Date: Thu, 17 Aug 1995 13:49:30 -0500 From: mar7@cec.wustl.edu (Matt Rosenberg) To: "Justin T. Gibbs" <gibbs@freefall.freebsd.org> Cc: questions@freebsd.org Subject: Re: Multiple Adaptec 2742W's Message-ID: <9508171849.AA05085@cec.wustl.edu>
next in thread | raw e-mail | index | archive | help
On Thu, 17 Aug 1995 at 07:54:26 -0700, "Justin T. Gibbs" <gibbs@freefall.freebsd.org> wrote: >>Apparently 2.0.5 is not compatable with more than 1 Adaptec >>2742W. I can add the second device line in the kernel >>configuration, and it will probe the bus for the second adaptor >>correctly (it probes both correctly). However, the kernel will >>continue to probe the remaining devices and at the end of the >>probing phase, right before the swap is turned on and the file >>systems mounted, the system locks up. I've tried different >>IRQ's for the two adaptors. I've also tried the same IRQ with >>the conflicts keyword added in the second device line. Same >>results - the adaptor and its bus are probed correctly but the >>system locks up before the file systems are mounted. Each >>adaptor has exactly one Seagate 43401N (Elite 3) drive. The >>first drive is assigned sd0 according to the messages and the >>second drive sd1, just as would be expected. > >Are you sure that the BIOS for the second card is disabled or >in a region that doesn't conflict with any other card in your >system. I've seen the same sort of symptom when the BIOS region >overlapped an ethernet cards shared mem location. The IRQs must >not be shared for EISA versions of this card (I know you've tried this). >What IRQs are you using? Are you sure that nothing else is using >those IRQs? Adaptec actually recommends using the same IRQ if the two cards are the model (which they are here). When trying the same IRQ, I've been using IRQ 11. When using the separate IRQ's, I've been using 11 and 12. I've checked closely for possible conflicts, but they do not exist. From the very beginning I've had the BIOS disabled for the second adaptor. >>BTW, although the Seagate 43401N _is_ a wide SCSI device if the >>2742W is configured to "initiate wide negotiation" for the >>43401N, FreeBSD 2.0.5 will not probe the device correctly. It >>doesn't make a difference to DOS whether that flag is turned >>on. > >Can you try the latest SNAP-shot floppy? Its not quite the latest >code, but it is much closer than 2.0.5. I'm using it as an active, up and running, web server, so I don't want to take it down many more times or much longer than possible. I did find a fix to that problem. Just turn off the "initiate wide negotiation" flag for the drive from the Adaptor configuration tool in DOS. When this flag is left on, FreeBSD reads the drive as a "S441 00427 1N" instead of "SEAGATE 43401N 0104"; the type is read as "type 0 fixed SCSI 3" instead of "type 0 fixed SCSI 2"; the size is read as 6528MB instead of the correct 2777MB; and this all causes the partition table to become undreadable, so I don't even have a chance to tell it the correct geometry. DOS doesn't seem to care if the flag is on or off, and the manual for the drive does say that it is a "wide SCSI" device, and its only I/O port is a 68-pin SCSI 3 port. *shrug* I don't know, I found a work-around. Give this to the hackers who wrote the driver for the adaptor. ============================================= Matt Rosenberg Washington University School of Law St. Louis, MO, USA mar7@cec.wustl.edu ; rosenbem@wulaw.wustl.edu http://www.cec.wustl.edu/~mar7/ =============================================
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9508171849.AA05085>