[lkml]   [2006]   [Sep]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    SubjectRe: patch [0/2]: acpi: add generic removable drive bay support
    On Thu, Sep 07, 2006 at 04:13:05PM -0700, Kristen Carlson Accardi wrote:

    Firstly, thanks for this - I wrote some related code a while ago. A
    couple of questions...

    > can then be used by udev to unmount or rescan depending on the event. It will
    > create a proc entry under /proc/acpi/bay for "eject" and for "status". Writing

    Do we really want it under /proc? It would seem to make more sense for
    it to be under /sys.

    > a 1 to the eject call will cause the drive bays acpi eject method to be called,
    > and should be done prior to removing the device. Reading the "status" entry
    > will tell you either "present" or "removed" depending on the state of the
    > device. User space scripts can use the status entry to determine if a device
    > has been either inserted or removed in the case of some laptops who generate
    > the exact same event for either insertion or removal (i.e. the Dell M65 for
    > example). Same scripts for using these events and udev can be found on the
    > thinkwiki website:

    What gets generated if I rip a drive out without notifying the system
    beforehand? Dell hardware doesn't appear to send any event when poking
    the eject lever.

    The way I originally implemented this was to tie it into the libata
    layer directly. With a small amount of glue it's possible to tie a
    hotswap bay to a specific sata port, and use the event to trigger a
    hotplug rescan. Doing it in kernel space means that the device can be
    destroyed the moment it's removed, reducing the possibility that further
    writes will be made. There was some amount of resistance to acpi
    integration in the scsi layer, though I think we eventually reached a
    compromise about providing support for platform firmware hooks.

    Doing it this way also means that the bay itself can be represented in
    sysfs as an attribute under the appropriate port. That seems more
    natural than having the bay be entirely separate, despite ACPI providing
    us with the information for them to be tied together.

    Matthew Garrett |
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

     \ /
      Last update: 2006-09-08 22:05    [W:0.023 / U:12.068 seconds]
    ©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site