Recent changes (master)
The following changes since commit df597be63e26ef59c1538b3ce2026c83684ff7fb:
fio: really use LDFLAGS when linking dynamic engines (2022-02-08 09:28:30 -0700)
are available in the Git repository at:
git://git.kernel.dk/fio.git master
for you to fetch changes up to a1db4528a59a99c5e2aa66091c505fb60e3a70ca:
Merge branch 'fio-docs-ci' of https://github.com/vincentkfu/fio (2022-02-11 16:29:44 -0700)
----------------------------------------------------------------
Jens Axboe (4):
t/io_uring: avoid unused nr_batch warning
Add aarch64 cpu clock support
Merge branch 'fio_offload_fixes' of https://github.com/PCPartPicker/fio
Merge branch 'fio-docs-ci' of https://github.com/vincentkfu/fio
Vincent Fu (8):
docs: document cpumode option for the cpuio ioengine
docs: update Makefile in order to detect build failures
docs: rename HOWTO to HOWTO.rst
HOWTO: combine multiple pool option listings
HOWTO: combine separate hipri listings into a single one
HOWTO: combine two chunk_size listings into a single one
ci: install sphinx packages and add doc building to GitHub Actions
windows: update the installer build for renamed files
aggieNick02 (1):
Fix issues (assert or uninit var, hang) with check_min_rate and offloading
HOWTO => HOWTO.rst | 126 ++++++++++++++++++++++++++++-------------------- arch/arch-aarch64.h | 17 +++++++ backend.c | 9 +++- ci/actions-full-test.sh | 1 + ci/actions-install.sh | 3 +- doc/Makefile | 2 +- doc/fio_doc.rst | 2 +- doc/fio_man.rst | 2 +- fio.1 | 13 +++++ os/windows/install.wxs | 4 +- t/io_uring.c | 9 ++-- 11 files changed, 124 insertions(+), 64 deletions(-) rename HOWTO => HOWTO.rst (99%)
---
Diff of recent changes:
HOWTO.rst renamed from HOWTO +73 -53
@@ -2137,8 +2137,10 @@ I/O engine Asynchronous read and write using DDN's Infinite Memory Engine (IME). This engine will try to stack as much IOs as possible by creating requests for IME. FIO will then decide when to commit these requests. + **libiscsi** Read and write iscsi lun with libiscsi. + **nbd** Read and write a Network Block Device (NBD).
@@ -2149,6 +2151,7 @@ I/O engine unless :option:`verify` is set or :option:`cuda_io` is `posix`. :option:`iomem` must not be `cudamalloc`. This ioengine defines engine specific options. + **dfs** I/O engine supporting asynchronous read and write operations to the DAOS File System (DFS) via libdfs.
@@ -2175,8 +2178,8 @@ with the caveat that when used on the command line, they must come after the Set the percentage of I/O that will be issued with the highest priority. Default: 0. A single value applies to reads and writes. Comma-separated values may be specified for reads and writes. For this option to be - effective, NCQ priority must be supported and enabled, and `direct=1' - option must be used. fio must also be run as the root user. Unlike + effective, NCQ priority must be supported and enabled, and the :option:`direct` + option must be set. fio must also be run as the root user. Unlike slat/clat/lat stats, which can be tracked and reported independently, per priority stats only track and report a single type of latency. By default, completion latency (clat) will be reported, if :option:`lat_percentiles` is
@@ -2207,6 +2210,7 @@ with the caveat that when used on the command line, they must come after the meaning of priority may differ. See also the :option:`prio` option. .. option:: cmdprio_bssplit=str[,str] : [io_uring] [libaio] + To get a finer control over I/O priority, this option allows specifying the percentage of IOs that must have a priority set depending on the block size of the IO. This option is useful only
@@ -2243,14 +2247,6 @@ with the caveat that when used on the command line, they must come after the map and release for each IO. This is more efficient, and reduces the IO latency as well. -.. option:: hipri : [io_uring] - - If this option is set, fio will attempt to use polled IO completions. - Normal IO completions generate interrupts to signal the completion of - IO, polled completions do not. Hence they are require active reaping - by the application. The benefits are more efficient IO for high IOPS - scenarios, and lower latencies for low queue depth IO. - .. option:: registerfiles : [io_uring] With this option, fio registers the set of files being used with the
@@ -2271,6 +2267,33 @@ with the caveat that when used on the command line, they must come after the When :option:`sqthread_poll` is set, this option provides a way to define which CPU should be used for the polling thread. +.. option:: hipri + + [io_uring] + + If this option is set, fio will attempt to use polled IO completions. + Normal IO completions generate interrupts to signal the completion of + IO, polled completions do not. Hence they are require active reaping + by the application. The benefits are more efficient IO for high IOPS + scenarios, and lower latencies for low queue depth IO. + + [pvsync2] + + Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority + than normal. + + [sg] + + If this option is set, fio will attempt to use polled IO completions. + This will have a similar effect as (io_uring)hipri. Only SCSI READ and + WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor + VERIFY). Older versions of the Linux sg driver that do not support + hipri will simply ignore this flag and do normal IO. The Linux SCSI + Low Level Driver (LLD) that "owns" the device also needs to support + hipri (also known as iopoll and mq_poll). The MegaRAID driver is an + example of a SCSI LLD. Default: clear (0) which does normal + (interrupted based) IO. + .. option:: userspace_reap : [libaio] Normally, with the libaio engine in use, fio will use the
@@ -2279,11 +2302,6 @@ with the caveat that when used on the command line, they must come after the reap events. The reaping mode is only enabled when polling for a minimum of 0 events (e.g. when :option:`iodepth_batch_complete` `=0`). -.. option:: hipri : [pvsync2] - - Set RWF_HIPRI on I/O, indicating to the kernel that it's of higher priority - than normal. - .. option:: hipri_percentage : [pvsync2] When hipri is set this determines the probability of a pvsync2 I/O being high
@@ -2318,6 +2336,16 @@ with the caveat that when used on the command line, they must come after the Split the load into cycles of the given time. In microseconds. +.. option:: cpumode=str : [cpuio] + + Specify how to stress the CPU. It can take these two values: + + **noop** + This is the default where the CPU executes noop instructions. + **qsort** + Replace the default noop instructions loop with a qsort algorithm to + consume more energy. + .. option:: exit_on_io_done=bool : [cpuio] Detect when I/O threads are done, then exit.
@@ -2444,10 +2472,6 @@ with the caveat that when used on the command line, they must come after the Specifies the name of the RBD. -.. option:: pool=str : [rbd,rados] - - Specifies the name of the Ceph pool containing RBD or RADOS data. - .. option:: clientname=str : [rbd,rados] Specifies the username (without the 'client.' prefix) used to access the
@@ -2466,6 +2490,36 @@ with the caveat that when used on the command line, they must come after the Touching all objects affects ceph caches and likely impacts test results. Enabled by default. +.. option:: pool=str : + + [rbd,rados] + + Specifies the name of the Ceph pool containing RBD or RADOS data. + + [dfs] + + Specify the label or UUID of the DAOS pool to connect to. + +.. option:: cont=str : [dfs] + + Specify the label or UUID of the DAOS container to open. + +.. option:: chunk_size=int + + [dfs] + + Specificy a different chunk size (in bytes) for the dfs file. + Use DAOS container's chunk size by default. + + [libhdfs] + + The size of the chunk to use for each file. + +.. option:: object_class=str : [dfs] + + Specificy a different object class for the dfs file. + Use DAOS container's object class by default. + .. option:: skip_bad=bool : [mtd] Skip operations against known bad blocks.
@@ -2474,10 +2528,6 @@ with the caveat that when used on the command line, they must come after the libhdfs will create chunk in this HDFS directory. -.. option:: chunk_size : [libhdfs] - - The size of the chunk to use for each file. - .. option:: verb=str : [rdma] The RDMA verb to use on this side of the RDMA ioengine connection. Valid
@@ -2563,18 +2613,6 @@ with the caveat that when used on the command line, they must come after the a valid stream identifier) fio will open a stream and then close it when done. Default is 0. -.. option:: hipri : [sg] - - If this option is set, fio will attempt to use polled IO completions. - This will have a similar effect as (io_uring)hipri. Only SCSI READ and - WRITE commands will have the SGV4_FLAG_HIPRI set (not UNMAP (trim) nor - VERIFY). Older versions of the Linux sg driver that do not support - hipri will simply ignore this flag and do normal IO. The Linux SCSI - Low Level Driver (LLD) that "owns" the device also needs to support - hipri (also known as iopoll and mq_poll). The MegaRAID driver is an - example of a SCSI LLD. Default: clear (0) which does normal - (interrupted based) IO. - .. option:: http_host=str : [http] Hostname to connect to. For S3, this could be the bucket hostname.
@@ -2654,24 +2692,6 @@ with the caveat that when used on the command line, they must come after the GPU to RAM before a write and copied from RAM to GPU after a read. :option:`verify` does not affect use of cudaMemcpy. -.. option:: pool=str : [dfs] - - Specify the label or UUID of the DAOS pool to connect to. - -.. option:: cont=str : [dfs] - - Specify the label or UUID of the DAOS container to open. - -.. option:: chunk_size=int : [dfs] - - Specificy a different chunk size (in bytes) for the dfs file. - Use DAOS container's chunk size by default. - -.. option:: object_class=str : [dfs] - - Specificy a different object class for the dfs file. - Use DAOS container's object class by default. - .. option:: nfs_url=str : [nfs] URL in libnfs format, eg nfs://<server|ipv4|ipv6>/path[?arg=val[&arg=val]*]
arch/arch-aarch64.h +17 -0
@@ -27,4 +27,21 @@ static inline int arch_ffz(unsigned long bitmask)#define ARCH_HAVE_FFZ +static inline unsigned long long get_cpu_clock (void ) +{ +unsigned long val; + + asmvolatile ("mrs %0, cntvct_el0" :"=r" (val)); +return val; +} +#define ARCH_HAVE_CPU_CLOCK + +#define ARCH_HAVE_INIT +extern bool tsc_reliable; +static inline int arch_init (char *envp[]) +{ + tsc_reliable = true; +return 0 ; +} +#endif
backend.c +7 -2
@@ -1091,8 +1091,10 @@ static void do_io(struct thread_data *td, uint64_t *bytes_done) td->rate_io_issue_bytes[__ddir] += blen; } -if (should_check_rate (td)) +if (should_check_rate (td)) { td->rate_next_io_time[__ddir] =usec_for_io (td, __ddir); +fio_gettime (&comp_time,NULL ); + } }else { ret =io_u_submit (td, io_u);
@@ -1172,8 +1174,11 @@ reap: f->file_name); } } - }else + }else { +if (td->o.io_submit_mode ==IO_MODE_OFFLOAD ) +workqueue_flush (&td->io_wq);cleanup_pending_aio (td); + }/* * stop job if we failed doing any IO
ci/actions-full-test.sh +1 -0
@@ -10,6 +10,7 @@ main() { else sudo python3 t/run-fio-tests.py --skip 6 1007 1008 --debug fi + make -C doc html } main
ci/actions-install.sh +2 -1
@@ -60,6 +60,7 @@ DPKGCFG # care about the architecture. pkgs+=( python3-scipy + python3-sphinx ) echo "Updating APT..."
@@ -78,7 +79,7 @@ install_macos() { #brew update >/dev/null 2>&1 echo "Installing packages..." HOMEBREW_NO_AUTO_UPDATE=1 brew install cunit - pip3 install scipy six + pip3 install scipy six sphinx } main() {
doc/Makefile +1 -1
@@ -2,7 +2,7 @@ # # You can set these variables from the command line. -SPHINXOPTS = +SPHINXOPTS = -W --keep-going SPHINXBUILD = sphinx-build PAPER = BUILDDIR = output
doc/fio_doc.rst +1 -1
@@ -5,7 +5,7 @@ fio - Flexible I/O tester rev. |version| .. include:: ../README.rst -.. include:: ../HOWTO +.. include:: ../HOWTO.rst
doc/fio_man.rst +1 -1
@@ -9,4 +9,4 @@ Fio Manpage .. include:: ../README.rst -.. include:: ../HOWTO +.. include:: ../HOWTO.rst
fio.1 +13 -0
@@ -2091,6 +2091,19 @@ option when using cpuio I/O engine. .BI (cpuio)cpuchunks \fR=\fPint Split the load into cycles of the given time. In microseconds. .TP +.BI (cpuio)cpumode \fR=\fPstr +Specify how to stress the CPU. It can take these two values: +.RS +.RS +.TP +.B noop +This is the default and directs the CPU to execute noop instructions. +.TP +.B qsort +Replace the default noop instructions with a qsort algorithm to consume more energy. +.RE +.RE +.TP .BI (cpuio)exit_on_io_done \fR=\fPbool Detect when I/O threads are done, then exit. .TP
os/windows/install.wxs +2 -2
@@ -33,13 +33,13 @@ </Component> <?endif?> <Component> - <File Id="README" Name="README.txt" Source="..\..\README"/> + <File Id="README" Name="README.txt" Source="..\..\README.rst"/> </Component> <Component> <File Id="REPORTING_BUGS" Name="REPORTING-BUGS.txt" Source="..\..\REPORTING-BUGS"/> </Component> <Component> - <File Id="HOWTO" Name="HOWTO.txt" Source="..\..\HOWTO"/> + <File Id="HOWTO" Name="HOWTO.txt" Source="..\..\HOWTO.rst"/> </Component> <Component> <File Id="COPYING" Name="COPYING.txt" Source="..\..\COPYING"/>
t/io_uring.c +6 -3
@@ -714,12 +714,15 @@ static int reap_events_aio(struct submitter *s, struct io_event *events, int evsstatic void *submitter_aio_fn (void *data) {struct submitter *s = data; -int i, ret, prepped, nr_batch; +int i, ret, prepped;struct iocb **iocbsptr;struct iocb *iocbs;struct io_event *events; - - nr_batch =submitter_init (s); +#ifdef ARCH_HAVE_CPU_CLOCK +int nr_batch =submitter_init (s); +#else +submitter_init (s); +#endif iocbsptr =calloc (depth,sizeof (struct iocb *)); iocbs =calloc (depth,sizeof (struct iocb ));