Messages in this thread | | | Date | Mon, 26 Oct 1998 21:59:12 +0200 (IST) | From | Gadi Oxman <> | Subject | Re: IDE tape and SMP |
| |
On Mon, 26 Oct 1998, Andre M. Hedrick wrote:
> Gadi, > > I know that you wrote the original source, and we both know that I am > trying to play pick and learn in weeks and months what you did in years. > The reality is that it will be a long time before I can move through the > source like you and Mark can. So what will it take to get you to pick up > an play again?
I'm thinking about the latest ide-tape + SMP problems; it's hard for me to fix, as I'm not very familiar with 2.1.x SMP. Mark/Scott/Erik and myself developed the drivers without access to SMP systems, and although I upgraded to a SMP system very recently, unfortunately I was not able to repeat the latest SMP problems on my system yet.
I can certainly imagine that the IDE tape driver is not SMP safe, though; it was not designed to support 2.1.x fine-grained SMP, and as cli() doesn't seem to actually disable interrupts on all processors when local interrupts were already disabled, the prevention of race conditions on UP seems to no longer work unmodified on 2.1.x SMP.
That can potentially explain the locking of the backup process in the 'D' state, as Matthew reported.
I currently don't have an explanation for the "DSC timeouts" and "IRQ timeouts" problems, though. It seems as if the timeout function is called pre-maturely, and reminds me of the problem which Geert reported a while ago, in which the IDE driver timeout handler was called before the intended expiry. We worked around that one by replacing a "add_timer()" with "mod_timer()" in ide_do_request(), but at this point no timer should have been active in the first place.
> Second, I will follow your suggested direction. As always, you will be > needed to point out assumption follies until more knowledge is hammered > into my skull. > > Just by chance, would you send me copies of any and all > docs/whitepages/etc and or forward the locations where I can get them?? > I have piles of stuff already and my wife wants me to stop the piles, but > to quote a silly movie......"MORE INPUT, MORE INPUT"........ > > > Cheers, > Andre Hedrick > The IDE-FNG for Linux > The APC UPS Specialist for Linux > > http://www.dyer.vanderbilt.edu/server/udma/ > http://www.dyer.vanderbilt.edu/server/apcupsd/ BINARIES
The IDE/ATAPI specifications are availabe on ftp://fission.dt.wdc.com, under /pub/standards:
IDE ATA DISK: ata/ata-2/0948dr3.doc IDE/ATAPI FLOPPY: SFF/specs/8070ir11.pdf IDE/ATAPI CDROM: SFF_atapi/spec/SFF8020-r2.6/PDF/8020r26.pdf IDE/ATAPI TAPE: QIC/QIC157/QIC157D.ZIP
DSC overlap is described in QIC157, while ATAPI overlap is described in SFF8020. Linux currently doesn't support ATAPI overlap, but to date I received just a single report of an ATAPI device which actually implements it.
Gadi
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |