Windows 98 Professional Reference

Previous chapterNext chapterContents

- 36 -
The Windows 98 Boot Process and Emergency Recovery

Sometimes you find yourself in a jam--a real jam, where your system won't start. What's the problem? What to do? How to handle it? That's part of the subject of this chapter: Emergency Recovery. Chapter 37, "Tools and Strategies for Troubleshooting Windows 98" covers dealing with individual hardware devices, software that isn't working correctly, or other aberrant behavior on your system. But Chapter 37 assumes that your system is starting (more or less). Not here, though: This chapter will get you on the road to recovery when you can't even get the darn thing booted. And because emergency recovery is closely related to how Windows 98 boots, you'll also learn about the Windows 98 boot process in this chapter.

In this chapter you learn about the following:

Startup problems can be among the most frustrating because you have so little information with which to work. There are no friendly little trouble icons in Device Manager to show you where the trouble is, usually. As you will see, however, solving startup problems is more a matter of working through the different possibilities, one by one.

Understanding the Windows 98 Boot Process

Windows 98 goes through a series of steps as it boots the computer. If everything's working well, you won't care about the details about how Windows 98 gets the computer from an off state to when it displays the opening desktop. However, when troubleshooting boot problems, it's helpful to understand how booting works.

When a computer is powered up, the first thing that happens is that a program built into the computer's Basic Input/Output System (BIOS) runs and performs a Power-On Self Test (POST). The POST tests all the key system components to make sure that they're all there and that they're minimally functioning. Also, the RAM in the computer is briefly tested during the POST.

After the POST completes, the BIOS automatically loads a small program from the primary hard disk on the system (or from a disk drive, or in rare cases a CD-ROM). This small program is called a bootstrap because it helps the computer "pull itself up by its bootstraps." The bootstrap loads the operating system's own bootstrap program from special startup sectors on the hard disk and transfers control to the operating system. The operating system programs then proceed to initialize the hard disk's file system, and control is transferred to other programs, on down the line until the system is completely booted.

The first phase of Windows 98's startup process involves scanning the installed hardware in the system and selecting the appropriate Hardware Profile from the Registry. After the Hardware Profile is activated, Windows 98 processes any commands found first in the CONFIG.SYS file, and then the AUTOEXEC.BAT file, both located in the root directory of the primary boot device. that although CONFIG.SYS and AUTOEXEC.BAT are not required for Windows 98, they are still processed in case they are needed for compatibility with older software or hardware on the system. Any drivers loaded by CONFIG.SYS and AUTOEXEC.BAT are real-mode drivers; the operating system has not yet switched the system into protected mode.

Windows 98 then proceeds to load WIN.COM and any VxD device drivers needed for the system. There are both static VxDs that are loaded during each boot and dynamically loaded VxDs that may or may not be loaded, depending on the selected Hardware Profile. The individual VxD device drivers are all contained in a file called VMM32.VXD.

The VxDs that are part of VMM32.VXD are loaded based on entries found in the Registry at HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\VxD. If there are conflicting entries in the SYSTEM.INI file, the entries in SYSTEM.INI will take precedence over the Registry entries.

After the VxDs are loaded, Windows 98 switches the computer into protected mode to proceed. The first thing that happens during this phase is that the Configuration Manager is loaded and started, which then queries all Plug and Play devices in the system; if the system's BIOS doesn't support Plug and Play, then the various buses in the system are scanned for installed devices. Based on the information gleaned from the Plug and Play queries or the bus scanning, dynamic protected-mode drivers are loaded for the installed devices. After the device drivers are loaded, any resource conflicts between the devices (such as IRQ assignments) are resolved by the Configuration Manager.

Next, a variety of files are loaded: KERNEL32.DLL and KRNL386.EXE (contain the basic components of Windows and the Windows device drivers, respectively), GDI.EXE and GDI32.EXE (handle the graphical interface duties for the operating system), and USER.EXE and USER32.EXE (provide the user-interface programming for Windows 98). Any related resources, such as icons, images, and fonts, are then loaded. Next, WIN.INI is processed and the desktop software is loaded and started. When User Profiles are active on the system, a logon prompt for the system then displays. After the user logs on, any settings specified for the User Profile are processed. Finally, any programs listed in the Startup folder are loaded for the user while the desktop initializes.

Windows 98 Startup Files

There are seven key files that form the base of Windows 98's startup behavior. They are all (except for WIN.INI and SYSTEM.INI) located in the root directory of the primary boot device, usually C:.

Understanding Startup Problems

If a Windows 98 computer won't boot, the trouble is most likely due to one of the following reasons:

You'll take different paths for resolving each of these errors. The following sections discuss each type of problem in more detail.

WARNING: This author can't count the number of times he's worked through a failure with someone, only to determine that the failure required a restoration of data from a backup, and then discovering from a panicked user that there are no backups--or only very old backups--available. The very best emergency recovery tool anyone can ever have is a current, reliable backup of the system. Computers can and do fail, and sometimes spectacularly. A good backup system in place will prevent you from losing thousands of hours of accumulated work that is irreplaceable. If you're in an environment that relies on users to perform their own backups, you owe it to them to hammer this point home as forcefully as you can. Even better, find a way to take backups out of their hands and set up a reliable system to take care of this necessary chore.

Resolving Hard Disk Problems

Hard disk drives are much more reliable than in years past. It used to be that they had a Mean Time Between Failure (MTBF) of about five to seven years. Now, most drives quote MTBFs of 20 years or more. Still, even with this increased reliability (or more optimistic marketing--we're not sure), hard drives can still fail. Sometimes they fail slowly, accumulating bad sectors at a gradual rate, which then accelerate over time, and sometimes fail suddenly, going dead for no apparent reason. The first order of business is to ascertain whether the hard disk has failed. Afterward, if you determine that the primary boot drive has indeed failed, you can take these steps to replace it and rebuild your system:

1. Leave the system powered on for an extended period (at least an hour) and then try to restart the system with a warm boot. If you can do this in a warm environment, it may help.

2. If the drive still isn't starting after going through this warm-up phase, try a cool-down phase and leave the system powered off overnight, and try it again first thing in the morning.

3. This is really a desperate move, but the author has seen it work before: Warm up the system for an hour by leaving it on, and then give the hard drive a couple of sturdy whacks with the heel of your hand, then do a warm boot. (Make sure you've tried everything else first, because this can damage the drive!)

If all the preceding steps fail, you've most likely got a failed primary drive on your hands. You'll need to replace it, reinstall Windows 98, reinstall all of your applications, and then restore your data from backup.

TIP: Most hard disk drives have at least a one-year warranty. If the drive has failed, you might be able to get a free replacement from the drive's manufacturer or your computer's manufacturer.

Assuming you are getting normal POST behavior from the drive (the hard disk lights flash normally during startup and normal sounds emanate from the drive), then you start to get into other configuration problem possibilities. Check the following:

1. Start the system with the EBD and see if you can do a DIR of your C: drive. If so, perform a SCANDISK on the C: drive, and then use the SYS command that is part of the EBD to reinstall the system files on the drive. Continue with the "Resolving System File Problems" section.
2. Start the system with the EBD and use FDISK to see if the disk partition data has been lost for some reason. You may be able to repartition the disk and have it start functioning again, with your data mostly intact. However, this requires that the partition be created exactly as it was when the system was last working, and this is a chancy proposition at best. Chances are you'll have to repartition the drive and reinstall Windows 98 and your applications.

Resolving Other Hardware Problems

Sometimes a system won't boot and the problem isn't due to the hard disk drive. It might be due to another device in the system that's failing in some way and preventing a startup, or it may be a number of different possible file problems. This section deals with the former possibility, while the next two deal with the latter.

If the system starts to boot, displays the message Loading Windows 98 and possibly the Windows 98 splash screen, but then stops booting, try these steps to find out if there's other hardware in the system preventing startup: