[Intel-gfx] [PATCH] prime_self_import: Assure no pending requests before object counting

[Ben Widawsky]

On Fri, Nov 01, 2013 at 12:53:42PM +0000, oscar.mateo at intel.com wrote:
> From: Oscar Mateo <oscar.mateo at intel.com>
> 
> We don't want a previously used object to be freed in the middle of a
> before/after object counting operation (or we would get a "-1 objects
> leaked" message). We have seen this happening, e.g., when a context
> from a previous run dies, but its backing object is alive waiting for
> a retire_work to kick in.
> 
> Signed-off-by: Oscar Mateo <oscar.mateo at intel.com>
> Cc: Ben Widawsky <ben at bwidawsk.net>

Since as far as we know this can't happen until the 'context per fd'
code lands, I think we may as well hold off.

Instead how about 3 patches:

1. Introduce a library version of drop caches in igt. Move existing
callers.
2. Update prime_self_import.c
3. Update gem_flink_close

Meanwhile, I will merge this patch to the PPGTT IGT branch, so QA will
use it.

What do you think? If you don't want to do that, then simply merging
this afterwards is also fine with me.

[snip]

-- 
Ben Widawsky, Intel Open Source Technology Center