Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 25 Jan 2000 09:38:14 -0800 (PST)
From:      Matthew Dillon <dillon@apollo.backplane.com>
To:        Warner Losh <imp@village.org>
Cc:        Brett Glass <brett@lariat.org>, security@FreeBSD.ORG
Subject:   Re: Merged patches 
Message-ID:  <200001251738.JAA04802@apollo.backplane.com>
References:  <4.2.2.20000125095042.01a5aba0@localhost>   <200001251722.KAA04527@harmony.village.org>

next in thread | previous in thread | raw e-mail | index | archive | help

:By what code paths is this possible?  Please be specific.
:
:: Also, in at least one place (maybe more), the code does
:: multiple tests of the TCP option flags in succession.
:: Several tests of this kind should generally be merged
:: into a switch for speed (the many conditional jumps
:: cause pipeline stalls on many processors, especially 
:: older ones) and readability.
:
:It does?  If so, it certainly doesn't ADD them.

    So we do multiple tests, so what?  Not only will GCC potentially
    optimize the code, but doing multiple tests means the memory references 
    are already in the L1 cache so, frankly, I doubt you would save more
    then a few nanoseconds glomming it all together into a switch.  In fact,
    it's quite possible that attempting to optimize it in this fashion will
    actually make it slower since you have no control over the critical path
    when you glom things into a switch statement.  It makes no sense at all 
    to mess around with working code just to save a few nanoseconds.

					-Matt
					Matthew Dillon 
					<dillon@backplane.com>

:: In short, I'd only go with this patch as-is if my
:: purpose were to minimize the changes made before
:: release. If this were the goal, I'd go back to the
:: code immediately thereafter and try to tackle some
:: of the inefficiencies and holes in this key input 
:: path more aggressively.
:
:Yes.  That's exactly the goal.
:
:Like I said in my initial mail, I may remove the ICMP_BANDLIM option
:as an option, but bump the rate limiter to 1000.  But that's about as
:far as I'd be willing to go at this time.
:
:Warner



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-security" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200001251738.JAA04802>