kern/93761: iir driver is broken for >=4GB system RAM

V. T. Mueller freebsd-amd64 at datafarm.de
Thu Feb 23 10:20:12 PST 2006


>Number:         93761
>Category:       kern
>Synopsis:       iir driver is broken for >=4GB system RAM
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Thu Feb 23 18:20:06 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator:     V. T. Mueller
>Release:        i386-5.4, i386-6.0, amd64-5.4, amd64-6.0, amd64-6.1b2
>Organization:
Continum AG
>Environment:
any of the above install and/or PAE kernels            
>Description:
The iir driver appears to be not 64bit clean and/or has serious problems with ressource usage when there are 4GB+ in the system.

This applies to any kernel found on amd64 iso images from 5.4 to 6.1b2 and has been thoroughly tested on two supermicro H8DA8 with two opteron 245 and 8x1GB of RAM. The controllers tested were ICP-vortex GDT8500RZ and GDT8114RZ which both produced identical results. All hardware has latest firmware, was extensively tested and can run Debian and Windows flawlessly.

Using i386, the problem is easily reproducible with a PAE kernel. Without PAE, it takes a random amount of time for the system to crash. A PAE-kernel system just dies at the end of bootup.

I can't tell if other iir-supported controllers are affected, but it looks as if this is a generic driver problem.

The current situation in plain words: using FreeBSD is impossible with hardware like described above. Worse, people who have a running system now will render their system unusable when upgrading memory over 4G.
>How-To-Repeat:
Using one of the controllers mentioned above with either amd64 or i386/PAE when there are more than 4GB of RAM available.
>Fix:
No way.
>Release-Note:
>Audit-Trail:
>Unformatted:


More information about the freebsd-bugs mailing list