Date: Wed, 26 Jul 2006 13:59:07 GMT From: Hans Petter Selasky <hselasky@FreeBSD.org> To: Perforce Change Reviews <perforce@FreeBSD.org> Subject: PERFORCE change 102456 for review Message-ID: <200607261359.k6QDx7dL022649@repoman.freebsd.org>
next in thread | raw e-mail | index | archive | help
http://perforce.freebsd.org/chv.cgi?CH=102456 Change 102456 by hselasky@hselasky_mini_itx on 2006/07/26 13:58:20 There is a big problem in the keyboard system: If scroll lock is pressed and "printf()" is called, the Giant lock must be acquired by "printf()" before un-scroll-locking the keyboard by an IOCTL call. Hence that will cause a locking reversal problem, I see no other solution than to disable the IOCTL call, leaving the keyboard scroll-locked. Affected files ... .. //depot/projects/usb/src/sys/dev/usb/ukbd.c#7 edit Differences ... ==== //depot/projects/usb/src/sys/dev/usb/ukbd.c#7 (text+ko) ==== @@ -1162,6 +1162,21 @@ struct ukbd_softc *sc = kbd->kb_data; int i; + if (!mtx_owned(&Giant)) { + /* XXX big problem: + * If scroll lock is pressed and + * "printf()" is called, the CPU will + * get here, to un-scroll lock the + * keyboard. But if "printf()" acquires + * the "Giant" lock, there will be a + * locking order reversal problem, + * so the keyboard system must get + * out of "Giant" first, before + * the CPU can proceed here ... + */ + return EINVAL; + } + mtx_assert(&Giant, MA_OWNED); switch (cmd) {
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200607261359.k6QDx7dL022649>