|
Re: 2.6.12 / udev / pcmcia: msg#00018handhelds.linux.kernel
On Mon, 2005-09-19 at 15:06 -0400, Todd Blumer wrote: > Since we went to kernel 2.6.12, we've had problems with CF memory cards. > After doing some tracking, it seems that udev is removing the hda device > in /dev before /etc/pcmcia/ide has a chance to run (to unmount the > filesystem). Is the correct solution to inform udev to ignore pcmcia? > That seems to work (e.g. mv /sbin/udevd /sbin/junk). What is the proper > way to do this? I'm in the process of trying to knock OE's udev scripts into shape. udev in OE has recently started to disable hotplug and this has exposed several problems. The replacement of pcmcia-cs with pcmciautils will probably finish things off. Does pcmciautils support the /etc/pcmcia scripts? Regardless, we can probably do the mounting of disk devices (be them pcmcia, mmc etc.) through udev scripts calling pmount (which is the current plan). > Second, if a process is in the midst of reading the directory structure > (or file) from CF and one removes the card, a panic will result. Has > anyone looked at that problem? > Thanks for your insights. I've seen this but don't know what to do about it. I guess a lazy unmount won't help as the device is gone. I seem to remember Alan Cox saying we needed to kill all the IDE ops (make then null) and let the VFS fail gracefully rather that the current pull the feet from under it approach. I've not really looked into it though. Richard |
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| Previous by Date: | 2.6.12 / udev / pcmcia: 00018, Todd Blumer |
|---|---|
| Next by Date: | Re: 2.6.12 / udev / pcmcia: 00018, pHilipp Zabel |
| Previous by Thread: | 2.6.12 / udev / pcmciai: 00018, Todd Blumer |
| Next by Thread: | Re: 2.6.12 / udev / pcmcia: 00018, pHilipp Zabel |
| Indexes: | [Date] [Thread] [Top] [All Lists] |
| News | FAQ | advertise |