Date: Thu, 06 Feb 1997 10:35:32 -0800 (PST) From: Simon Shapiro <Shimon@i-Connect.Net> To: "Justin T. Gibbs" <gibbs@narnia.plutotech.com> Cc: freebsd-hackers@FreeBSD.ORG, freebsd-scsi@FreeBSD.ORG Subject: Re: Contigious (spelling?) allocation in kernel Message-ID: <XFMail.970206131700.Shimon@i-Connect.Net> In-Reply-To: <199702060533.VAA00624@narnia.plutotech.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Justin T. Gibbs; On 06-Feb-97 you wrote: ... > You really think you're going to ever transfer a full 32MB of data in a single > transaction? Even if you were, you'd have to do something special to the > kernel since it currently only generates requests of at most 64k (its a silly > restriction and it should die a horrible death). I agree. It does funny things to tape drives, CD writers, etc. > Assuming you did have a way > to stuff 32MB through all of the layers, you'd probably be better off chopping > that I/O into smaller chunks so as to reduce the amount of SG space you must > allocate. At anything over 8k, the transaction overhead starts to become an > increasingly small percentage of your I/O time, so dropping even down as low as > 1MB per transaction may better utilize your resources. Agree. The board vendor would like to see these large transfers. But a 1MB limit is not unreasonable. ... > You don't want to continuously malloc your SG list. This will kill your > performance. If it is acceptable for the system to run with all 256 > requests active at once, you have to be able to deal with all of the requests > being allocated anyway, so you might as well pool them. I agree. We never thought to dynamically allocate them, per request. What I have done is to allocate them, in blocks, into a free list, every time the free list is empty. I do not have a mechanism to ever free them, but that can be added later by comparing the current freelist size with a usage high water mark. All this brings me to another question. How can I create another kernel thread? What I am trying to do is have several execution threads in the kernel, each responsible for a given task. All I see in the drivers I read so far is a single thread, associated wit hthe calling user process. Another thread is invoked by the interrupt handler. I want a couple more. One ugly way to do it is to have a ``daemon'' which makes a known system call (ioctl) which is never returned. Aside from being ugly, it has a problem at boot time; The daemon is not there until way after init has started. What I am after is something that could resemble the swapper. Linux has few of them and they become very handy. Especially on an SMP machine. Thanx, Simon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.970206131700.Shimon>