As an example of where this "logging" could come in handy:
My "netscape" starts up taking about 12 seconds before I see the screen.
This is about 10 seconds of disk IO and 2 seconds of CPU.
I now have a script that does:
dd if=netscape.bin of=/dev/null bs=1024k
netscape.bin $*
This linearly reads netscape into memory, and costs a little over a
second. Starting up netscape now only takes 5 seconds total... Thus
the "demand paging" is not efficient in this case: the seeks cost much
more than is gained by not reading part of the binary.
The ideal situation would be that those parts of the binary that get
loaded at startup would be linearly on the disk, those parts that
don't get used at startup, just behind that:
now: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
referenced at startup 1 - - 4 - 2 - 5 - - - 3 6 - -
fancy defrag: 1 6 12 4 8 13 2 3 5 7 9 10 11 14 15
referenced at startup 1 2 3 4 5 6 - - - - - - - - -
The "block accessed logger" would come in very handy to show you which
blocks do get referenced while starting up a program.
By the way, has anybody tried it yet?
Roger.
-- If it's there and you can see it, it's REAL |___R.E.Wolff@BitWizard.nl | If it's there and you can't see it, it's TRANSPARENT | Tel: +31-15-2137555 | If it's not there and you can see it, it's VIRTUAL |__FAX:_+31-15-2138217 | If it's not there and you can't see it, it's GONE! -- Roy Wilks, 1983 |_____|- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu