Toshiba Satellite M45-S359 THE MACHINE THAT REFUSED TO BE FORGOTTEN Old does not mean forgotten. The Toshiba Satellite M45-S359 was released in 2005. Years later, it reached our laboratory in a rather unusual condition. It was described as a machine that had been written off, except that technically it had never even been written off because it had disappeared from the paperwork long before anybody remembered to do so. It came from a university research laboratory. By the time we first looked at it, the machine had already spent years outside normal service. Around eight years before its current restoration, an initial inspection had produced a simple conclusion: this was not a dead laptop. It was something considerably more complicated. It was a survivor. The first thing that became obvious was that the hardware itself was still fundamentally sound. For its generation, the M45 was a serious working machine. The Pentium M platform was already old by modern standards, but it had one particularly stubborn quality: it did not need the newest operating system to demonstrate that it could still work. The real problem was not performance. It was booting. The original BIOS simply did not provide USB boot support. In 2005 this was not considered a problem. The machine had been designed around the storage and boot media of its time. Press F12 and the machine offered roughly the following universe: 1. CD/DVD 2. floppy disk 3. network boot 4. internal hard disk There was one small historical joke hidden inside this menu. The floppy drive appeared as a boot option even though this particular machine did not have one. Twenty years later, the CD/DVD option was not exactly convenient either. Finding a suitable blank disc is one problem. Finding one that can still be written is another. Finding another working machine capable of writing a 32-bit service disc is yet another. The same problem applies to floppy disks, except that after finding the disk you still have to find a functioning floppy drive and convince the modern world to cooperate with it. Network boot was technically possible, but it was not what we wanted. So there was one option left. The hard disk. The solution began with a piece of storage that had already survived several other experiments. We had a collection of small SSDs and adapters that had gradually become part of the laboratory ecosystem. One 16 GB M.2 SATA SSD could be connected through a SATA adapter. Another could be connected through an IDE 44-pin adapter. Which one would fit depended on the particular revision of the motherboard. Our M45 happened to have the SATA-compatible arrangement. The little SSD went in. The machine now had modern solid-state storage attached to a platform designed almost two decades earlier. But storage was not yet the interesting part. We first needed to make the machine boot something. The first serious attempt brought us back to Windows. At that time the main workstation still had a Windows installation sitting quietly beside 64-bit Ubuntu in a dual-boot configuration. Windows had already become something we rarely entered. It had mainly survived because, once upon a time, it had been useful for creating the first Linux installation media. Now it had another job. Rufus was used to prepare the 16 GB drive with Hiren's BootCD 15.2, the 32-bit service environment appropriate for this generation of hardware. The result was almost absurdly effective. The Toshiba booted from the internal drive. Hiren's environment appeared. We were inside a portable Windows XP environment with an entire collection of service and recovery tools available without installing a full operating system. Linux SystemRescue was available as another environment, and local SSH worked as expected. The old laptop had suddenly acquired something that the original engineers in 2005 certainly did not design into it: a portable service workstation that could recover other computers, inspect filesystems, work over the network and carry a substantial collection of tools without permanently installing them. The first problem had been solved. The second problem was more interesting. How do you make a 2005 laptop boot from USB when its BIOS simply refuses to do so? We tried the obvious paths first. Linux itself was not particularly difficult to install. External storage could be partitioned. Bootloaders could be placed into partitions or the MBR. antiX could be installed. The machine could run a small Linux environment. But the purpose of the Toshiba was slowly becoming clearer. We did not want another machine that merely happened to run Linux. We wanted a console terminal that could survive almost anything. And at this point the history of the machine began to matter. It had spent years sitting in a university laboratory without its memory modules and without its hard disk. What exactly had happened to it during that period is known only to the gods of old hardware, chemical laboratories and unattended storage rooms. Before rebuilding it properly, we ordered a basic hardware stack. Two 1 GB memory modules arrived. One eventually found its way into the black Acer Aspire One ZG5. The remaining hardware went into the Toshiba. The compatibility was unexpectedly useful. The same general memory family used in the Acer could also work in the Toshiba, and the M45 accepted 1.5 GB of RAM with no need for heroic expansion. Memory testing was not entirely peaceful. Different combinations of modules, slots and testing environments produced different results. Some configurations passed without complaint. Others reported suspicious addresses. The machine could still boot and operate with the working combination. That was becoming a recurring rule of the laboratory. Old hardware does not always fail in a clean and polite way. Sometimes it gives you three different answers and expects you to discover which one is useful. The Pentium M helped considerably. The processor could reduce its operating frequency when the workload was light, and the machine remained perfectly comfortable while doing so. Once the power supply was found, the Toshiba came alive without smoke, drama or any of the other traditional signs of archaeological failure. The BIOS opened. The hardware responded. The battery arrived a few days later. It was not a twenty-year-old battery that had spent its entire life in storage. It was a replacement unit, properly charged and capable of holding the machine alive for more than three hours under the service tools we used. The resulting configuration was modest: Pentium M 760 1.5 GB RAM 16 GB M.2 SATA SSD through an adapter replacement battery replacement power adapter It was enough. More than enough, in fact. But the USB problem remained. Then Hiren's BootCD gave us the missing piece. Plop Boot Manager. The program is tiny. Ridiculously tiny. A piece of software weighing only a few dozen kilobytes turned out to be the bridge between a BIOS designed for 2005 and the removable storage of the modern world. Plop does not magically teach the BIOS to boot USB. It inserts itself between the BIOS and the next stage of the boot process. That distinction matters. The Toshiba boots from its internal disk because the BIOS understands that. Plop starts. Plop then examines the USB controllers and provides its own method of accessing the USB device. On this particular machine we discovered another peculiar detail. USB detection was much more reliable if Plop was started once with the ports empty. After the controllers had been initialized, a USB flash drive could be inserted, Plop could be invoked again, and the machine could then see the removable device. The first attempt could simply produce a message saying that no USB device was present. That was not failure. It was information. We had to learn the machine's timing. We had to learn Plop's behavior. We had to read the documentation. For something that occupies only a few dozen kilobytes, Plop generated several days of investigation. Eventually it became obvious that speed was not the important part. The important part was having a layer that understood something the original BIOS did not. The Toshiba could now reach USB storage. The old BIOS had not been replaced. It had been given an interpreter. That changed the entire machine. We then built a small internal storage architecture around the 16 GB SSD. The first partition contained Windows XP. The second contained antiX 23 Base. The third was a small shared partition for moving files between the two systems. The final arrangement was approximately: sda1 Windows XP ~6 GB sda2 antiX 23 Base ~5 GB sda3 shared storage ~4 GB Each operating system had its own space. Windows remained Windows. antiX remained antiX. The shared partition provided a small bridge between them. Plop was finally installed into the MBR, turning the entire boot process into a small menu-driven system of its own. BIOS first. Plop second. Then the operating system of our choice. Or, if necessary, USB. The machine had finally learned a trick its original firmware had never been taught. The BIOS was also updated to the final available Toshiba release. The firmware dated from 2008, considerably newer than the original machine, but it did not suddenly acquire USB boot support. That feature simply was not part of the machine's design. So we stopped trying to make the BIOS become something it was never intended to be. We worked around it. That became the more interesting solution anyway. Hiren's BootCD became more than an emergency disc during this process. It turned into a portable service environment that could live on the machine without permanently occupying its operating systems. One particularly useful discovery came later. If the Hiren's ISO image was unpacked from antiX into the Windows environment, many of the service utilities could be launched directly without installing them into the system. After a reboot, they disappeared again. No permanent software pollution. No enormous collection of utilities consuming the little space available. Around 700 MB of properly organized service tools turned out to be a remarkably good trade for a machine with only 16 GB of storage. Windows XP eventually became the practical choice for the hardware- specific side of the machine. Toshiba's own recovery media was important here. Finding the correct recovery package was not as simple as searching for "Toshiba M45 recovery". There were multiple series, regional variants and hardware revisions. The correct combination depended on the model, release, country, Windows version and motherboard configuration. Driver Pack Solution 17 also proved useful as a final-era Windows XP resource. The rule was simple. If the hardware was identified correctly, the system usually knew where to look. Toshiba's multimedia keyboard was another reminder that hardware support is not always universal. Linux did not fully understand all of the special multimedia controls provided by the machine's original design. Windows XP, with the proper Toshiba utilities and drivers, handled them much better. Brightness control was similarly strange. Windows could restore the intended Toshiba behavior after the manufacturer utilities were installed. antiX 23 Base could not properly discover the hardware brightness controller and simply preferred to operate at the maximum brightness level. It was not elegant. It was usable. Sometimes that is the correct definition of success on twenty-year-old hardware. After replacing the thermal paste, the Pentium M settled into a remarkably comfortable thermal range. Ordinary operation generally kept the processor around 40 to 47 degrees Celsius. Idle CPU usage could remain somewhere around 2 to 13 percent depending on what the machine was doing. Streaming radio. Reading documentation. Browsing forums through text interfaces. Monitoring services. Running small utilities. The machine simply worked. This was one of the most important discoveries. The Toshiba did not need to be fast. It needed to be given tasks that matched its nature. A modern graphical desktop could turn the machine into a struggle. A terminal could turn it into a workstation. Text browsers, SSH, small scripts, network utilities and service tools consumed remarkably little of the available resources. A few hundred megabytes of RAM was enough for a surprising amount of work. The 16 GB SSD was not a compromise anymore. It was the correct scale. Windows XP had its own small world. antiX had its own. The shared partition connected them. Hiren's could appear when needed. Plop could reach USB storage when the BIOS could not. The old Toshiba had become a small boot laboratory. It was also becoming something much simpler. A terminal. A very large terminal. The M45 has a large display for its generation, two Harman/Kardon speakers and a full multimedia keyboard. It is an entirely different physical object from the tiny Acer netbooks that came later. It feels like a machine. A heavy, slightly ridiculous, wonderfully overbuilt machine. A cyberpunk carriage on wheels, going somewhere without ever explaining exactly where. Most of the time, however, it is simply sitting there doing useful work. Documents. Forums. Manuals. Text-based Internet services. And radio. The little machine can run PMRP-ng and stream from hundreds of radio servers while continuing to behave like it has nothing particularly important to do. The fan quietly moves air. The Pentium M quietly changes frequency. The SSD remains silent. The network remains connected. The terminal waits. That is the strange part. Twenty-one years after its manufacture, the machine is not being kept alive by nostalgia. It is being used. The Toshiba did not become a museum piece. It became a wide-screen console window into the modern network. It also became proof that the 16 GB storage experiment was not a one-machine trick. The same basic idea had already worked in other old laptops. A tiny industrial-grade SSD removed from a smart-home system can become the system disk of a twenty-year-old notebook. A Chinese adapter can bridge generations of storage interfaces. A thirty-six-kilobyte boot manager can solve a problem that the original firmware never anticipated. A collection of old service utilities can still be more useful than a modern rescue environment if the hardware is old enough. A 2005 Pentium M can still spend an evening connected to the Internet. The important part is not that the machine survived. The important part is that we stopped asking whether it deserved to survive. We simply gave it something useful to do. There are no grand plans for the M45 at the moment. It does not need them. It has a working dual-boot environment, USB access through Plop, a service environment, a functioning battery, solid-state storage, a stable thermal profile and enough memory to handle the work we actually give it. It has already survived one laboratory, years of neglect, a missing disk, missing memory, obsolete firmware, incompatible boot expectations, several operating systems and the strange economics of twenty-first-century replacement parts. Now it mostly sits on the desk and waits for the next command. Sometimes it plays radio. Sometimes it watches the network. Sometimes it opens a manual. Sometimes it becomes a service machine. Sometimes it simply sits there with a terminal on the screen, consuming almost nothing. And that is perhaps the best possible ending for a computer from It was not rescued so that somebody could say: "Look, an old Toshiba still boots." It was rescued because it could still answer. Old does not mean forgotten. It means that the machine has already had twenty-one years to learn how to survive. And apparently, it is still learning.