Date: Thu, 23 Jul 2009 18:41:20 +0300 From: Andriy Gapon <avg@icyb.net.ua> To: "Peter C. Lai" <peter@simons-rock.edu> Cc: Sagara Wijetunga <sagara@tomahawk.com.sg>, Eugene Grosbein <eugen@kuzbass.ru>, freebsd-stable@freebsd.org Subject: Re: SCSI device not created upon a CF card plug in Message-ID: <4A6884A0.5000906@icyb.net.ua> In-Reply-To: <20090723153338.GA31463@cesium.hyperfine.info> References: <20090723055317.GA72627@svzserv.kemerovo.su> <20090723141440.A69EF1CC0B@ptavv.es.net> <4A6880A1.3010003@icyb.net.ua> <20090723153338.GA31463@cesium.hyperfine.info>
next in thread | previous in thread | raw e-mail | index | archive | help
on 23/07/2009 18:33 Peter C. Lai said the following: > GEOM should pick up the state change, but the bug may continue to exist > because the CF reader is tagged as a storage device with invalid media > when umass attaches to it on startup; there wasn't a really good way to > actually enumerate the slice table upon insertion because the device > does not send a disconnect/reconnect to the bus when you insert the > card. At the time, the workaround was to tickle the device by attempting > a mount against the whole disk, then the slice table will be enumerated. It definitely doesn't do disconnect/reconnect, the question is does it send other kind of notification on media insertion/change? > Example, CF card reader on boot will be picked up as /dev/da4 > You insert card, no state change can be detected because the device > doesn't get reset on the bus. > mount /dev/da0 /mnt <-- will fail for obvious reasons, but now slices > are enumerated > mount /dev/da0s1f /mnt/mycfcard <-- ok $ true > /dev/da0 looks cleaner and is safer way to trigger the re-tasting. Still, this is a sub-case of polling. -- Andriy Gapon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4A6884A0.5000906>