dri-devel July 2018

dri-devel@lists.freedesktop.org

277 participants
1883 discussions

[PATCH] drm: Convert drm_atomic_helper_suspend/resume()
by Souptick Joarder 13 Aug '18

13 Aug '18

convert drm_atomic_helper_suspend/resume() to use drm_mode_config_helper_suspend/resume(). With this conversion, drm_fbdev_cma_set_suspend_unlocked() will left with no consumer. So this function can be removed. Signed-off-by: Souptick Joarder <jrdr.linux(a)gmail.com> Signed-off-by: Ajit Negi <ajitn.linux(a)gmail.com> --- drivers/gpu/drm/arm/hdlcd_drv.c | 26 ++------------------------ drivers/gpu/drm/drm_fb_cma_helper.c | 18 ------------------ drivers/gpu/drm/rcar-du/rcar_du_drv.c | 21 ++------------------- include/drm/drm_fb_cma_helper.h | 2 -- 4 files changed, 4 insertions(+), 63 deletions(-) diff --git a/drivers/gpu/drm/arm/hdlcd_drv.c b/drivers/gpu/drm/arm/hdlcd_drv.c index feaa8bc..4e617e0 100644 --- a/drivers/gpu/drm/arm/hdlcd_drv.c +++ b/drivers/gpu/drm/arm/hdlcd_drv.c @@ -427,37 +427,15 @@ static int hdlcd_remove(struct platform_device *pdev) static int __maybe_unused hdlcd_pm_suspend(struct device *dev) { struct drm_device *drm = dev_get_drvdata(dev); - struct hdlcd_drm_private *hdlcd = drm ? drm->dev_private : NULL; - if (!hdlcd) - return 0; - - drm_kms_helper_poll_disable(drm); - drm_fbdev_cma_set_suspend_unlocked(hdlcd->fbdev, 1); - - hdlcd->state = drm_atomic_helper_suspend(drm); - if (IS_ERR(hdlcd->state)) { - drm_fbdev_cma_set_suspend_unlocked(hdlcd->fbdev, 0); - drm_kms_helper_poll_enable(drm); - return PTR_ERR(hdlcd->state); - } - - return 0; + return drm_mode_config_helper_suspend(drm); } static int __maybe_unused hdlcd_pm_resume(struct device *dev) { struct drm_device *drm = dev_get_drvdata(dev); - struct hdlcd_drm_private *hdlcd = drm ? drm->dev_private : NULL; - - if (!hdlcd) - return 0; - drm_atomic_helper_resume(drm, hdlcd->state); - drm_fbdev_cma_set_suspend_unlocked(hdlcd->fbdev, 0); - drm_kms_helper_poll_enable(drm); - - return 0; + return drm_mode_config_helper_resume(drm); } static SIMPLE_DEV_PM_OPS(hdlcd_pm_ops, hdlcd_pm_suspend, hdlcd_pm_resume); diff --git a/drivers/gpu/drm/drm_fb_cma_helper.c b/drivers/gpu/drm/drm_fb_cma_helper.c index 186d00a..f604a84 100644 --- a/drivers/gpu/drm/drm_fb_cma_helper.c +++ b/drivers/gpu/drm/drm_fb_cma_helper.c @@ -529,21 +529,3 @@ void drm_fbdev_cma_set_suspend(struct drm_fbdev_cma *fbdev_cma, bool state) drm_fb_helper_set_suspend(&fbdev_cma->fb_helper, state); } EXPORT_SYMBOL(drm_fbdev_cma_set_suspend); - -/** - * drm_fbdev_cma_set_suspend_unlocked - wrapper around - * drm_fb_helper_set_suspend_unlocked - * @fbdev_cma: The drm_fbdev_cma struct, may be NULL - * @state: desired state, zero to resume, non-zero to suspend - * - * Calls drm_fb_helper_set_suspend, which is a wrapper around - * fb_set_suspend implemented by fbdev core. - */ -void drm_fbdev_cma_set_suspend_unlocked(struct drm_fbdev_cma *fbdev_cma, - bool state) -{ - if (fbdev_cma) - drm_fb_helper_set_suspend_unlocked(&fbdev_cma->fb_helper, - state); -} -EXPORT_SYMBOL(drm_fbdev_cma_set_suspend_unlocked); diff --git a/drivers/gpu/drm/rcar-du/rcar_du_drv.c b/drivers/gpu/drm/rcar-du/rcar_du_drv.c index 02aee6c..288220f 100644 --- a/drivers/gpu/drm/rcar-du/rcar_du_drv.c +++ b/drivers/gpu/drm/rcar-du/rcar_du_drv.c @@ -357,32 +357,15 @@ static void rcar_du_lastclose(struct drm_device *dev) static int rcar_du_pm_suspend(struct device *dev) { struct rcar_du_device *rcdu = dev_get_drvdata(dev); - struct drm_atomic_state *state; - drm_kms_helper_poll_disable(rcdu->ddev); - drm_fbdev_cma_set_suspend_unlocked(rcdu->fbdev, true); - - state = drm_atomic_helper_suspend(rcdu->ddev); - if (IS_ERR(state)) { - drm_fbdev_cma_set_suspend_unlocked(rcdu->fbdev, false); - drm_kms_helper_poll_enable(rcdu->ddev); - return PTR_ERR(state); - } - - rcdu->suspend_state = state; - - return 0; + return drm_mode_config_helper_suspend(rcdu->ddev); } static int rcar_du_pm_resume(struct device *dev) { struct rcar_du_device *rcdu = dev_get_drvdata(dev); - drm_atomic_helper_resume(rcdu->ddev, rcdu->suspend_state); - drm_fbdev_cma_set_suspend_unlocked(rcdu->fbdev, false); - drm_kms_helper_poll_enable(rcdu->ddev); - - return 0; + return drm_mode_config_helper_resume(rcdu->ddev); } #endif diff --git a/include/drm/drm_fb_cma_helper.h b/include/drm/drm_fb_cma_helper.h index d532f88a..15c4569 100644 --- a/include/drm/drm_fb_cma_helper.h +++ b/include/drm/drm_fb_cma_helper.h @@ -33,8 +33,6 @@ struct drm_fbdev_cma *drm_fbdev_cma_init(struct drm_device *dev, void drm_fbdev_cma_restore_mode(struct drm_fbdev_cma *fbdev_cma); void drm_fbdev_cma_hotplug_event(struct drm_fbdev_cma *fbdev_cma); void drm_fbdev_cma_set_suspend(struct drm_fbdev_cma *fbdev_cma, bool state); -void drm_fbdev_cma_set_suspend_unlocked(struct drm_fbdev_cma *fbdev_cma, - bool state); struct drm_gem_cma_object *drm_fb_cma_get_gem_obj(struct drm_framebuffer *fb, unsigned int plane); -- 1.9.1

3 8

[PATCH v5 10/10] drm/rcar-du/crc: Implement get_crc_sources callback
by Mahesh Kumar 13 Aug '18

13 Aug '18

This patch implements get_crc_sources callback, which returns list of all the crc sources supported by driver in current platform. Changes Since V1: - move sources list per-crtc - init sources-list only for gen3 Changes Since V2: - Adopt to driver style - Address other review comments from Laurent Pinchart Signed-off-by: Mahesh Kumar <mahesh1.kumar(a)intel.com> Cc: dri-devel(a)lists.freedesktop.org Cc: Laurent Pinchart <laurent.pinchart(a)ideasonboard.com> --- drivers/gpu/drm/rcar-du/rcar_du_crtc.c | 85 +++++++++++++++++++++++++++++++++- drivers/gpu/drm/rcar-du/rcar_du_crtc.h | 3 ++ 2 files changed, 87 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/rcar-du/rcar_du_crtc.c b/drivers/gpu/drm/rcar-du/rcar_du_crtc.c index 43e67cffdee0..39981ce422e1 100644 --- a/drivers/gpu/drm/rcar-du/rcar_du_crtc.c +++ b/drivers/gpu/drm/rcar-du/rcar_du_crtc.c @@ -691,6 +691,68 @@ static const struct drm_crtc_helper_funcs crtc_helper_funcs = { .atomic_disable = rcar_du_crtc_atomic_disable, }; +static void rcar_du_crtc_crc_sources_list_init(struct rcar_du_crtc *rcrtc) +{ + struct rcar_du_device *rcdu = rcrtc->group->dev; + const char **sources; + unsigned int count; + int i = -1; + + /* CRC available only on Gen3 HW. */ + if (rcdu->info->gen < 3) + return; + + /* Reserve 1 for "auto" source. */ + count = rcrtc->vsp->num_planes + 1; + + sources = kmalloc_array(count, sizeof(*sources), GFP_KERNEL); + if (!sources) + return; + + sources[0] = kstrdup("auto", GFP_KERNEL); + if (!sources[0]) + goto error; + + for (i = 0; i < rcrtc->vsp->num_planes; ++i) { + struct drm_plane *plane = &rcrtc->vsp->planes[i].plane; + char name[16]; + + sprintf(name, "plane%u", plane->base.id); + sources[i + 1] = kstrdup(name, GFP_KERNEL); + if (!sources[i + 1]) + goto error; + } + + rcrtc->sources = sources; + rcrtc->sources_count = count; + return; + +error: + while (i >= 0) { + kfree(sources[i]); + i--; + } + kfree(sources); + + rcrtc->sources = NULL; + rcrtc->sources_count = 0; +} + +static void rcar_du_crtc_crc_sources_list_uninit(struct rcar_du_crtc *rcrtc) +{ + unsigned int i; + + if (!rcrtc->sources) + return; + + for (i = 0; i < rcrtc->sources_count; i++) + kfree(rcrtc->sources[i]); + kfree(rcrtc->sources); + + rcrtc->sources = NULL; + rcrtc->sources_count = 0; +} + static struct drm_crtc_state * rcar_du_crtc_atomic_duplicate_state(struct drm_crtc *crtc) { @@ -717,6 +779,15 @@ static void rcar_du_crtc_atomic_destroy_state(struct drm_crtc *crtc, kfree(to_rcar_crtc_state(state)); } +static void rcar_du_crtc_cleanup(struct drm_crtc *crtc) +{ + struct rcar_du_crtc *rcrtc = to_rcar_crtc(crtc); + + rcar_du_crtc_crc_sources_list_uninit(rcrtc); + + return drm_crtc_cleanup(crtc); +} + static void rcar_du_crtc_reset(struct drm_crtc *crtc) { struct rcar_du_crtc_state *state; @@ -809,6 +880,15 @@ static int rcar_du_crtc_verify_crc_source(struct drm_crtc *crtc, return 0; } +const char *const *rcar_du_crtc_get_crc_sources(struct drm_crtc *crtc, + size_t *count) +{ + struct rcar_du_crtc *rcrtc = to_rcar_crtc(crtc); + + *count = rcrtc->sources_count; + return rcrtc->sources; +} + static int rcar_du_crtc_set_crc_source(struct drm_crtc *crtc, const char *source_name) { @@ -879,7 +959,7 @@ static const struct drm_crtc_funcs crtc_funcs_gen2 = { static const struct drm_crtc_funcs crtc_funcs_gen3 = { .reset = rcar_du_crtc_reset, - .destroy = drm_crtc_cleanup, + .destroy = rcar_du_crtc_cleanup, .set_config = drm_atomic_helper_set_config, .page_flip = drm_atomic_helper_page_flip, .atomic_duplicate_state = rcar_du_crtc_atomic_duplicate_state, @@ -888,6 +968,7 @@ static const struct drm_crtc_funcs crtc_funcs_gen3 = { .disable_vblank = rcar_du_crtc_disable_vblank, .set_crc_source = rcar_du_crtc_set_crc_source, .verify_crc_source = rcar_du_crtc_verify_crc_source, + .get_crc_sources = rcar_du_crtc_get_crc_sources, }; /* ----------------------------------------------------------------------------- @@ -1026,5 +1107,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int swindex, return ret; } + rcar_du_crtc_crc_sources_list_init(rcrtc); + return 0; } diff --git a/drivers/gpu/drm/rcar-du/rcar_du_crtc.h b/drivers/gpu/drm/rcar-du/rcar_du_crtc.h index 7680cb2636c8..592c79993e08 100644 --- a/drivers/gpu/drm/rcar-du/rcar_du_crtc.h +++ b/drivers/gpu/drm/rcar-du/rcar_du_crtc.h @@ -67,6 +67,9 @@ struct rcar_du_crtc { struct rcar_du_group *group; struct rcar_du_vsp *vsp; unsigned int vsp_pipe; + + const char *const *sources; + unsigned int sources_count; }; #define to_rcar_crtc(c) container_of(c, struct rcar_du_crtc, crtc) -- 2.16.2

3 3

[PATCH 00/21] omapdrm: Rework the HPD-related operations
by Laurent Pinchart 13 Aug '18

13 Aug '18

Hello, This patch series reworks all the HPD-related operations (detection, EDID read, HPD callback (un)registration and HPD enable/disable) as a step toward moving from omap_dss_device to drm_bridge. All HPD-related operations are called by the omapdrm driver on the omap_dss_device at the end of display pipeline, and the operations are then handled directly or forwarded to the previous omap_dss_device in the pipeline. This causes an issue in our quest to move to drm_bridge: there are currently no HPD-related operations in the drm_bridge API, as they are implemented internally by bridge drivers for the drm_connector created by the bridges. To solve this, we will need to extend the drm_bridge API with HPD-related operations. This patch series is a prototype of such an API, still local to omap_dss_device for now. HPD operations can be implemented by multiple components in a display pipeline. For instance, the DDC bus of an HDMI connector can be wired to an I2C bus of the SoC, or to the DDC pins of an HDMI encoder. In the first case the HDMI connector component will provide EDID access, while in the second case EDID will be read through the HDMI encoder component. Both components will thus implement the EDID read operation, but on a particular system only one of them will be able to provide the feature. Determining which component provides a feature at runtime is thus required, and is currently performed through recursive calls. This requires all components to cooperate in a way that can be implemented locally in the omapdrm driver but would be too complex for the more generic drm_bridge API. Our solution is to use flags in the omap_dss_device structure to tell, at runtime, which features are available. The top-level code can then locate the component providing a particular feature and then use that component directly. The series starts with removal of dead or unneeded code in patches 01/21 to 03/21. Patch 04/21 prepares for the merge of the omap_dss_driver and omap_dss_device_ops structures. Patches then 05/21 to 10/21 convert the driver from the GPIO API to the GPIO descriptors API (drive-by cleanup). Patch 11/21 is where the real refactoring starts. It moves most operations from the omap_dss_driver structure to the omap_dss_device_ops structure in order to simplify code iterating over pipelines. Patch 12/21 defines the operation flags, and patches 13/21 and 14/21 make use of them to rework the .detect(), .register_hpd_cb() and .unregister_hpd_cb() operations. Patches 15/21 and 16/21 remove the unneeded .enable_hpd() and .disable_hpd() operations. Patch 17/21 moves DSS-specific HPD-related code from omap_dss_device instances to the omapdrm driver, and patch 18/21 reworks the .read_edid() operation. Patch 19/21 simplifies the component handling in the CRTC mode set code, and patches 20/21 and 21/21 finally rework the .set_hdmi_mode() and .set_infoframe() operations. Laurent Pinchart (21): drm/omap: dss: Remove unused omap_dss_driver operations drm/omap: dss: Remove omap_dss_driver .[gs]et_mirror operations drm/omap: Remove unnecessary display output sanity checks drm/omap: Check omap_dss_device type based on the output_type field drm/omap: connector-hdmi: Convert to the GPIO descriptors API drm/omap: encoder-tfp410: Convert to the GPIO descriptors API drm/omap: panel-nec-nl8048hl11: Convert to the GPIO descriptors API drm/omap: panel-sony-acx565akm: Convert to the GPIO descriptors API drm/omap: panel-tpo-td028ttec1: Drop unneeded linux/gpio.h header drm/omap: panel-tpo-td043mtea1: Convert to the GPIO descriptors API drm/omap: Move most omap_dss_driver operations to omap_dss_device_ops drm/omap: dss: Add device operations flags drm/omap: Don't call .detect() operation recursively drm/omap: Don't call HPD registration operations recursively drm/omap: Remove unneeded safety checks in the HPD operations drm/omap: Merge HPD enable operation with HPD callback registration drm/omap: Move HPD disconnection handling to omap_connector drm/omap: Don't call EDID read operation recursively drm/omap: Get from CRTC to display device directly drm/omap: Pass both output and display omap_dss_device to encoder init drm/omap: Don't call HDMI mode and infoframe operations recursively .../gpu/drm/omapdrm/displays/connector-analog-tv.c | 4 +- drivers/gpu/drm/omapdrm/displays/connector-dvi.c | 52 +--- drivers/gpu/drm/omapdrm/displays/connector-hdmi.c | 160 +++---------- drivers/gpu/drm/omapdrm/displays/encoder-tfp410.c | 51 +--- .../gpu/drm/omapdrm/displays/encoder-tpd12s015.c | 79 +----- drivers/gpu/drm/omapdrm/displays/panel-dpi.c | 4 +- drivers/gpu/drm/omapdrm/displays/panel-dsi-cm.c | 12 +- .../omapdrm/displays/panel-lgphilips-lb035q02.c | 4 +- .../drm/omapdrm/displays/panel-nec-nl8048hl11.c | 58 +---- .../drm/omapdrm/displays/panel-sharp-ls037v7dw01.c | 4 +- .../drm/omapdrm/displays/panel-sony-acx565akm.c | 60 ++--- .../drm/omapdrm/displays/panel-tpo-td028ttec1.c | 5 +- .../drm/omapdrm/displays/panel-tpo-td043mtea1.c | 81 ++----- drivers/gpu/drm/omapdrm/dss/base.c | 16 +- drivers/gpu/drm/omapdrm/dss/dss.c | 5 +- drivers/gpu/drm/omapdrm/dss/hdmi4.c | 4 +- drivers/gpu/drm/omapdrm/dss/hdmi5.c | 4 +- drivers/gpu/drm/omapdrm/dss/omapdss.h | 81 +++---- drivers/gpu/drm/omapdrm/omap_connector.c | 265 +++++++++++++-------- drivers/gpu/drm/omapdrm/omap_connector.h | 2 + drivers/gpu/drm/omapdrm/omap_crtc.c | 25 +- drivers/gpu/drm/omapdrm/omap_drv.c | 28 +-- drivers/gpu/drm/omapdrm/omap_encoder.c | 49 ++-- drivers/gpu/drm/omapdrm/omap_encoder.h | 6 +- 24 files changed, 386 insertions(+), 673 deletions(-) -- Regards, Laurent Pinchart

3 43

[PATCH 1/2] drm/scheduler: only kill entity if last user is killed v2
by Christian König 10 Aug '18

10 Aug '18

Note which task is using the entity and only kill it if the last user of the entity is killed. This should prevent problems when entities are leaked to child processes. v2: add missing kernel doc Signed-off-by: Christian König <christian.koenig(a)amd.com> --- drivers/gpu/drm/scheduler/gpu_scheduler.c | 6 +++++- include/drm/gpu_scheduler.h | 2 ++ 2 files changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/scheduler/gpu_scheduler.c b/drivers/gpu/drm/scheduler/gpu_scheduler.c index 3f2fc5e8242a..f563e4fbb4b6 100644 --- a/drivers/gpu/drm/scheduler/gpu_scheduler.c +++ b/drivers/gpu/drm/scheduler/gpu_scheduler.c @@ -275,6 +275,7 @@ static void drm_sched_entity_kill_jobs_cb(struct dma_fence *f, long drm_sched_entity_flush(struct drm_sched_entity *entity, long timeout) { struct drm_gpu_scheduler *sched; + struct task_struct *last_user; long ret = timeout; sched = entity->rq->sched; @@ -295,7 +296,9 @@ long drm_sched_entity_flush(struct drm_sched_entity *entity, long timeout) /* For killed process disable any more IBs enqueue right now */ - if ((current->flags & PF_EXITING) && (current->exit_code == SIGKILL)) + last_user = cmpxchg(&entity->last_user, current->group_leader, NULL); + if ((!last_user || last_user == current->group_leader) && + (current->flags & PF_EXITING) && (current->exit_code == SIGKILL)) drm_sched_entity_set_rq(entity, NULL); return ret; @@ -541,6 +544,7 @@ void drm_sched_entity_push_job(struct drm_sched_job *sched_job, trace_drm_sched_job(sched_job, entity); + WRITE_ONCE(entity->last_user, current->group_leader); first = spsc_queue_push(&entity->job_queue, &sched_job->queue_node); /* first job wakes up scheduler */ diff --git a/include/drm/gpu_scheduler.h b/include/drm/gpu_scheduler.h index 091b9afcd184..21c648b0b2a1 100644 --- a/include/drm/gpu_scheduler.h +++ b/include/drm/gpu_scheduler.h @@ -66,6 +66,7 @@ enum drm_sched_priority { * @guilty: points to ctx's guilty. * @fini_status: contains the exit status in case the process was signalled. * @last_scheduled: points to the finished fence of the last scheduled job. + * @last_user: last group leader pushing a job into the entity. * * Entities will emit jobs in order to their corresponding hardware * ring, and the scheduler will alternate between entities based on @@ -85,6 +86,7 @@ struct drm_sched_entity { struct dma_fence_cb cb; atomic_t *guilty; struct dma_fence *last_scheduled; + struct task_struct *last_user; }; /** -- 2.14.1

4 25

[PATCH] drm: qxl: Fix error handling at qxl_device_init
by Anton Vasilyev 10 Aug '18

10 Aug '18

If qxl_device_init fails on creating resources and does not report it, then qxl module will catch null pointer exception on remove, or on probe's error path. The patch adds error path with resources release into qxl_device_init. Found by Linux Driver Verification project (linuxtesting.org). Signed-off-by: Anton Vasilyev <vasilyev(a)ispras.ru> --- drivers/gpu/drm/qxl/qxl_kms.c | 80 ++++++++++++++++++++++++++++++++--- 1 file changed, 73 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/qxl/qxl_kms.c b/drivers/gpu/drm/qxl/qxl_kms.c index 771250aed78d..e25c589d5f50 100644 --- a/drivers/gpu/drm/qxl/qxl_kms.c +++ b/drivers/gpu/drm/qxl/qxl_kms.c @@ -102,8 +102,10 @@ int qxl_device_init(struct qxl_device *qdev, int r, sb; r = drm_dev_init(&qdev->ddev, drv, &pdev->dev); - if (r) - return r; + if (r) { + pr_err("Unable to init drm dev"); + goto error; + } qdev->ddev.pdev = pdev; pci_set_drvdata(pdev, &qdev->ddev); @@ -121,6 +123,11 @@ int qxl_device_init(struct qxl_device *qdev, qdev->io_base = pci_resource_start(pdev, 3); qdev->vram_mapping = io_mapping_create_wc(qdev->vram_base, pci_resource_len(pdev, 0)); + if (!qdev->vram_mapping) { + pr_err("Unable to create vram_mapping"); + r = -ENOMEM; + goto error; + } if (pci_resource_len(pdev, 4) > 0) { /* 64bit surface bar present */ @@ -139,6 +146,11 @@ int qxl_device_init(struct qxl_device *qdev, qdev->surface_mapping = io_mapping_create_wc(qdev->surfaceram_base, qdev->surfaceram_size); + if (!qdev->surface_mapping) { + pr_err("Unable to create surface_mapping"); + r = -ENOMEM; + goto vram_mapping_free; + } } DRM_DEBUG_KMS("qxl: vram %llx-%llx(%dM %dk), surface %llx-%llx(%dM %dk, %s)\n", @@ -155,20 +167,29 @@ int qxl_device_init(struct qxl_device *qdev, qdev->rom = ioremap(qdev->rom_base, qdev->rom_size); if (!qdev->rom) { pr_err("Unable to ioremap ROM\n"); - return -ENOMEM; + r = -ENOMEM; + goto surface_mapping_free; } - qxl_check_device(qdev); + if (!qxl_check_device(qdev)) { + r = -ENODEV; + goto surface_mapping_free; + } r = qxl_bo_init(qdev); if (r) { DRM_ERROR("bo init failed %d\n", r); - return r; + goto rom_unmap; } qdev->ram_header = ioremap(qdev->vram_base + qdev->rom->ram_header_offset, sizeof(*qdev->ram_header)); + if (!qdev->ram_header) { + DRM_ERROR("Unable to ioremap RAM header\n"); + r = -ENOMEM; + goto bo_fini; + } qdev->command_ring = qxl_ring_create(&(qdev->ram_header->cmd_ring_hdr), sizeof(struct qxl_command), @@ -176,6 +197,11 @@ int qxl_device_init(struct qxl_device *qdev, qdev->io_base + QXL_IO_NOTIFY_CMD, false, &qdev->display_event); + if (!qdev->command_ring) { + DRM_ERROR("Unable to create command ring\n"); + r = -ENOMEM; + goto ram_header_unmap; + } qdev->cursor_ring = qxl_ring_create( &(qdev->ram_header->cursor_ring_hdr), @@ -185,12 +211,23 @@ int qxl_device_init(struct qxl_device *qdev, false, &qdev->cursor_event); + if (!qdev->cursor_ring) { + DRM_ERROR("Unable to create cursor ring\n"); + r = -ENOMEM; + goto command_ring_free; + } + qdev->release_ring = qxl_ring_create( &(qdev->ram_header->release_ring_hdr), sizeof(uint64_t), QXL_RELEASE_RING_SIZE, 0, true, NULL); + if (!qdev->release_ring) { + DRM_ERROR("Unable to create release ring\n"); + r = -ENOMEM; + goto cursor_ring_free; + } /* TODO - slot initialization should happen on reset. where is our * reset handler? */ qdev->n_mem_slots = qdev->rom->slots_end; @@ -203,6 +240,12 @@ int qxl_device_init(struct qxl_device *qdev, kmalloc_array(qdev->n_mem_slots, sizeof(struct qxl_memslot), GFP_KERNEL); + if (!qdev->mem_slots) { + DRM_ERROR("Unable to alloc mem slots\n"); + r = -ENOMEM; + goto release_ring_free; + } + idr_init(&qdev->release_idr); spin_lock_init(&qdev->release_idr_lock); spin_lock_init(&qdev->release_lock); @@ -218,8 +261,10 @@ int qxl_device_init(struct qxl_device *qdev, /* must initialize irq before first async io - slot creation */ r = qxl_irq_init(qdev); - if (r) - return r; + if (r) { + DRM_ERROR("Unable to init qxl irq\n"); + goto mem_slots_free; + } /* * Note that virtual is surface0. We rely on the single ioremap done @@ -243,6 +288,27 @@ int qxl_device_init(struct qxl_device *qdev, INIT_WORK(&qdev->gc_work, qxl_gc_work); return 0; + +mem_slots_free: + kfree(qdev->mem_slots); +release_ring_free: + qxl_ring_free(qdev->release_ring); +cursor_ring_free: + qxl_ring_free(qdev->cursor_ring); +command_ring_free: + qxl_ring_free(qdev->command_ring); +ram_header_unmap: + iounmap(qdev->ram_header); +bo_fini: + qxl_bo_fini(qdev); +rom_unmap: + iounmap(qdev->rom); +surface_mapping_free: + io_mapping_free(qdev->surface_mapping); +vram_mapping_free: + io_mapping_free(qdev->vram_mapping); +error: + return r; } void qxl_device_fini(struct qxl_device *qdev) -- 2.18.0

2 1

[PATCH] drm/cirrus: Use drm_framebuffer_put to avoid kernel oops in clean-up
by Thomas Zimmermann 10 Aug '18

10 Aug '18

In the Cirrus driver, the regular clean-up code also performs the clean-up of a failed initialization. If the fbdev's framebuffer was not initialized, the clean-up will fail within drm_framebuffer_unregister_private. Booting with cirrus.bpp=16 triggers this bug. The framebuffer is currently stored directly within struct cirrus_fbdev. To fix the bug, we turn it into a pointer that is only set for initialized framebuffers. The fbdev's clean-up code skips uninitialized framebuffers. The memory for struct drm_framebuffer is allocated dynamically. This requires additional error handling within cirrusfb_create. The framebuffer clean-up is now performed by drm_framebuffer_put, which also frees the data strcuture's memory. Link: https://bugzilla.suse.com/show_bug.cgi?id=1101822 Signed-off-by: Thomas Zimmermann <tzimmermann(a)suse.de> --- drivers/gpu/drm/cirrus/cirrus_drv.h | 2 +- drivers/gpu/drm/cirrus/cirrus_fbdev.c | 48 +++++++++++++++------------ drivers/gpu/drm/cirrus/cirrus_mode.c | 2 +- 3 files changed, 29 insertions(+), 23 deletions(-) diff --git a/drivers/gpu/drm/cirrus/cirrus_drv.h b/drivers/gpu/drm/cirrus/cirrus_drv.h index ce9db7aab225..a29f87e98d9d 100644 --- a/drivers/gpu/drm/cirrus/cirrus_drv.h +++ b/drivers/gpu/drm/cirrus/cirrus_drv.h @@ -146,7 +146,7 @@ struct cirrus_device { struct cirrus_fbdev { struct drm_fb_helper helper; - struct drm_framebuffer gfb; + struct drm_framebuffer *gfb; void *sysram; int size; int x1, y1, x2, y2; /* dirty rect */ diff --git a/drivers/gpu/drm/cirrus/cirrus_fbdev.c b/drivers/gpu/drm/cirrus/cirrus_fbdev.c index b643ac92801c..82cc82e0bd80 100644 --- a/drivers/gpu/drm/cirrus/cirrus_fbdev.c +++ b/drivers/gpu/drm/cirrus/cirrus_fbdev.c @@ -22,14 +22,14 @@ static void cirrus_dirty_update(struct cirrus_fbdev *afbdev, struct drm_gem_object *obj; struct cirrus_bo *bo; int src_offset, dst_offset; - int bpp = afbdev->gfb.format->cpp[0]; + int bpp = afbdev->gfb->format->cpp[0]; int ret = -EBUSY; bool unmap = false; bool store_for_later = false; int x2, y2; unsigned long flags; - obj = afbdev->gfb.obj[0]; + obj = afbdev->gfb->obj[0]; bo = gem_to_cirrus_bo(obj); /* @@ -82,7 +82,7 @@ static void cirrus_dirty_update(struct cirrus_fbdev *afbdev, } for (i = y; i < y + height; i++) { /* assume equal stride for now */ - src_offset = dst_offset = i * afbdev->gfb.pitches[0] + (x * bpp); + src_offset = dst_offset = i * afbdev->gfb->pitches[0] + (x * bpp); memcpy_toio(bo->kmap.virtual + src_offset, afbdev->sysram + src_offset, width * bpp); } @@ -192,23 +192,26 @@ static int cirrusfb_create(struct drm_fb_helper *helper, return -ENOMEM; info = drm_fb_helper_alloc_fbi(helper); - if (IS_ERR(info)) - return PTR_ERR(info); + if (IS_ERR(info)) { + ret = PTR_ERR(info); + goto err_vfree; + } info->par = gfbdev; - ret = cirrus_framebuffer_init(cdev->dev, &gfbdev->gfb, &mode_cmd, gobj); + fb = kzalloc(sizeof(*fb), GFP_KERNEL); + if (!fb) { + ret = -ENOMEM; + goto err_drm_gem_object_put_unlocked; + } + + ret = cirrus_framebuffer_init(cdev->dev, fb, &mode_cmd, gobj); if (ret) - return ret; + goto err_kfree; gfbdev->sysram = sysram; gfbdev->size = size; - - fb = &gfbdev->gfb; - if (!fb) { - DRM_INFO("fb is NULL\n"); - return -EINVAL; - } + gfbdev->gfb = fb; /* setup helper */ gfbdev->helper.fb = fb; @@ -241,24 +244,27 @@ static int cirrusfb_create(struct drm_fb_helper *helper, DRM_INFO(" pitch is %d\n", fb->pitches[0]); return 0; + +err_kfree: + kfree(fb); +err_drm_gem_object_put_unlocked: + drm_gem_object_put_unlocked(gobj); +err_vfree: + vfree(sysram); + return ret; } static int cirrus_fbdev_destroy(struct drm_device *dev, struct cirrus_fbdev *gfbdev) { - struct drm_framebuffer *gfb = &gfbdev->gfb; + struct drm_framebuffer *gfb = gfbdev->gfb; drm_fb_helper_unregister_fbi(&gfbdev->helper); - if (gfb->obj[0]) { - drm_gem_object_put_unlocked(gfb->obj[0]); - gfb->obj[0] = NULL; - } - vfree(gfbdev->sysram); drm_fb_helper_fini(&gfbdev->helper); - drm_framebuffer_unregister_private(gfb); - drm_framebuffer_cleanup(gfb); + if (gfb) + drm_framebuffer_put(gfb); return 0; } diff --git a/drivers/gpu/drm/cirrus/cirrus_mode.c b/drivers/gpu/drm/cirrus/cirrus_mode.c index 336bfda40125..90a4e641d3fb 100644 --- a/drivers/gpu/drm/cirrus/cirrus_mode.c +++ b/drivers/gpu/drm/cirrus/cirrus_mode.c @@ -127,7 +127,7 @@ static int cirrus_crtc_do_set_base(struct drm_crtc *crtc, return ret; } - if (&cdev->mode_info.gfbdev->gfb == crtc->primary->fb) { + if (cdev->mode_info.gfbdev->gfb == crtc->primary->fb) { /* if pushing console in kmap it */ ret = ttm_bo_kmap(&bo->bo, 0, bo->bo.num_pages, &bo->kmap); if (ret) -- 2.18.0

2 1

[PATCH 0/8] Add support for drm/rockchip to dynamically control the DDR frequency.
by Enric Balletbo i Serra 09 Aug '18

09 Aug '18

Dear all, The rk3399 platform has a DFI controller that can monitor DDR load and a DMC driver that talks with the TF-A (Trusted Firmware-A) to dynamically set the DDR frequency with following flow. kernel Trusted Firmware-A (bl31) monitor ddr load | | get_target_rate | | pass rate to TF-A clk_set_rate(ddr) --------------------->run ddr dvs flow | | <------------------------------end ddr dvs flow | | return These patches add support for devfreq to dynamically control the DDR frequency into the drm rockchip driver. By default it uses the 'simple_ondemand' governor which can adjust the frequency based on the DDR load. Waiting for your feedback. Best regards, Enric Changes in v1: - [RFC 1/10] Add Acked-by: Chanwoo Choi <cw00.choi(a)samsung.com> - [RFC 1/10] s/Generic/General/ (Robin Murphy) - [RFC 2/10] Add reviewed and acked tags from Chanwoo Choi and Rob Herring - [RFC 3/10] Add an explanation for platform SIP calls. - [RFC 3/10] Change if statement for a switch. - [RFC 3/10] Rename ddr_flag to odt_enable to be more clear. - [RFC 4/10] Removed from the series. I did not found a use case where not holding the mutex causes the issue. - [RFC 7/10] Removed from the series. I did not found a use case where this matters. - [RFC 8/10] Move rk3399-dram.h to dt-includes. - [RFC 8/10] Put sdram default values under the dmc node. - [RFC 8/10] Removed rk3399-dram-default-timing.dts Derek Basehore (2): devfreq: rk3399_dmc / clk: rockchip: Sync with vblank in the kernel for DDRfreq. devfreq: rk3399_dmc / clk: rockchip: Disable DDR clk timeout on suspend. Enric Balletbo i Serra (3): devfreq: rockchip-dfi: Move GRF definitions to a common place. dt-bindings: devfreq: rk3399_dmc: Add rockchip,pmu phandle. devfreq: rk3399_dmc: Pass ODT and auto power down parameters to TF-A. Lin Huang (2): arm64: dts: rk3399: Add dfi and dmc nodes. arm64: dts: rockchip: Enable dmc and dfi nodes on gru. Sean Paul (1): drm: rockchip: Add DDR devfreq support. .../bindings/devfreq/rk3399_dmc.txt | 2 + arch/arm64/boot/dts/rockchip/rk3399-gru.dtsi | 21 ++ .../boot/dts/rockchip/rk3399-op1-opp.dtsi | 29 +++ arch/arm64/boot/dts/rockchip/rk3399.dtsi | 51 ++++- drivers/clk/rockchip/clk-ddr.c | 157 +++++++++++-- drivers/clk/rockchip/clk.c | 2 +- drivers/clk/rockchip/clk.h | 3 +- drivers/devfreq/event/rockchip-dfi.c | 23 +- drivers/devfreq/rk3399_dmc.c | 209 +++++++++++++++--- drivers/devfreq/rk3399_dmc_priv.h | 16 ++ drivers/gpu/drm/rockchip/rockchip_drm_drv.c | 46 ++++ drivers/gpu/drm/rockchip/rockchip_drm_drv.h | 9 + drivers/gpu/drm/rockchip/rockchip_drm_fb.c | 35 +++ drivers/gpu/drm/rockchip/rockchip_drm_vop.c | 34 +++ include/dt-bindings/power/rk3399-dram.h | 73 ++++++ include/soc/rockchip/rk3399_dmc.h | 42 ++++ include/soc/rockchip/rk3399_grf.h | 21 ++ include/soc/rockchip/rockchip_sip.h | 1 + 18 files changed, 704 insertions(+), 70 deletions(-) create mode 100644 drivers/devfreq/rk3399_dmc_priv.h create mode 100644 include/dt-bindings/power/rk3399-dram.h create mode 100644 include/soc/rockchip/rk3399_dmc.h create mode 100644 include/soc/rockchip/rk3399_grf.h -- 2.18.0

3 13

[PATCH] drm/panel: simple: fix -Wliteral-conversion warning
by Nick Desaulniers 09 Aug '18

09 Aug '18

Fixes commit 8cfe83419cdb ("drm/panel: simple: Add support for KEO TX31D200VM0BAA") drivers/gpu/drm/panel/panel-simple.c:1250:27: warning: implicit conversion from 'double' to 'u32' (aka 'unsigned int') changes value from 33.5 to 33 [-Wliteral-conversion] .vfront_porch = { 6, 21, 33.5 }, ~ ^~~~ drivers/gpu/drm/panel/panel-simple.c:1251:26: warning: implicit conversion from 'double' to 'u32' (aka 'unsigned int') changes value from 33.5 to 33 [-Wliteral-conversion] .vback_porch = { 6, 21, 33.5 }, ~ ^~~~ Signed-off-by: Nick Desaulniers <ndesaulniers(a)google.com> --- Alternatively, should these be rounded up to 34? I'm guessing the current behaviour (truncation) is correct since that's how the patch was operating. drivers/gpu/drm/panel/panel-simple.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/panel/panel-simple.c b/drivers/gpu/drm/panel/panel-simple.c index cbf1ab404ee7..12bcbd1dd77b 100644 --- a/drivers/gpu/drm/panel/panel-simple.c +++ b/drivers/gpu/drm/panel/panel-simple.c @@ -1247,8 +1247,8 @@ static const struct display_timing koe_tx31d200vm0baa_timing = { .hback_porch = { 16, 36, 56 }, .hsync_len = { 8, 8, 8 }, .vactive = { 480, 480, 480 }, - .vfront_porch = { 6, 21, 33.5 }, - .vback_porch = { 6, 21, 33.5 }, + .vfront_porch = { 6, 21, 33 }, + .vback_porch = { 6, 21, 33 }, .vsync_len = { 8, 8, 8 }, .flags = DISPLAY_FLAGS_DE_HIGH, }; -- 2.18.0.rc1.244.gcf134e6275-goog

2 3

[PATCH] drm/qxl: Replace ttm_bo_unref with ttm_bo_put
by Thomas Zimmermann 09 Aug '18

09 Aug '18

The function ttm_bo_put releases a reference to a TTM buffer object. The function's name is more aligned to the Linux kernel convention of naming ref-counting function _get and _put. A call to ttm_bo_unref takes the address of the TTM BO object's pointer and clears the pointer's value to NULL. This is not necessary in most cases and sometimes even worked around by the calling code. A call to ttm_bo_put only releases the reference without clearing the pointer. The current behaviour of cleaning the pointer is kept in the calling code, but should be removed if not required in a later patch. Signed-off-by: Thomas Zimmermann <tzimmermann(a)suse.de> --- drivers/gpu/drm/qxl/qxl_gem.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/qxl/qxl_gem.c b/drivers/gpu/drm/qxl/qxl_gem.c index f5c1e7872e92..89606c819d82 100644 --- a/drivers/gpu/drm/qxl/qxl_gem.c +++ b/drivers/gpu/drm/qxl/qxl_gem.c @@ -40,7 +40,7 @@ void qxl_gem_object_free(struct drm_gem_object *gobj) qxl_surface_evict(qdev, qobj, false); tbo = &qobj->tbo; - ttm_bo_unref(&tbo); + ttm_bo_put(tbo); } int qxl_gem_object_create(struct qxl_device *qdev, int size, -- 2.18.0

2 1

[PATCH] drm/cirrus: Replace ttm_bo_unref with ttm_bo_put
by Thomas Zimmermann 09 Aug '18

09 Aug '18

The function ttm_bo_put releases a reference to a TTM buffer object. The function's name is more aligned to the Linux kernel convention of naming ref-counting function _get and _put. A call to ttm_bo_unref takes the address of the TTM BO object's pointer and clears the pointer's value to NULL. This is not necessary in most cases and sometimes even worked around by the calling code. A call to ttm_bo_put only releases the reference without clearing the pointer. The current behaviour of cleaning the pointer is kept in the calling code, but should be removed if not required in a later patch. Signed-off-by: Thomas Zimmermann <tzimmermann(a)suse.de> --- drivers/gpu/drm/cirrus/cirrus_main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/cirrus/cirrus_main.c b/drivers/gpu/drm/cirrus/cirrus_main.c index 60d54e10a34d..57f8fe6d020b 100644 --- a/drivers/gpu/drm/cirrus/cirrus_main.c +++ b/drivers/gpu/drm/cirrus/cirrus_main.c @@ -269,7 +269,7 @@ static void cirrus_bo_unref(struct cirrus_bo **bo) return; tbo = &((*bo)->bo); - ttm_bo_unref(&tbo); + ttm_bo_put(tbo); *bo = NULL; } -- 2.18.0

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel July 2018