Hi Les and Mikkel: I would like to object to your objection to my description of what occurs with GRUB. On Fri, 2007-10-19 at 13:07 -0500, Les Mikesell wrote: > William Case wrote: > > >> They can't mesh other than by convention. When updating the config, > >> kernel, initrd, etc., files you are working with the OS (doesn't have to > >> be linux) view of the filesystem and it's relative locations. When grub > >> boots, it only has the bios view of things and only sees one partition > >> through the bios conventions. > >> > > > > Grub's job is to find partitions and file systems; not to use them. > > Grub doesn't 'find' partitions, you tell it where the one it will use is > from the bios perspective with the root (hdx,x) directive before > installing. > Whether GRUB finds, goes to, or loads stage 1.5 or stage 2 are moot and/or pedantic points. The OP maintained and continues to maintain "There is nothing simple about what Grub does. It is the tiny software that let us boot our Linux or Windows operating systems". I was responding by saying, in plain language, that it is in fact quite simple. Grub in its own way is just a specialized very small Operating System that has one job -- to load other full Operating Systems. Your responses unnecessarily re-complicate an issue that can conceptually be easily understood. Once the idea of a separate small independent system is absorbed by a reader, then the peculiarities of the GRUB operating system (I know it is never called that) can be discussed. Similarly, the fact that the viewable source configuration file may reside within a Linux partition (eg /boot) does not limit GRUB's independence from all OS's that exist on a given hard disk. As a matter of necessary convenience it is made visible within a full operating system so that it can be amended or changed before the next boot, but nonetheless remains unconnected to that (Linux or other) Operating System. It is easy to confuse the Grub operating system with the full system (Linux or others) Therefore, manuals that discuss and explain GRUB refer to (or at least mean) only the Grub operating system. Confusion abounds when directions, names and commands for the Grub system get confused with directions, names and commands of the full operating system (whether Linux or other). > > Grub is so powerful exactly because it can find so many different kinds > > of partitions/filesystems and load them and their kernels with their > > various forms of init into memory. > > That's true, and it makes it possible to change the configuration, > although not the partition to be used, without reinstalling grub itself > as you would with lilo. However, it still doesn't know/care what the OS > it is booting will call those partitions or if they are subsequently > mounted at all. > If one thinks of Grub as an independent and separate Operating System, there is no reason to assume it does care what some other system calls partitions or if they are subsequently mounted at all. In fact, it can't. The name and location of partitions and files are stored within the as yet unloaded Linux or other kernel. > > To do that Grub needs its own generic file system (rudimentary though it > > is) and it's own small kernel that allows it to operate on *any* > > machine. The manuals (although I agree they could have some basic (less > > technical) explanations included) deal only with the GRUB file system > > and kernel making no assumptions about the partitions/files/kernel(s) > > that will eventually be loaded. > > And the missing piece is that you - or the distro-packaged scripts - > update those locations through the OS concepts of where that partition > is mounted. This part is OS/distro specific and doesn't have a lot to > do with grub. That's the point. Why confuse the issue? I have taken the time to respond here, because by your responses you have touched on a point that has become important to me. I have spent three years digging deeper and deeper into the functioning of computers and Operating Systems, particularly the RedHat/Fedora version of Linux. Always after hours and hours of digging through technical and often pedantic descriptions, I find that a two or three sentence overview in plain understandable language would have sufficed to let someone new to a concept have a eureka moment. Then all the rest easily slides into place. -- Regards Bill
