re: Deadlock on fragmented memory?

[matthew green <mrg%eterna.com.au@localhost>]

> Thanks for doing this.
> 
> How difficult you gather it would be to convert the execargs to use 4K
> pages rathern then continuous KVA, are there any particular gotchas? I
> would be interested to look at this - it sounds like nice high-kernel only
> project for me, to make a break from the ATA oddities.

oh..  if you want to fix this properly...  :-)

please make this code use dynamically allocated sizes that only
rely upon contiguous VAs inside the users address spaces, not
the kernels, and at the same time, remove the MAXARGS restriction,
replacing it with something like linux's 1/3rd of RLIMIT_AS (i
*think* this is the way they do it.)

this basically means argument lists can grow to machine limits
and 'grep foo *' in a really large dir on a system with so much
free or cached ram that it's really shameful it requires using
traditional unix methods instead of just working.

the _how_ to manage in the kernel is less relevant than actually
making it able to be huge in my mind, but fixing the KVA issue
enables this a lot better it seems to me, so it feels like a
right step to be making.

thanks.


.mrg.