Date: Wed, 20 Sep 2006 14:14:22 -0500 From: "M. L. Dodson" <mldodson@houston.rr.com> To: John-Mark Gurney <gurney_j@resnet.uoregon.edu> Cc: freebsd-firewire@freebsd.org Subject: Re: devfs and hot unplugging firewire device Message-ID: <200609201414.22322.mldodson@houston.rr.com> In-Reply-To: <20060919171950.GD23915@funkthat.com> References: <200609191005.17015.mldodson@houston.rr.com> <200609191125.35128.mldodson@houston.rr.com> <20060919171950.GD23915@funkthat.com>
next in thread | previous in thread | raw e-mail | index | archive | help
More experimental results on firewire disks and hot unplugging: Transcripts of two sessions can be retrieved from http://siegfried.utmb.edu/bdodson/firewire.on.histidine.txt and http://siegfried.utmb.edu/bdodson/firewire.on.serine.txt To summarize, on two different machines (histidine and serine) using two different brands of firewire PCI cards, but both using Texas Instruments chips, plugging and unplugging an external firewire device (Ultra ATA Seagate Barracuda 7200.7, 120GB disks in a combo USB 2.0/Firewire device) in no case gave funky /dev/da0s1a? devices, if the devices were never mounted (see earlier postings in this thread). In addition plugging and unplugging never destroyed the /dev/da0s1a - /dev/da0s1g nodes created the first time the device was plugged in. The nodes always persisted for at least 10 minutes past the hot unplug time. This may be the expected behaviour, but I had thought that the nodes in /dev for unplugged devices were deleted after some timeout. I guess I was wrong. However, after a lot of plugging/unplugging actions (see the serine transcript), an attempt to mount one of the partitions caused the machine to become mostly unresponsive: It responded to the keyboard with the echo, but the contents of the input buffer did not seem to get further. Attempts to ssh into the machine were unsuccessful (unresponsive). Attempts to log on as root at the physical console did not succeed (it went into catatonia after typing the password and pressing return). So something is not right with devfs and at least these firewire cards and device, with multiple hot plug/unplug events. My hypothesis is that the /dev/da0s1a? devices only appear if you hot unplug one disk, then immediately plug in another disk (in the same combo ATA->firewire adapter?). Mounting a partition may also be necessary for their subsequent creation. Multiple plugging/unplugging events cause the lockups when a subsequent attempt to mount one of the partitions is tried. Any ideas to clue me in? Who should I talk to about firewire drivers and devfs on -STABLE/-RELEASE? Bud Dodson -- M. L. Dodson Email: mldodson-at-houston-dot-rr-dot-com Phone: eight_three_two-56_three-386_one
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200609201414.22322.mldodson>