Also, if there's a way to have USB devices have multiple
personalities, i.e. the Novatel WCDMA card which shows up as a serial port
and as an SD  card reader.  This may be orthogonal to the autoconf project
you're discussing, but if it's relevant, itmight be useful to address
the issue here.
-Brian

On Dec 10,  9:33pm, Quentin Garnier wrote:
} Subject: Re: Renovating autoconf(9)
} 
} --AkbCVLjbJ9qUtAXD
} Content-Type: text/plain; charset=us-ascii
} Content-Disposition: inline
} Content-Transfer-Encoding: quoted-printable
} 
} On Mon, Dec 10, 2007 at 03:08:14PM -0500, Michael Lorenz wrote:
} > I'm not sure if this can be done in a sane and easy way but who knows.
} > Sometimes it would be useful for a device to have more than one =20
} > parent, like some Apple audio chips which attach to obio for DMA, =20
} > gpio etc. and a mixer part that attaches to an i2c bus. Right now the =20
} > DMA part attaches and defers further configuration, then digs though =20
} > the device list looking for its long lost cousin, errm - mixer, which =20
} > is kind of ugly.
} 
} The problem with your example is that there is information about the
} identity (in the sense that they're the same) of the two devices is
} only known through God's advice (i.e., reading the docs about the
} hardware).
} 
} A better example of multi-parenting would be a SCSI device seen through
} two different HBAs.  In this case there is a way to tell the device is
} the same because it carries the same identification.  We could thus
} imagine adding a parameter to ac_alloc_child() e.g., hwident (which
} would only make sense for a given device class) which would help
} retrieve the existing device_t.
} 
} If in your issue you have such a gateway between the two sides of the
} audio device, then it's possible.  Of course there are a few other
} things to change in the API...
} 
} Also, I've been thinking about adding the possibility to have multiple
} drivers attaching to a device, a main driver, and several feature-
} specific ones.  One situation where it is an issue is the PCI-to-ISA
} bridges on x86:  if I create an acpipcib(4), then we lose the ability
} to attach ichlpcib(4).  It would be better to keep the existing pcib(4)
} and let ichlpcib and apcipcib be persona of the pcib(4) device.
} 
} --=20
} Quentin Garnier - cube@cubidou.net - cube@NetBSD.org
} "See the look on my face from staying too long in one place
} [...] every time the morning breaks I know I'm closer to falling"
} KT Tunstall, Saving My Face, Drastic Fantastic, 2007.
} 
} --AkbCVLjbJ9qUtAXD
} Content-Type: application/pgp-signature
} Content-Disposition: inline
} 
} -----BEGIN PGP SIGNATURE-----
} Version: GnuPG v1.4.6 (NetBSD)
} 
} iQEVAwUBR12ihNgoQloHrPnoAQLy3gf/erj2Rn9PAdPdvb75C2ZDFDpTh/ZjdycJ
} 4L+xhcvIwOIfeeGcDEZS+W20u/6rK06bK3c/oRp9oYGoNR1zDmtZKoaVds29uThj
} 9WMxW4xpP0XDAuLeOjC1s4EqA1pxmQaLpPV+rPbBxKqz3/Qz2azUpCojMEttgMRt
} a3QI8YlnDO4SdQM2LZjeoRju8v8ELjVozY9NQRZj0Tc9HHQB9Ipnlf7b+M6D/xSB
} RUtypvcT/IP1RlINTe1fq7bG2p9ncIYVmh/BoXcN0pz8OqiaNZyfNTOOp3zrahOF
} 0dtC3hi3atJvivZrR8/fi145OvnMpHQB1BNw+jYtr6l/KXT49KM0Gg==
} =1RHb
} -----END PGP SIGNATURE-----
} 
} --AkbCVLjbJ9qUtAXD--
>-- End of excerpt from Quentin Garnier