On Thu, Oct 13, 2016 at 04:57:39PM +0200, Daniel Vetter wrote:
> On Wed, Oct 12, 2016 at 10:05:19AM +0100, Chris Wilson wrote:
> > The error state is purposefully racy as we expect it to be called at any
> > time and so have avoided any locking whilst capturing the crash dump.
> > However, with multi-engine GPUs and multiple CPUs, those races can
> > manifest into OOPSes as we attempt to chase dangling pointers freed on
> > other CPUs. Under discussion are lots of ways to slow down normal
> > operation in order to protect the post-mortem error capture, but what it
> > we take the opposite approach and freeze the machine whilst the error
> > capture runs (note the GPU may still running, but as long as we don't
> > process any of the results the driver's bookkeeping will be static).
> > Note that by of itself, this is not a complete fix. It also depends on
> > the compiler barriers in list_add/list_del to prevent traversing the
> > lists into the void. We also depend that we only require state from
> > carefully controlled sources - i.e. all the state we require for
> > post-mortem debugging should be reachable from the request itself so
> > that we only have to worry about retrieving the request carefully. Once
> > we have the request, we know that all pointers from it are intact.
> > v2: Avoid drm_clflush_pages() inside stop_machine() as it may use
> > stop_machine() itself for its wbinvd fallback.
> Hm, won't this hurt us real bad on any atom with ppgtt? Maybe a big check
> gen check with a scary comment about why we can't call drm_clflush_pages
> on old machines? Iirc gen3+ should all be able to flush without
:) Patch 2 switched to using coherent reads through the GTT for all.
Everyone is now equal (and the nice part about that was that it
uncovered the WC bug from kernel 4.0!)
Chris Wilson, Intel Open Source Technology Centre
Intel-gfx mailing list