Re: [gomp4] kernels offload fns

2016-01-07 Thread Nathan Sidwell
On 01/06/16 19:11, Cesar Philippidis wrote: On 12/29/2015 06:19 AM, Nathan Sidwell wrote: @@ -19385,7 +19415,7 @@ new_oacc_loop_routine (oacc_loop *parent { oacc_loop *loop = new_oacc_loop_raw (parent, gimple_location (call)); int dims[GOMP_DIM_MAX]; - int level = oacc_validate_dims

Re: [gomp4] kernels offload fns

2016-01-06 Thread Cesar Philippidis
On 12/29/2015 06:19 AM, Nathan Sidwell wrote: > @@ -19385,7 +19415,7 @@ new_oacc_loop_routine (oacc_loop *parent > { >oacc_loop *loop = new_oacc_loop_raw (parent, gimple_location (call)); >int dims[GOMP_DIM_MAX]; > - int level = oacc_validate_dims (decl, attrs, dims); > + int level = oa

[gomp4] kernels offload fns

2015-12-29 Thread Nathan Sidwell
In developing a non-unity default partition mechanism I discovered there was no mechanism to reliably determine whether an offload was for a kernels region or not. The tree-ssa pass uses a heuristic that is sufficient for its needs, but not very clear. This patch adjusts set_oacc_fn_attrib t