Subject: Re: have scsibus sleep only one time
To: None <firstname.lastname@example.org>
From: Christoph Badura <email@example.com>
Date: 03/10/2003 01:31:30
Bill Studenmund wrote:
>On Sun, 9 Mar 2003, Manuel Bouyer wrote:
>> would anyone object to the attached change ?
>> It makes all the attached scsibusses sleep "in parallel" instead of
>> sequencially, which is a real win for machines with several SCSI busses.
>> On my DS20, which has 8 scsi busses, this saves 14 seconds on boot.
>I'll be honest, I don't know that part of the code well. I assume that the
>resetting of all busses occured above where you moved the code to?
>i.e. we reset all busses, sleep once, then probe all busses? One question,
>won't that make the per-bus check, (chan->chan_flags &
>SCSIPI_CHAN_NOSETTLE) == 0, useless?
Actually, what business do we have resetting the SCSI busses?
1) this wreaks havoc with scsibusses with multiple initiators on the bus
(vulgo: HA clusters). We'll have to deal with that anyway at one point.
2) most, if not all, of the host adapters/firmware already do a reset on
boot, sometimes controlled by per-adapter configuration settings. why bother
a second time.
(The same goes for resetting ATA disks. The only effects I seem to be able
to witness is that a) the bios sleep and suspend setting get lost. b) the
bios-programmed geometry gets lost. Both aren't exactly bonus points in my