[Intel-gfx] [PATCH v2 17/22] drm/i915: error capture with no ggtt slot
Matthew Auld
matthew.auld at intel.com
Thu Oct 3 19:24:39 UTC 2019
From: Daniele Ceraolo Spurio <daniele.ceraolospurio at intel.com>
If the aperture is not available in HW we can't use a ggtt slot and wc
copy, so fall back to regular kmap.
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio at intel.com>
Signed-off-by: Abdiel Janulgue <abdiel.janulgue at linux.intel.com>
Signed-off-by: Matthew Auld <matthew.auld at intel.com>
---
drivers/gpu/drm/i915/i915_gem_gtt.c | 19 ++++----
drivers/gpu/drm/i915/i915_gpu_error.c | 65 ++++++++++++++++++++++-----
2 files changed, 64 insertions(+), 20 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 64bf47eaa2a1..096f50f1dda5 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -2661,7 +2661,8 @@ static void ggtt_release_guc_top(struct i915_ggtt *ggtt)
static void cleanup_init_ggtt(struct i915_ggtt *ggtt)
{
ggtt_release_guc_top(ggtt);
- drm_mm_remove_node(&ggtt->error_capture);
+ if (drm_mm_node_allocated(&ggtt->error_capture))
+ drm_mm_remove_node(&ggtt->error_capture);
}
static int init_ggtt(struct i915_ggtt *ggtt)
@@ -2692,13 +2693,15 @@ static int init_ggtt(struct i915_ggtt *ggtt)
if (ret)
return ret;
- /* Reserve a mappable slot for our lockless error capture */
- ret = drm_mm_insert_node_in_range(&ggtt->vm.mm, &ggtt->error_capture,
- PAGE_SIZE, 0, I915_COLOR_UNEVICTABLE,
- 0, ggtt->mappable_end,
- DRM_MM_INSERT_LOW);
- if (ret)
- return ret;
+ if (ggtt->mappable_end) {
+ /* Reserve a mappable slot for our lockless error capture */
+ ret = drm_mm_insert_node_in_range(&ggtt->vm.mm, &ggtt->error_capture,
+ PAGE_SIZE, 0, I915_COLOR_UNEVICTABLE,
+ 0, ggtt->mappable_end,
+ DRM_MM_INSERT_LOW);
+ if (ret)
+ return ret;
+ }
/*
* The upper portion of the GuC address space has a sizeable hole
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c
index 6384a06aa5bf..c6c96f0c6b28 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.c
+++ b/drivers/gpu/drm/i915/i915_gpu_error.c
@@ -40,6 +40,7 @@
#include "display/intel_overlay.h"
#include "gem/i915_gem_context.h"
+#include "gem/i915_gem_lmem.h"
#include "i915_drv.h"
#include "i915_gpu_error.h"
@@ -235,6 +236,7 @@ struct compress {
struct pagevec pool;
struct z_stream_s zstream;
void *tmp;
+ bool wc;
};
static bool compress_init(struct compress *c)
@@ -292,7 +294,7 @@ static int compress_page(struct compress *c,
struct z_stream_s *zstream = &c->zstream;
zstream->next_in = src;
- if (c->tmp && i915_memcpy_from_wc(c->tmp, src, PAGE_SIZE))
+ if (c->wc && c->tmp && i915_memcpy_from_wc(c->tmp, src, PAGE_SIZE))
zstream->next_in = c->tmp;
zstream->avail_in = PAGE_SIZE;
@@ -367,6 +369,7 @@ static void err_compression_marker(struct drm_i915_error_state_buf *m)
struct compress {
struct pagevec pool;
+ bool wc;
};
static bool compress_init(struct compress *c)
@@ -389,7 +392,7 @@ static int compress_page(struct compress *c,
if (!ptr)
return -ENOMEM;
- if (!i915_memcpy_from_wc(ptr, src, PAGE_SIZE))
+ if (!(c->wc && i915_memcpy_from_wc(ptr, src, PAGE_SIZE)))
memcpy(ptr, src, PAGE_SIZE);
dst->pages[dst->page_count++] = ptr;
@@ -970,7 +973,6 @@ i915_error_object_create(struct drm_i915_private *i915,
struct drm_i915_error_object *dst;
unsigned long num_pages;
struct sgt_iter iter;
- dma_addr_t dma;
int ret;
might_sleep();
@@ -996,17 +998,54 @@ i915_error_object_create(struct drm_i915_private *i915,
dst->page_count = 0;
dst->unused = 0;
+ compress->wc = i915_gem_object_is_lmem(vma->obj) ||
+ drm_mm_node_allocated(&ggtt->error_capture);
+
ret = -EINVAL;
- for_each_sgt_daddr(dma, iter, vma->pages) {
+ if (drm_mm_node_allocated(&ggtt->error_capture)) {
void __iomem *s;
+ dma_addr_t dma;
- ggtt->vm.insert_page(&ggtt->vm, dma, slot, I915_CACHE_NONE, 0);
+ for_each_sgt_daddr(dma, iter, vma->pages) {
+ ggtt->vm.insert_page(&ggtt->vm, dma, slot,
+ I915_CACHE_NONE, 0);
- s = io_mapping_map_wc(&ggtt->iomap, slot, PAGE_SIZE);
- ret = compress_page(compress, (void __force *)s, dst);
- io_mapping_unmap(s);
- if (ret)
- break;
+ s = io_mapping_map_wc(&ggtt->iomap, slot, PAGE_SIZE);
+ ret = compress_page(compress, (void __force *)s, dst);
+ io_mapping_unmap(s);
+ if (ret)
+ break;
+ }
+ } else if (i915_gem_object_is_lmem(vma->obj)) {
+ struct intel_memory_region *mem = vma->obj->mm.region;
+ dma_addr_t dma;
+
+ for_each_sgt_daddr(dma, iter, vma->pages) {
+ void __iomem *s;
+
+ s = io_mapping_map_atomic_wc(&mem->iomap, dma);
+ ret = compress_page(compress, s, dst);
+ io_mapping_unmap_atomic(s);
+ if (ret)
+ break;
+ }
+ } else {
+ struct page *page;
+
+ for_each_sgt_page(page, iter, vma->pages) {
+ void *s;
+
+ drm_clflush_pages(&page, 1);
+
+ s = kmap_atomic(page);
+ ret = compress_page(compress, s, dst);
+ kunmap_atomic(s);
+
+ drm_clflush_pages(&page, 1);
+
+ if (ret)
+ break;
+ }
}
if (ret || compress_flush(compress, dst)) {
@@ -1675,9 +1714,11 @@ static unsigned long capture_find_epoch(const struct i915_gpu_state *error)
static void capture_finish(struct i915_gpu_state *error)
{
struct i915_ggtt *ggtt = &error->i915->ggtt;
- const u64 slot = ggtt->error_capture.start;
- ggtt->vm.clear_range(&ggtt->vm, slot, PAGE_SIZE);
+ if (drm_mm_node_allocated(&ggtt->error_capture)) {
+ const u64 slot = ggtt->error_capture.start;
+ ggtt->vm.clear_range(&ggtt->vm, slot, PAGE_SIZE);
+ }
}
#define DAY_AS_SECONDS(x) (24 * 60 * 60 * (x))
--
2.20.1
More information about the Intel-gfx
mailing list