Date: Thu, 20 Jan 2000 15:46:38 -0700 From: Brett Glass <brett@lariat.org> To: Terry Lambert <tlambert@primenet.com> Cc: tlambert@primenet.com (Terry Lambert), jcm@dogma.freebsd-uk.eu.org (Jonathon McKitrick), chat@FreeBSD.ORG (freebsd-chat) Subject: Re: IBM Message-ID: <4.2.2.20000120145704.01a24100@localhost> In-Reply-To: <200001202145.OAA18179@usr01.primenet.com> References: <4.2.2.20000118162503.0193bc60@localhost>
next in thread | previous in thread | raw e-mail | index | archive | help
At 02:45 PM 1/20/2000 , Terry Lambert wrote: > > It's under Windows; after all, the modem doesn't work AT ALL > > under BSD. But while Windows is unstable, the modem driver > > should not be coded so badly as to lock up the machine. > >So you are complaining about WinModems in general, and using a >particular instance as an example. This one has a particularly bad driver. Most Winmodems do not crash the machine. So, I do fault IBM for this malfunction. They wrote the driver. > > While Frank's an excellent programmer, I somehow do not think that > > Sourcer will disassemble the MWave code. ;-) > >It disassembled a network card driver for me, so that I could hack >it and reassemble it to fix a bug. > >It will certainly disassemble and comment OS entry points for >x86 code that is used to load the MWave CODEC. If you are talking >about the CODEC itself, then I think you will need to contact >the company that licenses it, probably in binary form. This is >probably _not_ IBM, but instead a third party vendor. The thing is that the MWave is reloaded on the fly, just like the TMS320 in the old IBM Voice Communications Option boards. (IBM's MWave strategy was launched by the people who developed those boards, in cooperation with TI and Dragon Systems. I worked with the VCO a great deal, and so know how it was architected internally. It was VERY complex and you needed to know a lot of detail.) So you won't just need to load code once, but will need to swap code at very specific times. >It is illegal in the US everywhere shrink-wrap licensing is >legally binding. The "Millenium Copyright Act" will incidently >make this "everywhere in the US", and will additionally apply it >to things like videos and CDs. So be ready to say goodbye to >used videos and CDs if it passes. I think you're overly pessimistic about the current situation, but there's surely a drive to make it that way in the future. UCITA is actually the biggest threat, not the Millennium Copyright Act. But the latter is significant too. >The Stac lawsuit came about by virtue of them naming thier >compresstion code "DRVSPACE.SYS", taking advantage of the fact >that Microsoft engineered the IO.SYS to load it by name (an >anti-competitive practice, in itself) to give their code a >special hook to allow booting from a compressed disk. As I understand it, the court was going to rule against them because of a shrinkwrap prohibition against disassembly of the MS-DOS boot code. One of the reasons Microsoft settled is that they didn't want to lose this point -- the ONLY point they would have won at trial -- on appeal. > There wasn't any other option on that machine. >PCMCIA. I know for a fact that modems were an add-on option, >unless they were hooked through a sound card, in which case >they are a non-option (e.g. don't expect the phone connector >to work just because you paid for it). But what about sound? I could add a modem, but the laptop would still be mute. Unfortunately, no PCMCIA sound card works with BSD. And they're almost impossible to find now that virtually all laptops integrate sound. I have an old Media Vision one, but it was equivalent to a Sound Blaster Pro in quality and no one supports it anymore. I hear that Roland did a PCMCIA Sound Canvas card but has now dropped it. > > If I'd known that > > IBM would do such a poor job of writing drivers, or about > > the OTHER flaws in that unit (which none of the reviewers had > > documented), I wouldn't have bought it. > >The chip vendor wrote the driver. IBM only wrote the loader (if >that). I expect that it was really the vendor. MWave was IBM's own thing. At first, some of the design was licensed from TI; IBM then branched off on its own. But IBM wrote the drivers. See http://watson.mbb.sfu.ca/mwave.html. >Tell me, if you bought a PC with a Diamond video board or an >Adaptec SCSI controller during the black days of both those >vendors, would you have expected it to work under a non-Windows >OS? I did both! Diamond supported OS/2. And Adaptec published interfaces for its 1542 boards. I used them with Coherent and QNX. >Let's narrow that down: Winmodems post-date OS/2 as a >supported product on anything but POS systems. Not the MWave. In fact, if you look on IBM's ThinkPad site, they specifically list the model I have as supporting OS/2. You're probably thinking of the USR/Rockwell devices that were actually CALLED "WinModems." >You are looking too deep, then. You don't want to disassemble >the MWave code, you only want to disassemble the Windows loader >for the MWave code, and write a FreeBSD loader that loads the >code. That won't work. The modem code is partly in the MWave and partly on the host, and modules get swapped dynamically (more on this below). I don't even think that all of the modem speeds fit at once; when you move from Bell 303 to Bell 212 to FAX, etc., code is brought in for that modulation scheme. It is said to be possible to load the sound code and keep it there, but it will only emulate a cheap Sound Blaster. >The MWave code that is non-x86 code is going to be the same for >all supported OSs, regardless, just like the SCO version and >Windows NT version of the Adaptec RAID controller microcode are >the same, and only the loader code for it differs. If that is so, then why doesn't IBM simply publish the binaries and instructions for loading them? The answer, it appears, is that it isn't QUITE that simple. There's a "discriminator" program, for example, that answers phone calls and attempts to quickly load the module that handles the type of call that's come in -- voice, data, etc. So, you don't just load them; you must communicate with them through sophisticated APIs. And swap them at the right moments. >The one concession on this point which I will have to make (and >you will have to deal with) is that there is likely a fixed >amount of main CPU required in order to run the CODEC, Not the CODEC but other functions. I believe that when the modem runs V.32/V.42bis, V.32 runs on the MWave and V.42bis on the host. > > Most often, the vendor of the hardware has already paid the > > required royalty. Rockwell pays royalties on the Heatherington > > patent for all buyers of its hardware. I'm sure that, in the > > case of the ThinkPad, the royalty has been paid before the > > computer leaves the shop. > >Read the fine print; it is likely restricted to a license for >Windows as the OS. If it's not, go for it. We'd have to see. As mentioned above, though, it may not be feasible. >> After I lost the use of it for a month and a half and had > > to purchase a replacement in the interim. > >I had the same thing happen with a car. Welcome to the new >so-called "service economy", where we can all produce no >tangible results or goods, and get paid anyway. Sigh. >[ ... IBM service philosophy ... ] > > > No, I'm afraid that my opinions are based solely on actual > > experience, not indoctrination! ;-) > >Apparently not including your "valiant" Customer Care Representative. Well, she TRIED valiantly, but the repair contractor delivered no faster than before. The replacement part they put in this time does not seem to be defective, but it is unclear whether it was luck or skill that caused this. (Pardon me for sounding cynical, but the repair people were REALLY lame.) >Linux gets you press. Linux is nothing more than a vehicle for >getting press. The trouble is that Linux is a direct competitor, which means that for every bit of press it gets BSD, it gets much more for itself. BSD should not rely on Linux for publicity! >IBM had opportunity to buy other Whistle-like companies that >used Linux, but bought Whistle instead because of the intellectual >property rights dilution effects of the GPL. It's good to see that they're aware of this! I noted that their license for ViaVoice was more BSD-like than GPL-like -- another positive sign. >Believe me, there is no love of Linux in the legal departments of >any of the apparent Linux boosters out there; if you have a patent >at all, you let the marketing department hold Linux up to get >their press and to keep them from whining, but you make them hold >it at arms reach, much in the same way you would hold a dead skunk. Interesting. This does not seem to be the case at at least some of them, though. Corel seems to have bought Raymond's rhetoric hook, line, and sinker. (Of course, Corel may be a poster child for why NOT to do Linux. They do embedded stuff and value-added software; they really could benefit from moving to BSD.) >Adaptec drivers are integrated into Windows. If there is a GUI, >it is a third-party add-on, and not an integral part of the driver. No, the driver can (and often does!) contain GUI components. Have you ever developed a Windows printer driver? There are hooks for GUI applets which are part of the driver. In fact, the new "Winprinters" rely on this for ink monitoring, etc. >Executives of large companies only talk in terms of the >dollars or in terams of "deals". If you can't speak their >language, then they can't hear you; they see your lips moving, >but that's all they see. I know. Alas, they seem to think that only Microsoft products are associated with dollars, so it's a tough sell. --Brett To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4.2.2.20000120145704.01a24100>