Thanks for the dump. On Wed, 15 Feb 2006, Fredrik Roubert wrote: > This is the output from dmesg, from boot until second failed scan: ... > [ 344.852017] dc395x: reselect: w/o disconnected cmds <02-0> > [ 344.852038] dc395x: disconnect: Unexpected reselection <02-0> > [ 464.630090] dc395x: eh_abort: (pid#2002) target=<02-0> cmd=ffff8100743c04c0 > [ 464.630096] dc395x: eh_abort: Command in progress<6>dc395x: eh_bus_reset: (pid#2002) target=<02-0> cmd=ffff8100743c04c0 > [ 464.630623] dc395x: doing_srb_done: pids G:2002(02-0) > [ 528.066303] dc395x: reselect: w/o disconnected cmds <02-0> > [ 528.066324] dc395x: disconnect: Unexpected reselection <02-0> > [ 647.842343] dc395x: eh_abort: (pid#2130) target=<02-0> cmd=ffff81007cc1b5c0 > [ 647.842350] dc395x: eh_abort: Command in progress<6>dc395x: eh_abort: (pid#2131) target=<02-0> cmd=ffff81007cc1b780 > [ 647.842356] dc395x: eh_abort: Command was waiting > [ 657.823181] dc395x: eh_abort: (pid#2131) target=<02-0> cmd=ffff81007cc1b780 > [ 657.823188] dc395x: eh_abort: Command was waiting > [ 657.823196] dc395x: eh_bus_reset: (pid#2130) target=<02-0> cmd=ffff81007cc1b5c0 > [ 657.823717] dc395x: doing_srb_done: pids G:2130(02-0) Ok, I will assume that this dump includes the KERN_DEBUG messages too now. Still, could you, please, just verify that you don't have anything else from dc395x in your /var/log/kern.log (if you have one)? If there's nothing new there you don't have to send it. This means, PIO is not entered now any more, although the same "reselect without disconnect" events happen. The only reason for this I can think about is my addition of dma_mask to drivers/scsi/dc395x.c: @@ -4834,6 +4904,9 @@ irq = dev->irq; dprintkdbg(DBG_0, "IO_PORT=0x%04lx, IRQ=0x%x\n", io_port_base, dev->irq); + if ((ret = pci_set_dma_mask(dev, DMA_32BIT_MASK))) + goto fail; + /* allocate scsi host information (includes out adapter) */ scsi_host = scsi_host_alloc(&dc395x_driver_template, sizeof(struct AdapterCtlBlk)); Could you, please, remove / comment out these 2 lines and try again? If then PIO comes back, I would first try to verify / debug / fix it, and then, as it sems not to be the main problem, we'll study the 2nd problem. Thanks Guennadi --- Guennadi Liakhovetski - : send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
