Date: Sun, 13 Apr 2014 17:20:28 +0300 From: Konstantin Belousov <kostikbel@gmail.com> To: Baptiste Daroussin <bapt@freebsd.org> Cc: desrt@desrt.ca, John-Mark Gurney <jmg@funkthat.com>, "freebsd-hackers@freebsd.org" <hackers@freebsd.org> Subject: Re: [CFR] Kevent timer improvements Message-ID: <20140413142028.GD4016@kib.kiev.ua> In-Reply-To: <20140413131550.GD17898@ivaldir.etoilebsd.net> References: <20140310131632.GI6900@ithaqua.etoilebsd.net> <CAJ-Vmok7ODK7FfJaHiQej8Za4-JH6Aryctk577vM9cYpeUcHew@mail.gmail.com> <20140310180404.GI32089@funkthat.com> <20140413131550.GD17898@ivaldir.etoilebsd.net>
next in thread | previous in thread | raw e-mail | index | archive | help
--dc+cDN39EJAMEtIO Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sun, Apr 13, 2014 at 03:15:51PM +0200, Baptiste Daroussin wrote: > I have splitted my patch in multiple parts, let start with the first one: > adding NOTE_NSECONDS, NOTE_USECONDS, NOTE_NSECONDS >=20 > http://people.freebsd.org/~bapt/kevent_timer.diff > Index: sys/kern/kern_event.c > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > --- sys/kern/kern_event.c (revision 264413) > +++ sys/kern/kern_event.c (working copy) > @@ -524,14 +524,24 @@ > * interval timer support code. > */ > static __inline sbintime_t > -timer2sbintime(intptr_t data) > +timer2sbintime(intptr_t data, int flags) > { > + sbintime_t modifier; > =20 > + modifier =3D SBT_1MS; > + > + if (flags & NOTE_SECONDS) > + modifier =3D SBT_1S; > + else if (flags & NOTE_USECONDS) > + modifier =3D SBT_1US; > + else if (flags & NOTE_NSECONDS) > + modifier =3D SBT_1NS; It is better to put the 'modifier =3D SBT_1MS;' statement as the else part. That said, IMO it would be sometimes beneficial to have real flag to specify milliseconds precision, in addition to milliseconds be the default. > + > #ifdef __LP64__ > - if (data > SBT_MAX / SBT_1MS) > + if (data > SBT_MAX / modifier) > return (SBT_MAX); > #endif > - return (SBT_1MS * data); > + return (modifier * data); > } > =20 > static void > @@ -547,7 +557,7 @@ > if ((kn->kn_flags & EV_ONESHOT) !=3D EV_ONESHOT) { > calloutp =3D (struct callout *)kn->kn_hook; > callout_reset_sbt_on(calloutp, > - timer2sbintime(kn->kn_sdata), 0 /* 1ms? */, > + timer2sbintime(kn->kn_sdata, kn->kn_sfflags), 0 /* 1ms? */, There, at least the comment about precision should be updated. But, it seems that for the seconds precision, it makes sense to specify e.g. 1/2 sec as precision; or add an API flag to allow imprecise callout scheduling. > filt_timerexpire, kn, PCPU_GET(cpuid), 0); > } > } > @@ -566,7 +576,7 @@ > return (EINVAL); > if ((intptr_t)kn->kn_sdata =3D=3D 0 && (kn->kn_flags & EV_ONESHOT) =3D= =3D 0) > kn->kn_sdata =3D 1; > - to =3D timer2sbintime(kn->kn_sdata); > + to =3D timer2sbintime(kn->kn_sdata, kn->kn_sfflags); > if (to < 0) > return (EINVAL); > =20 > Index: sys/sys/event.h > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > --- sys/sys/event.h (revision 264413) > +++ sys/sys/event.h (working copy) > @@ -133,6 +133,11 @@ > #define NOTE_TRACKERR 0x00000002 /* could not track child */ > #define NOTE_CHILD 0x00000004 /* am a child process */ > =20 > +/* additional flags for EVFILE_TIMER */ > +#define NOTE_SECONDS 0x00000001 /* data is seconds */ > +#define NOTE_USECONDS 0x00000002 /* data is microseconds */ > +#define NOTE_NSECONDS 0x00000004 /* data is nanoseconds */ > + > struct knote; > SLIST_HEAD(klist, knote); > struct kqueue; --dc+cDN39EJAMEtIO Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (FreeBSD) iQIcBAEBAgAGBQJTSp0rAAoJEJDCuSvBvK1BjLwP/j8i6SlKTRYAh8s0DivRFCLd IRNPTxy6s2EHsos5c+DztwfY1LPbEy/XLm97qE90QuL83JF814+Q0pFwwuqySEb6 sLePohvzH51nFHOzqafe2E942BjNGLXC/sNcRDByS3+j/4cMCyUfsaWj4PpQ7QI/ 7MUS+SzeR/zF8eVvbYB9K4RB0B7qKBwZaODX4Uw0GH6CaD7dPOu2ekbt2opdXdKf a57ZhFaxb5NvuMeGoBdrfsdivHPanDI6rpE64/NWOgBmEiFdzuonudkGIfCJWtk2 n6y3q1SKSPSoBYxO9wEalbTce+wEwukwf9oouE+UErZDdUGFmwsqRpyauAVFtYum 56V8oocn+qonaxESMdpqMbBYc/KQWPy+widvYRVgZZgE1Dcl+w8ZILDhx12IOKXu V2OIXkjefiT243pcEtWx59p3cG5M1/QNid094nbRLPvO3ggesFRqy+As4siLe+OZ K6ruVfaQhOYeYJoj7djG+KpAGvOOZiyS9XuSr8Me1qirg5eo5uPkw5LwuqF6YcM1 IftVOinVWLMBt009AZO3R2GiSzNiuCPDRTqLZTEA4DTVx+HpgI6oYHEHgcfL+O9G ME0O1fPirkiTDnezpqiYmaV9kp46bA5RrrrXatnujJr7AvOw4svNOX6PHhJs4jeq kR9FgkTutAuqH6yjXVF/ =Aa+6 -----END PGP SIGNATURE----- --dc+cDN39EJAMEtIO--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20140413142028.GD4016>