|
|
 |
 |
 |
 |
 |
 |
|
|
|
From: Jay Miller (jnmiller, colorado dot edu) Date: 2002.01.06 - 20.25 MST
X-Mailer: Mutt 1.2.5i - Words by Perry and Lorae Merritt <plmerritt, hypermall dot net> [020106 17:08]: > I stand corrected. Jay is right, there is no fioctl, I was thinking of > fcntl. So now the question is do we need to use fcntl or can we use ioctl? Um, I looked into this a while back and I decided that fcntl() wouldn't help. It doesn't talk to drivers, per se, it just performs actions on the file descriptors themselves.. > I was thinking that a CLI is different from a script in that only one > command is called. Is that wrong? Um, if we're being picky, yes I'd say that's wrong.. (of course) Beyond that, though, my point was not that CLI's and scripts weren't different. My point was that, in communicating with the driver (or portal, or whatever), a script isn't functionally different from a CLI.. because a script would just *use* a CLI for communicating. So from the system's perspective, scripts and a CLI are the same. > In your comment regarding (10) aren't we calling the functions pointed to in > the file operations and inode operations tables? Yes - but you said we're 'calling pointers', not 'calling functions pointed to by...'. See what I mean? Or am I nit-picking? > There aren't multiple inodes per file??? What happens when a single inode > can't describe the whole file like in the case of a very fragmented file? No, there aren't multiple inodes per file. If a single inode can't describe an entire file, then you have reached the file system's per file size limit. There are multiple *data blocks* per file, of course, and they may be as fragmented as you like. The inode may also extend *into* data blocks in the cases where indirection is needed. But the file-inode relationship is still one to one. Do an 'ls -i' - the number to the left of the file is its inode #! > Ill fix the drawing numbers after I get all of the feedback. I don't want to > renumber everything more than once. Oh yeah, I forgot my comments about the drawings, too. Just a couple more typos. - In (10), I am concerned (though not overly) that 'filter driver' might be somehow trademarked, particularly by Microsoft.. anyone know anything about this? - In (17), I think it should be 'inode_operation*s*' if you're looking for the actual Linux structure name. - In (18), for consistency, should be use 'file_operations', the actual name of the structure in Linux? (17) and (18) should be the same, whatever we choose. - Should your decision boxes (21) and (23) have question marks? - I think (24)'s 'return' should be capitalized like the other boxes. - Ditto for (34). That's it for now! -- Jay Miller -- This is the antera mailing list. To unsubscribe, email majordomo, cryptofreak dot org with message body `unsubscribe antera'. Or, for more information, visit http://www.cryptofreak.org/.
This archive was generated by hypermail 2b30 : 2002.01.07 - 04.02 MST |
