Date: Sun, 9 Jan 2000 20:29:50 +0100 (MET) From: Gerard Roudier <groudier@club-internet.fr> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: "David O'Brien" <obrien@NUXI.com>, Peter Wemm <peter@netplex.com.au>, "Jordan K. Hubbard" <jkh@FreeBSD.org>, cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/i386/conf GENERIC LINT Message-ID: <Pine.LNX.3.95.1000109200045.1447A-100000@localhost> In-Reply-To: <200001091819.KAA19795@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 9 Jan 2000, Matthew Dillon wrote: > :On Sun, Jan 09, 2000 at 10:44:03PM +0800, Peter Wemm wrote: > :> > A) Let the 'sym' take precedence over 'ncr' (default) > :>=20 > :> Yes! > : > :I'm personally against this, as I feel it is the wrong thing to do in > :-STABLE. Changing a kernel driver out from under deployed stable system= s > :is something we state we don't do. Don't forget it also causes people > :with wired down drive numbering to have to change things if they don't b= e > :extra carful when merging GENERIC changes into their kernel config files= =2E > : > :Making it so that `sym' handles things that `ncr' does, is of our course > :a good thing to do. Then the user could have both `ncr' and `sym' in th= e > :config file and keep `ncr' as the driver for what they have today. For > :those with cards that both `ncr' and `sym' support, users can comment ou= t > :`ncr' if they wish to use `sym'. > : > :--=20 > :-- David (obrien@NUXI.com) >=20 > Don't we still have the issue of the ncr<->fxp interaction causing=20 > crashes? I was under the impression that this problem didn't occur > with the sym driver and because of that the sym driver was given > precedence... ncr<->fxp interactions and sym over ncr or vice-versa must be considered as different issues, in my opinion. It seems that the 'sym' driver has had some good effects on some systems that used to crash (don't know if they were affected by the ncr<->fxp syndrome), but I didn't read anything that proves that a sym<->fxp issue actually exists ot not. Anyway, I have asked for the PCI device id and PCI revision id of SYMBIOS/NCR chips that was involved in proven ncr<->fxp crashes (or sym involved) but never received anything clear about. Another reason for some committers seeming to prefer the sym over the ncr when both can be used could well be due to the difference in features and the sym driver never having failed when the ncr succeeds, at least for the moment.=20 If the 'sym' driver will help fix the ncr/fxp problem, I would be very glad of that. Sad I would be if this issue got mixed with default selection of a driver when both fit, unless experience proves things actually run significantly better with a given driver on systems using the offending combination of devices. We must try to fix what is actually broken.=20 G=E9rard. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.LNX.3.95.1000109200045.1447A-100000>