Re: MPSAFE CAM?

From: M. Warner Losh (imp_at_bsdimp.com)
Date: 05/31/05

  • Next message: Jacques Vidrine: "Re: [CFR] correct type of addrinfo.ai_addrlen and netent.n_net" 
    Date: Mon, 30 May 2005 20:30:26 -0600 (MDT)
To: bde@zeta.org.au

    In message: <20050531113008.R91505@delplex.bde.org>
                Bruce Evans <bde@zeta.org.au> writes:
    : On Mon, 30 May 2005, Robert Watson wrote:
    :
    : > On Mon, 30 May 2005, M. Warner Losh wrote:
    : >
    : >> In message: <20050529121224.L52379@fledge.watson.org>
    : >> Robert Watson <rwatson@FreeBSD.org> writes:
    : >> : driver, we've gained many of the benefits of making MPSAFE. Especially
    : >> if
    : >> : we can knock Giant off a lot of the remaining non-CAM device drivers that
    : >> : it's still stuck over.
    : >>
    : >> usb, psm and atkbd are important ones for interactive performance...
    : >
    : > And the tty subsystem becomes an immediate dependency for several of these.
    :
    : Interactive performance is irrelevant here. You can can lose efficiency
    : by competing for Giant a few hundred thousand times in the time that
    : it takes an interactive user to notice. The problem is that non-interactive
    : performance is impaired by using Giant locking anywhere. Using it for
    : interactive devices is only a problem if the interaction often requires
    : holding Giant for a long time. It doesn't for most input devices
    : including keyboards, but it might for slow bulk output devices like
    : syscons consoles.

    Also in the case of usb, it means that all devices sharing interrupts
    with the usb bridges are effectively under giant...

    Warner
    _______________________________________________
    freebsd-arch@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-arch
    To unsubscribe, send any mail to "freebsd-arch-unsubscribe@freebsd.org"

  • Next message: Jacques Vidrine: "Re: [CFR] correct type of addrinfo.ai_addrlen and netent.n_net"

    Relevant Pages

    • Re: Fwd: Removing Giant from a driver
      ... pass that mutex instead of Giant... ... I doubt this driver is using a fast interrupt handler... ... A fast interrupt handler doesn't context switch, ... don't understand what Giant is good for. ...
      (freebsd-hackers)
    • Re: New TTY layer: condvar(9) and Giant
      ... to reimplement a new TTY layer that fixes a lot of architectural ... I'm using a new PTY driver called pts. ... per-class mutex to lock all related TTY devices down at once (i.e. ... When set to Giant, all TTY instances will lock down the ...
      (freebsd-arch)
    • Re: Pen Drive & Win98
      ... > drives are controlled by the same driver. ... >> "Are you still wasting your time with spam?... ... >> Protected by GIANT Company's Spam Inspector ...
      (microsoft.public.development.device.drivers)
    • Re: New != Faster
      ... 6.x equals or outperforms 4.x on disk ... I/O (depends on driver) and filesystem I/O. ... all now SMP-safe or do some still live under GIANT? ...
      (freebsd-questions)