There is no working IDE hotswap support available for most hot-swappable IDE
controllers. For instance, many laptops include drive bays where either a cdrom
or an hdd can be inserted. hdparm and idectl are supposed to have ide channel
register/unregister functionality, but on my system both utilities have been
broken for a long time (due to changes in the kernel), not to mention that the
functionality is undocumented and marked experimental.
There is a utility called hotswap with motif and qt frontends and no homepage I
could find, but as it relies on a custom version of idectl, it also rarely works
(though on my system it was the only one that has worked a few times).
Another problem with ide in the kernel is that if the controller has an hdd
plugged in at boot and it is later swapped for a cdrom, the cdrom driver is not
loaded. Also if there was nothing in the bay at boot starting the controller is
problematic. A potential solution is discussed in a forum post below.
I don't know how hotswap works on SATA, but any solution for this would benefit
from the ability to work on SATA. Unfortunately SATA hotswap hardware seems to
be even less common than IDE.
Some laptops (e.g. IBM Thinkpads) may need extra hotswap support beyond IDE
controller management. On modern Thinkpads, ibm-acpi provides functionality for
powering the bay down/up before swapping. Custom acpid scripts are probably
needed to execute all the necessary actions. Also, spinning down the drives is
necessary before unregistering the controller.
Sorry, but there is virtually zero what could Gentoo do about this. I have
absolutely no clue why are you filing this bug with Gentoo.
Ask hardware manufacturers to support such feature and release the
specifications so that *upstream* kernel developers are in turn able to support
these features in kernel. For most IDE hardware hotplug is a definitive evil(tm)
that may damage your hardware.
IDE hardware can be hotplugged as long as the device is parked and turned off,
the controller is turned off, and the pins are disconnected in order. Like I
said, this technology is in use in laptops as well as some servers.
I filed this bug with Gentoo because minimal hotplug support exists in the
kernel (the controller can be turned off, the bus can be rescanned, and I think
the necessary modules can be loaded as needed, though like I mentioned this has
problems) and userspace work is needed to see if any problems exist on the
kernel side. There is no active project handling this, and this is a definite
need for laptop users.
(In reply to comment #2)
> There is no active project handling this, and this is a definite
> need for laptop users.
Sorry, Gentoo has no such project either.