Messages in this thread | | | Date | Sun, 22 Apr 2012 22:05:54 +0200 | From | Sam Ravnborg <> | Subject | Re: Weirdness in __alloc_bootmem_node_high |
| |
On Sun, Apr 22, 2012 at 03:22:10PM -0400, David Miller wrote: > From: Sam Ravnborg <sam@ravnborg.org> > Date: Fri, 20 Apr 2012 21:43:09 +0200 > > > I have it almost finished - except that it does not work :-( > > We have limitations in what area we can allocate very early, > > and here I had to use the alloc_bootmem_low() variant. > > I had preferred a variant that allowed me to allocate > > bottom-up in this case. > > I think you're going to have to bear down and map all of linear kernel > mappings before you start using the bootmem code rather than > afterwards.
Thanks - I will try to do so. But I must admit I do not (yet) see any big difference in the approach before/after.
I digged up an old todo regarding final link of vmlinux. Will look at this and then re-visit memblock at sparc32.
Sam
| |