Date: Tue, 14 Feb 2023 23:45:17 -0800 From: Mark Millard <marklmi@yahoo.com> To: Mateusz Guzik <mjguzik@gmail.com> Cc: dev-commits-src-main@freebsd.org Subject: Re: git: d9d5f2c042a5 - main - cpuset: add --count [changed/broken command line parsing] Message-ID: <EDD615B9-0F47-4D9B-8CD8-A5623AF08293@yahoo.com> In-Reply-To: <CAGudoHETv8i9pWS=x0sx%2B5TGLnWwA%2BsjmpD%2BOC5OZ9J9rThtVw@mail.gmail.com> References: <D2D940CD-B1A0-41B7-8C02-63CA63940525.ref@yahoo.com> <D2D940CD-B1A0-41B7-8C02-63CA63940525@yahoo.com> <CAGudoHETv8i9pWS=x0sx%2B5TGLnWwA%2BsjmpD%2BOC5OZ9J9rThtVw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Feb 14, 2023, at 23:08, Mateusz Guzik <mjguzik@gmail.com> wrote: > On 2/15/23, Mark Millard <marklmi@yahoo.com> wrote: >> Mateusz Guzik <mjg_at_FreeBSD.org> wrote on >> Date: Sat, 04 Feb 2023 17:51:27 UTC : >>=20 >>> The branch main has been updated by mjg: >>>=20 >>> URL: >>> = https://cgit.FreeBSD.org/src/commit/?id=3Dd9d5f2c042a51a9f0dd69eb1fc349efd= 81ffa483 >>>=20 >>> commit d9d5f2c042a51a9f0dd69eb1fc349efd81ffa483 >>> Author: Mateusz Guzik <mjg@FreeBSD.org> >>> AuthorDate: 2023-02-04 17:47:41 +0000 >>> Commit: Mateusz Guzik <mjg@FreeBSD.org> >>> CommitDate: 2023-02-04 17:50:41 +0000 >>>=20 >>> cpuset: add --count >>>=20 >>> Can be used to count the number of hardware threads in the cpu set. >>>=20 >>> For example: >>> $ cpuset -g -p $$ >>> pid 2440 mask: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, = 16, >>> 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, = 34, >>> 35, 36, 37, 38, 39 >>> pid 2440 domain policy: first-touch mask: 0, 1 >>> $ cpuset -g --count -p $$ >>> 40 >>>=20 >>> The intent is to replace calls to sysctl hw.ncpu and kern.smp.cpus = which >>> can be found in the tree, which are not adequate given existence of >>> cpusets. >>>=20 >>> Right now only -g -p combination is supported to reduce complexity. >>> As anything else errors out, this can be expanded later as needed. >>>=20 >>> Differential Revision: https://reviews.freebsd.org/D36351 >>=20 >> This broke long standing commands sequences in some makefiles >> that I use to build and run some benchmarks, the commands >> at issue involve use of cpuset. >>=20 >> However I can show the oddity in a much simpler context: >>=20 >> # cpuset echo "text" >> text >>=20 >> # cpuset echo "-text" >> usage: cpuset [-l cpu-list] [-n policy:domain-list] [-s setid] cmd = ... >> cpuset [-l cpu-list] [-n policy:domain-list] [-s setid] -p pid >> cpuset [-c] [-l cpu-list] [-n policy:domain-list] -C -p pid >> cpuset [-c] [-l cpu-list] [-n policy:domain-list] >> [-j jailid | -p pid | -t tid | -s setid | -x irq] >> cpuset -g [-cir] >> [-d domain | -j jailid | -p pid | -t tid | -s setid | -x = irq] >> cpuset -g --count -p pid >>=20 >>=20 >> By contrast, in my stable/13 context, so, showing the old behavior: >>=20 >> # cpuset echo "-text" >> -text FYI: the use of quotes was in my original text for other reasons. I should not have duplicated them into the above examples. (Same results without quotes.) > the idiomatic way to pass progs to spawn is to denote the end of any > opts with -- The cpuset man page gives no indication that it requires such notation. Programs with that requirement normally document it explicitly. (Programs that do not require the ntoation often allow the notation without documenting it.) There may be a consideration of how long it has allowed such a lack of use of "--" and how many things might break because it worked before. Lots of programs taking commands do not require such. For example: "time echo -text", "truss echo -text", and "ktrace echo -text". These have not and are not being changed to require such. They do allow the "--" notation. If it stays as it is, the change should probably be in UPDATING or other such place(s) that note where changes are required to keep things working. > so in particular you would: > cpuset -- echo -text Yep. > the fact that previous version worked without it is kind of a bug I disagree with that last statement, given the many programs that do not require such. Some programs have some specific notational complication lead to making it required. But, from what I've seen, most of the others do not make "--" required. =3D=3D=3D Mark Millard marklmi at yahoo.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?EDD615B9-0F47-4D9B-8CD8-A5623AF08293>