (test_hash_readwrite_functional(use_ext, use_htm) < 0)
> > > + /* htm = 0, rw_lf = 1, ext = 1 */
> > > + if (test_hash_readwrite_functional(0, 1, 1) < 0)
> > > return -1;
> The ordering of bits (0-0-0, 0-1-0, 0-0-1, 0-1-1) looks better here.
Ok, forget my comment.
I just want to get rid of this series and we stop getting random
timeout in the CI.
I will take it as is and cleanup if I find some time later.
--
David Marchand
next prev parent reply other threads:[~2020-02-05 16:41 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-06 5:49 [dpdk-dev] [PATCH 0/2] test/meson: fix hash readwrite timeout failure agupta3
2019-09-06 5:49 ` [dpdk-dev] [PATCH 1/2] test/meson: hash test split into shorter subtests agupta3
2019-09-11 17:05 ` Wang, Yipeng1
2019-10-17 5:02 ` Amit Gupta
2019-11-01 4:54 ` Amit Gupta
2019-11-01 17:04 ` Wang, Yipeng1
2019-11-05 16:37 ` Ferruh Yigit
2019-11-07 3:32 ` [dpdk-dev] [EXT] " Amit Gupta
2019-12-31 4:56 ` Amit Gupta
2019-09-06 5:49 ` [dpdk-dev] [PATCH 2/2] test/meson: hash lf test moved to dpdk perf testsuite agupta3
2019-09-11 17:13 ` Wang, Yipeng1
2019-09-12 15:00 ` Honnappa Nagarahalli
2019-09-13 8:24 ` Amit Gupta
2019-09-13 8:12 ` [dpdk-dev] [PATCH v2 1/1] " agupta3
2019-09-13 14:40 ` Aaron Conole
2019-09-13 15:09 ` Wang, Yipeng1
2019-09-13 15:46 ` Honnappa Nagarahalli
2019-09-16 4:39 ` Amit Gupta
2019-10-17 4:57 ` Amit Gupta
2019-10-17 13:16 ` Aaron Conole
2019-10-24 7:22 ` David Marchand
2019-09-13 8:15 ` agupta3
2019-09-11 5:55 ` [dpdk-dev] [PATCH 0/2] test/meson: fix hash readwrite timeout failure Amit Gupta
2020-02-03 19:49 ` [dpdk-dev] [PATCH v2 0/5] " Honnappa Nagarahalli
2020-02-03 19:49 ` [dpdk-dev] [PATCH v2 1/5] test/meson: hash test split into shorter subtests Honnappa Nagarahalli
2020-02-03 19:49 ` [dpdk-dev] [PATCH v2 2/5] test/hash: remove duplicated test code Honnappa Nagarahalli
2020-02-05 8:48 ` David Marchand
2020-02-05 16:42 ` David Marchand
2020-02-03 19:49 ` [dpdk-dev] [PATCH v2 3/5] test/hash: add lock free reader writer functional tests Honnappa Nagarahalli
2020-02-05 9:07 ` David Marchand
2020-02-05 16:22 ` Honnappa Nagarahalli
2020-02-05 16:41 ` David Marchand [this message]
2020-02-05 19:34 ` Wang, Yipeng1
2020-02-05 19:52 ` Honnappa Nagarahalli
2020-02-05 19:57 ` Wang, Yipeng1
2020-02-03 19:49 ` [dpdk-dev] [PATCH v2 4/5] test/hash: move reader writer lock free tests to perf tests Honnappa Nagarahalli
2020-02-03 19:49 ` [dpdk-dev] [PATCH v2 5/5] hash: correct lock free extendable table support Honnappa Nagarahalli
2020-02-05 18:41 ` [dpdk-dev] [PATCH v2 0/5] test/meson: fix hash readwrite timeout failure David Marchand
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='CAJFAV8wXe0mjT-9eT_wQ5Qh_qODbDSW2K77yM2e_=BeAGEa3sg@mail.gmail.com' \
--to=david.marchand@redhat.com \
--cc=Honnappa.Nagarahalli@arm.com \
--cc=agupta3@marvell.com \
--cc=dev@dpdk.org \
--cc=nd@arm.com \
--cc=sameh.gobriel@intel.com \
--cc=thomas@monjalon.net \
--cc=yipeng1.wang@intel.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).