Tiny Core Base > TCB Talk
Sugestion: XZ compression in kernel + tcz; squashfs with xz + biger block size
Rich:
Hi andyj
--- Quote ---so unless you are using copy2fs (because you have lots of RAM) then the compressed file (the squashfs archive) will be the only file in RAM. Thus, every access to any file in an extension will require decompression.
--- End quote ---
Maybe I'm missing something, but my understanding is if you don't use copy2fs then extensions get loop mounted. The mount
point is in the ramfs (/tmp/tcloop/) but the extension resides on the hard drive. The files in the extensions get linked into the ramfs
(/usr/local/bin for example) but the extensions files still reside on the hard drive. When you go to execute the program it follows
the link to read the file into RAM for execution. When you exit the program, it remains in RAM as long as that RAM doesn't need to
be freed up for something else.
curaga:
That's right, linux has a separate cache for (recently) opened files and executables. The uncompressed squashfs data stays in that until the memory is needed for other things.
andyj:
On a diskless system (PXE booted), the extensions are in RAM. But this is becoming a circular discussion. In most distros the files are not compressed because the assumption is that they are not resource constrained, thus the cost of decompression is seen as a liability because saving space isn't an asset. Compression is what helps resource constrained systems to function, so the space saved is an asset if it offsets the cost of decompressing on a low power CPU.
jazzbiker:
HI,all!
Maybe this new compression algorythm will be usefull for zswap?
hiro:
likely. if lz4 isn't even better...
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version