[Intel-gfx] [PATCH 2/2] drm/i915: Use a cached mapping for the physical HWS

[Chris Wilson]

Older gen use a physical address for the hardware status page, for which
we use cache-coherent writes. As the writes are into the cpu cache, we use
a normal WB mapped page to read the HWS, used for our seqno tracking.

Anecdotally, I observed lost breadcrumbs writes into the HWS on i965gm,
which so far have not reoccurred with this patch. How reliable that
evidence is remains to be seen.

v2: Explicitly pass the expected physical address to the hw
v3: Also remember the wild writes we once had for HWS above 4G.

Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Daniel Vetter <daniel at ffwll.ch>
Cc: Joonas Lahtinen <joonas.lahtinen at linux.intel.com>
Reviewed-by: Daniel Vetter <daniel at ffwll.ch>
---
 drivers/gpu/drm/i915/i915_drv.h         |  1 -
 drivers/gpu/drm/i915/intel_engine_cs.c  | 25 +++++++++++++------------
 drivers/gpu/drm/i915/intel_ringbuffer.c |  7 +++++--
 3 files changed, 18 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 9771f39d99b3..5a4da5b723fd 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1666,7 +1666,6 @@ struct drm_i915_private {
 	struct intel_engine_cs *engine_class[MAX_ENGINE_CLASS + 1]
 					    [MAX_ENGINE_INSTANCE + 1];
 
-	struct drm_dma_handle *status_page_dmah;
 	struct resource mch_res;
 
 	/* protects the irq masks */
diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index 292eae19fce2..10cd051ba29e 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -532,11 +532,11 @@ void intel_engine_cleanup_scratch(struct intel_engine_cs *engine)
 
 static void cleanup_status_page(struct intel_engine_cs *engine)
 {
-	struct drm_dma_handle *dmah;
+	if (HWS_NEEDS_PHYSICAL(engine->i915)) {
+		void *addr = fetch_and_zero(&engine->status_page.page_addr);
 
-	dmah = fetch_and_zero(&engine->i915->status_page_dmah);
-	if (dmah)
-		drm_pci_free(&engine->i915->drm, dmah);
+		__free_page(virt_to_page(addr));
+	}
 
 	i915_vma_unpin_and_release(&engine->status_page.vma,
 				   I915_VMA_RELEASE_MAP);
@@ -605,17 +605,18 @@ static int init_status_page(struct intel_engine_cs *engine)
 
 static int init_phys_status_page(struct intel_engine_cs *engine)
 {
-	struct drm_i915_private *dev_priv = engine->i915;
-
-	GEM_BUG_ON(engine->id != RCS);
+	struct page *page;
 
-	dev_priv->status_page_dmah =
-		drm_pci_alloc(&dev_priv->drm, PAGE_SIZE, PAGE_SIZE);
-	if (!dev_priv->status_page_dmah)
+	/*
+	 * Though the HWS register does support 36bit addresses, historically
+	 * we have had hangs and corruption reported due to wild writes if
+	 * the HWS is placed above 4G.
+	 */
+	page = alloc_page(GFP_KERNEL | __GFP_DMA32 | __GFP_ZERO);
+	if (!page)
 		return -ENOMEM;
 
-	engine->status_page.page_addr = dev_priv->status_page_dmah->vaddr;
-	memset(engine->status_page.page_addr, 0, PAGE_SIZE);
+	engine->status_page.page_addr = page_address(page);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 44432677160c..86604dd1c5a5 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -344,11 +344,14 @@ gen7_render_ring_flush(struct i915_request *rq, u32 mode)
 static void ring_setup_phys_status_page(struct intel_engine_cs *engine)
 {
 	struct drm_i915_private *dev_priv = engine->i915;
+	struct page *page = virt_to_page(engine->status_page.page_addr);
+	phys_addr_t phys = PFN_PHYS(page_to_pfn(page));
 	u32 addr;
 
-	addr = dev_priv->status_page_dmah->busaddr;
+	addr = lower_32_bits(phys);
 	if (INTEL_GEN(dev_priv) >= 4)
-		addr |= (dev_priv->status_page_dmah->busaddr >> 28) & 0xf0;
+		addr |= (phys >> 28) & 0xf0;
+
 	I915_WRITE(HWS_PGA, addr);
 }
 
-- 
2.19.0.rc1