dri-devel February 2013

dri-devel@lists.freedesktop.org

130 participants
696 discussions

Building DRM for i.MX fails with linux-next 20130212
by Fabio Estevam 12 Feb '13

12 Feb '13

Hi, Building imx_v6_v7_defconfig on linux-next 20130212 gives me the following build error: CC drivers/gpu/drm/drm_pci.o drivers/gpu/drm/drm_pci.c: In function ‘drm_pcie_get_speed_cap_mask’: drivers/gpu/drm/drm_pci.c:485:2: error: implicit declaration of function ‘pcie_capability_read_dword’ [-Werror=implicit-function-declaration] Then I tried to fix it as follows: diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile index 6f58c81..6a08165 100644 --- a/drivers/gpu/drm/Makefile +++ b/drivers/gpu/drm/Makefile @@ -8,7 +8,7 @@ drm-y := drm_auth.o drm_buffer.o drm_bufs.o drm_cache.o \ drm_context.o drm_dma.o \ drm_drv.o drm_fops.o drm_gem.o drm_ioctl.o drm_irq.o \ drm_lock.o drm_memory.o drm_proc.o drm_stub.o drm_vm.o \ - drm_agpsupport.o drm_scatter.o drm_pci.o \ + drm_agpsupport.o drm_scatter.o \ drm_platform.o drm_sysfs.o drm_hashtab.o drm_mm.o \ drm_crtc.o drm_modes.o drm_edid.o \ drm_info.o drm_debugfs.o drm_encoder_slave.o \ @@ -16,7 +16,7 @@ drm-y := drm_auth.o drm_buffer.o drm_bufs.o drm_cache.o \ drm-$(CONFIG_COMPAT) += drm_ioc32.o drm-$(CONFIG_DRM_GEM_CMA_HELPER) += drm_gem_cma_helper.o -drm-$(CONFIG_PCI) += ati_pcigart.o +drm-$(CONFIG_PCI) += ati_pcigart.o drm_pci.o drm-usb-y := drm_usb.o ,which caused another issue: GEN .version CHK include/generated/compile.h UPD include/generated/compile.h CC init/version.o LD init/built-in.o drivers/built-in.o: In function `drm_rmmap_locked': clk-vexpress-osc.c:(.text+0x4f370): undefined reference to `__drm_pci_free' drivers/built-in.o: In function `drm_addmap_core': clk-vexpress-osc.c:(.text+0x4f4f4): undefined reference to `drm_pci_alloc' drivers/built-in.o: In function `drm_cleanup_buf_error': clk-vexpress-osc.c:(.text+0x4f920): undefined reference to `drm_pci_free' drivers/built-in.o: In function `drm_addbufs_pci': clk-vexpress-osc.c:(.text+0x4fc00): undefined reference to `drm_pci_alloc' drivers/built-in.o: In function `drm_dma_takedown': clk-vexpress-osc.c:(.text+0x51c84): undefined reference to `drm_pci_free' drivers/built-in.o: In function `drm_vm_shm_close': clk-vexpress-osc.c:(.text+0x57d60): undefined reference to `__drm_pci_free' make: *** [vmlinux] Error 1 Any ideas or suggestions? Thanks, Fabio Estevam

2 2

[Bug 49981] [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation -12!
by bugzilla-daemon＠bugzilla.kernel.org 12 Feb '13

12 Feb '13

https://bugzilla.kernel.org/show_bug.cgi?id=49981 Jérôme Glisse <glisse(a)freedesktop.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |CODE_FIX --- Comment #7 from Jérôme Glisse <glisse(a)freedesktop.org> 2013-02-12 18:58:44 --- Reopen if you have issue with mesa 9.1 or mesa from git. A fix was pushed there. -- 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] drm/radeon: remove overzealous warning in hdmi handling
by alexdeucher＠gmail.com 12 Feb '13

12 Feb '13

From: Alex Deucher <alexander.deucher(a)amd.com> hdmi audio works fine. The warning just confuses users. fixes: https://bugzilla.kernel.org/show_bug.cgi?id=44341 Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Cc: stable(a)vger.kernel.org --- drivers/gpu/drm/radeon/r600_hdmi.c | 1 - 1 files changed, 0 insertions(+), 1 deletions(-) diff --git a/drivers/gpu/drm/radeon/r600_hdmi.c b/drivers/gpu/drm/radeon/r600_hdmi.c index ff80efe..95970ec 100644 --- a/drivers/gpu/drm/radeon/r600_hdmi.c +++ b/drivers/gpu/drm/radeon/r600_hdmi.c @@ -544,7 +544,6 @@ void r600_hdmi_disable(struct drm_encoder *encoder) /* Called for ATOM_ENCODER_MODE_HDMI only */ if (!dig || !dig->afmt) { - WARN_ON(1); return; } if (!dig->afmt->enabled) -- 1.7.7.5

2 1

[PATCH] drm: missing idr_preload_end in drm_gem_flink_ioctl
by Artem Savkov 12 Feb '13

12 Feb '13

Added missing idr_preload_end calls in drm_gem_flink_ioctl(). Without those preemption stays disabled resulting in lots of "scheduling while atomic" BUGs. Introduced in 4e486fae707d9a79c76f1be6dcacffc4a721cc1b (linux-next.git) Signed-off-by: Artem Savkov <artem.savkov(a)gmail.com> --- drivers/gpu/drm/drm_gem.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c index 6577514..f1700ef 100644 --- a/drivers/gpu/drm/drm_gem.c +++ b/drivers/gpu/drm/drm_gem.c @@ -456,6 +456,7 @@ drm_gem_flink_ioctl(struct drm_device *dev, void *data, obj->name = ret; args->name = (uint64_t) obj->name; spin_unlock(&dev->object_name_lock); + idr_preload_end(); if (ret < 0) goto err; @@ -465,6 +466,7 @@ drm_gem_flink_ioctl(struct drm_device *dev, void *data, } else { args->name = (uint64_t) obj->name; spin_unlock(&dev->object_name_lock); + idr_preload_end(); ret = 0; } -- 1.8.1.1

2 1

Re: [systemd-devel] Someone working on gui for multiseat?
by poma 12 Feb '13

12 Feb '13

On 02/11/13 21:30, Lennart Poettering wrote: […] > Of course, doing this via the command line is not user-frienldy, but > maybe this will one day get fixed and somebody writes a proper UI for > it... > Regarding 'loginctl attach seat0 <sysfs>', the only thing really missing is the 'sys' path completion, so one doesn't pars it via e.g. dmesg -t | grep -P '(?=.*devices)(?=.*USB)', grep -B1 sys /var/log/Xorg.0.log, tree /sys/class/drm, etc. Behind the wheel, dri/drm device 'sys' path doesn't show up in dmesg, nor in Xorg.*.log. - cc'ing dri-devel. Cheers, poma

1 0

[Bug 60723] New: Unable to compile Mesa 9.0.2 for radeon
by bugzilla-daemon＠freedesktop.org 12 Feb '13

12 Feb '13

https://bugs.freedesktop.org/show_bug.cgi?id=60723 Priority: medium Bug ID: 60723 Assignee: dri-devel(a)lists.freedesktop.org Summary: Unable to compile Mesa 9.0.2 for radeon Severity: normal Classification: Unclassified OS: Linux (All) Reporter: gogolander(a)gmail.com Hardware: x86-64 (AMD64) Status: NEW Version: 9.0 Component: Drivers/Gallium/r600 Product: Mesa Hi. I'm trying to compile Mesa 9.0.2 on my amd64 pc. After launching the following configure command: " ./configure --prefix=/usr \ --with-dri-drivers=radeon \ --enable-gles1 \ --enable-gles2 \ --enable-xa \ --enable-texture-float \ --enable-osmesa \ --enable-glx-tls \ --enable-opengl \ --enable-openvg \ --enable-gallium-egl \ --with-gallium-drivers="r300,r600,svga,swrast" " make fails to compile with this error message: "r300_dri.so.tmp: undefined reference to `_glapi_tls_Dispatch' r300_dri.so.tmp: undefined reference to `_glapi_tls_Context' collect2: error: ld returned 1 exit status make[3]: *** [r300_dri.so] Error 1" I haven't found any solution. How can I solve this? Thanks in advance -- You are receiving this mail because: You are the assignee for the bug.

1 5

[PATCH linux-next] drm/radeon: Avoid NULL pointer dereference from atom_index_iio() allocation failure
by Tim Gardner 12 Feb '13

12 Feb '13

Smatch anlysis: drivers/gpu/drm/radeon/atom.c:1242 atom_index_iio() error: potential null dereference 'ctx->iio'. (kzalloc returns null) Also cleaned up some checks before calls to kfree(). kfree(NULL) is OK. Cc: David Airlie <airlied(a)linux.ie> Cc: Alex Deucher <alexander.deucher(a)amd.com> Cc: "Michel Dänzer" <michel.daenzer(a)amd.com> Cc: Dave Airlie <airlied(a)redhat.com> Cc: "Christian König" <christian.koenig(a)amd.com> Cc: Jerome Glisse <jglisse(a)redhat.com> Cc: dri-devel(a)lists.freedesktop.org Signed-off-by: Tim Gardner <tim.gardner(a)canonical.com> --- drivers/gpu/drm/radeon/atom.c | 9 +++++++-- drivers/gpu/drm/radeon/radeon_device.c | 9 ++++++++- 2 files changed, 15 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/radeon/atom.c b/drivers/gpu/drm/radeon/atom.c index 5ce9bf5..46a9c37 100644 --- a/drivers/gpu/drm/radeon/atom.c +++ b/drivers/gpu/drm/radeon/atom.c @@ -1238,6 +1238,8 @@ static int atom_iio_len[] = { 1, 2, 3, 3, 3, 3, 4, 4, 4, 3 }; static void atom_index_iio(struct atom_context *ctx, int base) { ctx->iio = kzalloc(2 * 256, GFP_KERNEL); + if (!ctx->iio) + return; while (CU8(base) == ATOM_IIO_START) { ctx->iio[CU8(base + 1)] = base + 2; base += 2; @@ -1287,6 +1289,10 @@ struct atom_context *atom_parse(struct card_info *card, void *bios) ctx->cmd_table = CU16(base + ATOM_ROM_CMD_PTR); ctx->data_table = CU16(base + ATOM_ROM_DATA_PTR); atom_index_iio(ctx, CU16(ctx->data_table + ATOM_DATA_IIO_PTR) + 4); + if (!ctx->iio) { + atom_destroy(ctx); + return NULL; + } str = CSTR(CU16(base + ATOM_ROM_MSG_PTR)); while (*str && ((*str == '\n') || (*str == '\r'))) @@ -1335,8 +1341,7 @@ int atom_asic_init(struct atom_context *ctx) void atom_destroy(struct atom_context *ctx) { - if (ctx->iio) - kfree(ctx->iio); + kfree(ctx->iio); kfree(ctx); } diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c index 8794de1..44b8034 100644 --- a/drivers/gpu/drm/radeon/radeon_device.c +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -759,6 +759,11 @@ int radeon_atombios_init(struct radeon_device *rdev) atom_card_info->pll_write = cail_pll_write; rdev->mode_info.atom_context = atom_parse(atom_card_info, rdev->bios); + if (!rdev->mode_info.atom_context) { + radeon_atombios_fini(rdev); + return -ENOMEM; + } + mutex_init(&rdev->mode_info.atom_context->mutex); radeon_atom_initialize_bios_scratch_regs(rdev->ddev); atom_allocate_fb_scratch(rdev->mode_info.atom_context); @@ -778,9 +783,11 @@ void radeon_atombios_fini(struct radeon_device *rdev) { if (rdev->mode_info.atom_context) { kfree(rdev->mode_info.atom_context->scratch); - kfree(rdev->mode_info.atom_context); } + kfree(rdev->mode_info.atom_context); + rdev->mode_info.atom_context = NULL; kfree(rdev->mode_info.atom_card_info); + rdev->mode_info.atom_card_info = NULL; } /* COMBIOS */ -- 1.7.9.5

2 1

[PATCH v3 0/4] modetest enhancements
by Laurent Pinchart 12 Feb '13

12 Feb '13

Hello, Here's the third version of a small patch set that adds an argument to the modetest command line to specify the driver name instead of using the builtin list of drivers. This is particularly handy during development of new DRM/KMS drivers. This version enables compiler warnings for modetest and fix the many warning messages it creates. This should make sure that no v4 will be needed to fix a stupid issue. Laurent Pinchart (4): modetest: Fix warnings modetest: Remove extern declarations of opt(arg|ind|err|opt) modetest: Sort command line arguments modetest: Add a command line parameter to select the driver tests/modetest/Makefile.am | 4 +- tests/modetest/buffers.c | 13 ++--- tests/modetest/buffers.h | 5 +- tests/modetest/modetest.c | 132 +++++++++++++++++++++++++-------------------- 4 files changed, 88 insertions(+), 66 deletions(-) -- Regards, Laurent Pinchart

1 4

[PATCH] drm/exynos: release resources properly when fb creation is failed.
by Inki Dae 12 Feb '13

12 Feb '13

From: YoungJun Cho <yj44.cho(a)samsung.com> This patch releases allocated resources properly when exynos_user_fb_create() is failed. Signed-off-by: YoungJun Cho <yj44.cho(a)samsung.com> Signed-off-by: Inki Dae <inki.dae(a)samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park(a)samsung.com> --- drivers/gpu/drm/exynos/exynos_drm_fb.c | 51 +++++++++++++++++--------------- 1 files changed, 27 insertions(+), 24 deletions(-) diff --git a/drivers/gpu/drm/exynos/exynos_drm_fb.c b/drivers/gpu/drm/exynos/exynos_drm_fb.c index 31d4cb1..0e04f4e 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fb.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fb.c @@ -221,23 +221,25 @@ exynos_user_fb_create(struct drm_device *dev, struct drm_file *file_priv, struct drm_mode_fb_cmd2 *mode_cmd) { struct drm_gem_object *obj; + struct exynos_drm_gem_obj *exynos_gem_obj; struct exynos_drm_fb *exynos_fb; int i, ret; DRM_DEBUG_KMS("%s\n", __FILE__); - obj = drm_gem_object_lookup(dev, file_priv, mode_cmd->handles[0]); - if (!obj) { - DRM_ERROR("failed to lookup gem object\n"); - return ERR_PTR(-ENOENT); - } - exynos_fb = kzalloc(sizeof(*exynos_fb), GFP_KERNEL); if (!exynos_fb) { DRM_ERROR("failed to allocate exynos drm framebuffer\n"); return ERR_PTR(-ENOMEM); } + obj = drm_gem_object_lookup(dev, file_priv, mode_cmd->handles[0]); + if (!obj) { + DRM_ERROR("failed to lookup gem object\n"); + ret = -ENOENT; + goto err_free; + } + drm_helper_mode_fill_fb_struct(&exynos_fb->fb, mode_cmd); exynos_fb->exynos_gem_obj[0] = to_exynos_gem_obj(obj); exynos_fb->buf_cnt = exynos_drm_format_num_buffers(mode_cmd); @@ -245,43 +247,44 @@ exynos_user_fb_create(struct drm_device *dev, struct drm_file *file_priv, DRM_DEBUG_KMS("buf_cnt = %d\n", exynos_fb->buf_cnt); for (i = 1; i < exynos_fb->buf_cnt; i++) { - struct exynos_drm_gem_obj *exynos_gem_obj; - int ret; - obj = drm_gem_object_lookup(dev, file_priv, mode_cmd->handles[i]); if (!obj) { DRM_ERROR("failed to lookup gem object\n"); - kfree(exynos_fb); - return ERR_PTR(-ENOENT); + ret = -ENOENT; + exynos_fb->buf_cnt = i; + goto err_unreference; } exynos_gem_obj = to_exynos_gem_obj(obj); + exynos_fb->exynos_gem_obj[i] = exynos_gem_obj; ret = check_fb_gem_memory_type(dev, exynos_gem_obj); if (ret < 0) { DRM_ERROR("cannot use this gem memory type for fb.\n"); - kfree(exynos_fb); - return ERR_PTR(ret); + goto err_unreference; } - - exynos_fb->exynos_gem_obj[i] = to_exynos_gem_obj(obj); } ret = drm_framebuffer_init(dev, &exynos_fb->fb, &exynos_drm_fb_funcs); if (ret) { - for (i = 0; i < exynos_fb->buf_cnt; i++) { - struct exynos_drm_gem_obj *gem_obj; - - gem_obj = exynos_fb->exynos_gem_obj[i]; - drm_gem_object_unreference_unlocked(&gem_obj->base); - } - - kfree(exynos_fb); - return ERR_PTR(ret); + DRM_ERROR("failed to init framebuffer.\n"); + goto err_unreference; } return &exynos_fb->fb; + +err_unreference: + for (i = 0; i < exynos_fb->buf_cnt; i++) { + struct drm_gem_object *obj; + + obj = &exynos_fb->exynos_gem_obj[i]->base; + if (obj) + drm_gem_object_unreference_unlocked(obj); + } +err_free: + kfree(exynos_fb); + return ERR_PTR(ret); } struct exynos_drm_gem_buf *exynos_drm_fb_buffer(struct drm_framebuffer *fb, -- 1.7.4.1

1 0

[PATCH] drm/exynos: fix wrong pointer access at vm close.
by Inki Dae 12 Feb '13

12 Feb '13

From: YoungJun Cho <yj44.cho(a)samsung.com> This patch fixes wrong pointer access issue to filp->f_op and filp->private_data. The exynos_drm_gem_mmap_ioctl() changes filp->f_op and filp->private_data temporarily and restore them to use original ones in exynos_drm_gem_mmap_buffer() but there was no lock between the changing and the restoring so wrong pointer access to filp->f_op and filp->private_data was induced by vm close callback. So this patch uses mutex lock properly to resolve this issue. Signed-off-by: YoungJun Cho <yj44.cho(a)samsung.com> Signed-off-by: Inki Dae <inki.dae(a)samsung.com> Signed-off-by: Kyungmin Park <kyungmin.park(a)samsung.com> --- drivers/gpu/drm/exynos/exynos_drm_gem.c | 33 ++++++++++++++++++++----------- 1 file changed, 21 insertions(+), 12 deletions(-) diff --git a/drivers/gpu/drm/exynos/exynos_drm_gem.c b/drivers/gpu/drm/exynos/exynos_drm_gem.c index 4731807..67e17ce 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_gem.c +++ b/drivers/gpu/drm/exynos/exynos_drm_gem.c @@ -329,17 +329,11 @@ static struct drm_file *exynos_drm_find_drm_file(struct drm_device *drm_dev, { struct drm_file *file_priv; - mutex_lock(&drm_dev->struct_mutex); - /* find current process's drm_file from filelist. */ - list_for_each_entry(file_priv, &drm_dev->filelist, lhead) { - if (file_priv->filp == filp) { - mutex_unlock(&drm_dev->struct_mutex); + list_for_each_entry(file_priv, &drm_dev->filelist, lhead) + if (file_priv->filp == filp) return file_priv; - } - } - mutex_unlock(&drm_dev->struct_mutex); WARN_ON(1); return ERR_PTR(-EFAULT); @@ -400,9 +394,7 @@ static int exynos_drm_gem_mmap_buffer(struct file *filp, */ drm_gem_object_reference(obj); - mutex_lock(&drm_dev->struct_mutex); drm_vm_open_locked(drm_dev, vma); - mutex_unlock(&drm_dev->struct_mutex); return 0; } @@ -432,6 +424,16 @@ int exynos_drm_gem_mmap_ioctl(struct drm_device *dev, void *data, } /* + * We have to use gem object and its fops for specific mmaper, + * but vm_mmap() can deliver only filp. So we have to change + * filp->f_op and filp->private_data temporarily, then restore + * again. So it is important to keep lock until restoration the + * settings to prevent others from misuse of filp->f_op or + * filp->private_data. + */ + mutex_lock(&dev->struct_mutex); + + /* * Set specific mmper's fops. And it will be restored by * exynos_drm_gem_mmap_buffer to dev->driver->fops. * This is used to call specific mapper temporarily. @@ -448,13 +450,20 @@ int exynos_drm_gem_mmap_ioctl(struct drm_device *dev, void *data, addr = vm_mmap(file_priv->filp, 0, args->size, PROT_READ | PROT_WRITE, MAP_SHARED, 0); - drm_gem_object_unreference_unlocked(obj); + drm_gem_object_unreference(obj); if (IS_ERR((void *)addr)) { - file_priv->filp->private_data = file_priv; + /* check filp->f_op, filp->private_data are restored */ + if (file_priv->filp->f_op == &exynos_drm_gem_fops) { + file_priv->filp->f_op = fops_get(dev->driver->fops); + file_priv->filp->private_data = file_priv; + } + mutex_unlock(&dev->struct_mutex); return PTR_ERR((void *)addr); } + mutex_unlock(&dev->struct_mutex); + args->mapped = addr; DRM_DEBUG_KMS("mapped = 0x%lx\n", (unsigned long)args->mapped); -- 1.7.9.5

1 0

← Newer
1
...
57
58
59
60
61
62
63
...
70
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel February 2013