Date: Fri, 5 Apr 1996 16:23:28 +1000 From: Bruce Evans <bde@zeta.org.au> To: hackers@FreeBSD.ORG, jagnew@vtaix.cc.vt.edu Cc: hardware@FreeBSD.ORG Subject: Re: EIDE controler Message-ID: <199604050623.QAA21988@godzilla.zeta.org.au>
next in thread | raw e-mail | index | archive | help
>I have the CMD640b chip as my eide controller! There is a bug in this and a few >other chips. You can read in detail about it at >http://www.yahoo.com/pci/faq.html >it explains about a bug with true multi tasking OS's. The chips that are >bad is RZ-1000, and the CMD640(x) chips, and most likely the SMC set! The above URL doesn't exist, but I read the Intel web page about the rz1000 bug. FreeBSD doesn't have the bug because FreeBSD masks IDE and floppy interrupts while transferring the IDE sector buffer. In fact, it masks all disk interrupts (those in bio_imask). This is partly from good programming practice (don't allow yourself to be interrupted) and partly from being too stupid to handle disk interrupts independently. The bug has very little to do with multi-tasking. It has to do with reentrant interrupt handlers. I guess that very few IDE interrupt handlers allow themself to be interrupted, but many systems allow floppy interrupts while IDE interrupts are being handled. These systems get bitten by a another wart in the IDE design (if they start IDE and floppy i/o's concurrently): the digital input registers for the first IDE and floppy controllers have the same address (0x3F7), and reading this address in the floppy interrupt handler sometimes clobber IDE input. Anyway, FreeBSD is too stupid to use this register. It should be used by the floppy driver to determine if the media has changed. Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199604050623.QAA21988>