DragonFly BSD
DragonFly commits List (threaded) for 2005-10
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: cvs commit: src/sys/sys systm.h src/sys/kern kern_synch.c


From: Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>
Date: Mon, 24 Oct 2005 15:10:06 -0700 (PDT)

:eirikn      2005/10/24 14:57:55 PDT
:
:DragonFly src repository
:
:  Modified files:
:    sys/sys              systm.h 
:    sys/kern             kern_synch.c 
:  Log:
:  Make wakeup, tsleep and friends MP-safe.
:  
:  Reviewed by: hsu
:  
:  Revision  Changes    Path
:  1.32      +1 -0      src/sys/sys/systm.h
:  1.49      +140 -12   src/sys/kern/kern_synch.c
:
:
:http://www.dragonflybsd.org/cvsweb/src/sys/sys/systm.h.diff?r1=1.31&r2=1.32&f=u
:http://www.dragonflybsd.org/cvsweb/src/sys/kern/kern_synch.c.diff?r1=1.48&r2=1.49&f=u

    Urmm.... you can't safely malloc inside tsleep/wakeup, it will break
    the atomicy of the tsleep operation.  We'll have to redo this.  You
    also can't safely use the messaging functions... or at least, they
    weren't designed to be used in the tsleep/wakeup path.

    You can't safely pass a thread pointer with a message... the thread could
    very easily go away inbetween the time you sent the message and the time
    the other cpu would have otherwise received it.

    The only thing that can be safely used is the IPIQ routines, which you
    can use to pass the identifier to the other cpu's once you've determined
    that they 'might' have work to do.

    I would'a told ya that if you had posted it to submit first for review.

    I think you need to back this out temporarily, until we can fix all of
    these issues.  But, I *do* appreciate the attempt!

					-Matt
					Matthew Dillon 
					<dillon@xxxxxxxxxxxxx>



[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]