Date: Sat, 13 Jan 2001 16:07:16 -0800 (PST) From: Matt Dillon <dillon@earth.backplane.com> To: Doug Barton <DougB@FreeBSD.org> Cc: Mark Murray <mark@grondar.za>, Robert Watson <rwatson@FreeBSD.org>, cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/etc crontab rc src/etc/defaults rc.conf src/etc/mtree BSD.root.dist src/libexec Makefile src/libexec/save-entropy Makefile save-entropy.sh Message-ID: <200101140007.f0E07G336509@earth.backplane.com> References: <200101132228.f0DMSN135897@earth.backplane.com> <200101132237.f0DMbbI21558@gratis.grondar.za> <200101132251.f0DMpKS36028@earth.backplane.com> <3A60E8D6.6E7E3281@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
:
:Matt Dillon wrote:
:
:> If we are going to have to wait, then please commit a stopgap
:
: That stopgap was committed 10/27/00 in the form of the NOBLOCKRANDOM
:kernel config option which is the default in GENERIC. With this option
:enabled you will guarantee that /dev/random never blocks, including at boot
:time.
That is not an appropriate stopgap. It fixes NOBODY who config's up
custom kernels unless they know about the option and hack up their
configs. Forcing people to hack up their configs is not a stopgap.
A proper stopgap is for the system default to be non-blocking... THAT
is a proper stopgap, *NOT* adding an option that developers HAVE to set
in their custom kernels to unbreak themselves.
:> -- Warner's preloading patch, for example.
:
: Which, as has already been explained about 3 different times now will not
:cause /dev/random to unblock because it does not generate anywhere near
:enough entropy.
ls -lua, dmesg... it doesn't take much and you know it. It's good
enough for development work and it makes things look reasonable.
There is nothing wrong with making the kernel behave more like it does
:
:> There is absolutely no reason to leave the
:> tree in a broken state this long.
:
: The tree is not broken. There are numerous ways for you to avoid ever
:having to deal with this issue until the code is more polished, and in the
:case of the above mentioned option, the default would mean that you would
It is inappropriate to force the entire developer community to hack up
their configurations to 'avoid' a problem that should not exist by
default.
It IS appropriate to have options to enable the new /dev/random code...
so developers can turn it on in /etc/rc.conf while it's under development
if they want to mess with it. It is NOT appropriate to have it on by
default and to force developers to turn it off if they don't want to mess
with it.
There is a big difference. One creates havoc, one creates a smooth
development path. Doug and Mark have chosen the wrong path.
-Matt
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200101140007.f0E07G336509>