drm/i915: Move load time stolen memory init earlier
Subsystems (2)
cover drm/i915: Move load time stolen memory init earlier
For Sagar's BXT RC6 setup sanitization patch [1], we'll need the details about the GTT/stolen memory reservation early during driver loading, so this patchset moves the required init calls earlier accordingly. It also sanitizes a few MMIO/GEM init/clean-up steps on the way.
[1]
http://lists.freedesktop.org/archives/intel-gfx/2016-January/085440.html
Imre Deak (7):
drm/i915: Sanitize up DMC/CSR ucode cleanup code drm/i915: Sanitize i915_get_bridge_dev() error path drm/i915: Sanitize GEM shrinker init and clean-up drm/i915: Sanitize i915_gem_load() init and clean-up drm/i915: Move allocation of various workqueues earlier during init drm/i915: Move MCHBAR setup earlier during init drm/i915: Move stolen memory initialization earlier during loading
drivers/gpu/drm/i915/i915_dma.c | 218 ++++++++++++++++--------------- drivers/gpu/drm/i915/i915_drv.h | 4 +- drivers/gpu/drm/i915/i915_gem.c | 13 +- drivers/gpu/drm/i915/i915_gem_gtt.c | 15 +++ drivers/gpu/drm/i915/i915_gem_shrinker.c | 16 ++- 5 files changed, 158 insertions(+), 108 deletions(-)
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 1/7 drm/i915: Sanitize up DMC/CSR ucode cleanup code
commit ebae38d061df3deffa7c17b030ea14a5216ee55f
Author: Animesh Manna <[email protected]> Date: Wed Oct 28 23:58:55 2015 +0200
drm/i915/gen9: csr_init after runtime pm enable
moved the DMC/CSR initialization later during driver loading, but didn't move the cleanup earlier correspondingly during unloading. Fix this up.
--- drivers/gpu/drm/i915/i915_dma.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +7 -7
@@ -404,7 +404,7 @@ static int i915_load_modeset_init(struct drm_device *dev) ret =intel_irq_install (dev_priv);if (ret) - goto cleanup_gem_stolen; + goto cleanup_csr;intel_setup_gmbus (dev);
@@ -458,7 +458,8 @@ cleanup_irq:intel_guc_ucode_fini (dev);drm_irq_uninstall (dev);intel_teardown_gmbus (dev); -cleanup_gem_stolen: +cleanup_csr: +intel_csr_ucode_fini (dev_priv);i915_gem_cleanup_stolen (dev); cleanup_vga_switcheroo:vga_switcheroo_unregister_client (dev->pdev);
@@ -946,7 +947,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) ret =i915_gem_gtt_init (dev);if (ret) - goto out_freecsr; + goto out_uncore_fini;/* WARNING: Apparently we must kick fbdev drivers before vgacon, * otherwise the vga fbdev driver falls over. */
@@ -1116,8 +1117,7 @@ out_mtrrfree:io_mapping_free (dev_priv->gtt.mappable); out_gtt:i915_global_gtt_cleanup (dev); -out_freecsr: -intel_csr_ucode_fini (dev_priv); +out_uncore_fini:intel_uncore_fini (dev);pci_iounmap (dev->pdev, dev_priv->regs); put_bridge:
@@ -1183,6 +1183,8 @@ int i915_driver_unload(struct drm_device *dev)vga_switcheroo_unregister_client (dev->pdev);vga_client_register (dev->pdev,NULL ,NULL ,NULL ); +intel_csr_ucode_fini (dev_priv); +/* Free error state after interrupts are fully disabled. */ cancel_delayed_work_sync (&dev_priv->gpu_error.hangcheck_work);i915_destroy_error_state (dev);
@@ -1203,8 +1205,6 @@ int i915_driver_unload(struct drm_device *dev)intel_fbc_cleanup_cfb (dev_priv);i915_gem_cleanup_stolen (dev); -intel_csr_ucode_fini (dev_priv); -intel_teardown_mchbar (dev);destroy_workqueue (dev_priv->hotplug.dp_wq);
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 1/7] drm/i915: Sanitize up DMC/CSR ucode cleanup code
Quoted 10 lines · Imre Deak
commit
ebae38d061df3deffa7c17b030ea14a5216ee55fAuthor: Animesh Manna <[email protected]> Date: Wed Oct 28 23:58:55 2015 +0200drm/i915/gen9:
csr_initafter runtime pm enablemoved the DMC/CSR initialization later during driver loading, but didn't move the cleanup earlier correspondingly during unloading. Fix this up.
Signed-off-by Imre Deak <[email protected]>
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/7 drm/i915: Sanitize i915_get_bridge_dev() error path
In reply to Imre DeakClarify the name of the label on the error path, making it clear what's being cleaned up. The kmem_cache_destroy() calls are NOPs on the corresponding error path.
--- drivers/gpu/drm/i915/i915_dma.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +2 -3
@@ -917,7 +917,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)if (i915_get_bridge_dev (dev)) { ret = -EIO ; - goto free_priv; + goto out_runtime_pm_put; } mmio_bar =IS_GEN2 (dev) ?1 :0 ;
@@ -1122,11 +1122,10 @@ out_uncore_fini:pci_iounmap (dev->pdev, dev_priv->regs); put_bridge:pci_dev_put (dev_priv->bridge_dev); -free_priv:kmem_cache_destroy (dev_priv->requests);kmem_cache_destroy (dev_priv->vmas);kmem_cache_destroy (dev_priv->objects); - +out_runtime_pm_put:intel_runtime_pm_put (dev_priv);kfree (dev_priv);
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 2/7] drm/i915: Sanitize i915_get_bridge_dev() error path
Quoted 5 lines · Imre Deak
Clarify the name of the label on the error path, making it clear what's being cleaned up. The
kmem_cache_destroy()calls are NOPs on the corresponding error path.Signed-off-by Imre Deak <[email protected]>
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 3/7 drm/i915: Sanitize GEM shrinker init and clean-up
In reply to Imre DeakFactor out the common GEM shrinker clean-up code and call the shrinker init function from the same function from where the corresponding shrinker clean-up function is called. Also add sanity checking to the shrinker and OOM registration calls.
--- drivers/gpu/drm/i915/i915_dma.c | 7 +++---- drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 2 -- drivers/gpu/drm/i915/i915_gem_shrinker.c | 16 ++++++++++++++-- 4 files changed, 18 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +3 -4
@@ -1036,6 +1036,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)intel_opregion_setup (dev);i915_gem_load (dev); +i915_gem_shrinker_init (dev_priv);/* On the 945G/GM, the chipset reports the MSI capability on the * integrated graphics even though the support isn't actually there
@@ -1099,8 +1100,7 @@ out_power_well:intel_power_domains_fini (dev_priv);drm_vblank_cleanup (dev); out_gem_unload: -WARN_ON (unregister_oom_notifier (&dev_priv->mm.oom_notifier)); -unregister_shrinker (&dev_priv->mm.shrinker); +i915_gem_shrinker_cleanup (dev_priv);if (dev->pdev->msi_enabled)pci_disable_msi (dev->pdev);
@@ -1153,8 +1153,7 @@ int i915_driver_unload(struct drm_device *dev)i915_teardown_sysfs (dev); -WARN_ON (unregister_oom_notifier (&dev_priv->mm.oom_notifier)); -unregister_shrinker (&dev_priv->mm.shrinker); +i915_gem_shrinker_cleanup (dev_priv);io_mapping_free (dev_priv->gtt.mappable);arch_phys_wc_del (dev_priv->gtt.mtrr);
drivers/gpu/drm/i915/i915_drv.h +1 -0
@@ -3263,6 +3263,7 @@ unsigned long i915_gem_shrink(struct drm_i915_private *dev_priv,#define I915_SHRINK_ACTIVE 0x8unsigned long i915_gem_shrink_all (struct drm_i915_private *dev_priv);void i915_gem_shrinker_init (struct drm_i915_private *dev_priv); +void i915_gem_shrinker_cleanup (struct drm_i915_private *dev_priv);/* i915_gem_tiling.c */
drivers/gpu/drm/i915/i915_gem.c +0 -2
@@ -5063,8 +5063,6 @@ i915_gem_load(struct drm_device *dev) dev_priv->mm.interruptible = true; -i915_gem_shrinker_init (dev_priv); -mutex_init (&dev_priv->fb_tracking.lock); }
drivers/gpu/drm/i915/i915_gem_shrinker.c +14 -2
@@ -367,8 +367,20 @@ void i915_gem_shrinker_init(struct drm_i915_private *dev_priv) dev_priv->mm.shrinker.scan_objects = i915_gem_shrinker_scan; dev_priv->mm.shrinker.count_objects = i915_gem_shrinker_count; dev_priv->mm.shrinker.seeks =DEFAULT_SEEKS ; -register_shrinker (&dev_priv->mm.shrinker); +WARN_ON (register_shrinker (&dev_priv->mm.shrinker)); dev_priv->mm.oom_notifier.notifier_call = i915_gem_shrinker_oom; -register_oom_notifier (&dev_priv->mm.oom_notifier); +WARN_ON (register_oom_notifier (&dev_priv->mm.oom_notifier)); +} + +/** +* i915_gem_shrinker_cleanup - Clean up i915 shrinker +* @dev_priv: i915 device +* +* This function unregisters the i915 shrinker and OOM handler. +*/ +void i915_gem_shrinker_cleanup (struct drm_i915_private *dev_priv) +{ +WARN_ON (unregister_oom_notifier (&dev_priv->mm.oom_notifier)); +unregister_shrinker (&dev_priv->mm.shrinker); }
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 3/7] drm/i915: Sanitize GEM shrinker init and clean-up
Quoted 6 lines · Imre Deak
Factor out the common GEM shrinker clean-up code and call the shrinker init function from the same function from where the corresponding shrinker clean-up function is called. Also add sanity checking to the shrinker and OOM registration calls.
Signed-off-by Imre Deak <[email protected]>
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 4/7 drm/i915: Sanitize i915_gem_load() init and clean-up
In reply to Imre DeakFactor out common clean-up code for the GEM load time init function. Also rename i915_gem_load() to i915_gem_load_init() to have a better match with its new clean-up function.
No functional change.
--- drivers/gpu/drm/i915/i915_dma.c | 10 +++------- drivers/gpu/drm/i915/i915_drv.h | 3 ++- drivers/gpu/drm/i915/i915_gem.c | 11 ++++++++++- 3 files changed, 15 insertions(+), 9 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +3 -7
@@ -1035,7 +1035,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)intel_setup_mchbar (dev);intel_opregion_setup (dev); -i915_gem_load (dev); +i915_gem_load_init (dev);i915_gem_shrinker_init (dev_priv);/* On the 945G/GM, the chipset reports the MSI capability on the
@@ -1122,9 +1122,7 @@ out_uncore_fini:pci_iounmap (dev->pdev, dev_priv->regs); put_bridge:pci_dev_put (dev_priv->bridge_dev); -kmem_cache_destroy (dev_priv->requests); -kmem_cache_destroy (dev_priv->vmas); -kmem_cache_destroy (dev_priv->objects); +i915_gem_load_cleanup (dev); out_runtime_pm_put:intel_runtime_pm_put (dev_priv);
@@ -1216,9 +1214,7 @@ int i915_driver_unload(struct drm_device *dev)if (dev_priv->regs !=NULL )pci_iounmap (dev->pdev, dev_priv->regs); -kmem_cache_destroy (dev_priv->requests); -kmem_cache_destroy (dev_priv->vmas); -kmem_cache_destroy (dev_priv->objects); +i915_gem_load_cleanup (dev);pci_dev_put (dev_priv->bridge_dev);kfree (dev_priv);
drivers/gpu/drm/i915/i915_drv.h +2 -1
@@ -2850,7 +2850,8 @@ int i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv);int i915_gem_wait_ioctl (struct drm_device *dev,void *data,struct drm_file *file_priv); -void i915_gem_load (struct drm_device *dev); +void i915_gem_load_init (struct drm_device *dev); +void i915_gem_load_cleanup (struct drm_device *dev);void *i915_gem_object_alloc (struct drm_device *dev);void i915_gem_object_free (struct drm_i915_gem_object *obj); void i915_gem_object_init(struct drm_i915_gem_object *obj,
drivers/gpu/drm/i915/i915_gem.c +10 -1
@@ -4997,7 +4997,7 @@ init_ring_lists(struct intel_engine_cs *ring) }void -i915_gem_load (struct drm_device *dev) +i915_gem_load_init (struct drm_device *dev) {struct drm_i915_private *dev_priv = dev->dev_private;int i;
@@ -5066,6 +5066,15 @@ i915_gem_load(struct drm_device *dev)mutex_init (&dev_priv->fb_tracking.lock); } +void i915_gem_load_cleanup (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv =to_i915 (dev); + +kmem_cache_destroy (dev_priv->requests); +kmem_cache_destroy (dev_priv->vmas); +kmem_cache_destroy (dev_priv->objects); +} +void i915_gem_release (struct drm_device *dev,struct drm_file *file) {struct drm_i915_file_private *file_priv = file->driver_priv;
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/7] drm/i915: Sanitize i915_gem_load() init and clean-up
Quoted 7 lines · Imre Deak
Factor out common clean-up code for the GEM load time init function. Also rename
i915_gem_load()toi915_gem_load_init()to have a better match with its new clean-up function.No functional change.
Signed-off-by Imre Deak <[email protected]>
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 5/7 drm/i915: Move allocation of various workqueues earlier during init
In reply to Imre DeakWorkqueue initalization doesn't depend on any other device specific resource, so move it close to the beginning, so we don't need to consider them when thinking about dependencies for other resources.
Also factor out things to separate init/cleanup functions to make i915_driver_load()/unload() clearer, atm it's somewhat difficult to follow there in what order resources are inited/cleaned-up.
--- drivers/gpu/drm/i915/i915_dma.c | 101 ++++++++++++++++++++++------------------ 1 file changed, 56 insertions(+), 45 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +56 -45
@@ -856,6 +856,54 @@ static void intel_init_dpio(struct drm_i915_private *dev_priv) } } +static int i915_workqueues_init (struct drm_i915_private *dev_priv) +{ +/* +* The i915 workqueue is primarily used for batched retirement of +* requests (and thus managing bo) once the task has been completed +* by the GPU. i915_gem_retire_requests() is called directly when we +* need high-priority retirement, such as waiting for an explicit +* bo. +* +* It is also used for periodic low-priority events, such as +* idle-timers and recording error state. +* +* All tasks on the workqueue are expected to acquire the dev mutex +* so there is no point in running more than one instance of the +* workqueue at any time. Use an ordered one. +*/ + dev_priv->wq =alloc_ordered_workqueue ("i915" ,0 ); +if (dev_priv->wq ==NULL ) + goto out_err; + + dev_priv->hotplug.dp_wq =alloc_ordered_workqueue ("i915-dp" ,0 ); +if (dev_priv->hotplug.dp_wq ==NULL ) + goto out_free_wq; + + dev_priv->gpu_error.hangcheck_wq = +alloc_ordered_workqueue ("i915-hangcheck" ,0 ); +if (dev_priv->gpu_error.hangcheck_wq ==NULL ) + goto out_free_dp_wq; + +return 0 ; + +out_free_dp_wq: +destroy_workqueue (dev_priv->hotplug.dp_wq); +out_free_wq: +destroy_workqueue (dev_priv->wq); +out_err: +DRM_ERROR ("Failed to allocate workqueues.\n" ); + +return -ENOMEM ; +} + +static void i915_workqueues_cleanup (struct drm_i915_private *dev_priv) +{ +destroy_workqueue (dev_priv->gpu_error.hangcheck_wq); +destroy_workqueue (dev_priv->hotplug.dp_wq); +destroy_workqueue (dev_priv->wq); +} +/** * i915_driver_load - setup chip and create an initial config * @dev: DRM device
@@ -899,6 +947,10 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)mutex_init (&dev_priv->av_mutex);mutex_init (&dev_priv->wm.wm_mutex); + ret =i915_workqueues_init (dev_priv); +if (ret <0 ) + goto out_free_priv; +intel_pm_setup (dev);intel_runtime_pm_get (dev_priv);
@@ -993,41 +1045,6 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) dev_priv->gtt.mtrr =arch_phys_wc_add (dev_priv->gtt.mappable_base, aperture_size); -/* The i915 workqueue is primarily used for batched retirement of -* requests (and thus managing bo) once the task has been completed -* by the GPU. i915_gem_retire_requests() is called directly when we -* need high-priority retirement, such as waiting for an explicit -* bo. -* -* It is also used for periodic low-priority events, such as -* idle-timers and recording error state. -* -* All tasks on the workqueue are expected to acquire the dev mutex -* so there is no point in running more than one instance of the -* workqueue at any time. Use an ordered one. -*/ - dev_priv->wq =alloc_ordered_workqueue ("i915" ,0 ); -if (dev_priv->wq ==NULL ) { -DRM_ERROR ("Failed to create our workqueue.\n" ); - ret = -ENOMEM ; - goto out_mtrrfree; - } - - dev_priv->hotplug.dp_wq =alloc_ordered_workqueue ("i915-dp" ,0 ); -if (dev_priv->hotplug.dp_wq ==NULL ) { -DRM_ERROR ("Failed to create our dp workqueue.\n" ); - ret = -ENOMEM ; - goto out_freewq; - } - - dev_priv->gpu_error.hangcheck_wq = -alloc_ordered_workqueue ("i915-hangcheck" ,0 ); -if (dev_priv->gpu_error.hangcheck_wq ==NULL ) { -DRM_ERROR ("Failed to create our hangcheck workqueue.\n" ); - ret = -ENOMEM ; - goto out_freedpwq; - } -intel_irq_init (dev_priv);intel_uncore_sanitize (dev);
@@ -1107,12 +1124,6 @@ out_gem_unload:intel_teardown_mchbar (dev);pm_qos_remove_request (&dev_priv->pm_qos); -destroy_workqueue (dev_priv->gpu_error.hangcheck_wq); -out_freedpwq: -destroy_workqueue (dev_priv->hotplug.dp_wq); -out_freewq: -destroy_workqueue (dev_priv->wq); -out_mtrrfree:arch_phys_wc_del (dev_priv->gtt.mtrr);io_mapping_free (dev_priv->gtt.mappable); out_gtt:
@@ -1125,8 +1136,10 @@ put_bridge:i915_gem_load_cleanup (dev); out_runtime_pm_put:intel_runtime_pm_put (dev_priv); - +i915_workqueues_cleanup (dev_priv); +out_free_priv:kfree (dev_priv); +return ret; }
@@ -1203,9 +1216,6 @@ int i915_driver_unload(struct drm_device *dev)intel_teardown_mchbar (dev); -destroy_workqueue (dev_priv->hotplug.dp_wq); -destroy_workqueue (dev_priv->wq); -destroy_workqueue (dev_priv->gpu_error.hangcheck_wq);pm_qos_remove_request (&dev_priv->pm_qos);i915_global_gtt_cleanup (dev);
@@ -1216,6 +1226,7 @@ int i915_driver_unload(struct drm_device *dev)i915_gem_load_cleanup (dev);pci_dev_put (dev_priv->bridge_dev); +i915_workqueues_cleanup (dev_priv);kfree (dev_priv);return 0 ;
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 5/7] drm/i915: Move allocation of various workqueues earlier during init
Quoted 10 lines · Imre Deak
Workqueue initalization doesn't depend on any other device specific resource, so move it close to the beginning, so we don't need to consider them when thinking about dependencies for other resources.
Also factor out things to separate init/cleanup functions to make i915_driver_load()/unload() clearer, atm it's somewhat difficult to follow there in what order resources are inited/cleaned-up.
Suggested-by Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 6/7 drm/i915: Move MCHBAR setup earlier during init
In reply to Imre DeakMove the MCHBAR setup right after the MMIO setup, since the two things are logically related and the MCHBAR setup code doesn't depend on any other device specific resource. We'll also need MCHBAR to be ready earlier in an upcoming patch, so this is also a preparation for that.
Factor out the init/clean-up code to separate functions to make things clearer in the i915_driver_load()/unload() functions.
--- drivers/gpu/drm/i915/i915_dma.c | 71 ++++++++++++++++++++++++++--------------- 1 file changed, 45 insertions(+), 26 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +45 -26
@@ -904,6 +904,46 @@ static void i915_workqueues_cleanup(struct drm_i915_private *dev_priv)destroy_workqueue (dev_priv->wq); } +static int i915_mmio_setup (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv =to_i915 (dev); +int mmio_bar; +int mmio_size; + + mmio_bar =IS_GEN2 (dev) ?1 :0 ; +/* +* Before gen4, the registers and the GTT are behind different BARs. +* However, from gen4 onwards, the registers and the GTT are shared +* in the same BAR, so we want to restrict this ioremap from +* clobbering the GTT which we want ioremap_wc instead. Fortunately, +* the register BAR remains the same size for all the earlier +* generations up to Ironlake. +*/ +if (INTEL_INFO (dev)->gen <5 ) + mmio_size =512 *1024 ; +else + mmio_size =2 *1024 *1024 ; + dev_priv->regs =pci_iomap (dev->pdev, mmio_bar, mmio_size); +if (dev_priv->regs ==NULL ) { +DRM_ERROR ("failed to map registers\n" ); + +return -EIO ; + } + +/* Try to make sure MCHBAR is enabled before poking at it */ +intel_setup_mchbar (dev); + +return 0 ; +} + +static void i915_mmio_cleanup (struct drm_device *dev) +{ +struct drm_i915_private *dev_priv =to_i915 (dev); + +intel_teardown_mchbar (dev); +pci_iounmap (dev->pdev, dev_priv->regs); +} +/** * i915_driver_load - setup chip and create an initial config * @dev: DRM device
@@ -919,7 +959,7 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) {struct drm_i915_private *dev_priv;struct intel_device_info *info, *device_info; -int ret =0 , mmio_bar, mmio_size; +int ret =0 ;uint32_t aperture_size; info = (struct intel_device_info *) flags;
@@ -972,25 +1012,9 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags) goto out_runtime_pm_put; } - mmio_bar =IS_GEN2 (dev) ?1 :0 ; -/* Before gen4, the registers and the GTT are behind different BARs. -* However, from gen4 onwards, the registers and the GTT are shared -* in the same BAR, so we want to restrict this ioremap from -* clobbering the GTT which we want ioremap_wc instead. Fortunately, -* the register BAR remains the same size for all the earlier -* generations up to Ironlake. -*/ -if (info->gen <5 ) - mmio_size =512 *1024 ; -else - mmio_size =2 *1024 *1024 ; - - dev_priv->regs =pci_iomap (dev->pdev, mmio_bar, mmio_size); -if (!dev_priv->regs) { -DRM_ERROR ("failed to map registers\n" ); - ret = -EIO ; + ret =i915_mmio_setup (dev); +if (ret <0 ) goto put_bridge; - }/* This must be called before any calls to HAS_PCH_* */ intel_detect_pch (dev);
@@ -1048,8 +1072,6 @@ int i915_driver_load(struct drm_device *dev, unsigned long flags)intel_irq_init (dev_priv);intel_uncore_sanitize (dev); -/* Try to make sure MCHBAR is enabled before poking at it */ -intel_setup_mchbar (dev);intel_opregion_setup (dev);i915_gem_load_init (dev);
@@ -1130,7 +1152,7 @@ out_gtt:i915_global_gtt_cleanup (dev); out_uncore_fini:intel_uncore_fini (dev); -pci_iounmap (dev->pdev, dev_priv->regs); +i915_mmio_cleanup (dev); put_bridge:pci_dev_put (dev_priv->bridge_dev);i915_gem_load_cleanup (dev);
@@ -1214,15 +1236,12 @@ int i915_driver_unload(struct drm_device *dev)intel_fbc_cleanup_cfb (dev_priv);i915_gem_cleanup_stolen (dev); -intel_teardown_mchbar (dev); -pm_qos_remove_request (&dev_priv->pm_qos);i915_global_gtt_cleanup (dev);intel_uncore_fini (dev); -if (dev_priv->regs !=NULL ) -pci_iounmap (dev->pdev, dev_priv->regs); +i915_mmio_cleanup (dev);i915_gem_load_cleanup (dev);pci_dev_put (dev_priv->bridge_dev);
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 6/7] drm/i915: Move MCHBAR setup earlier during init
Quoted 10 lines · Imre Deak
Move the MCHBAR setup right after the MMIO setup, since the two things are logically related and the MCHBAR setup code doesn't depend on any other device specific resource. We'll also need MCHBAR to be ready earlier in an upcoming patch, so this is also a preparation for that.
Factor out the
init/clean-upcode to separate functions to make things clearer in the i915_driver_load()/unload() functions.Suggested-by Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 7/7 drm/i915: Move stolen memory initialization earlier during loading
In reply to Imre DeakThe only device specific dependency of the stolen memory setup is the MMIO mapping and the stolen memory size. Both are already available in i915_gtt_init(), so move the stolen initialization to there. The clean-up code for i915_gtt_init() is in i915_global_gtt_cleanup(), so move the stolen memory clean-up code there too.
This will be needed by an upcoming patch that needs the details of the memory we reserve, but the change is also part of our generic goal to move the initialization of resources with no or little dependencies on other device specific resources towards the beginning of the init sequence.
--- drivers/gpu/drm/i915/i915_dma.c | 10 ---------- drivers/gpu/drm/i915/i915_gem_gtt.c | 15 +++++++++++++++ 2 files changed, 15 insertions(+), 10 deletions(-)
drivers/gpu/drm/i915/i915_dma.c +0 -10
@@ -391,13 +391,6 @@ static int i915_load_modeset_init(struct drm_device *dev)if (ret) goto cleanup_vga_client; -/* Initialise stolen first so that we may reserve preallocated -* objects for the BIOS to KMS transition. -*/ - ret =i915_gem_init_stolen (dev); -if (ret) - goto cleanup_vga_switcheroo; -intel_power_domains_init_hw (dev_priv, false);intel_csr_ucode_init (dev_priv);
@@ -460,8 +453,6 @@ cleanup_irq:intel_teardown_gmbus (dev); cleanup_csr:intel_csr_ucode_fini (dev_priv); -i915_gem_cleanup_stolen (dev); -cleanup_vga_switcheroo:vga_switcheroo_unregister_client (dev->pdev); cleanup_vga_client:vga_client_register (dev->pdev,NULL ,NULL ,NULL );
@@ -1234,7 +1225,6 @@ int i915_driver_unload(struct drm_device *dev)i915_gem_context_fini (dev);mutex_unlock (&dev->struct_mutex);intel_fbc_cleanup_cfb (dev_priv); -i915_gem_cleanup_stolen (dev);pm_qos_remove_request (&dev_priv->pm_qos);
drivers/gpu/drm/i915/i915_gem_gtt.c +15 -0
@@ -2809,6 +2809,8 @@ void i915_global_gtt_cleanup(struct drm_device *dev) ppgtt->base.cleanup (&ppgtt->base); } +i915_gem_cleanup_stolen (dev); +if (drm_mm_initialized (&vm->mm)) {if (intel_vgpu_active (dev))intel_vgt_deballoon ();
@@ -3181,6 +3183,14 @@ int i915_gem_gtt_init(struct drm_device *dev)if (ret)return ret; +/* +* Initialise stolen early so that we may reserve preallocated +* objects for the BIOS to KMS transition. +*/ + ret =i915_gem_init_stolen (dev); +if (ret) + goto out_gtt_cleanup; +/* GMADR is the PCI mmio aperture into the global GTT. */ DRM_INFO ("Memory usable by graphics device = %lluM\n" , gtt->base.total >>20 );
@@ -3200,6 +3210,11 @@ int i915_gem_gtt_init(struct drm_device *dev)DRM_DEBUG_DRIVER ("ppgtt mode: %i\n" , i915.enable_ppgtt);return 0 ; + +out_gtt_cleanup: + gtt->base.cleanup (&dev_priv->gtt.base); + +return ret; }void i915_gem_restore_gtt_mappings (struct drm_device *dev)
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 7/7] drm/i915: Move stolen memory initialization earlier during loading
Quoted 6 lines · Imre Deak
The only device specific dependency of the stolen memory setup is the MMIO mapping and the stolen memory size. Both are already available in
i915_gtt_init(), so move the stolen initialization to there. The clean-up code fori915_gtt_init()is ini915_global_gtt_cleanup(), so move the stolen memory clean-up code there too.
Quoted 8 lines
This will be needed by an upcoming patch that needs the details of the memory we reserve, but the change is also part of our generic goal to move the initialization of resources with no or little dependencies on other device specific resources towards the beginning of the init sequence.
Suggested-by Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 7/7] drm/i915: Move stolen memory initialization earlier during loading
Quoted 19 lines · David Weinehall
Quoted 18 lines · Imre Deak
The only device specific dependency of the stolen memory setup is the
MMIO mapping and the stolen memory size. Both are already available ini915_gtt_init(), so move the stolen initialization to there. The clean-up code fori915_gtt_init()is ini915_global_gtt_cleanup(), so
move the stolen memory clean-up code there too.This will be needed by an upcoming patch that needs the details of the
memory we reserve, but the change is also part of our generic goal to
move the initialization of resources with no or little dependencies on
other device specific resources towards the beginning of the init sequence.
Quoted 4 lines
Quoted 2 lines
Suggested-by Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>Reviewed-by David Weinehall <[email protected]>
I pushed the series to dinq, thanks for the review.
--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for drm/i915: Move load time stolen memory init earlier
In reply to Imre Deak== Summary ==
== Summary ==
Built on 00a0c7d1ae09b1259c7af8e5a088b0b225d805df drm-intel-nightly: 2016y-01m-18d-16h-50m-37s UTC integration manifest
Test gem_ctx_basic:
pass -> FAIL (bdw-ultra)
Test kms_flip:
Subgroup basic-flip-vs-dpms:
dmesg-warn -> PASS (skl-i5k-2)
Test pm_rpm:
Subgroup basic-rte:
dmesg-warn -> PASS (byt-nuc) UNSTABLEbdw-nuci7 total:140 pass:131 dwarn:0 dfail:0 fail:0 skip:9 bdw-ultra total:140 pass:132 dwarn:0 dfail:1 fail:1 skip:6 byt-nuc total:143 pass:126 dwarn:2 dfail:0 fail:0 skip:15 hsw-brixbox total:143 pass:136 dwarn:0 dfail:0 fail:0 skip:7 hsw-gt2 total:143 pass:139 dwarn:0 dfail:0 fail:0 skip:4 ilk-hp8440p total:143 pass:102 dwarn:3 dfail:0 fail:0 skip:38 ivb-t430s total:137 pass:124 dwarn:3 dfail:4 fail:0 skip:6 skl-i5k-2 total:143 pass:134 dwarn:1 dfail:0 fail:0 skip:8 snb-dellxps total:143 pass:124 dwarn:5 dfail:0 fail:0 skip:14 snb-x220t total:143 pass:124 dwarn:5 dfail:0 fail:1 skip:13
Results at /archive/results/CI_IGT_test/Patchwork_1220/
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: ✗ Fi.CI.BAT: failure for drm/i915: Move load time stolen memory init earlier
Quoted 7 lines · Patchwork
== Summary ==
Built on
00a0c7d1ae09b1259c7af8e5a088b0b225d805dfdrm-intel-nightly:
2016y-01m-18d-16h-50m-37s UTC integration manifestTest
gem_ctx_basic:
pass -> FAIL (bdw-ultra)
Couldn't reproduce it on the same machine. To me it looks unrelated as it happened already several times for other patches, always on the same machine. These BAT results are not shown on the test's "long term" history btw. I filed a bug:
https://bugs.freedesktop.org/show_bug.cgi?id=93776
Quoted 37 lines
Test
kms_flip:
Subgroup basic-flip-vs-dpms:
dmesg-warn -> PASS (skl-i5k-2) Testpm_rpm:
Subgroup basic-rte:
dmesg-warn -> PASS (byt-nuc) UNSTABLEbdw-
nuci7 total:140 pass:131 dwarn:0 dfail:0 fail:0 skip:9
bdw-
ultra total:140 pass:132 dwarn:0 dfail:1 fail:1 skip:6
byt-
nuc total:143 pass:126 dwarn:2 dfail:0 fail:0 skip:1 5
hsw-
brixbox total:143 pass:136 dwarn:0 dfail:0 fail:0 skip:7
hsw-
gt2 total:143 pass:139 dwarn:0 dfail:0 fail:0 skip:4
ilk-
hp8440p total:143 pass:102 dwarn:3 dfail:0 fail:0 skip:3 8
ivb-
t430s total:137 pass:124 dwarn:3 dfail:4 fail:0 skip:6
skl-i5k-
2 total:143 pass:134 dwarn:1 dfail:0 fail:0 skip:8 snb-
dellxps total:143 pass:124 dwarn:5 dfail:0 fail:0 skip:1 4
snb-
x220t total:143 pass:124 dwarn:5 dfail:0 fail:1 skip:1 3
Quoted 2 lines
Results at /archive/results/CI_IGT_test/Patchwork_1220/
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: ✗ Fi.CI.BAT: failure for drm/i915: Move load time stolen memory init earlier
Quoted 6 lines · Imre Deak
Quoted 5 lines · Patchwork
== Summary ==
Built on
00a0c7d1ae09b1259c7af8e5a088b0b225d805dfdrm-intel-nightly:
2016y-01m-18d-16h-50m-37s UTC integration manifest
Quoted 7 lines
Quoted 2 lines
Test
gem_ctx_basic:
pass -> FAIL (bdw-ultra)Couldn't reproduce it on the same machine. To me it looks unrelated as it happened already several times for other patches, always on the same machine. These BAT results are not shown on the test's "long term" history btw. I filed a bug:
long term only shows changes, not all tests. Which means it didn't yet fail in -nightly, which is somewhat suspicious ... But I agree that this seems to have blown up a few times in other CI runs.
-Daniel
Quoted 45 lines
https://bugs.freedesktop.org/show_bug.cgi?id=93776
Quoted 37 lines
Test
kms_flip:
Subgroup basic-flip-vs-dpms:
dmesg-warn -> PASS (skl-i5k-2) Testpm_rpm:
Subgroup basic-rte:
dmesg-warn -> PASS (byt-nuc) UNSTABLEbdw-
nuci7 total:140 pass:131 dwarn:0 dfail:0 fail:0 skip:9
bdw-
ultra total:140 pass:132 dwarn:0 dfail:1 fail:1 skip:6
byt-
nuc total:143 pass:126 dwarn:2 dfail:0 fail:0 skip:1 5
hsw-
brixbox total:143 pass:136 dwarn:0 dfail:0 fail:0 skip:7
hsw-
gt2 total:143 pass:139 dwarn:0 dfail:0 fail:0 skip:4
ilk-
hp8440p total:143 pass:102 dwarn:3 dfail:0 fail:0 skip:3 8
ivb-
t430s total:137 pass:124 dwarn:3 dfail:4 fail:0 skip:6
skl-i5k-
2 total:143 pass:134 dwarn:1 dfail:0 fail:0 skip:8 snb-
dellxps total:143 pass:124 dwarn:5 dfail:0 fail:0 skip:1 4
snb-
x220t total:143 pass:124 dwarn:5 dfail:0 fail:1 skip:1 3Quoted 2 lines
Results at /archive/results/CI_IGT_test/Patchwork_1220/
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: ✗ Fi.CI.BAT: failure for drm/i915: Move load time stolen memory init earlier
Quoted 11 lines · Daniel Vetter
Quoted 10 lines · Imre Deak
Quoted 8 lines · Patchwork
== Summary ==
Built on
00a0c7d1ae09b1259c7af8e5a088b0b225d805dfdrm-intel- nightly:
2016y-01m-18d-16h-50m-37s UTC integration manifestTest
gem_ctx_basic:
pass -> FAIL (bdw-ultra)
Quoted 8 lines
Quoted 6 lines
Couldn't reproduce it on the same machine. To me it looks unrelated as
it happened already several times for other patches, always on the same
machine. These BAT results are not shown on the test's "long term" history btw. I filed a bug:long term only shows changes, not all tests.
Ok, so I take that's changes for CI runs, but not for patchwork initiated tests.
Quoted 4 lines
Which means it didn't yet
fail in -nightly, which is somewhat suspicious ... But I agree that this
seems to have blown up a few times in other CI runs.
Ok, so based on that this patchset is good to go I guess.
Still not sure why the test fails. One thing I noticed is a bunch of "gem_concurrent: drop caches" messages in dmesg before any test would be even started. So I think something in gem_concurrent is not guarded with igt_fixture{} and gets to run when piglit enumerates the subtests (to get the BAT subtests). Not sure though if this has a negative effect on anything.
--Imre
Quoted 58 lines
-Daniel
Quoted 55 lines
https://bugs.freedesktop.org/show_bug.cgi?id=93776
Quoted 47 lines
Test
kms_flip:
Subgroup basic-flip-vs-dpms:
dmesg-warn -> PASS (skl-i5k-2) Testpm_rpm:
Subgroup basic-rte:
dmesg-warn -> PASS (byt-nuc) UNSTABLEbdw-
nuci7 total:140 pass:131 dwarn:0 dfail:0 fail:0 sk ip:9
bdw-
ultra total:140 pass:132 dwarn:0 dfail:1 fail:1 sk ip:6
byt-
nuc total:143 pass:126 dwarn:2 dfail:0 fail:0 sk ip:1
5
hsw-
brixbox total:143 pass:136 dwarn:0 dfail:0 fail:0 sk ip:7
hsw-
gt2 total:143 pass:139 dwarn:0 dfail:0 fail:0 sk ip:4
ilk-
hp8440p total:143 pass:102 dwarn:3 dfail:0 fail:0 sk ip:3
8
ivb-
t430s total:137 pass:124 dwarn:3 dfail:4 fail:0 sk ip:6
skl-i5k-
2 total:143 pass:134 dwarn:1 dfail:0 fail:0 skip:8
snb-
dellxps total:143 pass:124 dwarn:5 dfail:0 fail:0 sk ip:1
4
snb-
x220t total:143 pass:124 dwarn:5 dfail:0 fail:1 sk ip:1
3Quoted 2 lines
Results at /archive/results/CI_IGT_test/Patchwork_1220/
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: ✗ Fi.CI.BAT: failure for drm/i915: Move load time stolen memory init earlier
Quoted 29 lines · Imre Deak
Quoted 11 lines · Daniel Vetter
Quoted 10 lines · Imre Deak
Quoted 8 lines · Patchwork
== Summary ==
Built on
00a0c7d1ae09b1259c7af8e5a088b0b225d805dfdrm-intel- nightly:
2016y-01m-18d-16h-50m-37s UTC integration manifestTest
gem_ctx_basic:
pass -> FAIL (bdw-ultra)Quoted 8 lines
Quoted 6 lines
Couldn't reproduce it on the same machine. To me it looks unrelated as
it happened already several times for other patches, always on the same
machine. These BAT results are not shown on the test's "long term" history btw. I filed a bug:long term only shows changes, not all tests.
Ok, so I take that's changes for CI runs, but not for patchwork initiated tests.
Quoted 4 lines
Which means it didn't yet
fail in -nightly, which is somewhat suspicious ... But I agree that this
seems to have blown up a few times in other CI runs.
Quoted 8 lines
Ok, so based on that this patchset is good to go I guess.
Still not sure why the test fails. One thing I noticed is a bunch of "
gem_concurrent: drop caches" messages in dmesg before any test would be even started. So I think something ingem_concurrentis not guarded with igt_fixture{} and gets to run when piglit enumerates the subtests (to get the BAT subtests). Not sure though if this has a negative effect on anything.
We iirc drop caches as part of gem_quiescent too.
-Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx