[dpdk-dev] [PATCH v4 3/6] build: optional NUMA and cpu counts detection

[Juraj Linkeš]

> -----Original Message-----
> From: Bruce Richardson <bruce.richardson at intel.com>
> Sent: Tuesday, October 27, 2020 12:21 PM
> To: Juraj Linkeš <juraj.linkes at pantheon.tech>
> Cc: Ruifeng.Wang at arm.com; Honnappa.Nagarahalli at arm.com;
> Phil.Yang at arm.com; vcchunga at amazon.com; Dharmik.Thakkar at arm.com;
> jerinjacobk at gmail.com; hemant.agrawal at nxp.com; dev at dpdk.org
> Subject: Re: [PATCH v4 3/6] build: optional NUMA and cpu counts detection
> 
> On Fri, Oct 23, 2020 at 04:48:05PM +0200, Juraj Linkeš wrote:
> > Add an option to automatically discover the host's numa and cpu counts
> > and use those values for a non cross-build.
> > Give users the option to override the per-arch default values or
> > values from cross files by specifying them on the command line with
> > -Dmax_lcores and -Dmax_numa_nodes.
> >
> > Signed-off-by: Juraj Linkeš <juraj.linkes at pantheon.tech>
> > ---
> >  buildtools/get_cpu_count.py  |  7 ++++++
> > buildtools/get_numa_count.py | 22 +++++++++++++++++
> >  buildtools/meson.build       |  2 ++
> >  config/meson.build           | 48 ++++++++++++++++++++++++++++++++++--
> >  meson_options.txt            |  8 +++---
> >  5 files changed, 81 insertions(+), 6 deletions(-)  create mode 100644
> > buildtools/get_cpu_count.py  create mode 100644
> > buildtools/get_numa_count.py
> >
> > diff --git a/buildtools/get_cpu_count.py b/buildtools/get_cpu_count.py
> > new file mode 100644 index 000000000..386f85f8b
> > --- /dev/null
> > +++ b/buildtools/get_cpu_count.py
> > @@ -0,0 +1,7 @@
> > +#!/usr/bin/python3
> > +# SPDX-License-Identifier: BSD-3-Clause # Copyright (c) 2020
> > +PANTHEON.tech s.r.o.
> > +
> > +import os
> > +
> > +print(os.cpu_count())
> > diff --git a/buildtools/get_numa_count.py
> > b/buildtools/get_numa_count.py new file mode 100644 index
> > 000000000..f0c49973a
> > --- /dev/null
> > +++ b/buildtools/get_numa_count.py
> > @@ -0,0 +1,22 @@
> > +#!/usr/bin/python3
> > +# SPDX-License-Identifier: BSD-3-Clause # Copyright (c) 2020
> > +PANTHEON.tech s.r.o.
> > +
> > +import ctypes
> > +import glob
> > +import os
> > +import subprocess
> > +
> > +if os.name == 'posix':
> > +    if os.path.isdir('/sys/devices/system/node'):
> > +        print(len(glob.glob('/sys/devices/system/node/node*')))
> > +    else:
> > +        print(subprocess.run(['sysctl', 'vm.ndomains'],
> > +capture_output=True).stdout)
> 
> I think you can shorten this, by just calling subprocess.run and not capturing
> anything, in which case the stdout will be printed as normal.
> 
> 	subprocess.run(['sysctl', 'vm.ndomains'])
> 

This will also print out the resulting object (e.g. CompletedProcess(args=['ls', '-ls', '/sys/devices/system/node'], returncode=0)), but an assignment will take care of that. I'll make the change.

> > +
> > +elif os.name == 'nt':
> > +    libkernel32 = ctypes.windll.kernel32
> > +
> > +    count = ctypes.c_ulong()
> > +
> > +    libkernel32.GetNumaHighestNodeNumber(ctypes.pointer(count))
> > +    print(count.value + 1)
> > diff --git a/buildtools/meson.build b/buildtools/meson.build index
> > 04808dabc..925e733b1 100644
> > --- a/buildtools/meson.build
> > +++ b/buildtools/meson.build
> > @@ -17,3 +17,5 @@ else
> >  endif
> >  map_to_win_cmd = py3 + files('map_to_win.py')  sphinx_wrapper = py3 +
> > files('call-sphinx-build.py')
> > +get_cpu_count_cmd = py3 + files('get_cpu_count.py')
> > +get_numa_count_cmd = py3 + files('get_numa_count.py')
> > diff --git a/config/meson.build b/config/meson.build index
> > 918ca2dfb..5c6fdcc7d 100644
> > --- a/config/meson.build
> > +++ b/config/meson.build
> > @@ -230,8 +230,14 @@ foreach arg: warning_flags  endforeach
> >
> >  # set other values pulled from the build options
> > -dpdk_conf.set('RTE_MAX_LCORE', get_option('max_lcores'))
> > -dpdk_conf.set('RTE_MAX_NUMA_NODES', get_option('max_numa_nodes'))
> > +if not meson.is_cross_build()
> > +	# set default values
> > +	# these defaults may be overwritten by meson.build in an arch subdir
> > +	# or later overwritten if a user specifies a value on the command line
> > +	# or the user chooses to use values discovered from the build machine
> > +	dpdk_conf.set('RTE_MAX_LCORE', 128)
> > +	dpdk_conf.set('RTE_MAX_NUMA_NODES', 4) endif
> 
> I think it might be better to remove this block, and instead set the values below
> where you do the additional checks. It's confusing having these settings
> managed in multiple places.
> 

It needs to be set before subdir(arch_subdir) which is why it's here.

The order goes like this:
Global defaults (or x86 defaults)
arch_subdir defaults overwrite those (or non-x86 defaults)
cmdline options or optional discovery overwrite those

What about I move these to config/x86/meson.build? It looks like the place where per-arch defaults should be. Then we'd only have the second part after arch_subdir.

> >  dpdk_conf.set('RTE_MAX_ETHPORTS', get_option('max_ethports'))
> > dpdk_conf.set('RTE_LIBEAL_USE_HPET', get_option('use_hpet'))
> > dpdk_conf.set('RTE_ENABLE_TRACE_FP', get_option('enable_trace_fp')) @@
> > -250,6 +256,44 @@ compile_time_cpuflags = []
> >  subdir(arch_subdir)
> >  dpdk_conf.set('RTE_COMPILE_TIME_CPUFLAGS',
> > ','.join(compile_time_cpuflags))
> >
> > +max_lcores = get_option('max_lcores') max_numa_nodes =
> > +get_option('max_numa_nodes') if max_lcores > 0
> > +	dpdk_conf.set('RTE_MAX_LCORE', max_lcores) elif max_lcores == -1
> > +	if not meson.is_cross_build()
> > +		max_lcores =
> run_command(get_cpu_count_cmd).stdout().to_int()
> > +		min_lcores = 2
> > +		if max_lcores < min_lcores
> > +			message('Found less than @0@ cores, building for
> @0@ cores'.format(min_lcores))
> > +			max_lcores = min_lcores
> > +		else
> > +			message('Found @0@ cores'.format(max_lcores))
> > +		endif
> > +		dpdk_conf.set('RTE_MAX_LCORE', max_lcores)
> > +	endif
> > +endif
> > +
> > +if max_numa_nodes > 0
> > +	dpdk_conf.set('RTE_MAX_NUMA_NODES', max_numa_nodes) elif
> > +max_numa_nodes == -1
> > +	if not meson.is_cross_build()
> > +		max_numa_nodes =
> run_command(get_numa_count_cmd).stdout().to_int()
> > +		message('Found @0@ numa nodes'.format(max_numa_nodes))
> > +		dpdk_conf.set('RTE_MAX_NUMA_NODES', max_numa_nodes)
> > +	endif
> > +endif
> > +
> > +# check that cpu and numa count is set in cross builds if
> > +meson.is_cross_build()
> > +	if not dpdk_conf.has('RTE_MAX_LCORE')
> > +		error('Number of cores for cross build not specified in @0@
> subdir (e.g. in a cross-file) nor on the cmdline'.format(arch_subdir))
> > +	endif
> > +	if not dpdk_conf.has('RTE_MAX_NUMA_NODES')
> > +		error('Number of numa nodes for cross build not specified in
> @0@ subdir (e.g. in a cross-file) nor on the cmdline'.format(arch_subdir))
> > +	endif
> > +endif
> > +
> 
> I think we need some comments explaining all the logic here, and probably some
> documentation updates.
> 

Will add comments. Where is this documented?

> >  # set the install path for the drivers
> > dpdk_conf.set_quoted('RTE_EAL_PMD_PATH', eal_pmd_path)
> >
> > diff --git a/meson_options.txt b/meson_options.txt index
> > 9bf18ab6b..a78c21b0d 100644
> > --- a/meson_options.txt
> > +++ b/meson_options.txt
> > @@ -26,10 +26,10 @@ option('machine', type: 'string', value: 'native',
> >  	description: 'set the target machine type')  option('max_ethports',
> > type: 'integer', value: 32,
> >  	description: 'maximum number of Ethernet devices')
> > -option('max_lcores', type: 'integer', value: 128,
> > -	description: 'maximum number of cores/threads supported by EAL')
> > -option('max_numa_nodes', type: 'integer', value: 4,
> > -	description: 'maximum number of NUMA nodes supported by EAL')
> > +option('max_lcores', type: 'integer', value: 0,
> > +	description: 'maximum number of cores/threads supported by EAL. Set
> > +to positive integer to overwrite per-arch or cross-compilation defaults. Set to
> -1 to use number of cores on the build machine.') option('max_numa_nodes',
> type: 'integer', value: 0,
> > +	description: 'maximum number of NUMA nodes supported by EAL. Set to
> > +positive integer to overwrite per-arch or cross-compilation defaults.
> > +Set to -1 to use number of numa nodes on the build machine.')
> >  option('enable_trace_fp', type: 'boolean', value: false,
> >  	description: 'enable fast path trace points.')  option('tests',
> > type: 'boolean', value: true,
> > --
> > 2.20.1
> >