From 8b0e378a20e48c691d374f39d8b0596e63598cfc Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Thu, 19 Feb 2009 14:40:50 -0800 Subject: drm/i915: Cut two args to set_to_gpu_domain that confused this tricky path. While not strictly required, it helped while thinking about the following change. This change should be invariant. Signed-off-by: Eric Anholt Signed-off-by: Dave Airlie --- drivers/gpu/drm/i915/i915_gem.c | 38 ++++++++++++++++---------------------- 1 file changed, 16 insertions(+), 22 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index ac534c9a2f81..02ef50d512d6 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -34,10 +34,6 @@ #define I915_GEM_GPU_DOMAINS (~(I915_GEM_DOMAIN_CPU | I915_GEM_DOMAIN_GTT)) -static void -i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj, - uint32_t read_domains, - uint32_t write_domain); static void i915_gem_object_flush_gpu_write_domain(struct drm_gem_object *obj); static void i915_gem_object_flush_gtt_write_domain(struct drm_gem_object *obj); static void i915_gem_object_flush_cpu_write_domain(struct drm_gem_object *obj); @@ -2021,30 +2017,28 @@ i915_gem_object_set_to_cpu_domain(struct drm_gem_object *obj, int write) * drm_agp_chipset_flush */ static void -i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj, - uint32_t read_domains, - uint32_t write_domain) +i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj) { struct drm_device *dev = obj->dev; struct drm_i915_gem_object *obj_priv = obj->driver_private; uint32_t invalidate_domains = 0; uint32_t flush_domains = 0; - BUG_ON(read_domains & I915_GEM_DOMAIN_CPU); - BUG_ON(write_domain == I915_GEM_DOMAIN_CPU); + BUG_ON(obj->pending_read_domains & I915_GEM_DOMAIN_CPU); + BUG_ON(obj->pending_write_domain == I915_GEM_DOMAIN_CPU); #if WATCH_BUF DRM_INFO("%s: object %p read %08x -> %08x write %08x -> %08x\n", __func__, obj, - obj->read_domains, read_domains, - obj->write_domain, write_domain); + obj->read_domains, obj->pending_read_domains, + obj->write_domain, obj->pending_write_domain); #endif /* * If the object isn't moving to a new write domain, * let the object stay in multiple read domains */ - if (write_domain == 0) - read_domains |= obj->read_domains; + if (obj->pending_write_domain == 0) + obj->pending_read_domains |= obj->read_domains; else obj_priv->dirty = 1; @@ -2054,15 +2048,17 @@ i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj, * any read domains which differ from the old * write domain */ - if (obj->write_domain && obj->write_domain != read_domains) { + if (obj->write_domain && + obj->write_domain != obj->pending_read_domains) { flush_domains |= obj->write_domain; - invalidate_domains |= read_domains & ~obj->write_domain; + invalidate_domains |= + obj->pending_read_domains & ~obj->write_domain; } /* * Invalidate any read caches which may have * stale data. That is, any new read domains. */ - invalidate_domains |= read_domains & ~obj->read_domains; + invalidate_domains |= obj->pending_read_domains & ~obj->read_domains; if ((flush_domains | invalidate_domains) & I915_GEM_DOMAIN_CPU) { #if WATCH_BUF DRM_INFO("%s: CPU domain flush %08x invalidate %08x\n", @@ -2071,9 +2067,9 @@ i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj, i915_gem_clflush_object(obj); } - if ((write_domain | flush_domains) != 0) - obj->write_domain = write_domain; - obj->read_domains = read_domains; + if ((obj->pending_write_domain | flush_domains) != 0) + obj->write_domain = obj->pending_write_domain; + obj->read_domains = obj->pending_read_domains; dev->invalidate_domains |= invalidate_domains; dev->flush_domains |= flush_domains; @@ -2583,9 +2579,7 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, struct drm_gem_object *obj = object_list[i]; /* Compute new gpu domains and update invalidate/flush */ - i915_gem_object_set_to_gpu_domain(obj, - obj->pending_read_domains, - obj->pending_write_domain); + i915_gem_object_set_to_gpu_domain(obj); } i915_verify_inactive(dev, __FILE__, __LINE__); -- cgit v1.2.3 From efbeed96f7e20783b22d9529ef536b61f7ea8637 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Thu, 19 Feb 2009 14:54:51 -0800 Subject: drm/i915: Don't let a device flush to prepare buffers clear new write_domains. The problem was that object_set_to_gpu_domain would set the new write_domains that are getting set by this batchbuffer, then the accumulated flushes required for all the objects in preparation for this batchbuffer were posted, and the brand new write domain would get cleared by the flush being posted. Instead, hang on to the new (or old if we're not changing it) value and set it after the flush is queued. Results from this noticably included conformance test failures from reads shortly after writes (where the new write domain had been lost and thus not flushed and waited on), but is a suspected cause of hangs in some apps when a write domain is lost on a buffer that gets reused for instruction or commmand state. Signed-off-by: Eric Anholt Signed-off-by: Dave Airlie --- drivers/gpu/drm/i915/i915_gem.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 02ef50d512d6..0f50574076b1 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -2067,8 +2067,14 @@ i915_gem_object_set_to_gpu_domain(struct drm_gem_object *obj) i915_gem_clflush_object(obj); } - if ((obj->pending_write_domain | flush_domains) != 0) - obj->write_domain = obj->pending_write_domain; + /* The actual obj->write_domain will be updated with + * pending_write_domain after we emit the accumulated flush for all + * of our domain changes in execbuffers (which clears objects' + * write_domains). So if we have a current write domain that we + * aren't changing, set pending_write_domain to that. + */ + if (flush_domains == 0 && obj->pending_write_domain == 0) + obj->pending_write_domain = obj->write_domain; obj->read_domains = obj->pending_read_domains; dev->invalidate_domains |= invalidate_domains; @@ -2598,6 +2604,12 @@ i915_gem_execbuffer(struct drm_device *dev, void *data, (void)i915_add_request(dev, dev->flush_domains); } + for (i = 0; i < args->buffer_count; i++) { + struct drm_gem_object *obj = object_list[i]; + + obj->write_domain = obj->pending_write_domain; + } + i915_verify_inactive(dev, __FILE__, __LINE__); #if WATCH_COHERENCY -- cgit v1.2.3 From 5669fcacc58bf3a7386057addffd280d75380858 Mon Sep 17 00:00:00 2001 From: Jesse Barnes Date: Tue, 17 Feb 2009 15:13:31 -0800 Subject: drm/i915: suspend/resume GEM when KMS is active In the KMS case, we need to suspend/resume GEM as well. So on suspend, make sure we idle GEM and stop any new rendering from coming in, and on resume, re-init the framebuffer and clear the suspended flag. Signed-off-by: Jesse Barnes Signed-off-by: Eric Anholt Signed-off-by: Dave Airlie --- drivers/gpu/drm/i915/i915_drv.c | 23 ++++++++++++++++++++++- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 2 +- 3 files changed, 24 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index a31cbdbc3c54..0692622ee2b3 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -27,6 +27,7 @@ * */ +#include #include "drmP.h" #include "drm.h" #include "i915_drm.h" @@ -66,6 +67,12 @@ static int i915_suspend(struct drm_device *dev, pm_message_t state) i915_save_state(dev); + /* If KMS is active, we do the leavevt stuff here */ + if (drm_core_check_feature(dev, DRIVER_MODESET) && i915_gem_idle(dev)) { + dev_err(&dev->pdev->dev, "GEM idle failed, aborting suspend\n"); + return -EBUSY; + } + intel_opregion_free(dev); if (state.event == PM_EVENT_SUSPEND) { @@ -79,6 +86,9 @@ static int i915_suspend(struct drm_device *dev, pm_message_t state) static int i915_resume(struct drm_device *dev) { + struct drm_i915_private *dev_priv = dev->dev_private; + int ret = 0; + pci_set_power_state(dev->pdev, PCI_D0); pci_restore_state(dev->pdev); if (pci_enable_device(dev->pdev)) @@ -89,7 +99,18 @@ static int i915_resume(struct drm_device *dev) intel_opregion_init(dev); - return 0; + /* KMS EnterVT equivalent */ + if (drm_core_check_feature(dev, DRIVER_MODESET)) { + mutex_lock(&dev->struct_mutex); + dev_priv->mm.suspended = 0; + + ret = i915_gem_init_ringbuffer(dev); + if (ret != 0) + ret = -1; + mutex_unlock(&dev->struct_mutex); + } + + return ret; } static struct vm_operations_struct i915_gem_vm_ops = { diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 135a08f615cd..17fa40858d26 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -618,6 +618,7 @@ int i915_gem_init_ringbuffer(struct drm_device *dev); void i915_gem_cleanup_ringbuffer(struct drm_device *dev); int i915_gem_do_init(struct drm_device *dev, unsigned long start, unsigned long end); +int i915_gem_idle(struct drm_device *dev); int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf); int i915_gem_object_set_to_gtt_domain(struct drm_gem_object *obj, int write); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 0f50574076b1..58c789da95a3 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -2973,7 +2973,7 @@ i915_gem_evict_from_list(struct drm_device *dev, struct list_head *head) return 0; } -static int +int i915_gem_idle(struct drm_device *dev) { drm_i915_private_t *dev_priv = dev->dev_private; -- cgit v1.2.3 From f21289b355cee8738d80c2ae5cbd272c3f7b5689 Mon Sep 17 00:00:00 2001 From: Eric Anholt Date: Wed, 18 Feb 2009 09:44:56 -0800 Subject: drm/i915: Retire requests from i915_gem_busy_ioctl. This ensures that the user gets the latest information from the hardware on whether the buffer is busy, potentially reducing the working set of objects that the user chooses. Signed-off-by: Eric Anholt Signed-off-by: Dave Airlie --- drivers/gpu/drm/i915/i915_gem.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 58c789da95a3..8b50d4820389 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -2872,6 +2872,13 @@ i915_gem_busy_ioctl(struct drm_device *dev, void *data, return -EBADF; } + /* Update the active list for the hardware's current position. + * Otherwise this only updates on a delayed timer or when irqs are + * actually unmasked, and our working set ends up being larger than + * required. + */ + i915_gem_retire_requests(dev); + obj_priv = obj->driver_private; /* Don't count being on the flushing list against the object being * done. Otherwise, a buffer left on the flushing list but not getting -- cgit v1.2.3 From bab2d1f6531657e37dc84f26184f3f64e1e73ecd Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Fri, 20 Feb 2009 17:52:20 +0000 Subject: drm/i915: Fix regression in 95ca9d The object is dereferenced before the NULL check. Oops. Fixes http://bugs.freedesktop.org/show_bug.cgi?id=20235 Signed-off-by: Chris Wilson Signed-off-by: Eric Anholt Signed-off-by: Dave Airlie --- drivers/gpu/drm/i915/i915_gem.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 8b50d4820389..25b337438ca7 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -3143,16 +3143,20 @@ static void i915_gem_cleanup_hws(struct drm_device *dev) { drm_i915_private_t *dev_priv = dev->dev_private; - struct drm_gem_object *obj = dev_priv->hws_obj; - struct drm_i915_gem_object *obj_priv = obj->driver_private; + struct drm_gem_object *obj; + struct drm_i915_gem_object *obj_priv; if (dev_priv->hws_obj == NULL) return; + obj = dev_priv->hws_obj; + obj_priv = obj->driver_private; + kunmap(obj_priv->page_list[0]); i915_gem_object_unpin(obj); drm_gem_object_unreference(obj); dev_priv->hws_obj = NULL; + memset(&dev_priv->hws_map, 0, sizeof(dev_priv->hws_map)); dev_priv->hw_status_page = NULL; -- cgit v1.2.3 From 6fb88588555a18792a27f483887fe1f2af5f9c9b Mon Sep 17 00:00:00 2001 From: Jesse Barnes Date: Mon, 23 Feb 2009 10:08:21 +1000 Subject: drm/i915: fix WC mapping in non-GEM i915 code. [airlied - taken from mailing list posting] Signed-off-by: Dave Airlie --- drivers/gpu/drm/i915/i915_dma.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c index 81f1cff56fd5..2d797ffe8137 100644 --- a/drivers/gpu/drm/i915/i915_dma.c +++ b/drivers/gpu/drm/i915/i915_dma.c @@ -202,7 +202,7 @@ static int i915_initialize(struct drm_device * dev, drm_i915_init_t * init) dev_priv->ring.map.flags = 0; dev_priv->ring.map.mtrr = 0; - drm_core_ioremap(&dev_priv->ring.map, dev); + drm_core_ioremap_wc(&dev_priv->ring.map, dev); if (dev_priv->ring.map.handle == NULL) { i915_dma_cleanup(dev); -- cgit v1.2.3 From 5004417d840e6dcb0052061fd04569b9c9f037a8 Mon Sep 17 00:00:00 2001 From: Pierre Willenbrock Date: Mon, 23 Feb 2009 10:12:15 +1000 Subject: drm/i915: Add missing mutex_lock(&dev->struct_mutex) there might be a nicer way to fix this but this is the simplest for now. Signed-off-by: Pierre Willenbrock Signed-off-by: Dave Airlie --- drivers/gpu/drm/i915/intel_display.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c index 4d2baf7b00be..65b635ce28c8 100644 --- a/drivers/gpu/drm/i915/intel_display.c +++ b/drivers/gpu/drm/i915/intel_display.c @@ -1008,6 +1008,7 @@ static int intel_crtc_cursor_set(struct drm_crtc *crtc, temp = CURSOR_MODE_DISABLE; addr = 0; bo = NULL; + mutex_lock(&dev->struct_mutex); goto finish; } -- cgit v1.2.3