Date: Wed, 27 Apr 2005 19:59:53 -0700 From: Kris Kennaway <kris@obsecurity.org> To: Bruce Evans <bde@zeta.org.au> Cc: cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/vm vm_init.c Message-ID: <20050428025953.GA50766@xor.obsecurity.org> In-Reply-To: <20050428124206.C85412@delplex.bde.org> References: <200504280029.j3S0TOXp078389@repoman.freebsd.org> <20050428124206.C85412@delplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--0F1p//8PRICkK4MW Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Thu, Apr 28, 2005 at 12:52:06PM +1000, Bruce Evans wrote: > On Thu, 28 Apr 2005, Kris Kennaway wrote: >=20 > >kris 2005-04-28 00:29:24 UTC > > > > FreeBSD src repository > > > > Modified files: (Branch: RELENG_5) > > sys/vm vm_init.c > > Log: > > MFC: > > > > Add the vm.exec_map_entries tunable and read-only sysctl, which contr= ols > > the number of entries in exec_map (maximum number of simultaneous exe= cs > > that can be handled by the kernel). The default value of 16 is > > insufficient on heavily loaded machines (particularly SMP machines), = and > > if it is exceeded then executing further processes will generate a=20 > > SIGABRT. > > > > This is a workaround until a better solution can be implemented. >=20 > It seems to be insufficient on all reasonably fast machines. Just > fork-execing 16+ processes in the following benchmark seems to trigger it: >=20 > %%% > #!/bin/sh >=20 > for i in 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 > do > nice -$i sh -c "while :; do echo -n;done" & > done > top -o time > %%% >=20 > Lately, this has started failing with a strange "Abort trap" messages > printed by the shell after I exit top. The 3 processes with niceness > 1, 2 and 3 tend to get killed more than others. I didn't change the default value of 16 simultaneous execs to be conservative and because of the proximity of 5.4, but on my SMP package machines I typically need to increase it. Perhaps the default should be increased - this change was a minimal bandaid to at least allow the problem to be worked around for now. Alan has a patch that addresses 2 of the 3 failure cases, so hopefully he'll be able to fix the remaining one soon. Kris --0F1p//8PRICkK4MW Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (FreeBSD) iD8DBQFCcFGoWry0BWjoQKURAokpAJ9gJxlh9ac2RideoKS6EeEOzuDYTgCfRGfV wSMJC3sxNIk8ojAZI8kRGeg= =WQeD -----END PGP SIGNATURE----- --0F1p//8PRICkK4MW--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050428025953.GA50766>