Indeed, "grave" may have been too high, but then the Severity description forgot to include the possibility of actually killing hardware, which should be rated rather high IMHO.
I didn't know about the Sid tagging (Sid here, with a custom 2.6.11-ck8 kernel), sorry. The LKML thread indicates that FAT sync behaviour has been changed by Colin Leroy (http://readlist.com/lists/vger.kernel.org/linux-kernel/22/112590.html). The discussion thread for this patch is at http://www.ussg.iu.edu/hypermail/linux/kernel/0411.3/0000.html For the record, I'd like to publicly state here: - pmount should get rid of default sync ASAP for the moment, in order to avoid too many killed USB sticks - contrary to the strange belief of certain people: "we don't need a new sync mode", http://readlist.com/lists/vger.kernel.org/linux-kernel/22/112509.html) , I do think we want a special mount mode (if possible), to indicate that we want fast, but not ultra-immediate (sync) writeout of dirty buffers. The usual caching in async mode will take too long for people to notice that things aren't fully written yet (the USB flash LED will switch off again!), thus we need faster processing (almost constant writing without many dropouts would be desireable, yet not as crazy as updates on every single-byte change as with the sync flag). Plus, thinking of *manually* using "umount" or even something esoteric as "sync" is absolutely unthinkable for "less educated" (ahem) people who unplug things on a whim. Such a new mount flag could be called "nowritecache" or "nowritebuffer" to indicate that we don't want any write caching in the OS. But this is not a very accurate description of what we really want to have with USB sticks, so maybe "fastwrite" (preferred) or "instantwrite" or "immediatewrite" (hmm) are better. Thanks, Andreas Mohr -- No programming skills!? Why not help translate many Linux applications! https://launchpad.ubuntu.com/rosetta -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
