OpenBSD Stories: Strange Medieval Devices
This piece delves into the intricate world of retrocomputing, specifically chronicling the author's journey to keep OpenBSD running on a vintage Sun-4/260 workstation equipped with ancient Storage Module Device (SMD) hard drives. It highlights the unique challenges of dealing with decades-old hardware, from mysterious boot errors to obscure driver bugs and failing physical components. The popularity on HN stems from the deep technical dive, the nostalgia for classic systems, and the perseverance required to maintain them.
The Lowdown
The article "Strange Medieval Devices" by Miod Vallat offers a captivating look into the challenges and triumphs of working with computing hardware from the late 1980s and early 1990s. Specifically, Vallat details his extensive efforts to get and keep OpenBSD running on a Sun-4/260 system, which utilizes archaic Storage Module Device (SMD) hard drives. His narrative is a blend of historical context, technical problem-solving, and a deep appreciation for vintage technology.
- SMD Technology Explained: The author introduces SMD disks as the high-capacity, high-performance storage solution of their era (late 1970s-1980s), noting their massive physical size, significant power draw, and unique dual-cable interface (separate for command and data). Capacities topped out around 1.25GB, dwarfed by later SCSI drives.
- Acquiring the Sun-4/260: Vallat recounts a year-long quest to obtain a decommissioned Sun-4/260, a SPARC-based system, complete with its original SMD disk cabinet, underscoring the dedication required for such retrocomputing endeavors.
- Initial Hurdles: Early setup involved troubleshooting defective memory boards and a non-functional SMD disk, eventually getting the system operational with 48MB of RAM and one 280MB SMD drive.
- The Boot Block Bug: A crucial problem arose during a reinstallation, preventing the system from booting. This was meticulously traced to a conflict between OpenBSD's updated 16KB filesystem block size and an undocumented, 8216-byte I/O limit in the Sun PROM monitor. The author devised a fix to perform I/O in smaller, compliant chunks.
- Unearthing Legacy Bugs: The process of resolving the boot issue led to the discovery and rectification of 19-year-old bugs within OpenBSD's Xylogics SMD controller driver, particularly concerning bad sector handling and disk geometry calculations.
- Disk Deterioration and Disassembly: Despite fixes and low-level formatting, one SMD disk began to fail rapidly due to bad sectors. Vallat documents its eventual retirement and provides detailed internal photographs, showcasing the physical mechanics of these "medieval devices."
- Future Endeavors with Ciprico Rimfire: Looking ahead, the author plans to port the SunOS driver for the more advanced Ciprico Rimfire SMD controller to OpenBSD, aiming for better performance and utilizing its Xylogics compatibility mode.
- SMD Controller Compatibility Nuances: A technical sidebar explains that SMD disks often require reformatting when moved between different controllers due to controller-specific format details and the absence of a standard method for querying disk geometry.
Vallat's account serves as a testament to the passion of retrocomputing enthusiasts, who go to extraordinary lengths to breathe life into historical hardware, revealing the intricate technical details and surprising challenges hidden within seemingly obsolete systems.