dri-devel August 2012

dri-devel@lists.freedesktop.org

145 participants
811 discussions

[Bug 16375] missing blit_vb check in r600_prepare_blit_copy
by bugzilla-daemon＠bugzilla.kernel.org 09 Aug '12

09 Aug '12

https://bugzilla.kernel.org/show_bug.cgi?id=16375 Alan <alan(a)lxorguk.ukuu.org.uk> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED CC| |alan(a)lxorguk.ukuu.org.uk Resolution| |CODE_FIX -- Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching the assignee of the bug.

1 0

[Bug 16295] radeon conflict/interference with rt2800pci
by bugzilla-daemon＠bugzilla.kernel.org 09 Aug '12

09 Aug '12

https://bugzilla.kernel.org/show_bug.cgi?id=16295 Alan <alan(a)lxorguk.ukuu.org.uk> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED CC| |alan(a)lxorguk.ukuu.org.uk Resolution| |OBSOLETE -- Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching the assignee of the bug.

1 0

[Bug 16273] suspend/resume failure with radeon kms driver
by bugzilla-daemon＠bugzilla.kernel.org 09 Aug '12

09 Aug '12

https://bugzilla.kernel.org/show_bug.cgi?id=16273 Alan <alan(a)lxorguk.ukuu.org.uk> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED CC| |alan(a)lxorguk.ukuu.org.uk Resolution| |OBSOLETE --- Comment #8 from Alan <alan(a)lxorguk.ukuu.org.uk> 2012-08-09 13:59:44 --- The driver has changed so much that this bug report is really obsolete. If you see the problem with a modern (3.2/4/5 kernel) feel free to update it -- Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching the assignee of the bug.

1 0

[PATCH 1/1] omap: include omap_drm.h independently
by Víctor Manuel Jáquez Leal 09 Aug '12

09 Aug '12

omap_drm.h uses data type defined in stdint.h, but that header was not included. omap_drm.h includes drm.h as a local file when it is part of the compiler c flags. This two issues are fixed. New code can include omap_drm.h alone. Signed-off-by: Víctor Manuel Jáquez Leal <vjaquez(a)igalia.com> --- omap/omap_drm.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/omap/omap_drm.h b/omap/omap_drm.h index f677cd8..9c6c0e4 100644 --- a/omap/omap_drm.h +++ b/omap/omap_drm.h @@ -29,7 +29,8 @@ #ifndef __OMAP_DRM_H__ #define __OMAP_DRM_H__ -#include "drm.h" +#include <stdint.h> +#include <drm.h> /* Please note that modifications to all structs defined here are * subject to backwards-compatibility constraints. -- 1.7.10.4

1 0

[PATCH 1/3] dma-fence: dma-buf synchronization (v7)
by Maarten Lankhorst 09 Aug '12

09 Aug '12

A dma-fence can be attached to a buffer which is being filled or consumed by hw, to allow userspace to pass the buffer without waiting to another device. For example, userspace can call page_flip ioctl to display the next frame of graphics after kicking the GPU but while the GPU is still rendering. The display device sharing the buffer with the GPU would attach a callback to get notified when the GPU's rendering-complete IRQ fires, to update the scan-out address of the display, without having to wake up userspace. A dma-fence is transient, one-shot deal. It is allocated and attached to one or more dma-buf's. When the one that attached it is done, with the pending operation, it can signal the fence. + dma_fence_signal() The dma-buf-mgr handles tracking, and waiting on, the fences associated with a dma-buf. TODO maybe need some helper fxn for simple devices, like a display- only drm/kms device which simply wants to wait for exclusive fence to be signaled, and then attach a non-exclusive fence while scanout is in progress. The one pending on the fence can add an async callback: + dma_fence_add_callback() The callback can optionally be cancelled with remove_wait_queue() Or wait synchronously (optionally with timeout or interruptible): + dma_fence_wait() A default software-only implementation is provided, which can be used by drivers attaching a fence to a buffer when they have no other means for hw sync. But a memory backed fence is also envisioned, because it is common that GPU's can write to, or poll on some memory location for synchronization. For example: fence = dma_buf_get_fence(dmabuf); if (fence->ops == &bikeshed_fence_ops) { dma_buf *fence_buf; dma_bikeshed_fence_get_buf(fence, &fence_buf, &offset); ... tell the hw the memory location to wait on ... } else { /* fall-back to sw sync * / dma_fence_add_callback(fence, my_cb); } On SoC platforms, if some other hw mechanism is provided for synchronizing between IP blocks, it could be supported as an alternate implementation with it's own fence ops in a similar way. To facilitate other non-sw implementations, the enable_signaling callback can be used to keep track if a device not supporting hw sync is waiting on the fence, and in this case should arrange to call dma_fence_signal() at some point after the condition has changed, to notify other devices waiting on the fence. If there are no sw waiters, this can be skipped to avoid waking the CPU unnecessarily. The handler of the enable_signaling op should take a refcount until the fence is signaled, then release its ref. The intention is to provide a userspace interface (presumably via eventfd) later, to be used in conjunction with dma-buf's mmap support for sw access to buffers (or for userspace apps that would prefer to do their own synchronization). v1: Original v2: After discussion w/ danvet and mlankhorst on #dri-devel, we decided that dma-fence didn't need to care about the sw->hw signaling path (it can be handled same as sw->sw case), and therefore the fence->ops can be simplified and more handled in the core. So remove the signal, add_callback, cancel_callback, and wait ops, and replace with a simple enable_signaling() op which can be used to inform a fence supporting hw->hw signaling that one or more devices which do not support hw signaling are waiting (and therefore it should enable an irq or do whatever is necessary in order that the CPU is notified when the fence is passed). v3: Fix locking fail in attach_fence() and get_fence() v4: Remove tie-in w/ dma-buf.. after discussion w/ danvet and mlankorst we decided that we need to be able to attach one fence to N dma-buf's, so using the list_head in dma-fence struct would be problematic. v5: [ Maarten Lankhorst ] Updated for dma-bikeshed-fence and dma-buf-manager. v6: [ Maarten Lankhorst ] I removed dma_fence_cancel_callback and some comments about checking if fence fired or not. This is broken by design. waitqueue_active during destruction is now fatal, since the signaller should be holding a reference in enable_signalling until it signalled the fence. Pass the original dma_fence_cb along, and call __remove_wait in the dma_fence_callback handler, so that no cleanup needs to be performed. v7: [ Maarten Lankhorst ] Set cb->func and only enable sw signaling if fence wasn't signaled yet, for example for hardware fences that may choose to signal blindly. Signed-off-by: Maarten Lankhorst <maarten.lankhorst(a)canonical.com> --- drivers/base/Makefile | 2 drivers/base/dma-fence.c | 287 +++++++++++++++++++++++++++++++++++++++++++++ include/linux/dma-fence.h | 96 +++++++++++++++ 3 files changed, 384 insertions(+), 1 deletion(-) create mode 100644 drivers/base/dma-fence.c create mode 100644 include/linux/dma-fence.h diff --git a/drivers/base/Makefile b/drivers/base/Makefile index 5aa2d70..6e9f217 100644 --- a/drivers/base/Makefile +++ b/drivers/base/Makefile @@ -10,7 +10,7 @@ obj-$(CONFIG_CMA) += dma-contiguous.o obj-y += power/ obj-$(CONFIG_HAS_DMA) += dma-mapping.o obj-$(CONFIG_HAVE_GENERIC_DMA_COHERENT) += dma-coherent.o -obj-$(CONFIG_DMA_SHARED_BUFFER) += dma-buf.o +obj-$(CONFIG_DMA_SHARED_BUFFER) += dma-buf.o dma-fence.o obj-$(CONFIG_ISA) += isa.o obj-$(CONFIG_FW_LOADER) += firmware_class.o obj-$(CONFIG_NUMA) += node.o diff --git a/drivers/base/dma-fence.c b/drivers/base/dma-fence.c new file mode 100644 index 0000000..c280ee7 --- /dev/null +++ b/drivers/base/dma-fence.c @@ -0,0 +1,287 @@ +/* + * Fence mechanism for dma-buf to allow for asynchronous dma access + * + * Copyright (C) 2012 Texas Instruments + * Author: Rob Clark <rob.clark(a)linaro.org> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License version 2 as published by + * the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program. If not, see <http://www.gnu.org/licenses/>. + */ + +#include <linux/slab.h> +#include <linux/sched.h> +#include <linux/export.h> +#include <linux/dma-fence.h> + +/** + * dma_fence_signal - Signal a fence. + * + * @fence: The fence to signal + * + * All registered callbacks will be called directly (synchronously) and + * all blocked waters will be awoken. + * + * TODO: any value in adding a dma_fence_cancel(), for example to recov + * from hung gpu? It would behave like dma_fence_signal() but return + * an error to waiters and cb's to let them know that the condition they + * are waiting for will never happen. + */ +int dma_fence_signal(struct dma_fence *fence) +{ + unsigned long flags; + int ret = -EINVAL; + + if (WARN_ON(!fence)) + return -EINVAL; + + spin_lock_irqsave(&fence->event_queue.lock, flags); + if (!fence->signaled) { + fence->signaled = true; + __wake_up_locked_key(&fence->event_queue, TASK_NORMAL, + &fence->event_queue); + ret = 0; + } else WARN(1, "Already signaled"); + spin_unlock_irqrestore(&fence->event_queue.lock, flags); + + return ret; +} +EXPORT_SYMBOL_GPL(dma_fence_signal); + +static void release_fence(struct kref *kref) +{ + struct dma_fence *fence = + container_of(kref, struct dma_fence, refcount); + + BUG_ON(waitqueue_active(&fence->event_queue)); + + if (fence->ops->release) + fence->ops->release(fence); + + kfree(fence); +} + +/** + * dma_fence_put - Release a reference to the fence. + */ +void dma_fence_put(struct dma_fence *fence) +{ + WARN_ON(!fence); + kref_put(&fence->refcount, release_fence); +} +EXPORT_SYMBOL_GPL(dma_fence_put); + +/** + * dma_fence_get - Take a reference to the fence. + * + * In most cases this is used only internally by dma-fence. + */ +void dma_fence_get(struct dma_fence *fence) +{ + WARN_ON(!fence); + kref_get(&fence->refcount); +} +EXPORT_SYMBOL_GPL(dma_fence_get); + +static int check_signaling(struct dma_fence *fence) +{ + bool enable_signaling = false, signaled; + unsigned long flags; + + spin_lock_irqsave(&fence->event_queue.lock, flags); + signaled = fence->signaled; + if (!signaled && !fence->needs_sw_signal) + enable_signaling = fence->needs_sw_signal = true; + spin_unlock_irqrestore(&fence->event_queue.lock, flags); + + if (enable_signaling) { + int ret; + + /* At this point, if enable_signaling returns any error + * a wakeup has to be performanced regardless. + * -ENOENT signals fence was already signaled. Any other error + * inidicates a catastrophic hardware error. + * + * If any hardware error occurs, nothing can be done against + * it, so it's treated like the fence was already signaled. + * No synchronization can be performed, so we have to assume + * the fence was already signaled. + */ + ret = fence->ops->enable_signaling(fence); + if (ret) { + signaled = true; + dma_fence_signal(fence); + } + } + + if (!signaled) + return 0; + else + return -ENOENT; +} + +int __dma_fence_wake_func(wait_queue_t *wait, unsigned mode, + int flags, void *key) +{ + struct dma_fence_cb *cb = + container_of(wait, struct dma_fence_cb, base); + + __remove_wait_queue(key, wait); + return cb->func(cb, wait->private); +} + +/** + * dma_fence_add_callback - Add a callback to be called when the fence + * is signaled. + * + * @fence: The fence to wait on + * @cb: The callback to register + * + * Any number of callbacks can be registered to a fence, but a callback + * can only be registered to once fence at a time. + * + * Note that the callback can be called from an atomic context. If + * fence is already signaled, this function will return -ENOENT (and + * *not* call the callback) + */ +int dma_fence_add_callback(struct dma_fence *fence, struct dma_fence_cb *cb, + dma_fence_func_t func, void *priv) +{ + unsigned long flags; + int ret; + + if (WARN_ON(!fence || !func)) + return -EINVAL; + + ret = check_signaling(fence); + + spin_lock_irqsave(&fence->event_queue.lock, flags); + if (!ret && fence->signaled) + ret = -ENOENT; + + if (!ret) { + cb->base.flags = 0; + cb->base.func = __dma_fence_wake_func; + cb->base.private = priv; + cb->fence = fence; + cb->func = func; + __add_wait_queue(&fence->event_queue, &cb->base); + } + spin_unlock_irqrestore(&fence->event_queue.lock, flags); + + return ret; +} +EXPORT_SYMBOL_GPL(dma_fence_add_callback); + +/** + * dma_fence_wait - Wait for a fence to be signaled. + * + * @fence: The fence to wait on + * @interruptible: if true, do an interruptible wait + * @timeout: absolute time for timeout, in jiffies. + * + * Returns 0 on success, -EBUSY if a timeout occured, + * -ERESTARTSYS if the wait was interrupted by a signal. + */ +int dma_fence_wait(struct dma_fence *fence, bool interruptible, unsigned long timeout) +{ + unsigned long cur; + int ret; + + if (WARN_ON(!fence)) + return -EINVAL; + + cur = jiffies; + if (time_after_eq(cur, timeout)) + return -EBUSY; + + timeout -= cur; + + ret = check_signaling(fence); + if (ret == -ENOENT) + return 0; + else if (ret) + return ret; + + if (interruptible) + ret = wait_event_interruptible_timeout(fence->event_queue, + fence->signaled, + timeout); + else + ret = wait_event_timeout(fence->event_queue, + fence->signaled, timeout); + + if (ret > 0) + return 0; + else if (!ret) + return -EBUSY; + else + return ret; +} +EXPORT_SYMBOL_GPL(dma_fence_wait); + +/* + * Helpers intended to be used by the ops of the dma_fence implementation: + * + * NOTE: helpers and fxns intended to be used by other dma-fence + * implementations are not exported.. I'm not really sure if it makes + * sense to have a dma-fence implementation that is itself a module. + */ + +void __dma_fence_init(struct dma_fence *fence, struct dma_fence_ops *ops, void *priv) +{ + WARN_ON(!ops || !ops->enable_signaling); + + kref_init(&fence->refcount); + fence->ops = ops; + fence->priv = priv; + init_waitqueue_head(&fence->event_queue); +} +EXPORT_SYMBOL_GPL(__dma_fence_init); + +/* + * Pure sw implementation for dma-fence. The CPU always gets involved. + */ + +static int sw_enable_signaling(struct dma_fence *fence) +{ + /* + * pure sw, no irq's to enable, because the fence creator will + * always call dma_fence_signal() + */ + return 0; +} + +static struct dma_fence_ops sw_fence_ops = { + .enable_signaling = sw_enable_signaling, +}; + +/** + * dma_fence_create - Create a simple sw-only fence. + * + * This fence only supports signaling from/to CPU. Other implementations + * of dma-fence can be used to support hardware to hardware signaling, if + * supported by the hardware, and use the dma_fence_helper_* functions for + * compatibility with other devices that only support sw signaling. + */ +struct dma_fence *dma_fence_create(void) +{ + struct dma_fence *fence; + + fence = kzalloc(sizeof(struct dma_fence), GFP_KERNEL); + if (!fence) + return ERR_PTR(-ENOMEM); + + __dma_fence_init(fence, &sw_fence_ops, 0); + + return fence; +} +EXPORT_SYMBOL_GPL(dma_fence_create); diff --git a/include/linux/dma-fence.h b/include/linux/dma-fence.h new file mode 100644 index 0000000..70d12c0 --- /dev/null +++ b/include/linux/dma-fence.h @@ -0,0 +1,96 @@ +/* + * Fence mechanism for dma-buf to allow for asynchronous dma access + * + * Copyright (C) 2012 Texas Instruments + * Author: Rob Clark <rob.clark(a)linaro.org> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License version 2 as published by + * the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program. If not, see <http://www.gnu.org/licenses/>. + */ + +#ifndef __DMA_FENCE_H__ +#define __DMA_FENCE_H__ + +#include <linux/err.h> +#include <linux/list.h> +#include <linux/wait.h> +#include <linux/list.h> +#include <linux/dma-buf.h> + +struct dma_fence; +struct dma_fence_ops; +struct dma_fence_cb; + +struct dma_fence { + struct kref refcount; + struct dma_fence_ops *ops; + wait_queue_head_t event_queue; + void *priv; + + /* has this fence been signaled yet? */ + bool signaled : 1; + + /* do we have one or more waiters or callbacks? */ + bool needs_sw_signal : 1; +}; + +typedef int (*dma_fence_func_t)(struct dma_fence_cb *cb, void *priv); + +struct dma_fence_cb { + wait_queue_t base; + dma_fence_func_t func; + struct dma_fence *fence; +}; + +struct dma_fence_ops { + /** + * For fence implementations that have the capability for hw->hw + * signaling, they can implement this op to enable the necessary + * irqs, or insert commands into cmdstream, etc. This is called + * in the first wait() or add_callback() path to let the fence + * implementation know that there is another driver waiting on + * the signal (ie. hw->sw case). + * + * A return value of -ENOENT will indicate that the fence has + * already passed. Any other errors will be treated as -ENOENT, + * and can happen because of hardware failure. + */ + int (*enable_signaling)(struct dma_fence *fence); + void (*release)(struct dma_fence *fence); +}; + +/* + * TODO does it make sense to be able to enable dma-fence without dma-buf, + * or visa versa? + */ +#ifdef CONFIG_DMA_SHARED_BUFFER + +/* create a basic (pure sw) fence: */ +struct dma_fence *dma_fence_create(void); + +/* intended to be used by other dma_fence implementations: */ +void __dma_fence_init(struct dma_fence *fence, + struct dma_fence_ops *ops, void *priv); + +void dma_fence_get(struct dma_fence *fence); +void dma_fence_put(struct dma_fence *fence); + +int dma_fence_signal(struct dma_fence *fence); +int dma_fence_wait(struct dma_fence *fence, bool interruptible, unsigned long timeout); +int dma_fence_add_callback(struct dma_fence *fence, struct dma_fence_cb *cb, + dma_fence_func_t func, void *priv); + +#else +// TODO +#endif /* CONFIG_DMA_SHARED_BUFFER */ + +#endif /* __DMA_FENCE_H__ */

2 5

[PATCH] dma-buf: add reference counting for exporter module
by Tomasz Stanislawski 09 Aug '12

09 Aug '12

This patch adds reference counting on a module that exports dma-buf and implements its operations. This prevents the module from being unloaded while DMABUF file is in use. Signed-off-by: Tomasz Stanislawski <t.stanislaws(a)samsung.com> --- Documentation/dma-buf-sharing.txt | 3 ++- drivers/base/dma-buf.c | 10 +++++++++- drivers/gpu/drm/exynos/exynos_drm_dmabuf.c | 1 + drivers/gpu/drm/i915/i915_gem_dmabuf.c | 1 + drivers/gpu/drm/nouveau/nouveau_prime.c | 1 + drivers/gpu/drm/radeon/radeon_prime.c | 1 + drivers/staging/omapdrm/omap_gem_dmabuf.c | 1 + include/linux/dma-buf.h | 2 ++ 8 files changed, 18 insertions(+), 2 deletions(-) diff --git a/Documentation/dma-buf-sharing.txt b/Documentation/dma-buf-sharing.txt index ad86fb8..2613057 100644 --- a/Documentation/dma-buf-sharing.txt +++ b/Documentation/dma-buf-sharing.txt @@ -49,7 +49,8 @@ The dma_buf buffer sharing API usage contains the following steps: The buffer exporter announces its wish to export a buffer. In this, it connects its own private buffer data, provides implementation for operations that can be performed on the exported dma_buf, and flags for the file - associated with this buffer. + associated with this buffer. The operations structure has owner field. + You should initialize this to THIS_MODULE in most cases. Interface: struct dma_buf *dma_buf_export(void *priv, struct dma_buf_ops *ops, diff --git a/drivers/base/dma-buf.c b/drivers/base/dma-buf.c index c30f3e1..d14b2f5 100644 --- a/drivers/base/dma-buf.c +++ b/drivers/base/dma-buf.c @@ -27,6 +27,7 @@ #include <linux/dma-buf.h> #include <linux/anon_inodes.h> #include <linux/export.h> +#include <linux/module.h> static inline int is_dma_buf_file(struct file *); @@ -40,6 +41,7 @@ static int dma_buf_release(struct inode *inode, struct file *file) dmabuf = file->private_data; dmabuf->ops->release(dmabuf); + module_put(dmabuf->ops->owner); kfree(dmabuf); return 0; } @@ -96,6 +98,7 @@ struct dma_buf *dma_buf_export(void *priv, const struct dma_buf_ops *ops, struct file *file; if (WARN_ON(!priv || !ops + || !ops->owner || !ops->map_dma_buf || !ops->unmap_dma_buf || !ops->release @@ -105,9 +108,14 @@ struct dma_buf *dma_buf_export(void *priv, const struct dma_buf_ops *ops, return ERR_PTR(-EINVAL); } + if (!try_module_get(ops->owner)) + return ERR_PTR(-ENOENT); + dmabuf = kzalloc(sizeof(struct dma_buf), GFP_KERNEL); - if (dmabuf == NULL) + if (dmabuf == NULL) { + module_put(ops->owner); return ERR_PTR(-ENOMEM); + } dmabuf->priv = priv; dmabuf->ops = ops; diff --git a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c index 613bf8a..cf3bc6d 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c +++ b/drivers/gpu/drm/exynos/exynos_drm_dmabuf.c @@ -164,6 +164,7 @@ static void exynos_gem_dmabuf_kunmap(struct dma_buf *dma_buf, } static struct dma_buf_ops exynos_dmabuf_ops = { + .owner = THIS_MODULE, .map_dma_buf = exynos_gem_map_dma_buf, .unmap_dma_buf = exynos_gem_unmap_dma_buf, .kmap = exynos_gem_dmabuf_kmap, diff --git a/drivers/gpu/drm/i915/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/i915_gem_dmabuf.c index aa308e1..07ff03b 100644 --- a/drivers/gpu/drm/i915/i915_gem_dmabuf.c +++ b/drivers/gpu/drm/i915/i915_gem_dmabuf.c @@ -152,6 +152,7 @@ static int i915_gem_dmabuf_mmap(struct dma_buf *dma_buf, struct vm_area_struct * } static const struct dma_buf_ops i915_dmabuf_ops = { + .owner = THIS_MODULE, .map_dma_buf = i915_gem_map_dma_buf, .unmap_dma_buf = i915_gem_unmap_dma_buf, .release = i915_gem_dmabuf_release, diff --git a/drivers/gpu/drm/nouveau/nouveau_prime.c b/drivers/gpu/drm/nouveau/nouveau_prime.c index a25cf2c..8605033 100644 --- a/drivers/gpu/drm/nouveau/nouveau_prime.c +++ b/drivers/gpu/drm/nouveau/nouveau_prime.c @@ -127,6 +127,7 @@ static void nouveau_gem_prime_vunmap(struct dma_buf *dma_buf, void *vaddr) } static const struct dma_buf_ops nouveau_dmabuf_ops = { + .owner = THIS_MODULE, .map_dma_buf = nouveau_gem_map_dma_buf, .unmap_dma_buf = nouveau_gem_unmap_dma_buf, .release = nouveau_gem_dmabuf_release, diff --git a/drivers/gpu/drm/radeon/radeon_prime.c b/drivers/gpu/drm/radeon/radeon_prime.c index 6bef46a..4061fd3 100644 --- a/drivers/gpu/drm/radeon/radeon_prime.c +++ b/drivers/gpu/drm/radeon/radeon_prime.c @@ -127,6 +127,7 @@ static void radeon_gem_prime_vunmap(struct dma_buf *dma_buf, void *vaddr) mutex_unlock(&dev->struct_mutex); } const static struct dma_buf_ops radeon_dmabuf_ops = { + .owner = THIS_MODULE, .map_dma_buf = radeon_gem_map_dma_buf, .unmap_dma_buf = radeon_gem_unmap_dma_buf, .release = radeon_gem_dmabuf_release, diff --git a/drivers/staging/omapdrm/omap_gem_dmabuf.c b/drivers/staging/omapdrm/omap_gem_dmabuf.c index 42728e0..6a4dd67 100644 --- a/drivers/staging/omapdrm/omap_gem_dmabuf.c +++ b/drivers/staging/omapdrm/omap_gem_dmabuf.c @@ -179,6 +179,7 @@ out_unlock: } struct dma_buf_ops omap_dmabuf_ops = { + .owner = THIS_MODULE, .map_dma_buf = omap_gem_map_dma_buf, .unmap_dma_buf = omap_gem_unmap_dma_buf, .release = omap_gem_dmabuf_release, diff --git a/include/linux/dma-buf.h b/include/linux/dma-buf.h index eb48f38..22953de 100644 --- a/include/linux/dma-buf.h +++ b/include/linux/dma-buf.h @@ -37,6 +37,7 @@ struct dma_buf_attachment; /** * struct dma_buf_ops - operations possible on struct dma_buf + * @owner: the module that implements dma_buf operations * @attach: [optional] allows different devices to 'attach' themselves to the * given buffer. It might return -EBUSY to signal that backing storage * is already allocated and incompatible with the requirements @@ -70,6 +71,7 @@ struct dma_buf_attachment; * @vunmap: [optional] unmaps a vmap from the buffer */ struct dma_buf_ops { + struct module *owner; int (*attach)(struct dma_buf *, struct device *, struct dma_buf_attachment *); -- 1.7.9.5

4 4

[PATCH] drm/mgag200: fix G200ER pll picking algorithm
by Dave Airlie 09 Aug '12

09 Aug '12

The original code was misported from the X driver, a) an int went to unsigned int, breaking the downward counting testm code b) the port did the vco/computed clock bits completely wrong. This fixes an infinite loop on modprobe on some Dell servers with the G200ER chipset variant. Found in internal testing. Cc: stable(a)vger.kernel.org Signed-off-by: Dave Airlie <airlied(a)redhat.com> --- drivers/gpu/drm/mgag200/mgag200_mode.c | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/mgag200/mgag200_mode.c b/drivers/gpu/drm/mgag200/mgag200_mode.c index d303061..0201d1d 100644 --- a/drivers/gpu/drm/mgag200/mgag200_mode.c +++ b/drivers/gpu/drm/mgag200/mgag200_mode.c @@ -468,10 +468,11 @@ static int mga_g200er_set_plls(struct mga_device *mdev, long clock) { unsigned int vcomax, vcomin, pllreffreq; unsigned int delta, tmpdelta; - unsigned int testr, testn, testm, testo; + int testr, testn, testm, testo; unsigned int p, m, n; - unsigned int computed; + unsigned int computed, vco; int tmp; + const unsigned int m_div_val[] = { 1, 2, 4, 8 }; m = n = p = 0; vcomax = 1488000; @@ -490,12 +491,13 @@ static int mga_g200er_set_plls(struct mga_device *mdev, long clock) if (delta == 0) break; for (testo = 5; testo < 33; testo++) { - computed = pllreffreq * (testn + 1) / + vco = pllreffreq * (testn + 1) / (testr + 1); - if (computed < vcomin) + if (vco < vcomin) continue; - if (computed > vcomax) + if (vco > vcomax) continue; + computed = vco / (m_div_val[testm] * (testo + 1)); if (computed > clock) tmpdelta = computed - clock; else -- 1.7.10.4

1 0

[PATCH v4] DRM: add drm gem CMA helper
by Sascha Hauer 08 Aug '12

08 Aug '12

Many embedded drm devices do not have a IOMMU and no dedicated memory for graphics. These devices use CMA (Contiguous Memory Allocator) backed graphics memory. This patch provides helper functions to be able to share the code. The code technically does not depend on CMA as the backend allocator, the name has been chosen because cma makes for a nice, short but still descriptive function prefix. Signed-off-by: Sascha Hauer <s.hauer(a)pengutronix.de> --- changes since v3: - *really* use DIV_ROUND_UP(args->width * args->bpp, 8) to calculate the pitch changes since v2: - make drm_gem_cma_create_with_handle static - use DIV_ROUND_UP(args->width * args->bpp, 8) to calculate the pitch - make drm_gem_cma_vm_ops const - add missing #include <linux/export.h> changes since v1: - map whole buffer at mmap time, not page by page at fault time - several cleanups as suggested by Lars-Peter Clausen and Laurent Pinchart drivers/gpu/drm/Kconfig | 6 + drivers/gpu/drm/Makefile | 1 + drivers/gpu/drm/drm_gem_cma_helper.c | 251 ++++++++++++++++++++++++++++++++++ include/drm/drm_gem_cma_helper.h | 44 ++++++ 4 files changed, 302 insertions(+) create mode 100644 drivers/gpu/drm/drm_gem_cma_helper.c create mode 100644 include/drm/drm_gem_cma_helper.h diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig index dc5df12..8f362ec 100644 --- a/drivers/gpu/drm/Kconfig +++ b/drivers/gpu/drm/Kconfig @@ -53,6 +53,12 @@ config DRM_TTM GPU memory types. Will be enabled automatically if a device driver uses it. +config DRM_GEM_CMA_HELPER + tristate + depends on DRM + help + Choose this if you need the GEM CMA helper functions + config DRM_TDFX tristate "3dfx Banshee/Voodoo3+" depends on DRM && PCI diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile index 0487ff6..8759cda 100644 --- a/drivers/gpu/drm/Makefile +++ b/drivers/gpu/drm/Makefile @@ -15,6 +15,7 @@ drm-y := drm_auth.o drm_buffer.o drm_bufs.o drm_cache.o \ drm_trace_points.o drm_global.o drm_prime.o drm-$(CONFIG_COMPAT) += drm_ioc32.o +drm-$(CONFIG_DRM_GEM_CMA_HELPER) += drm_gem_cma_helper.o drm-usb-y := drm_usb.o diff --git a/drivers/gpu/drm/drm_gem_cma_helper.c b/drivers/gpu/drm/drm_gem_cma_helper.c new file mode 100644 index 0000000..1aa8fee --- /dev/null +++ b/drivers/gpu/drm/drm_gem_cma_helper.c @@ -0,0 +1,251 @@ +/* + * drm gem CMA (contiguous memory allocator) helper functions + * + * Copyright (C) 2012 Sascha Hauer, Pengutronix + * + * Based on Samsung Exynos code + * + * Copyright (c) 2011 Samsung Electronics Co., Ltd. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ + +#include <linux/mm.h> +#include <linux/slab.h> +#include <linux/mutex.h> +#include <linux/export.h> +#include <linux/dma-mapping.h> + +#include <drm/drmP.h> +#include <drm/drm.h> +#include <drm/drm_gem_cma_helper.h> + +static unsigned int get_gem_mmap_offset(struct drm_gem_object *obj) +{ + return (unsigned int)obj->map_list.hash.key << PAGE_SHIFT; +} + +static void drm_gem_cma_buf_destroy(struct drm_device *drm, + struct drm_gem_cma_object *cma_obj) +{ + dma_free_writecombine(drm->dev, cma_obj->base.size, cma_obj->vaddr, + cma_obj->paddr); +} + +/* + * drm_gem_cma_create - allocate an object with the given size + * + * returns a struct drm_gem_cma_object* on success or ERR_PTR values + * on failure. + */ +struct drm_gem_cma_object *drm_gem_cma_create(struct drm_device *drm, + unsigned int size) +{ + struct drm_gem_cma_object *cma_obj; + struct drm_gem_object *gem_obj; + int ret; + + size = round_up(size, PAGE_SIZE); + + cma_obj = kzalloc(sizeof(*cma_obj), GFP_KERNEL); + if (!cma_obj) + return ERR_PTR(-ENOMEM); + + cma_obj->vaddr = dma_alloc_writecombine(drm->dev, size, + &cma_obj->paddr, GFP_KERNEL | __GFP_NOWARN); + if (!cma_obj->vaddr) { + dev_err(drm->dev, "failed to allocate buffer with size %d\n", size); + ret = -ENOMEM; + goto err_dma_alloc; + } + + gem_obj = &cma_obj->base; + + ret = drm_gem_object_init(drm, gem_obj, size); + if (ret) + goto err_obj_init; + + ret = drm_gem_create_mmap_offset(gem_obj); + if (ret) + goto err_create_mmap_offset; + + return cma_obj; + +err_create_mmap_offset: + drm_gem_object_release(gem_obj); + +err_obj_init: + drm_gem_cma_buf_destroy(drm, cma_obj); + +err_dma_alloc: + kfree(cma_obj); + + return ERR_PTR(ret); +} +EXPORT_SYMBOL_GPL(drm_gem_cma_create); + +/* + * drm_gem_cma_create_with_handle - allocate an object with the given + * size and create a gem handle on it + * + * returns a struct drm_gem_cma_object* on success or ERR_PTR values + * on failure. + */ +static struct drm_gem_cma_object *drm_gem_cma_create_with_handle( + struct drm_file *file_priv, + struct drm_device *drm, unsigned int size, + unsigned int *handle) +{ + struct drm_gem_cma_object *cma_obj; + struct drm_gem_object *gem_obj; + int ret; + + cma_obj = drm_gem_cma_create(drm, size); + if (IS_ERR(cma_obj)) + return cma_obj; + + gem_obj = &cma_obj->base; + + /* + * allocate a id of idr table where the obj is registered + * and handle has the id what user can see. + */ + ret = drm_gem_handle_create(file_priv, gem_obj, handle); + if (ret) + goto err_handle_create; + + /* drop reference from allocate - handle holds it now. */ + drm_gem_object_unreference_unlocked(gem_obj); + + return cma_obj; + +err_handle_create: + drm_gem_cma_free_object(gem_obj); + + return ERR_PTR(ret); +} + +/* + * drm_gem_cma_free_object - (struct drm_driver)->gem_free_object callback + * function + */ +void drm_gem_cma_free_object(struct drm_gem_object *gem_obj) +{ + struct drm_gem_cma_object *cma_obj; + + if (gem_obj->map_list.map) + drm_gem_free_mmap_offset(gem_obj); + + drm_gem_object_release(gem_obj); + + cma_obj = to_drm_gem_cma_obj(gem_obj); + + drm_gem_cma_buf_destroy(gem_obj->dev, cma_obj); + + kfree(cma_obj); +} +EXPORT_SYMBOL_GPL(drm_gem_cma_free_object); + +/* + * drm_gem_cma_dumb_create - (struct drm_driver)->dumb_create callback + * function + * + * This aligns the pitch and size arguments to the minimum required. wrap + * this into your own function if you need bigger alignment. + */ +int drm_gem_cma_dumb_create(struct drm_file *file_priv, + struct drm_device *dev, struct drm_mode_create_dumb *args) +{ + struct drm_gem_cma_object *cma_obj; + int min_pitch = DIV_ROUND_UP(args->width * args->bpp, 8); + + if (args->pitch < min_pitch) + args->pitch = min_pitch; + + if (args->size < args->pitch * args->height) + args->size = args->pitch * args->height; + + cma_obj = drm_gem_cma_create_with_handle(file_priv, dev, + args->size, &args->handle); + if (IS_ERR(cma_obj)) + return PTR_ERR(cma_obj); + + return 0; +} +EXPORT_SYMBOL_GPL(drm_gem_cma_dumb_create); + +/* + * drm_gem_cma_dumb_map_offset - (struct drm_driver)->dumb_map_offset callback + * function + */ +int drm_gem_cma_dumb_map_offset(struct drm_file *file_priv, + struct drm_device *drm, uint32_t handle, uint64_t *offset) +{ + struct drm_gem_object *gem_obj; + + mutex_lock(&drm->struct_mutex); + + gem_obj = drm_gem_object_lookup(drm, file_priv, handle); + if (!gem_obj) { + dev_err(drm->dev, "failed to lookup gem object\n"); + mutex_unlock(&drm->struct_mutex); + return -EINVAL; + } + + *offset = get_gem_mmap_offset(gem_obj); + + drm_gem_object_unreference(gem_obj); + + mutex_unlock(&drm->struct_mutex); + + return 0; +} +EXPORT_SYMBOL_GPL(drm_gem_cma_dumb_map_offset); + +const struct vm_operations_struct drm_gem_cma_vm_ops = { + .open = drm_gem_vm_open, + .close = drm_gem_vm_close, +}; +EXPORT_SYMBOL_GPL(drm_gem_cma_vm_ops); + +/* + * drm_gem_cma_mmap - (struct file_operation)->mmap callback function + */ +int drm_gem_cma_mmap(struct file *filp, struct vm_area_struct *vma) +{ + struct drm_gem_object *gem_obj; + struct drm_gem_cma_object *cma_obj; + int ret; + + ret = drm_gem_mmap(filp, vma); + if (ret) + return ret; + + gem_obj = vma->vm_private_data; + cma_obj = to_drm_gem_cma_obj(gem_obj); + + ret = remap_pfn_range(vma, vma->vm_start, cma_obj->paddr >> PAGE_SHIFT, + vma->vm_end - vma->vm_start, vma->vm_page_prot); + if (ret) + drm_gem_vm_close(vma); + + return ret; +} +EXPORT_SYMBOL_GPL(drm_gem_cma_mmap); + +/* + * drm_gem_cma_dumb_destroy - (struct drm_driver)->dumb_destroy callback function + */ +int drm_gem_cma_dumb_destroy(struct drm_file *file_priv, + struct drm_device *drm, unsigned int handle) +{ + return drm_gem_handle_delete(file_priv, handle); +} +EXPORT_SYMBOL_GPL(drm_gem_cma_dumb_destroy); diff --git a/include/drm/drm_gem_cma_helper.h b/include/drm/drm_gem_cma_helper.h new file mode 100644 index 0000000..f0f6b1a --- /dev/null +++ b/include/drm/drm_gem_cma_helper.h @@ -0,0 +1,44 @@ +#ifndef __DRM_GEM_CMA_HELPER_H__ +#define __DRM_GEM_CMA_HELPER_H__ + +struct drm_gem_cma_object { + struct drm_gem_object base; + dma_addr_t paddr; + void *vaddr; +}; + +static inline struct drm_gem_cma_object * +to_drm_gem_cma_obj(struct drm_gem_object *gem_obj) +{ + return container_of(gem_obj, struct drm_gem_cma_object, base); +} + +/* free gem object. */ +void drm_gem_cma_free_object(struct drm_gem_object *gem_obj); + +/* create memory region for drm framebuffer. */ +int drm_gem_cma_dumb_create(struct drm_file *file_priv, + struct drm_device *drm, struct drm_mode_create_dumb *args); + +/* map memory region for drm framebuffer to user space. */ +int drm_gem_cma_dumb_map_offset(struct drm_file *file_priv, + struct drm_device *drm, uint32_t handle, uint64_t *offset); + +/* set vm_flags and we can change the vm attribute to other one at here. */ +int drm_gem_cma_mmap(struct file *filp, struct vm_area_struct *vma); + +/* + * destroy memory region allocated. + * - a gem handle and physical memory region pointed by a gem object + * would be released by drm_gem_handle_delete(). + */ +int drm_gem_cma_dumb_destroy(struct drm_file *file_priv, + struct drm_device *drm, unsigned int handle); + +/* allocate physical memory. */ +struct drm_gem_cma_object *drm_gem_cma_create(struct drm_device *drm, + unsigned int size); + +extern const struct vm_operations_struct drm_gem_cma_vm_ops; + +#endif /* __DRM_GEM_CMA_HELPER_H__ */ -- 1.7.10

4 7

[Bug 41231] New: GL_INVALID_ENUM when using glxinfo -l
by bugzilla-daemon＠freedesktop.org 08 Aug '12

08 Aug '12

https://bugs.freedesktop.org/show_bug.cgi?id=41231 Summary: GL_INVALID_ENUM when using glxinfo -l Product: Mesa Version: git Platform: x86-64 (AMD64) OS/Version: Linux (All) Status: NEW Severity: minor Priority: medium Component: Drivers/Gallium/r600 AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: stefano.teso(a)gmail.com Hi, using git 455a19b1b3d7ca0a40c65d89e45209dbd8ac7b35, and running: $ glxinfo -l > /dev/null makes mesa sad: Mesa: User error: GL_INVALID_ENUM in glGetIntegerv(pname=GL_MAX_COLOR_MATRIX_STACK_DEPTH) Mesa: User error: GL_INVALID_ENUM in glGetConvolutionParameteriv This doesn't happen with stock mesa 7.11, nor when using LIBGL_ALWAYS_SOFTWARE=1. The driver is r600g, the device is: 01:00.0 VGA compatible controller: ATI Technologies Inc Robson CE [AMD Radeon HD 6300 Series] I'm using debian unstable FWIW. -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 3

[Bug 50857] New: Invalid command stream with CAICOS when ColorTiling2D enabled
by bugzilla-daemon＠freedesktop.org 08 Aug '12

08 Aug '12

https://bugs.freedesktop.org/show_bug.cgi?id=50857 Bug #: 50857 Summary: Invalid command stream with CAICOS when ColorTiling2D enabled Classification: Unclassified Product: Mesa Version: 8.0 Platform: x86-64 (AMD64) OS/Version: Linux (All) Status: NEW Severity: normal Priority: medium Component: Drivers/Gallium/r600 AssignedTo: dri-devel(a)lists.freedesktop.org ReportedBy: rankincj(a)googlemail.com This might also be a bug in Linux kernel or radeon_drv.so. Just upgraded my 64 bit HD6450 machine to Fedora 17, and gnome-shell fails immediately (login screen). Disabling ColorTiling2D in xorg.conf fixes it. ColorTiling2D worked OK - or did nothing - in Fedora 16 with radeon_drv.so from git and 3.3.7 kernel. dmesg log and lspci output are attached. -- Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.

1 18

← Newer
1
...
70
71
72
73
74
75
76
...
82
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel August 2012