dri-devel May 2013

dri-devel@lists.freedesktop.org

116 participants
705 discussions

Radeon lockup on 3.8.5-201.fc18.x86_64
by Andy Lutomirski 18 Jun '13

18 Jun '13

Every day or so, I'll click something and my screens go blank for a second or two. dmesg complains about a lockup, and afterwards everything is painfully slow. (Even switching focus to other emacs windows takes a second or two.) Once this happens, if I restart X, I get a blank screen, although the mouse still works and I can switch VTs and use the console. My userspace driver is xorg-x11-drv-ati-7.0.0-0.9.20121015gitbd9e2c064.fc18.src.rpm. I have Option "ColorTiling2D" "true", but I'm not sure this matters. lspci says: 08:00.0 VGA compatible controller: Advanced Micro Devices [AMD] nee ATI Caicos [Radeon HD 6450] [ 7508.750753] radeon 0000:08:00.0: GPU lockup CP stall for more than 5102593msec [ 7508.750757] radeon 0000:08:00.0: GPU lockup (waiting for 0x0000000000000155 last fence id 0x0000000000000154) [ 7508.751832] radeon 0000:08:00.0: Saved 71 dwords of commands on ring 0. [ 7508.751837] radeon 0000:08:00.0: GPU softreset: 0x00000007 [ 7508.762035] radeon 0000:08:00.0: GRBM_STATUS = 0xA0003828 [ 7508.762037] radeon 0000:08:00.0: GRBM_STATUS_SE0 = 0x00000007 [ 7508.762040] radeon 0000:08:00.0: GRBM_STATUS_SE1 = 0x00000007 [ 7508.762042] radeon 0000:08:00.0: SRBM_STATUS = 0x200400C0 [ 7508.762045] radeon 0000:08:00.0: R_008674_CP_STALLED_STAT1 = 0x00000000 [ 7508.762047] radeon 0000:08:00.0: R_008678_CP_STALLED_STAT2 = 0x00000000 [ 7508.762050] radeon 0000:08:00.0: R_00867C_CP_BUSY_STAT = 0x00000004 [ 7508.762053] radeon 0000:08:00.0: R_008680_CP_STAT = 0x80008647 [ 7508.762054] radeon 0000:08:00.0: GRBM_SOFT_RESET=0x00007F6B [ 7508.762109] radeon 0000:08:00.0: GRBM_STATUS = 0x00003828 [ 7508.762112] radeon 0000:08:00.0: GRBM_STATUS_SE0 = 0x00000007 [ 7508.762114] radeon 0000:08:00.0: GRBM_STATUS_SE1 = 0x00000007 [ 7508.762117] radeon 0000:08:00.0: SRBM_STATUS = 0x200400C0 [ 7508.762120] radeon 0000:08:00.0: R_008674_CP_STALLED_STAT1 = 0x00000000 [ 7508.762122] radeon 0000:08:00.0: R_008678_CP_STALLED_STAT2 = 0x00000000 [ 7508.762125] radeon 0000:08:00.0: R_00867C_CP_BUSY_STAT = 0x00000000 [ 7508.762127] radeon 0000:08:00.0: R_008680_CP_STAT = 0x00000000 [ 7508.762131] radeon 0000:08:00.0: R_00D034_DMA_STATUS_REG = 0x4448C156 [ 7508.762187] radeon 0000:08:00.0: R_00D034_DMA_STATUS_REG = 0x44C83D57 [ 7508.779603] radeon 0000:08:00.0: GPU reset succeeded, trying to resume [ 7508.798041] [drm] probing gen 2 caps for device 8086:1d1a = 2/0 [ 7508.798047] [drm] enabling PCIE gen 2 link speeds, disable with radeon.pcie_gen2=0 [ 7508.805050] [drm] PCIE GART of 512M enabled (table at 0x0000000000040000). [ 7508.805186] radeon 0000:08:00.0: WB enabled [ 7508.805189] radeon 0000:08:00.0: fence driver on ring 0 use gpu addr 0x0000000040000c00 and cpu addr 0xffff880442ddcc00 [ 7508.805190] radeon 0000:08:00.0: fence driver on ring 3 use gpu addr 0x0000000040000c0c and cpu addr 0xffff880442ddcc0c [ 7508.821692] [drm] ring test on 0 succeeded in 2 usecs [ 7508.821765] [drm] ring test on 3 succeeded in 1 usecs [ 7508.821845] [drm] ib test on ring 0 succeeded in 0 usecs [ 7508.821923] [drm] ib test on ring 3 succeeded in 1 usecs --Andy

5 14

[PATCH v4 0/2] Enhance EDID quirks to allow forcing a mode
by Dylan Semler 18 Jun '13

18 Jun '13

Changes in this version * rename do_force_quirk_modes() -> do_force_quirk_mode() * use list_for_each_entry() instead of list_for_each_entry_safe() in do_force_quirk_mode() * remove num_modes from do_force_quirk_mode(), just return 1 or 0 as appropriate * remove unused quirks argument from add_force_quirk_modes() * fixes to allow cases of forcing multiple modes * adjusted comments to adhere closer to style guides Changes in version 3 * Uses drm_cvt_mode() instead of drm_gtf_mode() to build modeline * Adds bool to specify reduced blanking to edid_quirk_force_mode * Removes preferred bit from all other modes Changes in version 2 * none There is at least one monitor that doesn't report its native resolution in its EDID block. This enhancement extends the EDID quirk logic to make monitors like this "just work". The first patch in this series sets up a new quirk list where monitors' correct width, height, refresh rate, and reduced blanking parameters are specified. When a matching monitor is attached the full mode is calculated with drm_cvt_mode() and added to the connector. The DRM_MODE_TYPE_PREFERRED bit is set on the new mode and unset from all other modes. The first patch also defines a new quirk bit: EDID_QUIRK_FORCE_MODE. This bit needs to be set for the new quirk list described above to be checked. The second patch adds the offending monitor to the quirk lists. Dylan Semler (2): drm: Enhance EDID quirks to explicitly set a mode drm: Add EDID force quirk for MMT Monitor2Go HD+ drivers/gpu/drm/drm_edid.c | 89 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 89 insertions(+) -- 1.7.11.7

2 5

[PATCH v3, part2 14/20] PCI, DRM: use hotplug-safe iterators to walk PCI buses
by Jiang Liu 17 Jun '13

17 Jun '13

Enhance DRM drvier to use hotplug-safe iterators to walk PCI buses. Signed-off-by: Jiang Liu <jiang.liu(a)huawei.com> Cc: David Airlie <airlied(a)linux.ie> Cc: dri-devel(a)lists.freedesktop.org Cc: linux-kernel(a)vger.kernel.org --- drivers/gpu/drm/drm_fops.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/drm_fops.c b/drivers/gpu/drm/drm_fops.c index 429e07d..dc8c40b 100644 --- a/drivers/gpu/drm/drm_fops.c +++ b/drivers/gpu/drm/drm_fops.c @@ -359,9 +359,11 @@ static int drm_open_helper(struct inode *inode, struct file *filp, pci_dev_put(pci_dev); } if (!dev->hose) { - struct pci_bus *b = pci_bus_b(pci_root_buses.next); - if (b) + struct pci_bus *b = pci_get_next_root_bus(NULL); + if (b) { dev->hose = b->sysdata; + pci_bus_put(b); + } } } #endif -- 1.8.1.2

2 1

WARNING: at drivers/gpu/drm/nouveau/core/core/mm.c:242
by Jiri Slaby 17 Jun '13

17 Jun '13

Hi, we have a report of WARNING from 3.7.6 in nouveau at drivers/gpu/drm/nouveau/core/core/mm.c:242 here: https://bugzilla.novell.com/show_bug.cgi?id=802347#c11 There is an order 4 allocation failure in nouveau_drm_open -> nouveau_vm_create, i.e. this one failed: vm->pgt = kcalloc(vm->lpde - vm->fpde + 1, sizeof(*vm->pgt), GFP_KERNEL); Then, on the error path in still in nouveau_drm_open, it is followed by a call to nouveau_cli_destroy. But that one calls nouveau_vm_ref -> nouveau_mm_fini -> nouveau_vm_del -> nouveau_mm_fini which triggers the warning. Any ideas? thanks, -- js suse labs

4 10

[Bug 57281] New: Radeon: Bad performance and power consumption
by bugzilla-daemon＠bugzilla.kernel.org 16 Jun '13

16 Jun '13

https://bugzilla.kernel.org/show_bug.cgi?id=57281 Summary: Radeon: Bad performance and power consumption Product: Drivers Version: 2.5 Kernel Version: 3.8.2-pf Platform: All OS/Version: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: Video(DRI - non Intel) AssignedTo: drivers_video-dri(a)kernel-bugs.osdl.org ReportedBy: ka.nick(a)mail.ru Regression: No Created an attachment (id=100261) --> (https://bugzilla.kernel.org/attachment.cgi?id=100261) lspci -vv output After upgrading my AMD E-450-based notebook to a newer one (HP 4545s A4-3300-based) i noticed that in spite of noticeable higher clock rate the video performance is about 15-20% worse than on E-450. It also got much hotter initially, which I could cure by passing pcie_aspm=force to the kernel. In spite of this I still have a feeling that it probably eats more power than it should (which I can not prove with numbers, of course) comparing to E-450 (and fglrx) experience. (Yes, I know the proprietary driver performs better in terms of power management, too, but the question is to what extent?) Anyways, it shouldn't underperform the older and weaker model... It also shows the message at boot time: [drm:radeon_acpi_init] *ERROR* Cannot find backlight controller but it seems to work fine after that. (Shall I post another bug on this?) -- 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 13

[PATCH] drm: shmobile: Use devm_* managed functions
by Laurent Pinchart 14 Jun '13

14 Jun '13

This simplifies cleanup paths and fixes a probe time crash in the error path when trying to cleanup mode setting before it was initialized. Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas(a)ideasonboard.com> --- drivers/gpu/drm/shmobile/shmob_drm_drv.c | 28 +++++++++------------------- drivers/gpu/drm/shmobile/shmob_drm_plane.c | 7 +------ 2 files changed, 10 insertions(+), 25 deletions(-) diff --git a/drivers/gpu/drm/shmobile/shmob_drm_drv.c b/drivers/gpu/drm/shmobile/shmob_drm_drv.c index f6e0b53..29d15e3 100644 --- a/drivers/gpu/drm/shmobile/shmob_drm_drv.c +++ b/drivers/gpu/drm/shmobile/shmob_drm_drv.c @@ -90,7 +90,7 @@ static int shmob_drm_setup_clocks(struct shmob_drm_device *sdev, return -EINVAL; } - clk = clk_get(sdev->dev, clkname); + clk = devm_clk_get(sdev->dev, clkname); if (IS_ERR(clk)) { dev_err(sdev->dev, "cannot get dot clock %s\n", clkname); return PTR_ERR(clk); @@ -106,21 +106,12 @@ static int shmob_drm_setup_clocks(struct shmob_drm_device *sdev, static int shmob_drm_unload(struct drm_device *dev) { - struct shmob_drm_device *sdev = dev->dev_private; - drm_kms_helper_poll_fini(dev); drm_mode_config_cleanup(dev); drm_vblank_cleanup(dev); drm_irq_uninstall(dev); - if (sdev->clock) - clk_put(sdev->clock); - - if (sdev->mmio) - iounmap(sdev->mmio); - dev->dev_private = NULL; - kfree(sdev); return 0; } @@ -139,7 +130,7 @@ static int shmob_drm_load(struct drm_device *dev, unsigned long flags) return -EINVAL; } - sdev = kzalloc(sizeof(*sdev), GFP_KERNEL); + sdev = devm_kzalloc(&pdev->dev, sizeof(*sdev), GFP_KERNEL); if (sdev == NULL) { dev_err(dev->dev, "failed to allocate private data\n"); return -ENOMEM; @@ -156,29 +147,28 @@ static int shmob_drm_load(struct drm_device *dev, unsigned long flags) res = platform_get_resource(pdev, IORESOURCE_MEM, 0); if (res == NULL) { dev_err(&pdev->dev, "failed to get memory resource\n"); - ret = -EINVAL; - goto done; + return -EINVAL; } - sdev->mmio = ioremap_nocache(res->start, resource_size(res)); + sdev->mmio = devm_ioremap_nocache(&pdev->dev, res->start, + resource_size(res)); if (sdev->mmio == NULL) { dev_err(&pdev->dev, "failed to remap memory resource\n"); - ret = -ENOMEM; - goto done; + return -ENOMEM; } ret = shmob_drm_setup_clocks(sdev, pdata->clk_source); if (ret < 0) - goto done; + return ret; ret = shmob_drm_init_interface(sdev); if (ret < 0) - goto done; + return ret; ret = shmob_drm_modeset_init(sdev); if (ret < 0) { dev_err(&pdev->dev, "failed to initialize mode setting\n"); - goto done; + return ret; } for (i = 0; i < 4; ++i) { diff --git a/drivers/gpu/drm/shmobile/shmob_drm_plane.c b/drivers/gpu/drm/shmobile/shmob_drm_plane.c index e1eb899..6898f6f 100644 --- a/drivers/gpu/drm/shmobile/shmob_drm_plane.c +++ b/drivers/gpu/drm/shmobile/shmob_drm_plane.c @@ -221,11 +221,8 @@ static int shmob_drm_plane_disable(struct drm_plane *plane) static void shmob_drm_plane_destroy(struct drm_plane *plane) { - struct shmob_drm_plane *splane = to_shmob_plane(plane); - shmob_drm_plane_disable(plane); drm_plane_cleanup(plane); - kfree(splane); } static const struct drm_plane_funcs shmob_drm_plane_funcs = { @@ -251,7 +248,7 @@ int shmob_drm_plane_create(struct shmob_drm_device *sdev, unsigned int index) struct shmob_drm_plane *splane; int ret; - splane = kzalloc(sizeof(*splane), GFP_KERNEL); + splane = devm_kzalloc(sdev->dev, sizeof(*splane), GFP_KERNEL); if (splane == NULL) return -ENOMEM; @@ -261,8 +258,6 @@ int shmob_drm_plane_create(struct shmob_drm_device *sdev, unsigned int index) ret = drm_plane_init(sdev->ddev, &splane->plane, 1, &shmob_drm_plane_funcs, formats, ARRAY_SIZE(formats), false); - if (ret < 0) - kfree(splane); return ret; } -- Regards, Laurent Pinchart

2 1

[PATCH 0/6] Miscellaneous fixes to host1x
by Arto Merilainen 12 Jun '13

12 Jun '13

This patch series fixes two issues in the host1x driver: First, the command buffer validation routine had vulnerabilities that were not detected in earlier testing. Second, the return codes of some functions were misleading or completely missing. This caused the driver to give wrong return codes also to the userspace. The series is based on top of 3.10rc1. I have tested the patch series on cardhu by running host1x and gr2d test cases (available at [0]). I would appreciate any help in testing/reviewing these patches. [0] https://gitorious.org/linux-host1x/libdrm-host1x Arto Merilainen (5): gpu: host1x: Fix syncpoint wait return value gpu: host1x: Fix memory access in syncpt request gpu: host1x: Fix client_managed type gpu: host1x: Rework CPU syncpoint increment drm/tegra: Fix syncpoint increment return code Terje Bergstrom (1): gpu: host1x: Fixes to host1x firewall drivers/gpu/host1x/dev.h | 8 +-- drivers/gpu/host1x/drm/drm.c | 3 +- drivers/gpu/host1x/drm/gr2d.c | 2 +- drivers/gpu/host1x/hw/cdma_hw.c | 2 +- drivers/gpu/host1x/hw/syncpt_hw.c | 12 ++-- drivers/gpu/host1x/job.c | 120 ++++++++++++++++--------------------- drivers/gpu/host1x/syncpt.c | 29 +++------ drivers/gpu/host1x/syncpt.h | 13 ++-- 8 files changed, 79 insertions(+), 110 deletions(-) -- 1.7.9.5

5 24

[PATCH v4 0/4] add mutex wait/wound/style style locks
by Maarten Lankhorst 12 Jun '13

12 Jun '13

Version 4 already? Small api changes since v3: - Remove ww_mutex_unlock_single and ww_mutex_lock_single. - Rename ww_mutex_trylock_single to ww_mutex_trylock. - Remove separate implementations of ww_mutex_lock_slow*, normal functions can be used. Inline versions still exist for extra debugging, and to annotate. - Cleanup unneeded memory barriers, add comment to the remaining smp_mb(). Thanks to Daniel Vetter, Rob Clark and Peter Zijlstra for their feedback. --- Daniel Vetter (1): mutex: w/w mutex slowpath debugging Maarten Lankhorst (3): arch: make __mutex_fastpath_lock_retval return whether fastpath succeeded or not. mutex: add support for wound/wait style locks, v5 mutex: Add ww tests to lib/locking-selftest.c. v4 Documentation/ww-mutex-design.txt | 344 +++++++++++++++++++++++++++++++ arch/ia64/include/asm/mutex.h | 10 - arch/powerpc/include/asm/mutex.h | 10 - arch/sh/include/asm/mutex-llsc.h | 4 arch/x86/include/asm/mutex_32.h | 11 - arch/x86/include/asm/mutex_64.h | 11 - include/asm-generic/mutex-dec.h | 10 - include/asm-generic/mutex-null.h | 2 include/asm-generic/mutex-xchg.h | 10 - include/linux/mutex-debug.h | 1 include/linux/mutex.h | 363 +++++++++++++++++++++++++++++++++ kernel/mutex.c | 384 ++++++++++++++++++++++++++++++++--- lib/Kconfig.debug | 13 + lib/debug_locks.c | 2 lib/locking-selftest.c | 410 +++++++++++++++++++++++++++++++++++-- 15 files changed, 1492 insertions(+), 93 deletions(-) create mode 100644 Documentation/ww-mutex-design.txt -- ~Maarten

4 10

[PATCH 1/3] drm: Preserve the list head in drm_mode_copy
by ville.syrjala＠linux.intel.com 11 Jun '13

11 Jun '13

From: Ville Syrjälä <ville.syrjala(a)linux.intel.com> Preserve the destination mode's list head in drm_mode_copy. Just in case someone decides that it's a good idea to overwrite a mode which happens to be on some list, Signed-off-by: Ville Syrjälä <ville.syrjala(a)linux.intel.com> --- drivers/gpu/drm/drm_modes.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/drm_modes.c b/drivers/gpu/drm/drm_modes.c index a371ff8..ba86316 100644 --- a/drivers/gpu/drm/drm_modes.c +++ b/drivers/gpu/drm/drm_modes.c @@ -787,16 +787,17 @@ EXPORT_SYMBOL(drm_mode_set_crtcinfo); * LOCKING: * None. * - * Copy an existing mode into another mode, preserving the object id - * of the destination mode. + * Copy an existing mode into another mode, preserving the object id and + * list head of the destination mode. */ void drm_mode_copy(struct drm_display_mode *dst, const struct drm_display_mode *src) { int id = dst->base.id; + struct list_head head = dst->head; *dst = *src; dst->base.id = id; - INIT_LIST_HEAD(&dst->head); + dst->head = head; } EXPORT_SYMBOL(drm_mode_copy); -- 1.8.1.5

4 6

[PATCH] drm/tegra: add support for runtime pm
by Mayuresh Kulkarni 11 Jun '13

11 Jun '13

- as of now host1x and gr2d module's clocks are enabled in their driver's .probe and never disabled - this commit adds support for runtime pm in host1x and gr2d drivers - during boot-up clocks are enabled in .probe and disabled on its exit - when new work is submitted, clocks are enabled in submit and disabled when submit completes - parent->child relation between host1x and gr2d ensures that host1x clock is also enabled before enabling gr2d clock and it is turned off after gr2d clock is turned off Signed-off-by: Mayuresh Kulkarni <mkulkarni(a)nvidia.com> --- drivers/gpu/host1x/channel.c | 5 ++++ drivers/gpu/host1x/dev.c | 58 ++++++++++++++++++++++++++++++++++++++---- drivers/gpu/host1x/drm/gr2d.c | 59 ++++++++++++++++++++++++++++++++++++++----- drivers/gpu/host1x/intr.c | 9 +++++-- 4 files changed, 118 insertions(+), 13 deletions(-) diff --git a/drivers/gpu/host1x/channel.c b/drivers/gpu/host1x/channel.c index 83ea51b..b71caec 100644 --- a/drivers/gpu/host1x/channel.c +++ b/drivers/gpu/host1x/channel.c @@ -18,6 +18,7 @@ #include <linux/slab.h> #include <linux/module.h> +#include <linux/pm_runtime.h> #include "channel.h" #include "dev.h" @@ -41,6 +42,10 @@ int host1x_job_submit(struct host1x_job *job) { struct host1x *host = dev_get_drvdata(job->channel->dev->parent); + /* enable clocks + * they will be disabled in action_submit_complete */ + pm_runtime_get_sync(job->channel->dev); + return host1x_hw_channel_submit(host, job); } diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c index 28e28a2..93d9d72 100644 --- a/drivers/gpu/host1x/dev.c +++ b/drivers/gpu/host1x/dev.c @@ -23,6 +23,7 @@ #include <linux/of_device.h> #include <linux/clk.h> #include <linux/io.h> +#include <linux/pm_runtime.h> #define CREATE_TRACE_POINTS #include <trace/events/host1x.h> @@ -143,11 +144,11 @@ static int host1x_probe(struct platform_device *pdev) return err; } - err = clk_prepare_enable(host->clk); - if (err < 0) { - dev_err(&pdev->dev, "failed to enable clock\n"); - return err; - } + /* enable runtime pm */ + pm_runtime_enable(&pdev->dev); + + /* enable clocks */ + pm_runtime_get_sync(&pdev->dev); err = host1x_syncpt_init(host); if (err) { @@ -165,6 +166,9 @@ static int host1x_probe(struct platform_device *pdev) host1x_drm_alloc(pdev); + /* disable clocks */ + pm_runtime_put(&pdev->dev); + return 0; fail_deinit_syncpt: @@ -183,6 +187,47 @@ static int __exit host1x_remove(struct platform_device *pdev) return 0; } +#ifdef CONFIG_PM_RUNTIME +static int host1x_runtime_suspend(struct device *dev) +{ + struct host1x *host; + + host = dev_get_drvdata(dev); + if (IS_ERR_OR_NULL(host)) + return -EINVAL; + + clk_disable_unprepare(host->clk); + + return 0; +} + +static int host1x_runtime_resume(struct device *dev) +{ + int err = 0; + struct host1x *host; + + host = dev_get_drvdata(dev); + if (IS_ERR_OR_NULL(host)) + return -EINVAL; + + err = clk_prepare_enable(host->clk); + if (err < 0) + dev_err(dev, "failed to enable clock\n"); + + return err; +} +#endif /* CONFIG_PM_RUNTIME */ + +#ifdef CONFIG_PM +static const struct dev_pm_ops host1x_pm_ops = { +#ifdef CONFIG_PM_RUNTIME + .runtime_suspend = host1x_runtime_suspend, + .runtime_resume = host1x_runtime_resume, +#endif +}; + +#endif /* CONFIG_PM */ + static struct platform_driver tegra_host1x_driver = { .probe = host1x_probe, .remove = __exit_p(host1x_remove), @@ -190,6 +235,9 @@ static struct platform_driver tegra_host1x_driver = { .owner = THIS_MODULE, .name = "tegra-host1x", .of_match_table = host1x_of_match, +#ifdef CONFIG_PM + .pm = &host1x_pm_ops, +#endif }, }; diff --git a/drivers/gpu/host1x/drm/gr2d.c b/drivers/gpu/host1x/drm/gr2d.c index 6a45ae0..172a654 100644 --- a/drivers/gpu/host1x/drm/gr2d.c +++ b/drivers/gpu/host1x/drm/gr2d.c @@ -22,6 +22,7 @@ #include <linux/of.h> #include <linux/of_device.h> #include <linux/clk.h> +#include <linux/pm_runtime.h> #include "channel.h" #include "drm.h" @@ -271,11 +272,13 @@ static int gr2d_probe(struct platform_device *pdev) return PTR_ERR(gr2d->clk); } - err = clk_prepare_enable(gr2d->clk); - if (err) { - dev_err(dev, "cannot turn on clock\n"); - return err; - } + platform_set_drvdata(pdev, gr2d); + + /* enable runtime pm */ + pm_runtime_enable(&pdev->dev); + + /* enable clocks */ + pm_runtime_get_sync(&pdev->dev); gr2d->channel = host1x_channel_request(dev); if (!gr2d->channel) @@ -301,7 +304,8 @@ static int gr2d_probe(struct platform_device *pdev) gr2d_init_addr_reg_map(dev, gr2d); - platform_set_drvdata(pdev, gr2d); + /* disable clocks */ + pm_runtime_put(&pdev->dev); return 0; } @@ -328,6 +332,46 @@ static int __exit gr2d_remove(struct platform_device *pdev) return 0; } +#ifdef CONFIG_PM_RUNTIME +static int gr2d_runtime_suspend(struct device *dev) +{ + struct gr2d *gr2d; + + gr2d = dev_get_drvdata(dev); + if (IS_ERR_OR_NULL(gr2d)) + return -EINVAL; + + clk_disable_unprepare(gr2d->clk); + + return 0; +} + +static int gr2d_runtime_resume(struct device *dev) +{ + int err = 0; + struct gr2d *gr2d; + + gr2d = dev_get_drvdata(dev); + if (IS_ERR_OR_NULL(gr2d)) + return -EINVAL; + + err = clk_prepare_enable(gr2d->clk); + if (err < 0) + dev_err(dev, "failed to enable clock\n"); + + return err; +} +#endif /* CONFIG_PM_RUNTIME */ + +#ifdef CONFIG_PM +static const struct dev_pm_ops gr2d_pm_ops = { +#ifdef CONFIG_PM_RUNTIME + .runtime_suspend = gr2d_runtime_suspend, + .runtime_resume = gr2d_runtime_resume, +#endif +}; +#endif /* CONFIG_PM */ + struct platform_driver tegra_gr2d_driver = { .probe = gr2d_probe, .remove = __exit_p(gr2d_remove), @@ -335,5 +379,8 @@ struct platform_driver tegra_gr2d_driver = { .owner = THIS_MODULE, .name = "gr2d", .of_match_table = gr2d_match, +#ifdef CONFIG_PM + .pm = &gr2d_pm_ops, +#endif } }; diff --git a/drivers/gpu/host1x/intr.c b/drivers/gpu/host1x/intr.c index 2491bf8..c23fb64 100644 --- a/drivers/gpu/host1x/intr.c +++ b/drivers/gpu/host1x/intr.c @@ -20,6 +20,7 @@ #include <linux/interrupt.h> #include <linux/slab.h> #include <linux/irq.h> +#include <linux/pm_runtime.h> #include <trace/events/host1x.h> #include "channel.h" @@ -109,14 +110,18 @@ static void reset_threshold_interrupt(struct host1x *host, static void action_submit_complete(struct host1x_waitlist *waiter) { + int completed = waiter->count; struct host1x_channel *channel = waiter->data; + /* disable clocks for all the submits that got completed in this lot */ + while (completed--) + pm_runtime_put(channel->dev); + host1x_cdma_update(&channel->cdma); - /* Add nr_completed to trace */ + /* Add nr_completed to trace */ trace_host1x_channel_submit_complete(dev_name(channel->dev), waiter->count, waiter->thresh); - } static void action_wakeup(struct host1x_waitlist *waiter) -- 1.8.1.5

3 6

← Newer
1
...
45
46
47
48
49
50
51
...
71
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel May 2013