Date: Thu, 18 Nov 1999 22:16:37 -0500 (EST) From: Bosko Milekic <bmilekic@dsuper.net> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: freebsd-hackers@FreeBSD.ORG Subject: Re: mbuf wait code (revisited) -- review? Message-ID: <Pine.OSF.4.05.9911182204400.11057-100000@oracle.dsuper.net> In-Reply-To: <199911190102.RAA88661@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 18 Nov 1999, Matthew Dillon wrote: !> !> Sounds like a reasonable plan. I wonder if we should consider getting !> rid of the mbuf macros entirely and simply proceduralizing them. Then !> everything could be collected together into a single file. !> !> -Matt !> Matthew Dillon !> <dillon@backplane.com> !> Well, the macros are there, as commented in uipc_mbuf.c, "for critical paths." Ultimately, having them available as macros reduces some overhead. They are also, however, available as procedures (defined in uipc_mbuf.c). The procedures actually almost act as 'wrappers' for the macros because they really just end up running these macros. On what I believe emphasis should be placed upon is making sure that ALL the code using either the macros of the procedures doesn't automatically 'assume' that they will succeed, even if the call is made with M_WAIT. If the [new] diffs that I have sitting here are committed, then that should become the focus. There still seems to be plenty of code, especially around sys/nfs that 'assumes' this 'guaranteed success,' and because of that, problems arise. Bosko. -- Bosko Milekic <bmilekic@technokratis.com> To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.OSF.4.05.9911182204400.11057-100000>