From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from casper.infradead.org (casper.infradead.org [90.155.50.34]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 50D2917C91 for ; Thu, 25 Apr 2024 12:00:07 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=90.155.50.34 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1714046417; cv=none; b=fnhDsvXDr9knqy+ZPIZXwtKMvoKqOmrKqZvg6yxOIATooXid6cs7hquvLRBWPk3qFVGTQXCygQicT/O4f7DrKQH3w8+gVxnWXmPbleErPj37u6ege60Gmq8KRONpQ/nZzldsY+qyL7DuSCAM9OIzxH/t3jsjKAFnoCYxsSK1v84= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1714046417; c=relaxed/simple; bh=vUSQ1Qugb5ISOY5k5XcEbT2RvJIJHb7tGtwFpaeUyjw=; h=Subject:From:To:Message-Id:Date; b=e+Ui/WGltGPHuncppAyvEGo94j4JkWEXnMQObl44+fad5tHpG1voVq2aYH2nqYaoGWghzQ0icrq7InbGI/7b4LdILzsGCtbpBk7ND03miegt1cYsI12SxA7nmAvH0QgIqs9PxTL8+gGv04mv9eiUpV06LgDNQeb7Gyovn3wXJOo= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=kernel.dk; spf=fail smtp.mailfrom=kernel.dk; dkim=pass (2048-bit key) header.d=infradead.org header.i=@infradead.org header.b=X+PCXg43; arc=none smtp.client-ip=90.155.50.34 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=kernel.dk Authentication-Results: smtp.subspace.kernel.org; spf=fail smtp.mailfrom=kernel.dk Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=infradead.org header.i=@infradead.org header.b="X+PCXg43" DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=infradead.org; s=casper.20170209; h=Date:Message-Id:To:From:Subject:Sender: Reply-To:Cc:MIME-Version:Content-Type:Content-Transfer-Encoding:Content-ID: Content-Description:In-Reply-To:References; bh=9aRaBwMQ87bYeaz5Vid7MbtzG4QgvH9Ucy96e8Q1twg=; b=X+PCXg43UwEpwN5Ik+7SmGuyTJ lIyEJ0iYSI97KsEquYQwvtbS1JQu6oKS7x1T5MDyBpMPd9DMqjtZj5dT0IJoQqSmHWl8IGXgXxnwV yYBc7qkuem/kU2ous05xqBe1oBsAephpvDI8EU++LGKG7zkVRLUvkCPnyMkispBdTWGTHkHsd7eaC HncAZbOMEvvUYl7BnYe9H1EcXUX6zgZwITODGiYYLvBEK1x7NvjDk1AAP/YBxXwaOxNGIon5c6aPW 1p1cBzprutRppxHjAozysEsFy+ylOiaOQTXTkXPxqho7ZhUjwehNSjlyUKFBLaao/yJZv+zo8Uw8x 5nTboXDw==; Received: from [96.43.243.2] (helo=kernel.dk) by casper.infradead.org with esmtpsa (Exim 4.97.1 #2 (Red Hat Linux)) id 1rzxlk-00000002xsp-1Ihb for fio@vger.kernel.org; Thu, 25 Apr 2024 12:00:04 +0000 Received: by kernel.dk (Postfix, from userid 1000) id 654C31BC0128; Thu, 25 Apr 2024 06:00:01 -0600 (MDT) Subject: Recent changes (master) From: Jens Axboe To: X-Mailer: mail (GNU Mailutils 3.7) Message-Id: <20240425120001.654C31BC0128@kernel.dk> Date: Thu, 25 Apr 2024 06:00:01 -0600 (MDT) Precedence: bulk X-Mailing-List: fio@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: The following changes since commit c948ee34afde7eda14adf82512772b03f6fb1d69: Merge branch 'master' of https://github.com/celestinechen/fio (2024-04-19 13:51:05 -0600) are available in the Git repository at: git://git.kernel.dk/fio.git master for you to fetch changes up to 349bbcb2e36658db05d5247ef8cbcb285d58dfbf: docs: update for new data placement options (2024-04-24 13:44:09 -0400) ---------------------------------------------------------------- Vincent Fu (9): fio: rename fdp.[c,h] to dataplacement.[c,h] fio: create over-arching data placement option t/nvmept_fdp.py: test script for FDP fio: support NVMe streams options: reject placement IDs larger than the max options: parse placement IDs as unsigned values dataplacement: add a debug print for IOs t/nvmept_streams: test NVMe streams support docs: update for new data placement options HOWTO.rst | 36 ++- Makefile | 2 +- cconv.c | 18 +- fdp.c => dataplacement.c | 37 ++- dataplacement.h | 37 +++ engines/xnvme.c | 2 +- fdp.h | 29 -- filesetup.c | 4 +- fio.1 | 35 ++- init.c | 10 +- io_u.c | 4 +- ioengines.h | 2 +- options.c | 62 +++- server.h | 2 +- t/nvmept_fdp.py | 745 +++++++++++++++++++++++++++++++++++++++++++++++ t/nvmept_streams.py | 520 +++++++++++++++++++++++++++++++++ thread_options.h | 15 +- 17 files changed, 1466 insertions(+), 94 deletions(-) rename fdp.c => dataplacement.c (69%) create mode 100644 dataplacement.h delete mode 100644 fdp.h create mode 100755 t/nvmept_fdp.py create mode 100755 t/nvmept_streams.py --- Diff of recent changes: diff --git a/HOWTO.rst b/HOWTO.rst index 2f108e36..2f8ef6d4 100644 --- a/HOWTO.rst +++ b/HOWTO.rst @@ -2500,7 +2500,24 @@ with the caveat that when used on the command line, they must come after the Enable Flexible Data Placement mode for write commands. -.. option:: fdp_pli_select=str : [io_uring_cmd] [xnvme] +.. option:: dataplacement=str : [io_uring_cmd] [xnvme] + + Specifies the data placement directive type to use for write commands. + The following types are supported: + + **none** + Do not use a data placement directive. This is the + default. + + **fdp** + Use Flexible Data Placement directives for write + commands. This is equivalent to specifying + :option:`fdp` =1. + + **streams** + Use Streams directives for write commands. + +.. option:: plid_select=str, fdp_pli_select=str : [io_uring_cmd] [xnvme] Defines how fio decides which placement ID to use next. The following types are defined: @@ -2512,16 +2529,17 @@ with the caveat that when used on the command line, they must come after the Round robin over available placement IDs. This is the default. - The available placement ID index/indices is defined by the option - :option:`fdp_pli`. + The available placement ID (indices) are defined by the option + :option:`plids`. -.. option:: fdp_pli=str : [io_uring_cmd] [xnvme] +.. option:: plids=str, fdp_pli=str : [io_uring_cmd] [xnvme] - Select which Placement ID Index/Indicies this job is allowed to use for - writes. By default, the job will cycle through all available Placement - IDs, so use this to isolate these identifiers to specific jobs. If you - want fio to use placement identifier only at indices 0, 2 and 5 specify - ``fdp_pli=0,2,5``. + Select which Placement IDs (streams) or Placement ID Indices (FDP) this + job is allowed to use for writes. For FDP by default, the job will + cycle through all available Placement IDs, so use this to isolate these + identifiers to specific jobs. If you want fio to use FDP placement + identifiers only at indices 0, 2 and 5 specify ``plids=0,2,5``. For + streams this should be a comma-separated list of Stream IDs. .. option:: md_per_io_size=int : [io_uring_cmd] [xnvme] diff --git a/Makefile b/Makefile index cc8164b2..be57e296 100644 --- a/Makefile +++ b/Makefile @@ -62,7 +62,7 @@ SOURCE := $(sort $(patsubst $(SRCDIR)/%,%,$(wildcard $(SRCDIR)/crc/*.c)) \ gettime-thread.c helpers.c json.c idletime.c td_error.c \ profiles/tiobench.c profiles/act.c io_u_queue.c filelock.c \ workqueue.c rate-submit.c optgroup.c helper_thread.c \ - steadystate.c zone-dist.c zbd.c dedupe.c fdp.c + steadystate.c zone-dist.c zbd.c dedupe.c dataplacement.c ifdef CONFIG_LIBHDFS HDFSFLAGS= -I $(JAVA_HOME)/include -I $(JAVA_HOME)/include/linux -I $(FIO_LIBHDFS_INCLUDE) diff --git a/cconv.c b/cconv.c index ead47248..16112248 100644 --- a/cconv.c +++ b/cconv.c @@ -354,10 +354,11 @@ int convert_thread_options_to_cpu(struct thread_options *o, o->merge_blktrace_iters[i].u.f = fio_uint64_to_double(le64_to_cpu(top->merge_blktrace_iters[i].u.i)); o->fdp = le32_to_cpu(top->fdp); - o->fdp_pli_select = le32_to_cpu(top->fdp_pli_select); - o->fdp_nrpli = le32_to_cpu(top->fdp_nrpli); - for (i = 0; i < o->fdp_nrpli; i++) - o->fdp_plis[i] = le32_to_cpu(top->fdp_plis[i]); + o->dp_type = le32_to_cpu(top->dp_type); + o->dp_id_select = le32_to_cpu(top->dp_id_select); + o->dp_nr_ids = le32_to_cpu(top->dp_nr_ids); + for (i = 0; i < o->dp_nr_ids; i++) + o->dp_ids[i] = le32_to_cpu(top->dp_ids[i]); #if 0 uint8_t cpumask[FIO_TOP_STR_MAX]; uint8_t verify_cpumask[FIO_TOP_STR_MAX]; @@ -652,10 +653,11 @@ void convert_thread_options_to_net(struct thread_options_pack *top, top->merge_blktrace_iters[i].u.i = __cpu_to_le64(fio_double_to_uint64(o->merge_blktrace_iters[i].u.f)); top->fdp = cpu_to_le32(o->fdp); - top->fdp_pli_select = cpu_to_le32(o->fdp_pli_select); - top->fdp_nrpli = cpu_to_le32(o->fdp_nrpli); - for (i = 0; i < o->fdp_nrpli; i++) - top->fdp_plis[i] = cpu_to_le32(o->fdp_plis[i]); + top->dp_type = cpu_to_le32(o->dp_type); + top->dp_id_select = cpu_to_le32(o->dp_id_select); + top->dp_nr_ids = cpu_to_le32(o->dp_nr_ids); + for (i = 0; i < o->dp_nr_ids; i++) + top->dp_ids[i] = cpu_to_le32(o->dp_ids[i]); #if 0 uint8_t cpumask[FIO_TOP_STR_MAX]; uint8_t verify_cpumask[FIO_TOP_STR_MAX]; diff --git a/fdp.c b/dataplacement.c similarity index 69% rename from fdp.c rename to dataplacement.c index 49c80d2c..1d5b21ed 100644 --- a/fdp.c +++ b/dataplacement.c @@ -13,7 +13,7 @@ #include "file.h" #include "pshared.h" -#include "fdp.h" +#include "dataplacement.h" static int fdp_ruh_info(struct thread_data *td, struct fio_file *f, struct fio_ruhs_info *ruhs) @@ -49,6 +49,20 @@ static int init_ruh_info(struct thread_data *td, struct fio_file *f) if (!ruhs) return -ENOMEM; + /* set up the data structure used for FDP to work with the supplied stream IDs */ + if (td->o.dp_type == FIO_DP_STREAMS) { + if (!td->o.dp_nr_ids) { + log_err("fio: stream IDs must be provided for dataplacement=streams\n"); + return -EINVAL; + } + ruhs->nr_ruhs = td->o.dp_nr_ids; + for (int i = 0; i < ruhs->nr_ruhs; i++) + ruhs->plis[i] = td->o.dp_ids[i]; + + f->ruhs_info = ruhs; + return 0; + } + ret = fdp_ruh_info(td, f, ruhs); if (ret) { log_info("fio: ruh info failed for %s (%d)\n", @@ -59,13 +73,13 @@ static int init_ruh_info(struct thread_data *td, struct fio_file *f) if (ruhs->nr_ruhs > FDP_MAX_RUHS) ruhs->nr_ruhs = FDP_MAX_RUHS; - if (td->o.fdp_nrpli == 0) { + if (td->o.dp_nr_ids == 0) { f->ruhs_info = ruhs; return 0; } - for (i = 0; i < td->o.fdp_nrpli; i++) { - if (td->o.fdp_plis[i] >= ruhs->nr_ruhs) { + for (i = 0; i < td->o.dp_nr_ids; i++) { + if (td->o.dp_ids[i] >= ruhs->nr_ruhs) { ret = -EINVAL; goto out; } @@ -77,16 +91,16 @@ static int init_ruh_info(struct thread_data *td, struct fio_file *f) goto out; } - tmp->nr_ruhs = td->o.fdp_nrpli; - for (i = 0; i < td->o.fdp_nrpli; i++) - tmp->plis[i] = ruhs->plis[td->o.fdp_plis[i]]; + tmp->nr_ruhs = td->o.dp_nr_ids; + for (i = 0; i < td->o.dp_nr_ids; i++) + tmp->plis[i] = ruhs->plis[td->o.dp_ids[i]]; f->ruhs_info = tmp; out: sfree(ruhs); return ret; } -int fdp_init(struct thread_data *td) +int dp_init(struct thread_data *td) { struct fio_file *f; int i, ret = 0; @@ -107,7 +121,7 @@ void fdp_free_ruhs_info(struct fio_file *f) f->ruhs_info = NULL; } -void fdp_fill_dspec_data(struct thread_data *td, struct io_u *io_u) +void dp_fill_dspec_data(struct thread_data *td, struct io_u *io_u) { struct fio_file *f = io_u->file; struct fio_ruhs_info *ruhs = f->ruhs_info; @@ -119,7 +133,7 @@ void fdp_fill_dspec_data(struct thread_data *td, struct io_u *io_u) return; } - if (td->o.fdp_pli_select == FIO_FDP_RR) { + if (td->o.dp_id_select == FIO_DP_RR) { if (ruhs->pli_loc >= ruhs->nr_ruhs) ruhs->pli_loc = 0; @@ -129,6 +143,7 @@ void fdp_fill_dspec_data(struct thread_data *td, struct io_u *io_u) dspec = ruhs->plis[ruhs->pli_loc]; } - io_u->dtype = FDP_DIR_DTYPE; + io_u->dtype = td->o.dp_type == FIO_DP_FDP ? FDP_DIR_DTYPE : STREAMS_DIR_DTYPE; io_u->dspec = dspec; + dprint(FD_IO, "dtype set to 0x%x, dspec set to 0x%x\n", io_u->dtype, io_u->dspec); } diff --git a/dataplacement.h b/dataplacement.h new file mode 100644 index 00000000..b5718c86 --- /dev/null +++ b/dataplacement.h @@ -0,0 +1,37 @@ +#ifndef FIO_DATAPLACEMENT_H +#define FIO_DATAPLACEMENT_H + +#include "io_u.h" + +#define STREAMS_DIR_DTYPE 1 +#define FDP_DIR_DTYPE 2 +#define FDP_MAX_RUHS 128 +#define FIO_MAX_DP_IDS 16 + +/* + * How fio chooses what placement identifier to use next. Choice of + * uniformly random, or roundrobin. + */ +enum { + FIO_DP_RANDOM = 0x1, + FIO_DP_RR = 0x2, +}; + + +enum { + FIO_DP_NONE = 0x0, + FIO_DP_FDP = 0x1, + FIO_DP_STREAMS = 0x2, +}; + +struct fio_ruhs_info { + uint32_t nr_ruhs; + uint32_t pli_loc; + uint16_t plis[]; +}; + +int dp_init(struct thread_data *td); +void fdp_free_ruhs_info(struct fio_file *f); +void dp_fill_dspec_data(struct thread_data *td, struct io_u *io_u); + +#endif /* FIO_DATAPLACEMENT_H */ diff --git a/engines/xnvme.c b/engines/xnvme.c index a8137286..6ba4aa46 100644 --- a/engines/xnvme.c +++ b/engines/xnvme.c @@ -13,7 +13,7 @@ #include "fio.h" #include "verify.h" #include "zbd_types.h" -#include "fdp.h" +#include "dataplacement.h" #include "optgroup.h" static pthread_mutex_t g_serialize = PTHREAD_MUTEX_INITIALIZER; diff --git a/fdp.h b/fdp.h deleted file mode 100644 index accbac38..00000000 --- a/fdp.h +++ /dev/null @@ -1,29 +0,0 @@ -#ifndef FIO_FDP_H -#define FIO_FDP_H - -#include "io_u.h" - -#define FDP_DIR_DTYPE 2 -#define FDP_MAX_RUHS 128 - -/* - * How fio chooses what placement identifier to use next. Choice of - * uniformly random, or roundrobin. - */ - -enum { - FIO_FDP_RANDOM = 0x1, - FIO_FDP_RR = 0x2, -}; - -struct fio_ruhs_info { - uint32_t nr_ruhs; - uint32_t pli_loc; - uint16_t plis[]; -}; - -int fdp_init(struct thread_data *td); -void fdp_free_ruhs_info(struct fio_file *f); -void fdp_fill_dspec_data(struct thread_data *td, struct io_u *io_u); - -#endif /* FIO_FDP_H */ diff --git a/filesetup.c b/filesetup.c index 2d277a64..cb42a852 100644 --- a/filesetup.c +++ b/filesetup.c @@ -1411,8 +1411,8 @@ done: td_restore_runstate(td, old_state); - if (td->o.fdp) { - err = fdp_init(td); + if (td->o.dp_type != FIO_DP_NONE) { + err = dp_init(td); if (err) goto err_out; } diff --git a/fio.1 b/fio.1 index 5fd3603d..ee812494 100644 --- a/fio.1 +++ b/fio.1 @@ -2264,7 +2264,26 @@ file blocks are fully allocated and the disk request could be issued immediately .BI (io_uring_cmd,xnvme)fdp \fR=\fPbool Enable Flexible Data Placement mode for write commands. .TP -.BI (io_uring_cmd,xnvme)fdp_pli_select \fR=\fPstr +.BI (io_uring_cmd,xnvme)dataplacement \fR=\fPstr +Specifies the data placement directive type to use for write commands. The +following types are supported: +.RS +.RS +.TP +.B none +Do not use a data placement directive. This is the default. +.TP +.B fdp +Use Flexible Data placement directives for write commands. This is equivalent +to specifying \fBfdp\fR=1. +.TP +.B streams +Use Streams directives for write commands. +.TP +.RE +.RE +.TP +.BI (io_uring_cmd,xnvme)plid_select=str, fdp_pli_select \fR=\fPstr Defines how fio decides which placement ID to use next. The following types are defined: .RS @@ -2277,14 +2296,16 @@ Choose a placement ID at random (uniform). Round robin over available placement IDs. This is the default. .RE .P -The available placement ID index/indices is defined by \fBfdp_pli\fR option. +The available placement ID (indices) are defined by the \fBplids\fR option. .RE .TP -.BI (io_uring_cmd,xnvme)fdp_pli \fR=\fPstr -Select which Placement ID Index/Indicies this job is allowed to use for writes. -By default, the job will cycle through all available Placement IDs, so use this -to isolate these identifiers to specific jobs. If you want fio to use placement -identifier only at indices 0, 2 and 5 specify, you would set `fdp_pli=0,2,5`. +.BI (io_uring_cmd,xnvme)plids=str, fdp_pli \fR=\fPstr +Select which Placement IDs (streams) or Placement ID Indicies (FDP) this job is +allowed to use for writes. For FDP by default, the job will cycle through all +available Placement IDs, so use this to isolate these identifiers to specific +jobs. If you want fio to use placement identifier only at indices 0, 2 and 5 +specify, you would set `plids=0,2,5`. For streams this should be a +comma-separated list of Stream IDs. .TP .BI (io_uring_cmd,xnvme)md_per_io_size \fR=\fPint Size in bytes for separate metadata buffer per IO. Default: 0. diff --git a/init.c b/init.c index 7a0b14a3..ff3e9a90 100644 --- a/init.c +++ b/init.c @@ -1015,7 +1015,15 @@ static int fixup_options(struct thread_data *td) ret |= 1; } - + if (td->o.fdp) { + if (fio_option_is_set(&td->o, dp_type) && + (td->o.dp_type == FIO_DP_STREAMS || td->o.dp_type == FIO_DP_NONE)) { + log_err("fio: fdp=1 is not compatible with dataplacement={streams, none}\n"); + ret |= 1; + } else { + td->o.dp_type = FIO_DP_FDP; + } + } return ret; } diff --git a/io_u.c b/io_u.c index a499ff07..a090e121 100644 --- a/io_u.c +++ b/io_u.c @@ -1065,8 +1065,8 @@ static int fill_io_u(struct thread_data *td, struct io_u *io_u) } } - if (td->o.fdp) - fdp_fill_dspec_data(td, io_u); + if (td->o.dp_type != FIO_DP_NONE) + dp_fill_dspec_data(td, io_u); if (io_u->offset + io_u->buflen > io_u->file->real_file_size) { dprint(FD_IO, "io_u %p, off=0x%llx + len=0x%llx exceeds file size=0x%llx\n", diff --git a/ioengines.h b/ioengines.h index 4fe9bb98..d5b0cafe 100644 --- a/ioengines.h +++ b/ioengines.h @@ -7,7 +7,7 @@ #include "flist.h" #include "io_u.h" #include "zbd_types.h" -#include "fdp.h" +#include "dataplacement.h" #define FIO_IOOPS_VERSION 34 diff --git a/options.c b/options.c index de935efc..61ea41cc 100644 --- a/options.c +++ b/options.c @@ -270,12 +270,19 @@ static int str_fdp_pli_cb(void *data, const char *input) strip_blank_front(&str); strip_blank_end(str); - while ((v = strsep(&str, ",")) != NULL && i < FIO_MAX_PLIS) - td->o.fdp_plis[i++] = strtoll(v, NULL, 0); + while ((v = strsep(&str, ",")) != NULL && i < FIO_MAX_DP_IDS) { + unsigned long long id = strtoull(v, NULL, 0); + if (id > 0xFFFF) { + log_err("Placement IDs cannot exceed 0xFFFF\n"); + free(p); + return 1; + } + td->o.dp_ids[i++] = id; + } free(p); - qsort(td->o.fdp_plis, i, sizeof(*td->o.fdp_plis), fio_fdp_cmp); - td->o.fdp_nrpli = i; + qsort(td->o.dp_ids, i, sizeof(*td->o.dp_ids), fio_fdp_cmp); + td->o.dp_nr_ids = i; return 0; } @@ -3710,32 +3717,59 @@ struct fio_option fio_options[FIO_MAX_OPTS] = { .group = FIO_OPT_G_INVALID, }, { - .name = "fdp_pli_select", - .lname = "FDP Placement ID select", + .name = "dataplacement", + .alias = "data_placement", + .lname = "Data Placement interface", + .type = FIO_OPT_STR, + .off1 = offsetof(struct thread_options, dp_type), + .help = "Data Placement interface to use", + .def = "none", + .category = FIO_OPT_C_IO, + .group = FIO_OPT_G_INVALID, + .posval = { + { .ival = "none", + .oval = FIO_DP_NONE, + .help = "Do not specify a data placement interface", + }, + { .ival = "fdp", + .oval = FIO_DP_FDP, + .help = "Use Flexible Data Placement interface", + }, + { .ival = "streams", + .oval = FIO_DP_STREAMS, + .help = "Use Streams interface", + }, + }, + }, + { + .name = "plid_select", + .alias = "fdp_pli_select", + .lname = "Data Placement ID selection strategy", .type = FIO_OPT_STR, - .off1 = offsetof(struct thread_options, fdp_pli_select), - .help = "Select which FDP placement ID to use next", + .off1 = offsetof(struct thread_options, dp_id_select), + .help = "Strategy for selecting next Data Placement ID", .def = "roundrobin", .category = FIO_OPT_C_IO, .group = FIO_OPT_G_INVALID, .posval = { { .ival = "random", - .oval = FIO_FDP_RANDOM, + .oval = FIO_DP_RANDOM, .help = "Choose a Placement ID at random (uniform)", }, { .ival = "roundrobin", - .oval = FIO_FDP_RR, + .oval = FIO_DP_RR, .help = "Round robin select Placement IDs", }, }, }, { - .name = "fdp_pli", - .lname = "FDP Placement ID indicies", + .name = "plids", + .alias = "fdp_pli", + .lname = "Stream IDs/Data Placement ID indices", .type = FIO_OPT_STR, .cb = str_fdp_pli_cb, - .off1 = offsetof(struct thread_options, fdp_plis), - .help = "Sets which placement ids to use (defaults to all)", + .off1 = offsetof(struct thread_options, dp_ids), + .help = "Sets which Data Placement ids to use (defaults to all for FDP)", .hide = 1, .category = FIO_OPT_C_IO, .group = FIO_OPT_G_INVALID, diff --git a/server.h b/server.h index 6d2659b0..83ce449b 100644 --- a/server.h +++ b/server.h @@ -51,7 +51,7 @@ struct fio_net_cmd_reply { }; enum { - FIO_SERVER_VER = 103, + FIO_SERVER_VER = 104, FIO_SERVER_MAX_FRAGMENT_PDU = 1024, FIO_SERVER_MAX_CMD_MB = 2048, diff --git a/t/nvmept_fdp.py b/t/nvmept_fdp.py new file mode 100755 index 00000000..031b439c --- /dev/null +++ b/t/nvmept_fdp.py @@ -0,0 +1,745 @@ +#!/usr/bin/env python3 +# +# Copyright 2024 Samsung Electronics Co., Ltd All Rights Reserved +# +# For conditions of distribution and use, see the accompanying COPYING file. +# +""" +# nvmept_fdp.py +# +# Test fio's io_uring_cmd ioengine with NVMe pass-through FDP write commands. +# +# USAGE +# see python3 nvmept_fdp.py --help +# +# EXAMPLES +# python3 t/nvmept_fdp.py --dut /dev/ng0n1 +# python3 t/nvmept_fdp.py --dut /dev/ng1n1 -f ./fio +# +# REQUIREMENTS +# Python 3.6 +# Device formatted with LBA data size 4096 bytes +# Device with at least five placement IDs +# +# WARNING +# This is a destructive test +""" +import os +import sys +import json +import time +import locale +import logging +import argparse +import subprocess +from pathlib import Path +from fiotestlib import FioJobCmdTest, run_fio_tests +from fiotestcommon import SUCCESS_NONZERO + + +class FDPTest(FioJobCmdTest): + """ + NVMe pass-through test class. Check to make sure output for selected data + direction(s) is non-zero and that zero data appears for other directions. + """ + + def setup(self, parameters): + """Setup a test.""" + + fio_args = [ + "--name=nvmept-fdp", + "--ioengine=io_uring_cmd", + "--cmd_type=nvme", + "--randrepeat=0", + f"--filename={self.fio_opts['filename']}", + f"--rw={self.fio_opts['rw']}", + f"--output={self.filenames['output']}", + f"--output-format={self.fio_opts['output-format']}", + ] + for opt in ['fixedbufs', 'nonvectored', 'force_async', 'registerfiles', + 'sqthread_poll', 'sqthread_poll_cpu', 'hipri', 'nowait', + 'time_based', 'runtime', 'verify', 'io_size', 'num_range', + 'iodepth', 'iodepth_batch', 'iodepth_batch_complete', + 'size', 'rate', 'bs', 'bssplit', 'bsrange', 'randrepeat', + 'buffer_pattern', 'verify_pattern', 'offset', 'fdp', + 'fdp_pli', 'fdp_pli_select', 'dataplacement', 'plid_select', + 'plids', 'number_ios']: + if opt in self.fio_opts: + option = f"--{opt}={self.fio_opts[opt]}" + fio_args.append(option) + + super().setup(fio_args) + + + def check_result(self): + try: + self._check_result() + finally: + if not update_all_ruhs(self.fio_opts['filename']): + logging.error("Could not reset device") + if not check_all_ruhs(self.fio_opts['filename']): + logging.error("Reclaim units have inconsistent RUAMW values") + + + def _check_result(self): + + super().check_result() + + if 'rw' not in self.fio_opts or \ + not self.passed or \ + 'json' not in self.fio_opts['output-format']: + return + + job = self.json_data['jobs'][0] + + if self.fio_opts['rw'] in ['read', 'randread']: + self.passed = self.check_all_ddirs(['read'], job) + elif self.fio_opts['rw'] in ['write', 'randwrite']: + if 'verify' not in self.fio_opts: + self.passed = self.check_all_ddirs(['write'], job) + else: + self.passed = self.check_all_ddirs(['read', 'write'], job) + elif self.fio_opts['rw'] in ['trim', 'randtrim']: + self.passed = self.check_all_ddirs(['trim'], job) + elif self.fio_opts['rw'] in ['readwrite', 'randrw']: + self.passed = self.check_all_ddirs(['read', 'write'], job) + elif self.fio_opts['rw'] in ['trimwrite', 'randtrimwrite']: + self.passed = self.check_all_ddirs(['trim', 'write'], job) + else: + logging.error("Unhandled rw value %s", self.fio_opts['rw']) + self.passed = False + + if 'iodepth' in self.fio_opts: + # We will need to figure something out if any test uses an iodepth + # different from 8 + if job['iodepth_level']['8'] < 95: + logging.error("Did not achieve requested iodepth") + self.passed = False + else: + logging.debug("iodepth 8 target met %s", job['iodepth_level']['8']) + + +class FDPMultiplePLIDTest(FDPTest): + """ + Write to multiple placement IDs. + """ + + def setup(self, parameters): + mapping = { + 'nruhsd': FIO_FDP_NUMBER_PLIDS, + 'max_ruamw': FIO_FDP_MAX_RUAMW, + } + if 'number_ios' in self.fio_opts and isinstance(self.fio_opts['number_ios'], str): + self.fio_opts['number_ios'] = eval(self.fio_opts['number_ios'].format(**mapping)) + + super().setup(parameters) + + def _check_result(self): + if 'fdp_pli' in self.fio_opts: + plid_list = self.fio_opts['fdp_pli'].split(',') + elif 'plids' in self.fio_opts: + plid_list = self.fio_opts['plids'].split(',') + else: + plid_list = list(range(FIO_FDP_NUMBER_PLIDS)) + + plid_list = sorted([int(i) for i in plid_list]) + logging.debug("plid_list: %s", str(plid_list)) + + fdp_status = get_fdp_status(self.fio_opts['filename']) + + select = "roundrobin" + if 'fdp_pli_select' in self.fio_opts: + select = self.fio_opts['fdp_pli_select'] + elif 'plid_select' in self.fio_opts: + select = self.fio_opts['plid_select'] + + if select == "roundrobin": + self._check_robin(plid_list, fdp_status) + elif select == "random": + self._check_random(plid_list, fdp_status) + else: + logging.error("Unknown plid selection strategy %s", select) + self.passed = False + + super()._check_result() + + def _check_robin(self, plid_list, fdp_status): + """ + With round robin we can know exactly how many writes each PLID will + receive. + """ + ruamw = [FIO_FDP_MAX_RUAMW] * FIO_FDP_NUMBER_PLIDS + + remainder = int(self.fio_opts['number_ios'] % len(plid_list)) + whole = int((self.fio_opts['number_ios'] - remainder) / len(plid_list)) + logging.debug("PLIDs in the list should receive %d writes; %d PLIDs will receive one extra", + whole, remainder) + + for plid in plid_list: + ruamw[plid] -= whole + if remainder: + ruamw[plid] -= 1 + remainder -= 1 + logging.debug("Expected ruamw values: %s", str(ruamw)) + + for idx, ruhs in enumerate(fdp_status['ruhss']): + if ruhs['ruamw'] != ruamw[idx]: + logging.error("RUAMW mismatch with idx %d, pid %d, expected %d, observed %d", idx, + ruhs['pid'], ruamw[idx], ruhs['ruamw']) + self.passed = False + break + + logging.debug("RUAMW match with idx %d, pid %d: ruamw=%d", idx, ruhs['pid'], ruamw[idx]) + + def _check_random(self, plid_list, fdp_status): + """ + With random selection, a set of PLIDs will receive all the write + operations and the remainder will be untouched. + """ + + total_ruamw = 0 + for plid in plid_list: + total_ruamw += fdp_status['ruhss'][plid]['ruamw'] + + expected = len(plid_list) * FIO_FDP_MAX_RUAMW - self.fio_opts['number_ios'] + if total_ruamw != expected: + logging.error("Expected total ruamw %d for plids %s, observed %d", expected, + str(plid_list), total_ruamw) + self.passed = False + else: + logging.debug("Observed expected total ruamw %d for plids %s", expected, str(plid_list)) + + for idx, ruhs in enumerate(fdp_status['ruhss']): + if idx in plid_list: + continue + if ruhs['ruamw'] != FIO_FDP_MAX_RUAMW: + logging.error("Unexpected ruamw %d for idx %d, pid %d, expected %d", ruhs['ruamw'], + idx, ruhs['pid'], FIO_FDP_MAX_RUAMW) + self.passed = False + else: + logging.debug("Observed expected ruamw %d for idx %d, pid %d", ruhs['ruamw'], idx, + ruhs['pid']) + + +class FDPSinglePLIDTest(FDPTest): + """ + Write to a single placement ID only. + """ + + def _check_result(self): + if 'plids' in self.fio_opts: + plid = self.fio_opts['plids'] + elif 'fdp_pli' in self.fio_opts: + plid = self.fio_opts['fdp_pli'] + else: + plid = 0 + + fdp_status = get_fdp_status(self.fio_opts['filename']) + ruamw = fdp_status['ruhss'][plid]['ruamw'] + lba_count = self.fio_opts['number_ios'] + + if FIO_FDP_MAX_RUAMW - lba_count != ruamw: + logging.error("FDP accounting mismatch for plid %d; expected ruamw %d, observed %d", + plid, FIO_FDP_MAX_RUAMW - lba_count, ruamw) + self.passed = False + else: + logging.debug("FDP accounting as expected for plid %d; ruamw = %d", plid, ruamw) + + super()._check_result() + + +class FDPReadTest(FDPTest): + """ + Read workload test. + """ + + def _check_result(self): + ruamw = check_all_ruhs(self.fio_opts['filename']) + + if ruamw != FIO_FDP_MAX_RUAMW: + logging.error("Read workload affected FDP ruamw") + self.passed = False + else: + logging.debug("Read workload did not disturb FDP ruamw") + super()._check_result() + + +def get_fdp_status(dut): + """ + Run the nvme-cli command to obtain FDP status and return result as a JSON + object. + """ + + cmd = f"sudo nvme fdp status --output-format=json {dut}" + cmd = cmd.split(' ') + cmd_result = subprocess.run(cmd, capture_output=True, check=False, + encoding=locale.getpreferredencoding()) + + if cmd_result.returncode != 0: + logging.error("Error obtaining device %s FDP status: %s", dut, cmd_result.stderr) + return False + + return json.loads(cmd_result.stdout) + + +def update_ruh(dut, plid): + """ + Update reclaim unit handles with specified ID(s). This tells the device to + point the RUH to a new (empty) reclaim unit. + """ + + ids = ','.join(plid) if isinstance(plid, list) else plid + cmd = f"nvme fdp update --pids={ids} {dut}" + cmd = cmd.split(' ') + cmd_result = subprocess.run(cmd, capture_output=True, check=False, + encoding=locale.getpreferredencoding()) + + if cmd_result.returncode != 0: + logging.error("Error updating RUH %s ID(s) %s", dut, ids) + return False + + return True + + +def update_all_ruhs(dut): + """ + Update all reclaim unit handles on the device. + """ + + fdp_status = get_fdp_status(dut) + for ruhs in fdp_status['ruhss']: + if not update_ruh(dut, ruhs['pid']): + return False + + return True + + +def check_all_ruhs(dut): + """ + Check that all RUHs have the same value for reclaim unit available media + writes (RUAMW). Return the RUAMW value. + """ + + fdp_status = get_fdp_status(dut) + ruh_status = fdp_status['ruhss'] + + ruamw = ruh_status[0]['ruamw'] + for ruhs in ruh_status: + if ruhs['ruamw'] != ruamw: + logging.error("RUAMW mismatch: found %d, expected %d", ruhs['ruamw'], ruamw) + return False + + return ruamw + + +TEST_LIST = [ + # Write one LBA to one PLID using both the old and new sets of options + ## omit fdp_pli_select/plid_select + { + "test_id": 1, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "number_ios": 1, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 3, + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + { + "test_id": 2, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": 1, + "verify": "crc32c", + "dataplacement": "fdp", + "plids": 3, + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + ## fdp_pli_select/plid_select=roundrobin + { + "test_id": 3, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "number_ios": 1, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 3, + "fdp_pli_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + { + "test_id": 4, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": 1, + "verify": "crc32c", + "dataplacement": "fdp", + "plids": 3, + "plid_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + ## fdp_pli_select/plid_select=random + { + "test_id": 5, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "number_ios": 1, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 3, + "fdp_pli_select": "random", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + { + "test_id": 6, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": 1, + "verify": "crc32c", + "dataplacement": "fdp", + "plids": 3, + "plid_select": "random", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + # Write four LBAs to one PLID using both the old and new sets of options + ## omit fdp_pli_select/plid_select + { + "test_id": 7, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "number_ios": 4, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 1, + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + { + "test_id": 8, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": 4, + "verify": "crc32c", + "dataplacement": "fdp", + "plids": 1, + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + ## fdp_pli_select/plid_select=roundrobin + { + "test_id": 9, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "number_ios": 4, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 1, + "fdp_pli_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + { + "test_id": 10, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": 4, + "verify": "crc32c", + "dataplacement": "fdp", + "plids": 1, + "plid_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + ## fdp_pli_select/plid_select=random + { + "test_id": 11, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "number_ios": 4, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 1, + "fdp_pli_select": "random", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + { + "test_id": 12, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": 4, + "verify": "crc32c", + "dataplacement": "fdp", + "plids": 1, + "plid_select": "random", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + # Just a regular write without FDP directive--should land on plid 0 + { + "test_id": 13, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": 19, + "verify": "crc32c", + "output-format": "json", + }, + "test_class": FDPSinglePLIDTest, + }, + # Read workload + { + "test_id": 14, + "fio_opts": { + "rw": 'randread', + "bs": 4096, + "number_ios": 19, + "output-format": "json", + }, + "test_class": FDPReadTest, + }, + # write to multiple PLIDs using round robin to select PLIDs + ## write to all PLIDs using old and new sets of options + { + "test_id": 100, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "2*{nruhsd}+3", + "verify": "crc32c", + "fdp": 1, + "fdp_pli_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + { + "test_id": 101, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "2*{nruhsd}+3", + "verify": "crc32c", + "dataplacement": "fdp", + "plid_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + ## write to a subset of PLIDs using old and new sets of options + { + "test_id": 102, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "{nruhsd}+1", + "verify": "crc32c", + "fdp": 1, + "fdp_pli": "1,3", + "fdp_pli_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + { + "test_id": 103, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "{nruhsd}+1", + "verify": "crc32c", + "dataplacement": "fdp", + "plids": "1,3", + "plid_select": "roundrobin", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + # write to multiple PLIDs using random selection of PLIDs + ## write to all PLIDs using old and new sets of options + { + "test_id": 200, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "{max_ruamw}-1", + "verify": "crc32c", + "fdp": 1, + "fdp_pli_select": "random", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + { + "test_id": 201, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "{max_ruamw}-1", + "verify": "crc32c", + "dataplacement": "fdp", + "plid_select": "random", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + ## write to a subset of PLIDs using old and new sets of options + { + "test_id": 202, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "{max_ruamw}-1", + "verify": "crc32c", + "fdp": 1, + "fdp_pli": "1,3,4", + "fdp_pli_select": "random", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + { + "test_id": 203, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "number_ios": "{max_ruamw}-1", + "verify": "crc32c", + "dataplacement": "fdp", + "plids": "1,3,4", + "plid_select": "random", + "output-format": "json", + }, + "test_class": FDPMultiplePLIDTest, + }, + # Specify invalid options fdp=1 and dataplacement=none + { + "test_id": 300, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 4096, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 3, + "output-format": "normal", + "dataplacement": "none", + }, + "test_class": FDPTest, + "success": SUCCESS_NONZERO, + }, + # Specify invalid options fdp=1 and dataplacement=streams + { + "test_id": 301, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 4096, + "verify": "crc32c", + "fdp": 1, + "fdp_pli": 3, + "output-format": "normal", + "dataplacement": "streams", + }, + "test_class": FDPTest, + "success": SUCCESS_NONZERO, + }, +] + +def parse_args(): + """Parse command-line arguments.""" + + parser = argparse.ArgumentParser() + parser.add_argument('-d', '--debug', help='Enable debug messages', action='store_true') + parser.add_argument('-f', '--fio', help='path to file executable (e.g., ./fio)') + parser.add_argument('-a', '--artifact-root', help='artifact root directory') + parser.add_argument('-s', '--skip', nargs='+', type=int, + help='list of test(s) to skip') + parser.add_argument('-o', '--run-only', nargs='+', type=int, + help='list of test(s) to run, skipping all others') + parser.add_argument('--dut', help='target NVMe character device to test ' + '(e.g., /dev/ng0n1). WARNING: THIS IS A DESTRUCTIVE TEST', required=True) + args = parser.parse_args() + + return args + + +FIO_FDP_MAX_RUAMW = 0 +FIO_FDP_NUMBER_PLIDS = 0 + +def main(): + """Run tests using fio's io_uring_cmd ioengine to send NVMe pass through commands.""" + global FIO_FDP_MAX_RUAMW + global FIO_FDP_NUMBER_PLIDS + + args = parse_args() + + if args.debug: + logging.basicConfig(level=logging.DEBUG) + else: + logging.basicConfig(level=logging.INFO) + + artifact_root = args.artifact_root if args.artifact_root else \ + f"nvmept-fdp-test-{time.strftime('%Y%m%d-%H%M%S')}" + os.mkdir(artifact_root) + print(f"Artifact directory is {artifact_root}") + + if args.fio: + fio_path = str(Path(args.fio).absolute()) + else: + fio_path = 'fio' + print(f"fio path is {fio_path}") + + for test in TEST_LIST: + test['fio_opts']['filename'] = args.dut + + fdp_status = get_fdp_status(args.dut) + FIO_FDP_NUMBER_PLIDS = fdp_status['nruhsd'] + update_all_ruhs(args.dut) + FIO_FDP_MAX_RUAMW = check_all_ruhs(args.dut) + if not FIO_FDP_MAX_RUAMW: + sys.exit(-1) + + test_env = { + 'fio_path': fio_path, + 'fio_root': str(Path(__file__).absolute().parent.parent), + 'artifact_root': artifact_root, + 'basename': 'nvmept-fdp', + } + + _, failed, _ = run_fio_tests(TEST_LIST, test_env, args) + sys.exit(failed) + + +if __name__ == '__main__': + main() diff --git a/t/nvmept_streams.py b/t/nvmept_streams.py new file mode 100755 index 00000000..e5425506 --- /dev/null +++ b/t/nvmept_streams.py @@ -0,0 +1,520 @@ +#!/usr/bin/env python3 +# +# Copyright 2024 Samsung Electronics Co., Ltd All Rights Reserved +# +# For conditions of distribution and use, see the accompanying COPYING file. +# +""" +# nvmept_streams.py +# +# Test fio's NVMe streams support using the io_uring_cmd ioengine with NVMe +# pass-through commands. +# +# USAGE +# see python3 nvmept_streams.py --help +# +# EXAMPLES +# python3 t/nvmept_streams.py --dut /dev/ng0n1 +# python3 t/nvmept_streams.py --dut /dev/ng1n1 -f ./fio +# +# REQUIREMENTS +# Python 3.6 +# +# WARNING +# This is a destructive test +# +# Enable streams with +# nvme dir-send -D 0 -O 1 -e 1 -T 1 /dev/nvme0n1 +# +# See streams directive status with +# nvme dir-receive -D 0 -O 1 -H /dev/nvme0n1 +""" +import os +import sys +import time +import locale +import logging +import argparse +import subprocess +from pathlib import Path +from fiotestlib import FioJobCmdTest, run_fio_tests +from fiotestcommon import SUCCESS_NONZERO + + +class StreamsTest(FioJobCmdTest): + """ + NVMe pass-through test class for streams. Check to make sure output for + selected data direction(s) is non-zero and that zero data appears for other + directions. + """ + + def setup(self, parameters): + """Setup a test.""" + + fio_args = [ + "--name=nvmept-streams", + "--ioengine=io_uring_cmd", + "--cmd_type=nvme", + "--randrepeat=0", + f"--filename={self.fio_opts['filename']}", + f"--rw={self.fio_opts['rw']}", + f"--output={self.filenames['output']}", + f"--output-format={self.fio_opts['output-format']}", + ] + for opt in ['fixedbufs', 'nonvectored', 'force_async', 'registerfiles', + 'sqthread_poll', 'sqthread_poll_cpu', 'hipri', 'nowait', + 'time_based', 'runtime', 'verify', 'io_size', 'num_range', + 'iodepth', 'iodepth_batch', 'iodepth_batch_complete', + 'size', 'rate', 'bs', 'bssplit', 'bsrange', 'randrepeat', + 'buffer_pattern', 'verify_pattern', 'offset', 'dataplacement', + 'plids', 'plid_select' ]: + if opt in self.fio_opts: + option = f"--{opt}={self.fio_opts[opt]}" + fio_args.append(option) + + super().setup(fio_args) + + + def check_result(self): + try: + self._check_result() + finally: + release_all_streams(self.fio_opts['filename']) + + + def _check_result(self): + + super().check_result() + + if 'rw' not in self.fio_opts or \ + not self.passed or \ + 'json' not in self.fio_opts['output-format']: + return + + job = self.json_data['jobs'][0] + + if self.fio_opts['rw'] in ['read', 'randread']: + self.passed = self.check_all_ddirs(['read'], job) + elif self.fio_opts['rw'] in ['write', 'randwrite']: + if 'verify' not in self.fio_opts: + self.passed = self.check_all_ddirs(['write'], job) + else: + self.passed = self.check_all_ddirs(['read', 'write'], job) + elif self.fio_opts['rw'] in ['trim', 'randtrim']: + self.passed = self.check_all_ddirs(['trim'], job) + elif self.fio_opts['rw'] in ['readwrite', 'randrw']: + self.passed = self.check_all_ddirs(['read', 'write'], job) + elif self.fio_opts['rw'] in ['trimwrite', 'randtrimwrite']: + self.passed = self.check_all_ddirs(['trim', 'write'], job) + else: + logging.error("Unhandled rw value %s", self.fio_opts['rw']) + self.passed = False + + if 'iodepth' in self.fio_opts: + # We will need to figure something out if any test uses an iodepth + # different from 8 + if job['iodepth_level']['8'] < 95: + logging.error("Did not achieve requested iodepth") + self.passed = False + else: + logging.debug("iodepth 8 target met %s", job['iodepth_level']['8']) + + stream_ids = [int(stream) for stream in self.fio_opts['plids'].split(',')] + if not self.check_streams(self.fio_opts['filename'], stream_ids): + self.passed = False + logging.error("Streams not as expected") + else: + logging.debug("Streams created as expected") + + + def check_streams(self, dut, stream_ids): + """ + Confirm that the specified stream IDs exist on the specified device. + """ + + id_list = get_device_stream_ids(dut) + if not id_list: + return False + + for stream in stream_ids: + if stream in id_list: + logging.debug("Stream ID %d found active on device", stream) + id_list.remove(stream) + else: + if self.__class__.__name__ != "StreamsTestRand": + logging.error("Stream ID %d not found on device", stream) + else: + logging.debug("Stream ID %d not found on device", stream) + return False + + if len(id_list) != 0: + logging.error("Extra stream IDs %s found on device", str(id_list)) + return False + + return True + + +class StreamsTestRR(StreamsTest): + """ + NVMe pass-through test class for streams. Check to make sure output for + selected data direction(s) is non-zero and that zero data appears for other + directions. Check that Stream IDs are accessed in round robin order. + """ + + def check_streams(self, dut, stream_ids): + """ + The number of IOs is less than the number of stream IDs provided. Let N + be the number of IOs. Make sure that the device only has the first N of + the stream IDs provided. + + This will miss some cases where some other selection algorithm happens + to select the first N stream IDs. The solution would be to repeat this + test multiple times. Multiple trials passing would be evidence that + round robin is working correctly. + """ + + id_list = get_device_stream_ids(dut) + if not id_list: + return False + + num_streams = int(self.fio_opts['io_size'] / self.fio_opts['bs']) + stream_ids = sorted(stream_ids)[0:num_streams] + + return super().check_streams(dut, stream_ids) + + +class StreamsTestRand(StreamsTest): + """ + NVMe pass-through test class for streams. Check to make sure output for + selected data direction(s) is non-zero and that zero data appears for other + directions. Check that Stream IDs are accessed in random order. + """ + + def check_streams(self, dut, stream_ids): + """ + The number of IOs is less than the number of stream IDs provided. Let N + be the number of IOs. Confirm that the stream IDs on the device are not + the first N stream IDs. + + This will produce false positives because it is possible for the first + N stream IDs to be randomly selected. We can reduce the probability of + false positives by increasing N and increasing the number of streams + IDs to choose from, although fio has a max of 16 placement IDs. + """ + + id_list = get_device_stream_ids(dut) + if not id_list: + return False + + num_streams = int(self.fio_opts['io_size'] / self.fio_opts['bs']) + stream_ids = sorted(stream_ids)[0:num_streams] + + return not super().check_streams(dut, stream_ids) + + +def get_device_stream_ids(dut): + cmd = f"sudo nvme dir-receive -D 1 -O 2 -H {dut}" + logging.debug("check streams command: %s", cmd) + cmd = cmd.split(' ') + cmd_result = subprocess.run(cmd, capture_output=True, check=False, + encoding=locale.getpreferredencoding()) + + logging.debug(cmd_result.stdout) + + if cmd_result.returncode != 0: + logging.error("Error obtaining device %s stream IDs: %s", dut, cmd_result.stderr) + return False + + id_list = [] + for line in cmd_result.stdout.split('\n'): + if not 'Stream Identifier' in line: + continue + tokens = line.split(':') + id_list.append(int(tokens[1])) + + return id_list + + +def release_stream(dut, stream_id): + """ + Release stream on given device with selected ID. + """ + cmd = f"nvme dir-send -D 1 -O 1 -S {stream_id} {dut}" + logging.debug("release stream command: %s", cmd) + cmd = cmd.split(' ') + cmd_result = subprocess.run(cmd, capture_output=True, check=False, + encoding=locale.getpreferredencoding()) + + if cmd_result.returncode != 0: + logging.error("Error releasing %s stream %d", dut, stream_id) + return False + + return True + + +def release_all_streams(dut): + """ + Release all streams on specified device. + """ + + id_list = get_device_stream_ids(dut) + if not id_list: + return False + + for stream in id_list: + if not release_stream(dut, stream): + return False + + return True + + +TEST_LIST = [ + # 4k block size + # {seq write, rand write} x {single stream, four streams} + { + "test_id": 1, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "8", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + { + "test_id": 2, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "3", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + { + "test_id": 3, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "1,2,3,4", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + { + "test_id": 4, + "fio_opts": { + "rw": 'randwrite', + "bs": 4096, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "5,6,7,8", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + # 256KiB block size + # {seq write, rand write} x {single stream, four streams} + { + "test_id": 10, + "fio_opts": { + "rw": 'write', + "bs": 256*1024, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "88", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + { + "test_id": 11, + "fio_opts": { + "rw": 'randwrite', + "bs": 256*1024, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "20", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + { + "test_id": 12, + "fio_opts": { + "rw": 'write', + "bs": 256*1024, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "16,32,64,128", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + { + "test_id": 13, + "fio_opts": { + "rw": 'randwrite', + "bs": 256*1024, + "io_size": 256*1024*1024, + "verify": "crc32c", + "plids": "10,20,40,82", + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTest, + }, + # Test placement ID selection patterns + # default is round robin + { + "test_id": 20, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 8192, + "plids": '88,99,100,123,124,125,126,127,128,129,130,131,132,133,134,135', + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTestRR, + }, + { + "test_id": 21, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 8192, + "plids": '12,88,99,100,123,124,125,126,127,128,129,130,131,132,133,11', + "dataplacement": "streams", + "output-format": "json", + }, + "test_class": StreamsTestRR, + }, + # explicitly select round robin + { + "test_id": 22, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 8192, + "plids": '22,88,99,100,123,124,125,126,127,128,129,130,131,132,133,134', + "dataplacement": "streams", + "output-format": "json", + "plid_select": "roundrobin", + }, + "test_class": StreamsTestRR, + }, + # explicitly select random + { + "test_id": 23, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 8192, + "plids": '1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16', + "dataplacement": "streams", + "output-format": "json", + "plid_select": "random", + }, + "test_class": StreamsTestRand, + }, + # Error case with placement ID > 0xFFFF + { + "test_id": 30, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 8192, + "plids": "1,2,3,0x10000", + "dataplacement": "streams", + "output-format": "normal", + "plid_select": "random", + }, + "test_class": StreamsTestRand, + "success": SUCCESS_NONZERO, + }, + # Error case with no stream IDs provided + { + "test_id": 31, + "fio_opts": { + "rw": 'write', + "bs": 4096, + "io_size": 8192, + "dataplacement": "streams", + "output-format": "normal", + }, + "test_class": StreamsTestRand, + "success": SUCCESS_NONZERO, + }, + +] + +def parse_args(): + """Parse command-line arguments.""" + + parser = argparse.ArgumentParser() + parser.add_argument('-d', '--debug', help='Enable debug messages', action='store_true') + parser.add_argument('-f', '--fio', help='path to file executable (e.g., ./fio)') + parser.add_argument('-a', '--artifact-root', help='artifact root directory') + parser.add_argument('-s', '--skip', nargs='+', type=int, + help='list of test(s) to skip') + parser.add_argument('-o', '--run-only', nargs='+', type=int, + help='list of test(s) to run, skipping all others') + parser.add_argument('--dut', help='target NVMe character device to test ' + '(e.g., /dev/ng0n1). WARNING: THIS IS A DESTRUCTIVE TEST', required=True) + args = parser.parse_args() + + return args + + +def main(): + """Run tests using fio's io_uring_cmd ioengine to send NVMe pass through commands.""" + + args = parse_args() + + if args.debug: + logging.basicConfig(level=logging.DEBUG) + else: + logging.basicConfig(level=logging.INFO) + + artifact_root = args.artifact_root if args.artifact_root else \ + f"nvmept-streams-test-{time.strftime('%Y%m%d-%H%M%S')}" + os.mkdir(artifact_root) + print(f"Artifact directory is {artifact_root}") + + if args.fio: + fio_path = str(Path(args.fio).absolute()) + else: + fio_path = 'fio' + print(f"fio path is {fio_path}") + + for test in TEST_LIST: + test['fio_opts']['filename'] = args.dut + + release_all_streams(args.dut) + test_env = { + 'fio_path': fio_path, + 'fio_root': str(Path(__file__).absolute().parent.parent), + 'artifact_root': artifact_root, + 'basename': 'nvmept-streams', + } + + _, failed, _ = run_fio_tests(TEST_LIST, test_env, args) + sys.exit(failed) + + +if __name__ == '__main__': + main() diff --git a/thread_options.h b/thread_options.h index c2e71518..a36b7909 100644 --- a/thread_options.h +++ b/thread_options.h @@ -391,11 +391,11 @@ struct thread_options { fio_fp64_t zrt; fio_fp64_t zrf; -#define FIO_MAX_PLIS 16 unsigned int fdp; - unsigned int fdp_pli_select; - unsigned int fdp_plis[FIO_MAX_PLIS]; - unsigned int fdp_nrpli; + unsigned int dp_type; + unsigned int dp_id_select; + unsigned int dp_ids[FIO_MAX_DP_IDS]; + unsigned int dp_nr_ids; unsigned int log_entries; unsigned int log_prio; @@ -709,9 +709,10 @@ struct thread_options_pack { uint32_t log_prio; uint32_t fdp; - uint32_t fdp_pli_select; - uint32_t fdp_plis[FIO_MAX_PLIS]; - uint32_t fdp_nrpli; + uint32_t dp_type; + uint32_t dp_id_select; + uint32_t dp_ids[FIO_MAX_DP_IDS]; + uint32_t dp_nr_ids; uint32_t num_range; /*