[PATCH] mke2fs size reduction

[Rob Landley]

On Thursday 22 September 2005 15:23, Bernhard Fischer wrote:
> here (in test_disk() ) is an alleged candidate for
> {,UN}RESERVE_CONFIG_BUFFER(buf,1024)
> [snip]

I'm a little ambivalent about extending use of RESERVE_CONFIG_BUFFER before we 
nail down a couple issues.

What platforms is using a lot of stack space on bad?  (And how much is bad?)  
Busybox has developed a number of dependencies on the Linux kernel, and I 
don't know of any Linux systems where the stack doesn't grow dynamically.  
(What, mmu-less variants?)  If we have 64k to play with then allocating 1k on 
the stack isn't too alarming, and if we can _always_ allocate 1k on the stack 
then that's better (smaller and simpler) than having three possible codepaths 
to debug.

If we did care about this (and I'd like to establish first that we _do_), then 
we'd want to examine the existing stack usages to see which ones are too big.  
(If we're limited to 4k stacks there's bound to be places that violate that 
now...)

Rob