On Sat, Oct 13, 2018 at 2:04 AM Edgecombe, Rick P
[off-list ref] wrote:

On Fri, 2018-10-12 at 19:22 +0200, Jann Horn wrote:

quoted

On Fri, Oct 12, 2018 at 7:04 PM Edgecombe, Rick P
[off-list ref] wrote:

quoted

On Fri, 2018-10-12 at 02:35 +0200, Jann Horn wrote:

quoted

Why all the rbtree stuff instead of stashing a pointer in struct
vmap_area, or something like that?

Since the tracking was not for all vmalloc usage, the intention was to not
bloat
the structure for other usages likes stacks. I thought usually there
wouldn't be
nearly as much module space allocations as there would be kernel stacks, but
I
didn't do any actual measurements on the tradeoffs.

I imagine that one extra pointer in there - pointing to your struct
mod_alloc_user - would probably not be terrible. 8 bytes more per
kernel stack shouldn't be so bad?

I looked into this and it starts to look a little messy. The nommu.c version of
vmalloc doesn't use or expose access to vmap_area or vm_struct. So it starts to
look like a bunch of IFDEFs to remove the rlimit in the nommu case or making a
stand in that maintains pretend vm struct's in nommu.c. I had actually
previously tried to at least pull the allocations size from vmalloc structs, but it broke on nommu.

Thought I would check back and see. How important do you think this is?

I don't think it's important - I just thought that it would be nice to
avoid the extra complexity if it is easily avoidable.