[v3,0/4] Some cleanups for the KVA/vmalloc

Message

Uladzislau Rezki May 27, 2019, 9:38 a.m. UTC

Patch [1] removes an unused argument "node" from the __alloc_vmap_area()
function and that is it.

Patch [2] is not driven by any particular workload that fails or so,
it is just better approach to handle one specific split case.

Patch [3] some cleanups in merging path. Basically on a first step
the mergeable node is detached and there is no reason to "unlink" it.
The same concerns the second step unless it has been merged on first
one.

Patch [4] moves BUG_ON()/RB_EMPTY_NODE() checks under "unlink" logic.
After [3] merging path "unlink" only linked nodes. Therefore we can say
that removing detached object is a bug in all cases.

v2->v3:
    - remove the odd comment from the [3];

v1->v2:
    - update the commit message. [2] patch;
    - fix typos in comments. [2] patch;
    - do the "preload" for NUMA awareness. [2] patch;

Uladzislau Rezki (Sony) (4):
  mm/vmap: remove "node" argument
  mm/vmap: preload a CPU with one object for split purpose
  mm/vmap: get rid of one single unlink_va() when merge
  mm/vmap: move BUG_ON() check to the unlink_va()

 mm/vmalloc.c | 115 +++++++++++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 89 insertions(+), 26 deletions(-)

Comments

Roman Gushchin May 28, 2019, 10:45 p.m. UTC | #1

On Mon, May 27, 2019 at 11:38:41AM +0200, Uladzislau Rezki (Sony) wrote:
> It does not make sense to try to "unlink" the node that is
> definitely not linked with a list nor tree. On the first
> merge step VA just points to the previously disconnected
> busy area.
> 
> On the second step, check if the node has been merged and do
> "unlink" if so, because now it points to an object that must
> be linked.
> 
> Signed-off-by: Uladzislau Rezki (Sony) <[e-mail]>
> Acked-by: Hillf Danton <[e-mail]>

Reviewed-by: Roman Gushchin <[e-mail]>

Thanks!