Date: Mon, 03 Aug 1998 16:22:06 -0700 From: Mike Smith <mike@smith.net.au> To: chanders@timing.com (Craig Anderson) Cc: freebsd-hackers@FreeBSD.ORG Subject: Re: Device driver synchronization with spl? Message-ID: <199808032322.QAA01662@dingo.cdrom.com> In-Reply-To: Your message of "Mon, 03 Aug 1998 16:14:41 MDT." <199808032214.QAA02199@count.timing.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> > I would like an explanation of how the spl functions are used to > sychronize a driver that uses interrupts. Code fragments from a > driver I'm working on are at the end of the messages. The interrupt > handler is installed on irq 11. 'man 9 spl' on a -current system contains a fairly succinct answer. > Here are my assumptions (guesses) so far. > > 1) gt401read() calls splgt401() to add a mask for irq 11. No. I don't believe that creating new spl's is trivial anymore. In the first instance, I would simply add the 'tty' keyword to your device's ISA configuration entry (see eg. the 'sio' entries for examples). > 2) After splgt401() the interrupt function gt401intr will not run. > This prevents a wakeup being lost between checking the data count > and sleeping for data to show up. Not exactly. After spltty(), should an interrupt masked by the 'tty' mask occur, it will be remembered, and when the mask is cleared the interrupt handler will be run. > 3) The tsleep() causes another process to run with an interrupt > mask that will allow gt401intr() to run. A call to tsleep() will put the current process to sleep, preserving the current interrupt mask. When a new process is selected to run, the mask is set to that for the new process. > 4) irq 11 happens and the handler gt401intr() is run. Yes, although the interrupt may have happened during the peroid when the interrupt was masked. > 5) gt401intr() updates the data count and issues the wakeup. ... and returns. The woken process won't actually run until the interrupt handler has returned. > 6) gt401read() returns from the tsleep() and sees that there is now data. It should check to see if there is data; it may have been woken for other reasons (timeout, signal delivery, etc.). > > Some questions. > 1) Is my understanding of the spl usage good enough? Geting there. 8) > 2) If this is correct, what should splgt401() look like? Don't complicate things yet; stick with spltty(). > My guess here, after looking at /usr/src/sys/i386/include/spl.h is > GENSPL(splbio, cpl |= some_mask) > where some_mask is appropriate for irq11. If this is right what should > some_mask be? spl.h has been obsoleted in -current, you can find it in i386/isa/ ipl_funcs.c. Implementing a private interrupt mask is probably not a useful thing to do, as it's more work that necessary. -- \\ Sometimes you're ahead, \\ Mike Smith \\ sometimes you're behind. \\ mike@smith.net.au \\ The race is long, and in the \\ msmith@freebsd.org \\ end it's only with yourself. \\ msmith@cdrom.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?199808032322.QAA01662>