iB::Topic::September Extensions

	Damn Small Linux :: Damn Small Linux Board The DSL Forums

» Welcome Guest
[ Log In :: Register ]

Damn Small Linux Board » MyDSL Extensions » The Testing Area » September Extensions

		Mini-ITX Boards Sale, Fanless BareBones Mini-ITX, Bootable 1G DSL USBs, 533MHz Fanless PC <-- SALE $200 each!
		Get The Official Damn Small Linux Book. DSL Market , Great VPS hosting provided by Tektonic

Pages: (12) </ ... 4 5 6 7 8 [9] 10 11 12 >/

[ Track this topic :: Email this topic :: Print this topic ]

Topic: September Extensions

< Next Oldest | Next Newest >

curaga

Group: Members
Posts: 2163
Joined: Feb. 2007

Posted: Sep. 28 2007,12:33

The upx would be impractical: DSL is targeted to low machines, and upx must decompress the whole executable into memory!
Think 16mb without swap and OO.o.. crash or crash?

I don't think there's any need to remove docs, as the ones included are included for a reason (like my grub doc, or the XF86config examples)

In my opinion you should only remove the .po files, strip, and repack with advdef.

--------------
There's no such thing as life. Those mean little jocks invented it ;)
-
Windows is not a virus. A virus does something!

stupid_idiot Offline

Group: Members
Posts: 344
Joined: Oct. 2006

Posted: Sep. 28 2007,13:22

However, the upx website says this:
"Your executables suffer no memory overhead or other drawbacks because of in-place decompression."
- http://upx.sourceforge.net/#abstract

This is my (very unsophisticated) understanding of binary execution under Linux:
"Linux binary files are laid out in a way that pages on disk can be mapped into memory. Only the pages that are accessed ever hit memory." - http://mysqldump.azundris.com/archives/62-Notes-on-VM.html
Perhaps upx works this way also? That would justify the authors' claim of "no memory overhead".

If the above interpretation is correct, then the main determinant of upx performance would be cpu speed, when running the NRV decompression algorithm to load pages into memory. Judging by how the authors tout NRV as very fast, then a upx'ed binary probably has performance very close to a binary running from uci (since cloop also uses on-the-fly decompression).
If the following is to be believed -
"very fast decompression: ~10 MB/sec on an ancient Pentium 133, ~200 MB/sec on an Athlon XP 2000+." - http://upx.sourceforge.net/#overview
- then the speed difference between upx and cloop seems only academic.

curaga

Group: Members
Posts: 2163
Joined: Feb. 2007

Posted: Sep. 28 2007,14:53

Oh. The method for decompressing in memory is only used for scripts and a.out type apps. Sorry, didn't check..

There is still one major drawback with using upx: you cannot see the dependencies without a patched file.. and there is no patch for the busybox version

--------------
There's no such thing as life. Those mean little jocks invented it ;)
-
Windows is not a virus. A virus does something!

^thehatsrule^ Offline

Group: Members
Posts: 3275
Joined: July 2006

Posted: Sep. 28 2007,15:10

For 3: I believe using upx was decided against in general (i.e. for base), but for extensions I'm guessing it's up to the extension maker. How big is "bigger"?

EDIT: wow I did not see page 9 at all... I think the main reason was that performance was favoured in place of size.