Le Mardi 9 Octobre 2001 10:59, vous avez �crit : > En mettant un disque dans le lecteur au boot, �a fonctionne parfaitement > tant que je n'enl�ve pas le disque
Alors au choix : +refaire une install avec le zip branch� et une d7 zip dedans. +faire une entr�e � la main dans /lib/dev-state +patcher le kernel le pb est d�battu sur la ML cooker : Re: Possible bug in ide-floppy.c Re: [Cooker] Internal ide Zip devfs problems? De : Guillaume Cottenceau <[EMAIL PROTECTED]> � : [EMAIL PROTECTED] Cc : [EMAIL PROTECTED] Date : 08 Oct 2001 18:08:28 +0200 Chmouel Boudjnah <[EMAIL PROTECTED]> writes: > Borsenkow Andrej <[EMAIL PROTECTED]> writes: > > > nothing. IMHO grok_partitions should register at least .../disc node to > > allow access to media. Chmouel? > > look fine in that case Kevin, i'm not sure of the side effect of > setting a dummy value for empty floppy.. I tried to reproduce with 8.1 (2.4.8-26mdk) with a IDE ZIP100 and... it works for me. Actually it seems to be because I did the install with a disk in drive; it created the correct /lib/dev-state stuff and whenever I reboot without a disk or I software-eject the disk, the /dev/hdd4 an /dev/ide/... files always stay present (e.g. they are never unlink'ed). Now, when the dev-state stuff is removed, effectively I can reproduce the problem. Whenever you software-eject the disc, the files disappear. The only way to see the disk is to manually create the device node. The only solution apart patching the kernel code seems to be to use the ide-scsi interface for the ZIP drive, and to use the file "/dev/scsi/disc?/part4" as special device file, to "see" the partition table even when you boot with no disc in drive (/dev/sda4 and /dev/scsi/host0/... both don't work). Re: Possible bug in ide-floppy.c Re: [Cooker] Internal ide Zip devfs problems? De : Kevin Maciunas <[EMAIL PROTECTED]> � : [EMAIL PROTECTED] Cc : [EMAIL PROTECTED] Date : 09 Oct 2001 10:24:01 +0930 On Tue, 2001-10-09 at 01:38, Guillaume Cottenceau wrote: > Chmouel Boudjnah <[EMAIL PROTECTED]> writes: > > I tried to reproduce with 8.1 (2.4.8-26mdk) with a IDE ZIP100 and... it > works for me. > > Actually it seems to be because I did the install with a disk in drive; it > created the correct /lib/dev-state stuff and whenever I reboot without a > disk or I software-eject the disk, the /dev/hdd4 an /dev/ide/... files > always stay present (e.g. they are never unlink'ed). > But to do this is to obviate devfs. The whole point is that we don't need /dev/kitchen-sink - devfs builds what is needed. Granted, it is useful to cache state and restore it (which is what you've just said). However, it is a poor solution in this case - "When installing the O/S, make sure you have all your IDE removeable media installed in the drive, you can take it out after installation". Yeah, right :-) NOT having the disk inserted **WILL** be the standard configuration, and so all the newbie folk will trip over this one, and so believe that LM does not support Zips. > > The only solution apart patching the kernel code seems to be to use the > ide-scsi interface for the ZIP drive, and to use the file > "/dev/scsi/disc?/part4" as special device file, to "see" the partition > table even when you boot with no disc in drive (/dev/sda4 and > /dev/scsi/host0/... both don't work). > Is there something fundamentally wrong with Andrej's patch to ide-floppy.c? It looks innocent enough, although I don't have a ZIP-250 drive to see if this change will make the system think a Zip-250 is actually a Zip-100. Surely it would be easier to put out a new ide-floppy module as part of an update than turn on ide-scsi emulation layers for zip drives? For those who read this list, clearly either solution is fine, but what's the long term fix Chmouel? -- http://perso.wanadoo.fr/linux_wizard/index.html - Catherine Deneuve, C'est l'homme que je voudrais �tre. G�rard Depardieu.
Vous souhaitez acquerir votre Pack ou des Services MandrakeSoft? Rendez-vous sur "http://www.mandrakestore.com";
