From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <3ACBA866.50404@amberdata.demon.co.uk> Date: Thu, 05 Apr 2001 00:04:06 +0100 From: David Monro MIME-Version: 1.0 To: linuxppc-dev@lists.linuxppc.org Subject: asm-ppc/ide.h and the evil insw hack. Content-Type: text/plain; charset=us-ascii; format=flowed Sender: owner-linuxppc-dev@lists.linuxppc.org List-Id: Hi, first a couple of caveats: I'm using a vanilla kernel.org tree and I don't normally follow linuxppc-dev, so I may be saying something stupid, but here goes... Can we possibly get rid of the really evil hac in asm-ppc/ide.h where insw and outsw get redefined? Reasons for wanting to do this: 1) any unsuspecting soul who includes into some random file suddenly gets these redefined underneath them 2) because we only hack the "w" versions, setting the interface to 32 bit transfers (hdparm -c1) breaks horribly. 3) its an evil hack. In fact I have an ulterior motive; the only easy way for me to get ide working properly on my IBM 850 (PReP architecture) is to remove the hack and set ppc_ide_md.fix_driveid to be ppc_generiv_ide_fix_driveid() at which point the interface works absolutely perfectly (including 32 bit transfers and IDE cdroms which always break if the hack is in place at least for me). The problem is I am sure including these changes will break somebody else. I'm guessing that the chrp systems are probably broken in the same way as the PReP ones are at the moment (most of the chrp systems I know about seem to use SCSI so maybe noone has tested it). I'm really much more worried about the macs. Anyone care to comment? btw if you do some experiments and find that you end up in a state where the interface seems to work but your partition table isn't recognized you may need the hdx=swapdata option. Cheers, David ** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/