Date: Sat, 4 Dec 1999 19:47:18 -0800 (PST) From: Matthew Jacob <mjacob@feral.com> To: Jason Thorpe <thorpej@nas.nasa.gov> Cc: Wilko Bulte <wilko@yedi.iaf.nl>, gallatin@cs.duke.edu, freebsd-scsi@FreeBSD.ORG, freebsd-alpha@FreeBSD.ORG Subject: Re: ISP firmware compiled in as a default.... Message-ID: <Pine.BSF.4.05.9912041938020.368-100000@semuta.feral.com> In-Reply-To: <199912050332.TAA16773@lestat.nas.nasa.gov>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 4 Dec 1999, Jason Thorpe wrote: > On Sat, 4 Dec 1999 11:43:39 -0800 (PST) > Matthew Jacob <mjacob@feral.com> wrote: > > > Jason (bless his heart) Thorpe kept on claiming that NetBSD-alpha was > > completely broken without the f/w- I never saw such breakage at all and > > real active details were not provided, and in fact *you* (Wilko) are the > > only one who I know was completely blocked w/o the f/w. > > Oh, c'mon. The whole reason you started always downloading the firmware > into the ISP is because cgd reported to you that the SRM's ISP firmware > on his AlphaStation 500 didn't play nice with the NetBSD driver. I'm > pretty sure I have an archive of the e-mail conversation (which was all > CC'd to me). Nope, I don't think so. I pretty much always had been downloading f/w. There was a hop skip and dance with some f/w and Chris's machine and some stupid ass bugs in 7.55 f/w where you'd tell it to renegotiate and then ask it what it had done and it lied and gave back random values. > > And, I'm pretty sure there's actually a PR in the database about a > PC164 user having to back-rev his machine to before the firmware was > yanked because his ISP no longer worked after a *power-cycle* (i.e. the > RAM on the card lost power, and the SRM-loaded firmware was not functional > with the NetBSD driver). Yep. But I've concluded that this wasn't a Digital supplied board (and closed the PR) so the SRM wasn't starting it. > > Not only that, but users of CATS boards (arm32) were completely left out > in the cold; the firmware on those machines doesn't run the ISP BIOS, and > thus had no way of loading in the firmware into the card. The portmaster > went as far as to yank the "isp" driver out of the example kernel config > files in that port. So I saw. I never claimed it was not completely broken in some cases. For example, if there was a sparc64/PCI port going, it'd be mostly broken there because only PTI PCI cards have fcode that gets the f/w started. Read what I wrote above. You claimed NetBSD-alpha was completely broken. But neglected to provide details. It wasn't completely broken, but it's been problematic in a number of cases. > ...or don't you read the `source-changes' mailing list? Nope- the netbsd changes list is too hard to read. > > Anyhow, the arm32 case will happen on *ANY* platform who's firmware > doesn't natively understand the ISP. So, not loading the firmware by > default screws over anyone who tries to put it in an arm32, macppc, > Atari Hades, etc. Yup. Such is life and too bad. It is now corrected by being able to have the f/w back in. I've also said that if I can get more information out of Qlogic I can figure out how to extract the onboard firmware in flashram down into sram on the card and get it going if the BIOS hasn't. > > Now, you could do something like have #ifdefs for each firmware, > i.e. > > #ifdef ISP_1020_FIRMWARE > > ... > > #ifdef ISP_1080_FIRMWARE > > ... > > #ifdef ISP_2100_FIRMWARE > > ...actually, I just noticed... there's already ISP_DISABLE_..._SUPPORT > #ifdefs in there. Why not key on those? I do have "compile in these" f/w selectors, and an overall "compile in all f/w" selectors. The NetBSD integration, respecting your unexplained concerns, wraps *around* this by defining ISP_COMPILE_FW in isp_netbsd.h. If I had time to blow on it, it would be a lot better. The subject under discussion *here* is whether to have freebsd follow the settings for NetBSD and compile/load it by default. -matt To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.05.9912041938020.368-100000>