Q-bus to CF [was: IOmega]

Huh? The only sense I can think of in which the PDP-1 and VAX could be said to do read before write is in the bus transaction sense, and anything that could be called "IDE" was insulated from bus transactions. While the card didn't have much smarts, it did have that much; it looked enough like a wdc that I did the driver as an "attach wdc at uba with wdc_uba" attachment layer, rather than a completely new driver. I find I still have my copy of the dev directory, including a boot-time console log: uba0 at ibus0: Q22 ubasearch: cf_loc[0] 0171000, pa 0x20001200 [wdc_uba_probe: ua->ua_ioh 020500251000] st0=52 st1=50 wdc_uba_probe: setting vector 774 wdc0 at uba0 csr 171000 vec 774 ipl 17 st0=52 st1=50 wd0 at wdc0 channel 0 drive 0: <Samsung Electronics 120MB SHD-3062A> wd0: drive supports 16-sector pio transfers, chs addressing wd0: 114 MB, 917 cyl, 15 head, 17 sec, 512 bytes/sect x 233835 sectors wd1 at wdc0 channel 0 drive 1: <WDC AC31200F> wd1: drive supports 16-sector pio transfers, lba addressing wd1: 1222 MB, 2484 cyl, 16 head, 63 sec, 512 bytes/sect x 2503872 sectors wd1: drive supports PIO mode 3, DMA mode 1 wdc0: establishing interrupt, vector 1774 I'm not sure how it worked enough to query the disk like that but failed later. That serial-console log file seems to indicate that it worked fine for 512-byte transfers but attempts to do larger transfers led to an unending stream of "wdc_uba_intr: stray", meaning it was seeing interrupts that the wdc interrupt routine was declining to handle. /~\ The ASCII der Mouse \ / Ribbon Campaign X Against HTML mouse at rodents.montreal.qc.ca / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B