dri-devel November 2020

dri-devel@lists.freedesktop.org

316 participants
811 discussions

[PATCH V3 1/2] dt-bindings: Add DT bindings for Toshiba TC358762 DSI-to-DPI bridge
by Marek Vasut 03 Dec '20

03 Dec '20

Add DT bindings for Toshiba TC358762 DSI-to-DPI bridge, this one is used in the Raspberry Pi 7" touchscreen display unit. Signed-off-by: Marek Vasut <marex(a)denx.de> To: dri-devel(a)lists.freedesktop.org Cc: Eric Anholt <eric(a)anholt.net> Cc: Rob Herring <robh+dt(a)kernel.org> Cc: Sam Ravnborg <sam(a)ravnborg.org> Cc: devicetree(a)vger.kernel.org --- V2: Fix dt_binding_check errors V3: Add ... at the end of example --- .../display/bridge/toshiba,tc358762.yaml | … [View More]127 ++++++++++++++++++ 1 file changed, 127 insertions(+) create mode 100644 Documentation/devicetree/bindings/display/bridge/toshiba,tc358762.yaml diff --git a/Documentation/devicetree/bindings/display/bridge/toshiba,tc358762.yaml b/Documentation/devicetree/bindings/display/bridge/toshiba,tc358762.yaml new file mode 100644 index 000000000000..195025e6803c --- /dev/null +++ b/Documentation/devicetree/bindings/display/bridge/toshiba,tc358762.yaml @@ -0,0 +1,127 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/display/bridge/toshiba,tc358762.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Toshiba TC358762 MIPI DSI to MIPI DPI bridge + +maintainers: + - Marek Vasut <marex(a)denx.de> + +description: | + The TC358762 is bridge device which converts MIPI DSI to MIPI DPI. + +properties: + compatible: + enum: + - toshiba,tc358762 + + reg: + maxItems: 1 + description: virtual channel number of a DSI peripheral + + vddc-supply: + description: Regulator for 1.2V internal core power. + + ports: + type: object + + properties: + "#address-cells": + const: 1 + + "#size-cells": + const: 0 + + port@0: + type: object + additionalProperties: false + + description: | + Video port for MIPI DSI input + + properties: + reg: + const: 0 + + patternProperties: + endpoint: + type: object + additionalProperties: false + + properties: + remote-endpoint: true + + required: + - reg + + port@1: + type: object + additionalProperties: false + + description: | + Video port for MIPI DPI output (panel or connector). + + properties: + reg: + const: 1 + + patternProperties: + endpoint: + type: object + additionalProperties: false + + properties: + remote-endpoint: true + + required: + - reg + + required: + - "#address-cells" + - "#size-cells" + - port@0 + - port@1 + +required: + - compatible + - reg + - vddc-supply + - ports + +additionalProperties: false + +examples: + - | + i2c1 { + #address-cells = <1>; + #size-cells = <0>; + + bridge@0 { + reg = <0>; + compatible = "toshiba,tc358762"; + vddc-supply = <&vcc_1v2_reg>; + + ports { + #address-cells = <1>; + #size-cells = <0>; + + port@0 { + reg = <0>; + bridge_in: endpoint { + remote-endpoint = <&dsi_out>; + }; + }; + + port@1 { + reg = <1>; + bridge_out: endpoint { + remote-endpoint = <&panel_in>; + }; + }; + }; + }; + }; + +... -- 2.28.0 [View Less]

4 6

[Bug 78111] New: APU turbo core boost not working when radeon.dpm=1
by bugzilla-daemon＠bugzilla.kernel.org 03 Dec '20

03 Dec '20

https://bugzilla.kernel.org/show_bug.cgi?id=78111 Bug ID: 78111 Summary: APU turbo core boost not working when radeon.dpm=1 Product: Drivers Version: 2.5 Kernel Version: 3.14.6 Hardware: x86-64 OS: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: Video(DRI - non Intel) Assignee: drivers_video-dri(a)kernel-bugs.osdl.org … [View More]

1 14

[PATCH v2 1/3] drm/msm: adreno: Make speed-bin support generic
by Akhil P Oommen 03 Dec '20

03 Dec '20

So far a530v2 gpu has support for detecting its supported opps based on a fuse value called speed-bin. This patch makes this support generic across gpu families. This is in preparation to extend speed-bin support to a6x family. Signed-off-by: Akhil P Oommen <akhilpo(a)codeaurora.org> --- Changes from v1: 1. Added the changes to support a618 sku to the series. 2. Avoid failing probe in case of an unsupported sku. (Rob) drivers/gpu/drm/msm/adreno/a5xx_gpu.c | 34 -------------- … [View More]drivers/gpu/drm/msm/adreno/adreno_device.c | 4 ++ drivers/gpu/drm/msm/adreno/adreno_gpu.c | 71 ++++++++++++++++++++++++++++++ drivers/gpu/drm/msm/adreno/adreno_gpu.h | 5 +++ 4 files changed, 80 insertions(+), 34 deletions(-) diff --git a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c index 8fa5c91..7d42321 100644 --- a/drivers/gpu/drm/msm/adreno/a5xx_gpu.c +++ b/drivers/gpu/drm/msm/adreno/a5xx_gpu.c @@ -1531,38 +1531,6 @@ static const struct adreno_gpu_funcs funcs = { .get_timestamp = a5xx_get_timestamp, }; -static void check_speed_bin(struct device *dev) -{ - struct nvmem_cell *cell; - u32 val; - - /* - * If the OPP table specifies a opp-supported-hw property then we have - * to set something with dev_pm_opp_set_supported_hw() or the table - * doesn't get populated so pick an arbitrary value that should - * ensure the default frequencies are selected but not conflict with any - * actual bins - */ - val = 0x80; - - cell = nvmem_cell_get(dev, "speed_bin"); - - if (!IS_ERR(cell)) { - void *buf = nvmem_cell_read(cell, NULL); - - if (!IS_ERR(buf)) { - u8 bin = *((u8 *) buf); - - val = (1 << bin); - kfree(buf); - } - - nvmem_cell_put(cell); - } - - dev_pm_opp_set_supported_hw(dev, &val, 1); -} - struct msm_gpu *a5xx_gpu_init(struct drm_device *dev) { struct msm_drm_private *priv = dev->dev_private; @@ -1588,8 +1556,6 @@ struct msm_gpu *a5xx_gpu_init(struct drm_device *dev) a5xx_gpu->lm_leakage = 0x4E001A; - check_speed_bin(&pdev->dev); - ret = adreno_gpu_init(dev, pdev, adreno_gpu, &funcs, 4); if (ret) { a5xx_destroy(&(a5xx_gpu->base.base)); diff --git a/drivers/gpu/drm/msm/adreno/adreno_device.c b/drivers/gpu/drm/msm/adreno/adreno_device.c index 87c8b03..e0ff16c 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_device.c +++ b/drivers/gpu/drm/msm/adreno/adreno_device.c @@ -18,6 +18,8 @@ bool snapshot_debugbus = false; MODULE_PARM_DESC(snapshot_debugbus, "Include debugbus sections in GPU devcoredump (if not fused off)"); module_param_named(snapshot_debugbus, snapshot_debugbus, bool, 0600); +const u32 a530v2_speedbins[] = {0, 1, 2, 3, 4, 5, 6, 7}; + static const struct adreno_info gpulist[] = { { .rev = ADRENO_REV(2, 0, 0, 0), @@ -163,6 +165,8 @@ static const struct adreno_info gpulist[] = { ADRENO_QUIRK_FAULT_DETECT_MASK, .init = a5xx_gpu_init, .zapfw = "a530_zap.mdt", + .speedbins = a530v2_speedbins, + .speedbins_count = ARRAY_SIZE(a530v2_speedbins), }, { .rev = ADRENO_REV(5, 4, 0, 2), .revn = 540, diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c index f21561d..b342fa4 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c @@ -14,6 +14,7 @@ #include <linux/pm_opp.h> #include <linux/slab.h> #include <linux/soc/qcom/mdt_loader.h> +#include <linux/nvmem-consumer.h> #include <soc/qcom/ocmem.h> #include "adreno_gpu.h" #include "msm_gem.h" @@ -891,6 +892,69 @@ void adreno_gpu_ocmem_cleanup(struct adreno_ocmem *adreno_ocmem) adreno_ocmem->hdl); } +static int adreno_set_supported_hw(struct device *dev, + struct adreno_gpu *adreno_gpu) +{ + u8 speedbins_count = adreno_gpu->info->speedbins_count; + const u32 *speedbins = adreno_gpu->info->speedbins; + struct nvmem_cell *cell; + u32 bin, i; + u32 val = 0; + void *buf, *opp_table; + + cell = nvmem_cell_get(dev, "speed_bin"); + /* + * -ENOENT means that the platform doesn't support speedbin which is + * fine + */ + if (PTR_ERR(cell) == -ENOENT) + return 0; + else if (IS_ERR(cell)) + return PTR_ERR(cell); + + if (!speedbins) + goto done; + + buf = nvmem_cell_read(cell, NULL); + if (IS_ERR(buf)) { + nvmem_cell_put(cell); + return PTR_ERR(buf); + } + + bin = *((u32 *) buf); + + for (i = 0; i < speedbins_count; i++) { + if (bin == speedbins[i]) { + val = (1 << i); + break; + } + } + + kfree(buf); +done: + nvmem_cell_put(cell); + + if (!val) { + DRM_DEV_ERROR(dev, + "missing support for speed-bin: %u. Some OPPs may not be supported by hardware", + bin); + val = ~0U; + } + + opp_table = dev_pm_opp_set_supported_hw(dev, &val, 1); + if (IS_ERR(opp_table)) + return PTR_ERR(opp_table); + + adreno_gpu->opp_table = opp_table; + return 0; +} + +static void adreno_put_supported_hw(struct opp_table *opp_table) +{ + if (opp_table) + dev_pm_opp_put_supported_hw(opp_table); +} + int adreno_gpu_init(struct drm_device *drm, struct platform_device *pdev, struct adreno_gpu *adreno_gpu, const struct adreno_gpu_funcs *funcs, int nr_rings) @@ -899,6 +963,7 @@ int adreno_gpu_init(struct drm_device *drm, struct platform_device *pdev, struct adreno_platform_config *config = dev->platform_data; struct msm_gpu_config adreno_gpu_config = { 0 }; struct msm_gpu *gpu = &adreno_gpu->base; + int ret; adreno_gpu->funcs = funcs; adreno_gpu->info = adreno_info(config->rev); @@ -910,6 +975,10 @@ int adreno_gpu_init(struct drm_device *drm, struct platform_device *pdev, adreno_gpu_config.nr_rings = nr_rings; + ret = adreno_set_supported_hw(dev, adreno_gpu); + if (ret) + return ret; + adreno_get_pwrlevels(dev, gpu); pm_runtime_set_autosuspend_delay(dev, @@ -936,4 +1005,6 @@ void adreno_gpu_cleanup(struct adreno_gpu *adreno_gpu) icc_put(gpu->icc_path); icc_put(gpu->ocmem_icc_path); + + adreno_put_supported_hw(adreno_gpu->opp_table); } diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.h b/drivers/gpu/drm/msm/adreno/adreno_gpu.h index c3775f7..a756ad7 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.h +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.h @@ -55,6 +55,7 @@ struct adreno_reglist { }; extern const struct adreno_reglist a630_hwcg[], a640_hwcg[], a650_hwcg[]; +extern const u32 a618_speedbins[]; struct adreno_info { struct adreno_rev rev; @@ -67,6 +68,8 @@ struct adreno_info { const char *zapfw; u32 inactive_period; const struct adreno_reglist *hwcg; + const u32 *speedbins; + const u8 speedbins_count; }; const struct adreno_info *adreno_info(struct adreno_rev rev); @@ -112,6 +115,8 @@ struct adreno_gpu { * code (a3xx_gpu.c) and stored in this common location. */ const unsigned int *reg_offsets; + + struct opp_table *opp_table; }; #define to_adreno_gpu(x) container_of(x, struct adreno_gpu, base) -- 2.7.4 [View Less]

2 7

[PATCH v2 0/5] drm/omap: add color mgmt support
by Tomi Valkeinen 03 Dec '20

03 Dec '20

Hi, This is a resend of the v1, after adding Pekka's reviewed-by, rebasing to latest drm-misc, and re-testing. The cover letter from v1: This series is based on patches sent about a year ago: https://lists.freedesktop.org/archives/dri-devel/2019-September/233966.html https://lists.freedesktop.org/archives/dri-devel/2019-September/233967.html I've fixed the minor issues reported, and according to the recent discussions with Pekka and Daniel, I have changed how the gamma works. After these … [View More]

4 20

[PATCH RESEND 1/2] dma-fence: allow signaling drivers to set fence timestamp
by Veera Sundaram Sankaran 03 Dec '20

03 Dec '20

Some drivers have hardware capability to get the precise timestamp of certain events based on which the fences are triggered. This allows it to set accurate timestamp factoring out any software and IRQ latencies. Move the timestamp parameter out of union in dma_fence struct to allow signaling drivers to set it. If the parameter is not set, ktime_get is used to set the current time to fence timestamp during dma_fence_signal. Signed-off-by: Veera Sundaram Sankaran <veeras(a)codeaurora.org> … [View More]--- drivers/dma-buf/dma-fence.c | 18 ++++++++++-------- include/linux/dma-fence.h | 15 +++------------ 2 files changed, 13 insertions(+), 20 deletions(-) diff --git a/drivers/dma-buf/dma-fence.c b/drivers/dma-buf/dma-fence.c index 43624b4..7cef49a 100644 --- a/drivers/dma-buf/dma-fence.c +++ b/drivers/dma-buf/dma-fence.c @@ -4,6 +4,7 @@ * * Copyright (C) 2012 Canonical Ltd * Copyright (C) 2012 Texas Instruments + * Copyright (c) 2020 The Linux Foundation. All rights reserved. * * Authors: * Rob Clark <robdclark(a)gmail.com> @@ -329,7 +330,6 @@ void __dma_fence_might_wait(void) int dma_fence_signal_locked(struct dma_fence *fence) { struct dma_fence_cb *cur, *tmp; - struct list_head cb_list; lockdep_assert_held(fence->lock); @@ -337,16 +337,18 @@ int dma_fence_signal_locked(struct dma_fence *fence) &fence->flags))) return -EINVAL; - /* Stash the cb_list before replacing it with the timestamp */ - list_replace(&fence->cb_list, &cb_list); - - fence->timestamp = ktime_get(); + /* set current time, if not set by signaling driver */ + if (!fence->timestamp) + fence->timestamp = ktime_get(); set_bit(DMA_FENCE_FLAG_TIMESTAMP_BIT, &fence->flags); trace_dma_fence_signaled(fence); - list_for_each_entry_safe(cur, tmp, &cb_list, node) { - INIT_LIST_HEAD(&cur->node); - cur->func(fence, cur); + if (!list_empty(&fence->cb_list)) { + list_for_each_entry_safe(cur, tmp, &fence->cb_list, node) { + INIT_LIST_HEAD(&cur->node); + cur->func(fence, cur); + } + INIT_LIST_HEAD(&fence->cb_list); } return 0; diff --git a/include/linux/dma-fence.h b/include/linux/dma-fence.h index 09e23ad..a9eebaf 100644 --- a/include/linux/dma-fence.h +++ b/include/linux/dma-fence.h @@ -4,6 +4,7 @@ * * Copyright (C) 2012 Canonical Ltd * Copyright (C) 2012 Texas Instruments + * Copyright (c) 2020 The Linux Foundation. All rights reserved. * * Authors: * Rob Clark <robdclark(a)gmail.com> @@ -70,26 +71,16 @@ struct dma_fence { * release the fence it is unused. No one should be adding to the * cb_list that they don't themselves hold a reference for. * - * The lifetime of the timestamp is similarly tied to both the - * rcu freelist and the cb_list. The timestamp is only set upon - * signaling while simultaneously notifying the cb_list. Ergo, we - * only use either the cb_list of timestamp. Upon destruction, - * neither are accessible, and so we can use the rcu. This means - * that the cb_list is *only* valid until the signal bit is set, - * and to read either you *must* hold a reference to the fence, - * and not just the rcu_read_lock. - * * Listed in chronological order. */ union { struct list_head cb_list; - /* @cb_list replaced by @timestamp on dma_fence_signal() */ - ktime_t timestamp; - /* @timestamp replaced by @rcu on dma_fence_release() */ + /* @cb_list replaced by @rcu on dma_fence_release() */ struct rcu_head rcu; }; u64 context; u64 seqno; + ktime_t timestamp; unsigned long flags; struct kref refcount; int error; -- 2.7.4 [View Less]

3 8

[PATCH] drm/kmb: Fix possible oops in probe error handling
by Dan Carpenter 03 Dec '20

03 Dec '20

If kmb_dsi_init() fails the error handling will dereference an error pointer which will cause an Oops. Fixes: 7f7b96a8a0a1 ("drm/kmb: Add support for KeemBay Display") Signed-off-by: Dan Carpenter <dan.carpenter(a)oracle.com> --- drivers/gpu/drm/kmb/kmb_drv.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/kmb/kmb_drv.c b/drivers/gpu/drm/kmb/kmb_drv.c index a31a840ce634..8c43b136765c 100644 --- a/drivers/gpu/drm/kmb/kmb_drv.c +++ b/drivers/gpu/… [View More]

2 6

[PATCH v2 0/8] RFC Support hot device unplug in amdgpu
by Andrey Grodzovsky 02 Dec '20

02 Dec '20

This RFC is more of a proof of concept then a fully working solution as there are a few unresolved issues we are hoping to get advise on from people on the mailing list. Until now extracting a card either by physical extraction (e.g. eGPU with thunderbolt connection or by emulation through syfs -> /sys/bus/pci/devices/device_id/remove) would cause random crashes in user apps. The random crashes in apps were mostly due to the app having mapped a device backed BO into its address space was … [View More]still trying to access the BO while the backing device was gone. To answer this first problem Christian suggested to fix the handling of mapped memory in the clients when the device goes away by forcibly unmap all buffers the user processes has by clearing their respective VMAs mapping the device BOs. Then when the VMAs try to fill in the page tables again we check in the fault handler if the device is removed and if so, return an error. This will generate a SIGBUS to the application which can then cleanly terminate. This indeed was done but this in turn created a problem of kernel OOPs were the OOPSes were due to the fact that while the app was terminating because of the SIGBUS it would trigger use after free in the driver by calling to accesses device structures that were already released from the pci remove sequence. This was handled by introducing a 'flush' sequence during device removal were we wait for drm file reference to drop to 0 meaning all user clients directly using this device terminated. With this I was able to cleanly emulate device unplug with X and glxgears running and later emulate device plug back and restart of X and glxgears. v2: Based on discussions in the mailing list with Daniel and Pekka [1] and based on the document produced by Pekka from those discussions [2] the whole approach with returning SIGBUS and waiting for all user clients having CPU mapping of device BOs to die was dropped. Instead as per the document suggestion the device structures are kept alive until the last reference to the device is dropped by user client and in the meanwhile all existing and new CPU mappings of the BOs belonging to the device directly or by dma-buf import are rerouted to per user process dummy rw page. Also, I skipped the 'Requirements for KMS UAPI' section of [2] since i am trying to get the minimal set of requiremnts that still give useful solution to work and this is the 'Requirements for Render and Cross-Device UAPI' section and so my test case is removing a secondary device, which is render only and is not involved in KMS. This iteration is still more of a draft as I am still facing a few unsolved issues such as a crash in user client when trying to CPU map imported BO if the map happens after device was removed and HW failure to plug back a removed device. Also since i don't have real life setup with external GPU connected through TB I am using sysfs to emulate pci remove and i expect to encounter more issues once i try this on real life case. I am also expecting some help on this from a user who volunteered to test in the related gitlab ticket. So basically this is more of a way to get feedback if I am moving in the right direction. [1] - Discussions during v1 of the patchset https://lists.freedesktop.org/archives/dri-devel/2020-May/265386.html [2] - drm/doc: device hot-unplug for userspace https://www.spinics.net/lists/dri-devel/msg259755.html [3] - Related gitlab ticket https://gitlab.freedesktop.org/drm/amd/-/issues/1081 Andrey Grodzovsky (8): drm: Add dummy page per device or GEM object drm/ttm: Remap all page faults to per process dummy page. drm/ttm: Add unampping of the entire device address space drm/amdgpu: Split amdgpu_device_fini into early and late drm/amdgpu: Refactor sysfs removal drm/amdgpu: Unmap entire device address space on device remove. drm/amdgpu: Fix sdma code crash post device unplug drm/amdgpu: Prevent any job recoveries after device is unplugged. drivers/gpu/drm/amd/amdgpu/amdgpu.h | 19 +++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c | 7 ++- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 50 +++++++++++++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 23 ++++++++-- drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c | 12 +++-- drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 24 ++++++---- drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h | 1 + drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 8 ++++ drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 23 +++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c | 8 +++- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 3 ++ drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c | 21 ++++++--- drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 17 +++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c | 13 +++++- drivers/gpu/drm/amd/amdgpu/df_v3_6.c | 10 +++-- drivers/gpu/drm/drm_file.c | 8 ++++ drivers/gpu/drm/drm_prime.c | 10 +++++ drivers/gpu/drm/ttm/ttm_bo.c | 8 +++- drivers/gpu/drm/ttm/ttm_bo_vm.c | 65 ++++++++++++++++++++++++---- include/drm/drm_file.h | 2 + include/drm/drm_gem.h | 2 + include/drm/ttm/ttm_bo_driver.h | 7 +++ 22 files changed, 286 insertions(+), 55 deletions(-) -- 2.7.4 [View Less]

11 96

[PATCH] fbdev: Remove udlfb driver
by Thomas Zimmermann 02 Dec '20

02 Dec '20

Udlfb has been superseded by DRM's udl. The DRM driver is better by any means and actively maintained. Remove udlfb. Signed-off-by: Thomas Zimmermann <tzimmermann(a)suse.de> --- CREDITS | 5 + Documentation/fb/index.rst | 1 - Documentation/fb/udlfb.rst | 162 --- MAINTAINERS | 9 - drivers/video/fbdev/Kconfig | 17 +- drivers/video/fbdev/Makefile | 1 - drivers/video/fbdev/udlfb.c | 1994 ---------------------------------- 7 … [View More]files changed, 6 insertions(+), 2183 deletions(-) delete mode 100644 Documentation/fb/udlfb.rst delete mode 100644 drivers/video/fbdev/udlfb.c diff --git a/CREDITS b/CREDITS index 748301954ab7..fed3518c4dd8 100644 --- a/CREDITS +++ b/CREDITS @@ -3620,6 +3620,11 @@ S: 145 Howard St. S: Northborough, MA 01532 S: USA +N: Bernie Thompson +E: bernie(a)plugable.com +W: http://plugable.com/category/projects/udlfb/ +D: DisplayLink USB 2.0 framebuffer driver + N: Doug Thompson E: dougthompson(a)xmission.com D: EDAC diff --git a/Documentation/fb/index.rst b/Documentation/fb/index.rst index baf02393d8ee..8136c84328e2 100644 --- a/Documentation/fb/index.rst +++ b/Documentation/fb/index.rst @@ -36,7 +36,6 @@ Frame Buffer sstfb tgafb tridentfb - udlfb uvesafb vesafb viafb diff --git a/Documentation/fb/udlfb.rst b/Documentation/fb/udlfb.rst deleted file mode 100644 index 732b37db3504..000000000000 --- a/Documentation/fb/udlfb.rst +++ /dev/null @@ -1,162 +0,0 @@ -============== -What is udlfb? -============== - -This is a driver for DisplayLink USB 2.0 era graphics chips. - -DisplayLink chips provide simple hline/blit operations with some compression, -pairing that with a hardware framebuffer (16MB) on the other end of the -USB wire. That hardware framebuffer is able to drive the VGA, DVI, or HDMI -monitor with no CPU involvement until a pixel has to change. - -The CPU or other local resource does all the rendering; optionally compares the -result with a local shadow of the remote hardware framebuffer to identify -the minimal set of pixels that have changed; and compresses and sends those -pixels line-by-line via USB bulk transfers. - -Because of the efficiency of bulk transfers and a protocol on top that -does not require any acks - the effect is very low latency that -can support surprisingly high resolutions with good performance for -non-gaming and non-video applications. - -Mode setting, EDID read, etc are other bulk or control transfers. Mode -setting is very flexible - able to set nearly arbitrary modes from any timing. - -Advantages of USB graphics in general: - - * Ability to add a nearly arbitrary number of displays to any USB 2.0 - capable system. On Linux, number of displays is limited by fbdev interface - (FB_MAX is currently 32). Of course, all USB devices on the same - host controller share the same 480Mbs USB 2.0 interface. - -Advantages of supporting DisplayLink chips with kernel framebuffer interface: - - * The actual hardware functionality of DisplayLink chips matches nearly - one-to-one with the fbdev interface, making the driver quite small and - tight relative to the functionality it provides. - * X servers and other applications can use the standard fbdev interface - from user mode to talk to the device, without needing to know anything - about USB or DisplayLink's protocol at all. A "displaylink" X driver - and a slightly modified "fbdev" X driver are among those that already do. - -Disadvantages: - - * Fbdev's mmap interface assumes a real hardware framebuffer is mapped. - In the case of USB graphics, it is just an allocated (virtual) buffer. - Writes need to be detected and encoded into USB bulk transfers by the CPU. - Accurate damage/changed area notifications work around this problem. - In the future, hopefully fbdev will be enhanced with an small standard - interface to allow mmap clients to report damage, for the benefit - of virtual or remote framebuffers. - * Fbdev does not arbitrate client ownership of the framebuffer well. - * Fbcon assumes the first framebuffer it finds should be consumed for console. - * It's not clear what the future of fbdev is, given the rise of KMS/DRM. - -How to use it? -============== - -Udlfb, when loaded as a module, will match against all USB 2.0 generation -DisplayLink chips (Alex and Ollie family). It will then attempt to read the EDID -of the monitor, and set the best common mode between the DisplayLink device -and the monitor's capabilities. - -If the DisplayLink device is successful, it will paint a "green screen" which -means that from a hardware and fbdev software perspective, everything is good. - -At that point, a /dev/fb? interface will be present for user-mode applications -to open and begin writing to the framebuffer of the DisplayLink device using -standard fbdev calls. Note that if mmap() is used, by default the user mode -application must send down damage notifications to trigger repaints of the -changed regions. Alternatively, udlfb can be recompiled with experimental -defio support enabled, to support a page-fault based detection mechanism -that can work without explicit notification. - -The most common client of udlfb is xf86-video-displaylink or a modified -xf86-video-fbdev X server. These servers have no real DisplayLink specific -code. They write to the standard framebuffer interface and rely on udlfb -to do its thing. The one extra feature they have is the ability to report -rectangles from the X DAMAGE protocol extension down to udlfb via udlfb's -damage interface (which will hopefully be standardized for all virtual -framebuffers that need damage info). These damage notifications allow -udlfb to efficiently process the changed pixels. - -Module Options -============== - -Special configuration for udlfb is usually unnecessary. There are a few -options, however. - -From the command line, pass options to modprobe -modprobe udlfb fb_defio=0 console=1 shadow=1 - -Or modify options on the fly at /sys/module/udlfb/parameters directory via -sudo nano fb_defio -change the parameter in place, and save the file. - -Unplug/replug USB device to apply with new settings - -Or for permanent option, create file like /etc/modprobe.d/udlfb.conf with text -options udlfb fb_defio=0 console=1 shadow=1 - -Accepted boolean options: - -=============== ================================================================ -fb_defio Make use of the fb_defio (CONFIG_FB_DEFERRED_IO) kernel - module to track changed areas of the framebuffer by page faults. - Standard fbdev applications that use mmap but that do not - report damage, should be able to work with this enabled. - Disable when running with X server that supports reporting - changed regions via ioctl, as this method is simpler, - more stable, and higher performance. - default: fb_defio=1 - -console Allow fbcon to attach to udlfb provided framebuffers. - Can be disabled if fbcon and other clients - (e.g. X with --shared-vt) are in conflict. - default: console=1 - -shadow Allocate a 2nd framebuffer to shadow what's currently across - the USB bus in device memory. If any pixels are unchanged, - do not transmit. Spends host memory to save USB transfers. - Enabled by default. Only disable on very low memory systems. - default: shadow=1 -=============== ================================================================ - -Sysfs Attributes -================ - -Udlfb creates several files in /sys/class/graphics/fb? -Where ? is the sequential framebuffer id of the particular DisplayLink device - -======================== ======================================================== -edid If a valid EDID blob is written to this file (typically - by a udev rule), then udlfb will use this EDID as a - backup in case reading the actual EDID of the monitor - attached to the DisplayLink device fails. This is - especially useful for fixed panels, etc. that cannot - communicate their capabilities via EDID. Reading - this file returns the current EDID of the attached - monitor (or last backup value written). This is - useful to get the EDID of the attached monitor, - which can be passed to utilities like parse-edid. - -metrics_bytes_rendered 32-bit count of pixel bytes rendered - -metrics_bytes_identical 32-bit count of how many of those bytes were found to be - unchanged, based on a shadow framebuffer check - -metrics_bytes_sent 32-bit count of how many bytes were transferred over - USB to communicate the resulting changed pixels to the - hardware. Includes compression and protocol overhead - -metrics_cpu_kcycles_used 32-bit count of CPU cycles used in processing the - above pixels (in thousands of cycles). - -metrics_reset Write-only. Any write to this file resets all metrics - above to zero. Note that the 32-bit counters above - roll over very quickly. To get reliable results, design - performance tests to start and finish in a very short - period of time (one minute or less is safe). -======================== ======================================================== - -Bernie Thompson <bernie(a)plugable.com> diff --git a/MAINTAINERS b/MAINTAINERS index c5a399541d6a..2b445f87e1ef 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -5178,15 +5178,6 @@ F: fs/quota/ F: include/linux/quota*.h F: include/uapi/linux/quota*.h -DISPLAYLINK USB 2.0 FRAMEBUFFER DRIVER (UDLFB) -M: Bernie Thompson <bernie(a)plugable.com> -L: linux-fbdev(a)vger.kernel.org -S: Maintained -W: http://plugable.com/category/projects/udlfb/ -F: Documentation/fb/udlfb.rst -F: drivers/video/fbdev/udlfb.c -F: include/video/udlfb.h - DISTRIBUTED LOCK MANAGER (DLM) M: Christine Caulfield <ccaulfie(a)redhat.com> M: David Teigland <teigland(a)redhat.com> diff --git a/drivers/video/fbdev/Kconfig b/drivers/video/fbdev/Kconfig index 4f02db65dede..41189b31c5d6 100644 --- a/drivers/video/fbdev/Kconfig +++ b/drivers/video/fbdev/Kconfig @@ -558,7 +558,7 @@ config FB_STI BIOS routines contained in a ROM chip in HP PA-RISC based machines. Enabling this option will implement the linux framebuffer device using calls to the STI BIOS routines for initialisation. - + If you enable this option, you will get a planar framebuffer device /dev/fb which will work on the most common HP graphic cards of the NGLE family, including the artist chips (in the 7xx and Bxxx series), @@ -1939,21 +1939,6 @@ config FB_SMSCUFX (USB 3.0) devices. To compile as a module, choose M here: the module name is smscufx. -config FB_UDL - tristate "Displaylink USB Framebuffer support" - depends on FB && USB - select FB_MODE_HELPERS - select FB_SYS_FILLRECT - select FB_SYS_COPYAREA - select FB_SYS_IMAGEBLIT - select FB_SYS_FOPS - select FB_DEFERRED_IO - help - This is a kernel framebuffer driver for DisplayLink USB devices. - Supports fbdev clients like xf86-video-fbdev, kdrive, fbi, and - mplayer -vo fbdev. Supports all USB 2.0 era DisplayLink devices. - To compile as a module, choose M here: the module name is udlfb. - config FB_IBM_GXT4500 tristate "Framebuffer support for IBM GXT4000P/4500P/6000P/6500P adaptors" depends on FB diff --git a/drivers/video/fbdev/Makefile b/drivers/video/fbdev/Makefile index 477b9624b703..49b5dd0bef02 100644 --- a/drivers/video/fbdev/Makefile +++ b/drivers/video/fbdev/Makefile @@ -106,7 +106,6 @@ obj-$(CONFIG_FB_COBALT) += cobalt_lcdfb.o obj-$(CONFIG_FB_IBM_GXT4500) += gxt4500.o obj-$(CONFIG_FB_PS3) += ps3fb.o obj-$(CONFIG_FB_SM501) += sm501fb.o -obj-$(CONFIG_FB_UDL) += udlfb.o obj-$(CONFIG_FB_SMSCUFX) += smscufx.o obj-$(CONFIG_FB_XILINX) += xilinxfb.o obj-$(CONFIG_FB_SH_MOBILE_LCDC) += sh_mobile_lcdcfb.o diff --git a/drivers/video/fbdev/udlfb.c b/drivers/video/fbdev/udlfb.c deleted file mode 100644 index f9b3c1cb9530..000000000000 --- a/drivers/video/fbdev/udlfb.c +++ /dev/null @@ -1,1994 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0-only -/* - * udlfb.c -- Framebuffer driver for DisplayLink USB controller - * - * Copyright (C) 2009 Roberto De Ioris <roberto(a)unbit.it> - * Copyright (C) 2009 Jaya Kumar <jayakumar.lkml(a)gmail.com> - * Copyright (C) 2009 Bernie Thompson <bernie(a)plugable.com> - * - * Layout is based on skeletonfb by James Simmons and Geert Uytterhoeven, - * usb-skeleton by GregKH. - * - * Device-specific portions based on information from Displaylink, with work - * from Florian Echtler, Henrik Bjerregaard Pedersen, and others. - */ - -#include <linux/module.h> -#include <linux/kernel.h> -#include <linux/init.h> -#include <linux/usb.h> -#include <linux/uaccess.h> -#include <linux/mm.h> -#include <linux/fb.h> -#include <linux/vmalloc.h> -#include <linux/slab.h> -#include <linux/delay.h> -#include <asm/unaligned.h> -#include <video/udlfb.h> -#include "edid.h" - -static const struct fb_fix_screeninfo dlfb_fix = { - .id = "udlfb", - .type = FB_TYPE_PACKED_PIXELS, - .visual = FB_VISUAL_TRUECOLOR, - .xpanstep = 0, - .ypanstep = 0, - .ywrapstep = 0, - .accel = FB_ACCEL_NONE, -}; - -static const u32 udlfb_info_flags = FBINFO_DEFAULT | FBINFO_READS_FAST | - FBINFO_VIRTFB | - FBINFO_HWACCEL_IMAGEBLIT | FBINFO_HWACCEL_FILLRECT | - FBINFO_HWACCEL_COPYAREA | FBINFO_MISC_ALWAYS_SETPAR; - -/* - * There are many DisplayLink-based graphics products, all with unique PIDs. - * So we match on DisplayLink's VID + Vendor-Defined Interface Class (0xff) - * We also require a match on SubClass (0x00) and Protocol (0x00), - * which is compatible with all known USB 2.0 era graphics chips and firmware, - * but allows DisplayLink to increment those for any future incompatible chips - */ -static const struct usb_device_id id_table[] = { - {.idVendor = 0x17e9, - .bInterfaceClass = 0xff, - .bInterfaceSubClass = 0x00, - .bInterfaceProtocol = 0x00, - .match_flags = USB_DEVICE_ID_MATCH_VENDOR | - USB_DEVICE_ID_MATCH_INT_CLASS | - USB_DEVICE_ID_MATCH_INT_SUBCLASS | - USB_DEVICE_ID_MATCH_INT_PROTOCOL, - }, - {}, -}; -MODULE_DEVICE_TABLE(usb, id_table); - -/* module options */ -static bool console = true; /* Allow fbcon to open framebuffer */ -static bool fb_defio = true; /* Detect mmap writes using page faults */ -static bool shadow = true; /* Optionally disable shadow framebuffer */ -static int pixel_limit; /* Optionally force a pixel resolution limit */ - -struct dlfb_deferred_free { - struct list_head list; - void *mem; -}; - -static int dlfb_realloc_framebuffer(struct dlfb_data *dlfb, struct fb_info *info, u32 new_len); - -/* dlfb keeps a list of urbs for efficient bulk transfers */ -static void dlfb_urb_completion(struct urb *urb); -static struct urb *dlfb_get_urb(struct dlfb_data *dlfb); -static int dlfb_submit_urb(struct dlfb_data *dlfb, struct urb * urb, size_t len); -static int dlfb_alloc_urb_list(struct dlfb_data *dlfb, int count, size_t size); -static void dlfb_free_urb_list(struct dlfb_data *dlfb); - -/* - * All DisplayLink bulk operations start with 0xAF, followed by specific code - * All operations are written to buffers which then later get sent to device - */ -static char *dlfb_set_register(char *buf, u8 reg, u8 val) -{ - *buf++ = 0xAF; - *buf++ = 0x20; - *buf++ = reg; - *buf++ = val; - return buf; -} - -static char *dlfb_vidreg_lock(char *buf) -{ - return dlfb_set_register(buf, 0xFF, 0x00); -} - -static char *dlfb_vidreg_unlock(char *buf) -{ - return dlfb_set_register(buf, 0xFF, 0xFF); -} - -/* - * Map FB_BLANK_* to DisplayLink register - * DLReg FB_BLANK_* - * ----- ----------------------------- - * 0x00 FB_BLANK_UNBLANK (0) - * 0x01 FB_BLANK (1) - * 0x03 FB_BLANK_VSYNC_SUSPEND (2) - * 0x05 FB_BLANK_HSYNC_SUSPEND (3) - * 0x07 FB_BLANK_POWERDOWN (4) Note: requires modeset to come back - */ -static char *dlfb_blanking(char *buf, int fb_blank) -{ - u8 reg; - - switch (fb_blank) { - case FB_BLANK_POWERDOWN: - reg = 0x07; - break; - case FB_BLANK_HSYNC_SUSPEND: - reg = 0x05; - break; - case FB_BLANK_VSYNC_SUSPEND: - reg = 0x03; - break; - case FB_BLANK_NORMAL: - reg = 0x01; - break; - default: - reg = 0x00; - } - - buf = dlfb_set_register(buf, 0x1F, reg); - - return buf; -} - -static char *dlfb_set_color_depth(char *buf, u8 selection) -{ - return dlfb_set_register(buf, 0x00, selection); -} - -static char *dlfb_set_base16bpp(char *wrptr, u32 base) -{ - /* the base pointer is 16 bits wide, 0x20 is hi byte. */ - wrptr = dlfb_set_register(wrptr, 0x20, base >> 16); - wrptr = dlfb_set_register(wrptr, 0x21, base >> 8); - return dlfb_set_register(wrptr, 0x22, base); -} - -/* - * DisplayLink HW has separate 16bpp and 8bpp framebuffers. - * In 24bpp modes, the low 323 RGB bits go in the 8bpp framebuffer - */ -static char *dlfb_set_base8bpp(char *wrptr, u32 base) -{ - wrptr = dlfb_set_register(wrptr, 0x26, base >> 16); - wrptr = dlfb_set_register(wrptr, 0x27, base >> 8); - return dlfb_set_register(wrptr, 0x28, base); -} - -static char *dlfb_set_register_16(char *wrptr, u8 reg, u16 value) -{ - wrptr = dlfb_set_register(wrptr, reg, value >> 8); - return dlfb_set_register(wrptr, reg+1, value); -} - -/* - * This is kind of weird because the controller takes some - * register values in a different byte order than other registers. - */ -static char *dlfb_set_register_16be(char *wrptr, u8 reg, u16 value) -{ - wrptr = dlfb_set_register(wrptr, reg, value); - return dlfb_set_register(wrptr, reg+1, value >> 8); -} - -/* - * LFSR is linear feedback shift register. The reason we have this is - * because the display controller needs to minimize the clock depth of - * various counters used in the display path. So this code reverses the - * provided value into the lfsr16 value by counting backwards to get - * the value that needs to be set in the hardware comparator to get the - * same actual count. This makes sense once you read above a couple of - * times and think about it from a hardware perspective. - */ -static u16 dlfb_lfsr16(u16 actual_count) -{ - u32 lv = 0xFFFF; /* This is the lfsr value that the hw starts with */ - - while (actual_count--) { - lv = ((lv << 1) | - (((lv >> 15) ^ (lv >> 4) ^ (lv >> 2) ^ (lv >> 1)) & 1)) - & 0xFFFF; - } - - return (u16) lv; -} - -/* - * This does LFSR conversion on the value that is to be written. - * See LFSR explanation above for more detail. - */ -static char *dlfb_set_register_lfsr16(char *wrptr, u8 reg, u16 value) -{ - return dlfb_set_register_16(wrptr, reg, dlfb_lfsr16(value)); -} - -/* - * This takes a standard fbdev screeninfo struct and all of its monitor mode - * details and converts them into the DisplayLink equivalent register commands. - */ -static char *dlfb_set_vid_cmds(char *wrptr, struct fb_var_screeninfo *var) -{ - u16 xds, yds; - u16 xde, yde; - u16 yec; - - /* x display start */ - xds = var->left_margin + var->hsync_len; - wrptr = dlfb_set_register_lfsr16(wrptr, 0x01, xds); - /* x display end */ - xde = xds + var->xres; - wrptr = dlfb_set_register_lfsr16(wrptr, 0x03, xde); - - /* y display start */ - yds = var->upper_margin + var->vsync_len; - wrptr = dlfb_set_register_lfsr16(wrptr, 0x05, yds); - /* y display end */ - yde = yds + var->yres; - wrptr = dlfb_set_register_lfsr16(wrptr, 0x07, yde); - - /* x end count is active + blanking - 1 */ - wrptr = dlfb_set_register_lfsr16(wrptr, 0x09, - xde + var->right_margin - 1); - - /* libdlo hardcodes hsync start to 1 */ - wrptr = dlfb_set_register_lfsr16(wrptr, 0x0B, 1); - - /* hsync end is width of sync pulse + 1 */ - wrptr = dlfb_set_register_lfsr16(wrptr, 0x0D, var->hsync_len + 1); - - /* hpixels is active pixels */ - wrptr = dlfb_set_register_16(wrptr, 0x0F, var->xres); - - /* yendcount is vertical active + vertical blanking */ - yec = var->yres + var->upper_margin + var->lower_margin + - var->vsync_len; - wrptr = dlfb_set_register_lfsr16(wrptr, 0x11, yec); - - /* libdlo hardcodes vsync start to 0 */ - wrptr = dlfb_set_register_lfsr16(wrptr, 0x13, 0); - - /* vsync end is width of vsync pulse */ - wrptr = dlfb_set_register_lfsr16(wrptr, 0x15, var->vsync_len); - - /* vpixels is active pixels */ - wrptr = dlfb_set_register_16(wrptr, 0x17, var->yres); - - /* convert picoseconds to 5kHz multiple for pclk5k = x * 1E12/5k */ - wrptr = dlfb_set_register_16be(wrptr, 0x1B, - 200*1000*1000/var->pixclock); - - return wrptr; -} - -/* - * This takes a standard fbdev screeninfo struct that was fetched or prepared - * and then generates the appropriate command sequence that then drives the - * display controller. - */ -static int dlfb_set_video_mode(struct dlfb_data *dlfb, - struct fb_var_screeninfo *var) -{ - char *buf; - char *wrptr; - int retval; - int writesize; - struct urb *urb; - - if (!atomic_read(&dlfb->usb_active)) - return -EPERM; - - urb = dlfb_get_urb(dlfb); - if (!urb) - return -ENOMEM; - - buf = (char *) urb->transfer_buffer; - - /* - * This first section has to do with setting the base address on the - * controller * associated with the display. There are 2 base - * pointers, currently, we only * use the 16 bpp segment. - */ - wrptr = dlfb_vidreg_lock(buf); - wrptr = dlfb_set_color_depth(wrptr, 0x00); - /* set base for 16bpp segment to 0 */ - wrptr = dlfb_set_base16bpp(wrptr, 0); - /* set base for 8bpp segment to end of fb */ - wrptr = dlfb_set_base8bpp(wrptr, dlfb->info->fix.smem_len); - - wrptr = dlfb_set_vid_cmds(wrptr, var); - wrptr = dlfb_blanking(wrptr, FB_BLANK_UNBLANK); - wrptr = dlfb_vidreg_unlock(wrptr); - - writesize = wrptr - buf; - - retval = dlfb_submit_urb(dlfb, urb, writesize); - - dlfb->blank_mode = FB_BLANK_UNBLANK; - - return retval; -} - -static int dlfb_ops_mmap(struct fb_info *info, struct vm_area_struct *vma) -{ - unsigned long start = vma->vm_start; - unsigned long size = vma->vm_end - vma->vm_start; - unsigned long offset = vma->vm_pgoff << PAGE_SHIFT; - unsigned long page, pos; - - if (vma->vm_pgoff > (~0UL >> PAGE_SHIFT)) - return -EINVAL; - if (size > info->fix.smem_len) - return -EINVAL; - if (offset > info->fix.smem_len - size) - return -EINVAL; - - pos = (unsigned long)info->fix.smem_start + offset; - - dev_dbg(info->dev, "mmap() framebuffer addr:%lu size:%lu\n", - pos, size); - - while (size > 0) { - page = vmalloc_to_pfn((void *)pos); - if (remap_pfn_range(vma, start, page, PAGE_SIZE, PAGE_SHARED)) - return -EAGAIN; - - start += PAGE_SIZE; - pos += PAGE_SIZE; - if (size > PAGE_SIZE) - size -= PAGE_SIZE; - else - size = 0; - } - - return 0; -} - -/* - * Trims identical data from front and back of line - * Sets new front buffer address and width - * And returns byte count of identical pixels - * Assumes CPU natural alignment (unsigned long) - * for back and front buffer ptrs and width - */ -static int dlfb_trim_hline(const u8 *bback, const u8 **bfront, int *width_bytes) -{ - int j, k; - const unsigned long *back = (const unsigned long *) bback; - const unsigned long *front = (const unsigned long *) *bfront; - const int width = *width_bytes / sizeof(unsigned long); - int identical = width; - int start = width; - int end = width; - - for (j = 0; j < width; j++) { - if (back[j] != front[j]) { - start = j; - break; - } - } - - for (k = width - 1; k > j; k--) { - if (back[k] != front[k]) { - end = k+1; - break; - } - } - - identical = start + (width - end); - *bfront = (u8 *) &front[start]; - *width_bytes = (end - start) * sizeof(unsigned long); - - return identical * sizeof(unsigned long); -} - -/* - * Render a command stream for an encoded horizontal line segment of pixels. - * - * A command buffer holds several commands. - * It always begins with a fresh command header - * (the protocol doesn't require this, but we enforce it to allow - * multiple buffers to be potentially encoded and sent in parallel). - * A single command encodes one contiguous horizontal line of pixels - * - * The function relies on the client to do all allocation, so that - * rendering can be done directly to output buffers (e.g. USB URBs). - * The function fills the supplied command buffer, providing information - * on where it left off, so the client may call in again with additional - * buffers if the line will take several buffers to complete. - * - * A single command can transmit a maximum of 256 pixels, - * regardless of the compression ratio (protocol design limit). - * To the hardware, 0 for a size byte means 256 - * - * Rather than 256 pixel commands which are either rl or raw encoded, - * the rlx command simply assumes alternating raw and rl spans within one cmd. - * This has a slightly larger header overhead, but produces more even results. - * It also processes all data (read and write) in a single pass. - * Performance benchmarks of common cases show it having just slightly better - * compression than 256 pixel raw or rle commands, with similar CPU consumpion. - * But for very rl friendly data, will compress not quite as well. - */ -static void dlfb_compress_hline( - const uint16_t **pixel_start_ptr, - const uint16_t *const pixel_end, - uint32_t *device_address_ptr, - uint8_t **command_buffer_ptr, - const uint8_t *const cmd_buffer_end, - unsigned long back_buffer_offset, - int *ident_ptr) -{ - const uint16_t *pixel = *pixel_start_ptr; - uint32_t dev_addr = *device_address_ptr; - uint8_t *cmd = *command_buffer_ptr; - - while ((pixel_end > pixel) && - (cmd_buffer_end - MIN_RLX_CMD_BYTES > cmd)) { - uint8_t *raw_pixels_count_byte = NULL; - uint8_t *cmd_pixels_count_byte = NULL; - const uint16_t *raw_pixel_start = NULL; - const uint16_t *cmd_pixel_start, *cmd_pixel_end = NULL; - - if (back_buffer_offset && - *pixel == *(u16 *)((u8 *)pixel + back_buffer_offset)) { - pixel++; - dev_addr += BPP; - (*ident_ptr)++; - continue; - } - - *cmd++ = 0xAF; - *cmd++ = 0x6B; - *cmd++ = dev_addr >> 16; - *cmd++ = dev_addr >> 8; - *cmd++ = dev_addr; - - cmd_pixels_count_byte = cmd++; /* we'll know this later */ - cmd_pixel_start = pixel; - - raw_pixels_count_byte = cmd++; /* we'll know this later */ - raw_pixel_start = pixel; - - cmd_pixel_end = pixel + min3(MAX_CMD_PIXELS + 1UL, - (unsigned long)(pixel_end - pixel), - (unsigned long)(cmd_buffer_end - 1 - cmd) / BPP); - - if (back_buffer_offset) { - /* note: the framebuffer may change under us, so we must test for underflow */ - while (cmd_pixel_end - 1 > pixel && - *(cmd_pixel_end - 1) == *(u16 *)((u8 *)(cmd_pixel_end - 1) + back_buffer_offset)) - cmd_pixel_end--; - } - - while (pixel < cmd_pixel_end) { - const uint16_t * const repeating_pixel = pixel; - u16 pixel_value = *pixel; - - put_unaligned_be16(pixel_value, cmd); - if (back_buffer_offset) - *(u16 *)((u8 *)pixel + back_buffer_offset) = pixel_value; - cmd += 2; - pixel++; - - if (unlikely((pixel < cmd_pixel_end) && - (*pixel == pixel_value))) { - /* go back and fill in raw pixel count */ - *raw_pixels_count_byte = ((repeating_pixel - - raw_pixel_start) + 1) & 0xFF; - - do { - if (back_buffer_offset) - *(u16 *)((u8 *)pixel + back_buffer_offset) = pixel_value; - pixel++; - } while ((pixel < cmd_pixel_end) && - (*pixel == pixel_value)); - - /* immediately after raw data is repeat byte */ - *cmd++ = ((pixel - repeating_pixel) - 1) & 0xFF; - - /* Then start another raw pixel span */ - raw_pixel_start = pixel; - raw_pixels_count_byte = cmd++; - } - } - - if (pixel > raw_pixel_start) { - /* finalize last RAW span */ - *raw_pixels_count_byte = (pixel-raw_pixel_start) & 0xFF; - } else { - /* undo unused byte */ - cmd--; - } - - *cmd_pixels_count_byte = (pixel - cmd_pixel_start) & 0xFF; - dev_addr += (u8 *)pixel - (u8 *)cmd_pixel_start; - } - - if (cmd_buffer_end - MIN_RLX_CMD_BYTES <= cmd) { - /* Fill leftover bytes with no-ops */ - if (cmd_buffer_end > cmd) - memset(cmd, 0xAF, cmd_buffer_end - cmd); - cmd = (uint8_t *) cmd_buffer_end; - } - - *command_buffer_ptr = cmd; - *pixel_start_ptr = pixel; - *device_address_ptr = dev_addr; -} - -/* - * There are 3 copies of every pixel: The front buffer that the fbdev - * client renders to, the actual framebuffer across the USB bus in hardware - * (that we can only write to, slowly, and can never read), and (optionally) - * our shadow copy that tracks what's been sent to that hardware buffer. - */ -static int dlfb_render_hline(struct dlfb_data *dlfb, struct urb **urb_ptr, - const char *front, char **urb_buf_ptr, - u32 byte_offset, u32 byte_width, - int *ident_ptr, int *sent_ptr) -{ - const u8 *line_start, *line_end, *next_pixel; - u32 dev_addr = dlfb->base16 + byte_offset; - struct urb *urb = *urb_ptr; - u8 *cmd = *urb_buf_ptr; - u8 *cmd_end = (u8 *) urb->transfer_buffer + urb->transfer_buffer_length; - unsigned long back_buffer_offset = 0; - - line_start = (u8 *) (front + byte_offset); - next_pixel = line_start; - line_end = next_pixel + byte_width; - - if (dlfb->backing_buffer) { - int offset; - const u8 *back_start = (u8 *) (dlfb->backing_buffer - + byte_offset); - - back_buffer_offset = (unsigned long)back_start - (unsigned long)line_start; - - *ident_ptr += dlfb_trim_hline(back_start, &next_pixel, - &byte_width); - - offset = next_pixel - line_start; - line_end = next_pixel + byte_width; - dev_addr += offset; - back_start += offset; - line_start += offset; - } - - while (next_pixel < line_end) { - - dlfb_compress_hline((const uint16_t **) &next_pixel, - (const uint16_t *) line_end, &dev_addr, - (u8 **) &cmd, (u8 *) cmd_end, back_buffer_offset, - ident_ptr); - - if (cmd >= cmd_end) { - int len = cmd - (u8 *) urb->transfer_buffer; - if (dlfb_submit_urb(dlfb, urb, len)) - return 1; /* lost pixels is set */ - *sent_ptr += len; - urb = dlfb_get_urb(dlfb); - if (!urb) - return 1; /* lost_pixels is set */ - *urb_ptr = urb; - cmd = urb->transfer_buffer; - cmd_end = &cmd[urb->transfer_buffer_length]; - } - } - - *urb_buf_ptr = cmd; - - return 0; -} - -static int dlfb_handle_damage(struct dlfb_data *dlfb, int x, int y, int width, int height) -{ - int i, ret; - char *cmd; - cycles_t start_cycles, end_cycles; - int bytes_sent = 0; - int bytes_identical = 0; - struct urb *urb; - int aligned_x; - - start_cycles = get_cycles(); - - mutex_lock(&dlfb->render_mutex); - - aligned_x = DL_ALIGN_DOWN(x, sizeof(unsigned long)); - width = DL_ALIGN_UP(width + (x-aligned_x), sizeof(unsigned long)); - x = aligned_x; - - if ((width <= 0) || - (x + width > dlfb->info->var.xres) || - (y + height > dlfb->info->var.yres)) { - ret = -EINVAL; - goto unlock_ret; - } - - if (!atomic_read(&dlfb->usb_active)) { - ret = 0; - goto unlock_ret; - } - - urb = dlfb_get_urb(dlfb); - if (!urb) { - ret = 0; - goto unlock_ret; - } - cmd = urb->transfer_buffer; - - for (i = y; i < y + height ; i++) { - const int line_offset = dlfb->info->fix.line_length * i; - const int byte_offset = line_offset + (x * BPP); - - if (dlfb_render_hline(dlfb, &urb, - (char *) dlfb->info->fix.smem_start, - &cmd, byte_offset, width * BPP, - &bytes_identical, &bytes_sent)) - goto error; - } - - if (cmd > (char *) urb->transfer_buffer) { - int len; - if (cmd < (char *) urb->transfer_buffer + urb->transfer_buffer_length) - *cmd++ = 0xAF; - /* Send partial buffer remaining before exiting */ - len = cmd - (char *) urb->transfer_buffer; - dlfb_submit_urb(dlfb, urb, len); - bytes_sent += len; - } else - dlfb_urb_completion(urb); - -error: - atomic_add(bytes_sent, &dlfb->bytes_sent); - atomic_add(bytes_identical, &dlfb->bytes_identical); - atomic_add(width*height*2, &dlfb->bytes_rendered); - end_cycles = get_cycles(); - atomic_add(((unsigned int) ((end_cycles - start_cycles) - >> 10)), /* Kcycles */ - &dlfb->cpu_kcycles_used); - - ret = 0; - -unlock_ret: - mutex_unlock(&dlfb->render_mutex); - return ret; -} - -static void dlfb_init_damage(struct dlfb_data *dlfb) -{ - dlfb->damage_x = INT_MAX; - dlfb->damage_x2 = 0; - dlfb->damage_y = INT_MAX; - dlfb->damage_y2 = 0; -} - -static void dlfb_damage_work(struct work_struct *w) -{ - struct dlfb_data *dlfb = container_of(w, struct dlfb_data, damage_work); - int x, x2, y, y2; - - spin_lock_irq(&dlfb->damage_lock); - x = dlfb->damage_x; - x2 = dlfb->damage_x2; - y = dlfb->damage_y; - y2 = dlfb->damage_y2; - dlfb_init_damage(dlfb); - spin_unlock_irq(&dlfb->damage_lock); - - if (x < x2 && y < y2) - dlfb_handle_damage(dlfb, x, y, x2 - x, y2 - y); -} - -static void dlfb_offload_damage(struct dlfb_data *dlfb, int x, int y, int width, int height) -{ - unsigned long flags; - int x2 = x + width; - int y2 = y + height; - - if (x >= x2 || y >= y2) - return; - - spin_lock_irqsave(&dlfb->damage_lock, flags); - dlfb->damage_x = min(x, dlfb->damage_x); - dlfb->damage_x2 = max(x2, dlfb->damage_x2); - dlfb->damage_y = min(y, dlfb->damage_y); - dlfb->damage_y2 = max(y2, dlfb->damage_y2); - spin_unlock_irqrestore(&dlfb->damage_lock, flags); - - schedule_work(&dlfb->damage_work); -} - -/* - * Path triggered by usermode clients who write to filesystem - * e.g. cat filename > /dev/fb1 - * Not used by X Windows or text-mode console. But useful for testing. - * Slow because of extra copy and we must assume all pixels dirty. - */ -static ssize_t dlfb_ops_write(struct fb_info *info, const char __user *buf, - size_t count, loff_t *ppos) -{ - ssize_t result; - struct dlfb_data *dlfb = info->par; - u32 offset = (u32) *ppos; - - result = fb_sys_write(info, buf, count, ppos); - - if (result > 0) { - int start = max((int)(offset / info->fix.line_length), 0); - int lines = min((u32)((result / info->fix.line_length) + 1), - (u32)info->var.yres); - - dlfb_handle_damage(dlfb, 0, start, info->var.xres, - lines); - } - - return result; -} - -/* hardware has native COPY command (see libdlo), but not worth it for fbcon */ -static void dlfb_ops_copyarea(struct fb_info *info, - const struct fb_copyarea *area) -{ - - struct dlfb_data *dlfb = info->par; - - sys_copyarea(info, area); - - dlfb_offload_damage(dlfb, area->dx, area->dy, - area->width, area->height); -} - -static void dlfb_ops_imageblit(struct fb_info *info, - const struct fb_image *image) -{ - struct dlfb_data *dlfb = info->par; - - sys_imageblit(info, image); - - dlfb_offload_damage(dlfb, image->dx, image->dy, - image->width, image->height); -} - -static void dlfb_ops_fillrect(struct fb_info *info, - const struct fb_fillrect *rect) -{ - struct dlfb_data *dlfb = info->par; - - sys_fillrect(info, rect); - - dlfb_offload_damage(dlfb, rect->dx, rect->dy, rect->width, - rect->height); -} - -/* - * NOTE: fb_defio.c is holding info->fbdefio.mutex - * Touching ANY framebuffer memory that triggers a page fault - * in fb_defio will cause a deadlock, when it also tries to - * grab the same mutex. - */ -static void dlfb_dpy_deferred_io(struct fb_info *info, - struct list_head *pagelist) -{ - struct page *cur; - struct fb_deferred_io *fbdefio = info->fbdefio; - struct dlfb_data *dlfb = info->par; - struct urb *urb; - char *cmd; - cycles_t start_cycles, end_cycles; - int bytes_sent = 0; - int bytes_identical = 0; - int bytes_rendered = 0; - - mutex_lock(&dlfb->render_mutex); - - if (!fb_defio) - goto unlock_ret; - - if (!atomic_read(&dlfb->usb_active)) - goto unlock_ret; - - start_cycles = get_cycles(); - - urb = dlfb_get_urb(dlfb); - if (!urb) - goto unlock_ret; - - cmd = urb->transfer_buffer; - - /* walk the written page list and render each to device */ - list_for_each_entry(cur, &fbdefio->pagelist, lru) { - - if (dlfb_render_hline(dlfb, &urb, (char *) info->fix.smem_start, - &cmd, cur->index << PAGE_SHIFT, - PAGE_SIZE, &bytes_identical, &bytes_sent)) - goto error; - bytes_rendered += PAGE_SIZE; - } - - if (cmd > (char *) urb->transfer_buffer) { - int len; - if (cmd < (char *) urb->transfer_buffer + urb->transfer_buffer_length) - *cmd++ = 0xAF; - /* Send partial buffer remaining before exiting */ - len = cmd - (char *) urb->transfer_buffer; - dlfb_submit_urb(dlfb, urb, len); - bytes_sent += len; - } else - dlfb_urb_completion(urb); - -error: - atomic_add(bytes_sent, &dlfb->bytes_sent); - atomic_add(bytes_identical, &dlfb->bytes_identical); - atomic_add(bytes_rendered, &dlfb->bytes_rendered); - end_cycles = get_cycles(); - atomic_add(((unsigned int) ((end_cycles - start_cycles) - >> 10)), /* Kcycles */ - &dlfb->cpu_kcycles_used); -unlock_ret: - mutex_unlock(&dlfb->render_mutex); -} - -static int dlfb_get_edid(struct dlfb_data *dlfb, char *edid, int len) -{ - int i, ret; - char *rbuf; - - rbuf = kmalloc(2, GFP_KERNEL); - if (!rbuf) - return 0; - - for (i = 0; i < len; i++) { - ret = usb_control_msg(dlfb->udev, - usb_rcvctrlpipe(dlfb->udev, 0), 0x02, - (0x80 | (0x02 << 5)), i << 8, 0xA1, - rbuf, 2, USB_CTRL_GET_TIMEOUT); - if (ret < 2) { - dev_err(&dlfb->udev->dev, - "Read EDID byte %d failed: %d\n", i, ret); - i--; - break; - } - edid[i] = rbuf[1]; - } - - kfree(rbuf); - - return i; -} - -static int dlfb_ops_ioctl(struct fb_info *info, unsigned int cmd, - unsigned long arg) -{ - - struct dlfb_data *dlfb = info->par; - - if (!atomic_read(&dlfb->usb_active)) - return 0; - - /* TODO: Update X server to get this from sysfs instead */ - if (cmd == DLFB_IOCTL_RETURN_EDID) { - void __user *edid = (void __user *)arg; - if (copy_to_user(edid, dlfb->edid, dlfb->edid_size)) - return -EFAULT; - return 0; - } - - /* TODO: Help propose a standard fb.h ioctl to report mmap damage */ - if (cmd == DLFB_IOCTL_REPORT_DAMAGE) { - struct dloarea area; - - if (copy_from_user(&area, (void __user *)arg, - sizeof(struct dloarea))) - return -EFAULT; - - /* - * If we have a damage-aware client, turn fb_defio "off" - * To avoid perf imact of unnecessary page fault handling. - * Done by resetting the delay for this fb_info to a very - * long period. Pages will become writable and stay that way. - * Reset to normal value when all clients have closed this fb. - */ - if (info->fbdefio) - info->fbdefio->delay = DL_DEFIO_WRITE_DISABLE; - - if (area.x < 0) - area.x = 0; - - if (area.x > info->var.xres) - area.x = info->var.xres; - - if (area.y < 0) - area.y = 0; - - if (area.y > info->var.yres) - area.y = info->var.yres; - - dlfb_handle_damage(dlfb, area.x, area.y, area.w, area.h); - } - - return 0; -} - -/* taken from vesafb */ -static int -dlfb_ops_setcolreg(unsigned regno, unsigned red, unsigned green, - unsigned blue, unsigned transp, struct fb_info *info) -{ - int err = 0; - - if (regno >= info->cmap.len) - return 1; - - if (regno < 16) { - if (info->var.red.offset == 10) { - /* 1:5:5:5 */ - ((u32 *) (info->pseudo_palette))[regno] = - ((red & 0xf800) >> 1) | - ((green & 0xf800) >> 6) | ((blue & 0xf800) >> 11); - } else { - /* 0:5:6:5 */ - ((u32 *) (info->pseudo_palette))[regno] = - ((red & 0xf800)) | - ((green & 0xfc00) >> 5) | ((blue & 0xf800) >> 11); - } - } - - return err; -} - -/* - * It's common for several clients to have framebuffer open simultaneously. - * e.g. both fbcon and X. Makes things interesting. - * Assumes caller is holding info->lock (for open and release at least) - */ -static int dlfb_ops_open(struct fb_info *info, int user) -{ - struct dlfb_data *dlfb = info->par; - - /* - * fbcon aggressively connects to first framebuffer it finds, - * preventing other clients (X) from working properly. Usually - * not what the user wants. Fail by default with option to enable. - */ - if ((user == 0) && (!console)) - return -EBUSY; - - /* If the USB device is gone, we don't accept new opens */ - if (dlfb->virtualized) - return -ENODEV; - - dlfb->fb_count++; - - if (fb_defio && (info->fbdefio == NULL)) { - /* enable defio at last moment if not disabled by client */ - - struct fb_deferred_io *fbdefio; - - fbdefio = kzalloc(sizeof(struct fb_deferred_io), GFP_KERNEL); - - if (fbdefio) { - fbdefio->delay = DL_DEFIO_WRITE_DELAY; - fbdefio->deferred_io = dlfb_dpy_deferred_io; - } - - info->fbdefio = fbdefio; - fb_deferred_io_init(info); - } - - dev_dbg(info->dev, "open, user=%d fb_info=%p count=%d\n", - user, info, dlfb->fb_count); - - return 0; -} - -static void dlfb_ops_destroy(struct fb_info *info) -{ - struct dlfb_data *dlfb = info->par; - - cancel_work_sync(&dlfb->damage_work); - - mutex_destroy(&dlfb->render_mutex); - - if (info->cmap.len != 0) - fb_dealloc_cmap(&info->cmap); - if (info->monspecs.modedb) - fb_destroy_modedb(info->monspecs.modedb); - vfree(info->screen_base); - - fb_destroy_modelist(&info->modelist); - - while (!list_empty(&dlfb->deferred_free)) { - struct dlfb_deferred_free *d = list_entry(dlfb->deferred_free.next, struct dlfb_deferred_free, list); - list_del(&d->list); - vfree(d->mem); - kfree(d); - } - vfree(dlfb->backing_buffer); - kfree(dlfb->edid); - usb_put_dev(dlfb->udev); - kfree(dlfb); - - /* Assume info structure is freed after this point */ - framebuffer_release(info); -} - -/* - * Assumes caller is holding info->lock mutex (for open and release at least) - */ -static int dlfb_ops_release(struct fb_info *info, int user) -{ - struct dlfb_data *dlfb = info->par; - - dlfb->fb_count--; - - if ((dlfb->fb_count == 0) && (info->fbdefio)) { - fb_deferred_io_cleanup(info); - kfree(info->fbdefio); - info->fbdefio = NULL; - } - - dev_dbg(info->dev, "release, user=%d count=%d\n", user, dlfb->fb_count); - - return 0; -} - -/* - * Check whether a video mode is supported by the DisplayLink chip - * We start from monitor's modes, so don't need to filter that here - */ -static int dlfb_is_valid_mode(struct fb_videomode *mode, struct dlfb_data *dlfb) -{ - if (mode->xres * mode->yres > dlfb->sku_pixel_limit) - return 0; - - return 1; -} - -static void dlfb_var_color_format(struct fb_var_screeninfo *var) -{ - const struct fb_bitfield red = { 11, 5, 0 }; - const struct fb_bitfield green = { 5, 6, 0 }; - const struct fb_bitfield blue = { 0, 5, 0 }; - - var->bits_per_pixel = 16; - var->red = red; - var->green = green; - var->blue = blue; -} - -static int dlfb_ops_check_var(struct fb_var_screeninfo *var, - struct fb_info *info) -{ - struct fb_videomode mode; - struct dlfb_data *dlfb = info->par; - - /* set device-specific elements of var unrelated to mode */ - dlfb_var_color_format(var); - - fb_var_to_videomode(&mode, var); - - if (!dlfb_is_valid_mode(&mode, dlfb)) - return -EINVAL; - - return 0; -} - -static int dlfb_ops_set_par(struct fb_info *info) -{ - struct dlfb_data *dlfb = info->par; - int result; - u16 *pix_framebuffer; - int i; - struct fb_var_screeninfo fvs; - u32 line_length = info->var.xres * (info->var.bits_per_pixel / 8); - - /* clear the activate field because it causes spurious miscompares */ - fvs = info->var; - fvs.activate = 0; - fvs.vmode &= ~FB_VMODE_SMOOTH_XPAN; - - if (!memcmp(&dlfb->current_mode, &fvs, sizeof(struct fb_var_screeninfo))) - return 0; - - result = dlfb_realloc_framebuffer(dlfb, info, info->var.yres * line_length); - if (result) - return result; - - result = dlfb_set_video_mode(dlfb, &info->var); - - if (result) - return result; - - dlfb->current_mode = fvs; - info->fix.line_length = line_length; - - if (dlfb->fb_count == 0) { - - /* paint greenscreen */ - - pix_framebuffer = (u16 *) info->screen_base; - for (i = 0; i < info->fix.smem_len / 2; i++) - pix_framebuffer[i] = 0x37e6; - } - - dlfb_handle_damage(dlfb, 0, 0, info->var.xres, info->var.yres); - - return 0; -} - -/* To fonzi the jukebox (e.g. make blanking changes take effect) */ -static char *dlfb_dummy_render(char *buf) -{ - *buf++ = 0xAF; - *buf++ = 0x6A; /* copy */ - *buf++ = 0x00; /* from address*/ - *buf++ = 0x00; - *buf++ = 0x00; - *buf++ = 0x01; /* one pixel */ - *buf++ = 0x00; /* to address */ - *buf++ = 0x00; - *buf++ = 0x00; - return buf; -} - -/* - * In order to come back from full DPMS off, we need to set the mode again - */ -static int dlfb_ops_blank(int blank_mode, struct fb_info *info) -{ - struct dlfb_data *dlfb = info->par; - char *bufptr; - struct urb *urb; - - dev_dbg(info->dev, "blank, mode %d --> %d\n", - dlfb->blank_mode, blank_mode); - - if ((dlfb->blank_mode == FB_BLANK_POWERDOWN) && - (blank_mode != FB_BLANK_POWERDOWN)) { - - /* returning from powerdown requires a fresh modeset */ - dlfb_set_video_mode(dlfb, &info->var); - } - - urb = dlfb_get_urb(dlfb); - if (!urb) - return 0; - - bufptr = (char *) urb->transfer_buffer; - bufptr = dlfb_vidreg_lock(bufptr); - bufptr = dlfb_blanking(bufptr, blank_mode); - bufptr = dlfb_vidreg_unlock(bufptr); - - /* seems like a render op is needed to have blank change take effect */ - bufptr = dlfb_dummy_render(bufptr); - - dlfb_submit_urb(dlfb, urb, bufptr - - (char *) urb->transfer_buffer); - - dlfb->blank_mode = blank_mode; - - return 0; -} - -static const struct fb_ops dlfb_ops = { - .owner = THIS_MODULE, - .fb_read = fb_sys_read, - .fb_write = dlfb_ops_write, - .fb_setcolreg = dlfb_ops_setcolreg, - .fb_fillrect = dlfb_ops_fillrect, - .fb_copyarea = dlfb_ops_copyarea, - .fb_imageblit = dlfb_ops_imageblit, - .fb_mmap = dlfb_ops_mmap, - .fb_ioctl = dlfb_ops_ioctl, - .fb_open = dlfb_ops_open, - .fb_release = dlfb_ops_release, - .fb_blank = dlfb_ops_blank, - .fb_check_var = dlfb_ops_check_var, - .fb_set_par = dlfb_ops_set_par, - .fb_destroy = dlfb_ops_destroy, -}; - - -static void dlfb_deferred_vfree(struct dlfb_data *dlfb, void *mem) -{ - struct dlfb_deferred_free *d = kmalloc(sizeof(struct dlfb_deferred_free), GFP_KERNEL); - if (!d) - return; - d->mem = mem; - list_add(&d->list, &dlfb->deferred_free); -} - -/* - * Assumes &info->lock held by caller - * Assumes no active clients have framebuffer open - */ -static int dlfb_realloc_framebuffer(struct dlfb_data *dlfb, struct fb_info *info, u32 new_len) -{ - u32 old_len = info->fix.smem_len; - const void *old_fb = (const void __force *)info->screen_base; - unsigned char *new_fb; - unsigned char *new_back = NULL; - - new_len = PAGE_ALIGN(new_len); - - if (new_len > old_len) { - /* - * Alloc system memory for virtual framebuffer - */ - new_fb = vmalloc(new_len); - if (!new_fb) { - dev_err(info->dev, "Virtual framebuffer alloc failed\n"); - return -ENOMEM; - } - memset(new_fb, 0xff, new_len); - - if (info->screen_base) { - memcpy(new_fb, old_fb, old_len); - dlfb_deferred_vfree(dlfb, (void __force *)info->screen_base); - } - - info->screen_base = (char __iomem *)new_fb; - info->fix.smem_len = new_len; - info->fix.smem_start = (unsigned long) new_fb; - info->flags = udlfb_info_flags; - - /* - * Second framebuffer copy to mirror the framebuffer state - * on the physical USB device. We can function without this. - * But with imperfect damage info we may send pixels over USB - * that were, in fact, unchanged - wasting limited USB bandwidth - */ - if (shadow) - new_back = vzalloc(new_len); - if (!new_back) - dev_info(info->dev, - "No shadow/backing buffer allocated\n"); - else { - dlfb_deferred_vfree(dlfb, dlfb->backing_buffer); - dlfb->backing_buffer = new_back; - } - } - return 0; -} - -/* - * 1) Get EDID from hw, or use sw default - * 2) Parse into various fb_info structs - * 3) Allocate virtual framebuffer memory to back highest res mode - * - * Parses EDID into three places used by various parts of fbdev: - * fb_var_screeninfo contains the timing of the monitor's preferred mode - * fb_info.monspecs is full parsed EDID info, including monspecs.modedb - * fb_info.modelist is a linked list of all monitor & VESA modes which work - * - * If EDID is not readable/valid, then modelist is all VESA modes, - * monspecs is NULL, and fb_var_screeninfo is set to safe VESA mode - * Returns 0 if successful - */ -static int dlfb_setup_modes(struct dlfb_data *dlfb, - struct fb_info *info, - char *default_edid, size_t default_edid_size) -{ - char *edid; - int i, result = 0, tries = 3; - struct device *dev = info->device; - struct fb_videomode *mode; - const struct fb_videomode *default_vmode = NULL; - - if (info->dev) { - /* only use mutex if info has been registered */ - mutex_lock(&info->lock); - /* parent device is used otherwise */ - dev = info->dev; - } - - edid = kmalloc(EDID_LENGTH, GFP_KERNEL); - if (!edid) { - result = -ENOMEM; - goto error; - } - - fb_destroy_modelist(&info->modelist); - memset(&info->monspecs, 0, sizeof(info->monspecs)); - - /* - * Try to (re)read EDID from hardware first - * EDID data may return, but not parse as valid - * Try again a few times, in case of e.g. analog cable noise - */ - while (tries--) { - - i = dlfb_get_edid(dlfb, edid, EDID_LENGTH); - - if (i >= EDID_LENGTH) - fb_edid_to_monspecs(edid, &info->monspecs); - - if (info->monspecs.modedb_len > 0) { - dlfb->edid = edid; - dlfb->edid_size = i; - break; - } - } - - /* If that fails, use a previously returned EDID if available */ - if (info->monspecs.modedb_len == 0) { - dev_err(dev, "Unable to get valid EDID from device/display\n"); - - if (dlfb->edid) { - fb_edid_to_monspecs(dlfb->edid, &info->monspecs); - if (info->monspecs.modedb_len > 0) - dev_err(dev, "Using previously queried EDID\n"); - } - } - - /* If that fails, use the default EDID we were handed */ - if (info->monspecs.modedb_len == 0) { - if (default_edid_size >= EDID_LENGTH) { - fb_edid_to_monspecs(default_edid, &info->monspecs); - if (info->monspecs.modedb_len > 0) { - memcpy(edid, default_edid, default_edid_size); - dlfb->edid = edid; - dlfb->edid_size = default_edid_size; - dev_err(dev, "Using default/backup EDID\n"); - } - } - } - - /* If we've got modes, let's pick a best default mode */ - if (info->monspecs.modedb_len > 0) { - - for (i = 0; i < info->monspecs.modedb_len; i++) { - mode = &info->monspecs.modedb[i]; - if (dlfb_is_valid_mode(mode, dlfb)) { - fb_add_videomode(mode, &info->modelist); - } else { - dev_dbg(dev, "Specified mode %dx%d too big\n", - mode->xres, mode->yres); - if (i == 0) - /* if we've removed top/best mode */ - info->monspecs.misc - &= ~FB_MISC_1ST_DETAIL; - } - } - - default_vmode = fb_find_best_display(&info->monspecs, - &info->modelist); - } - - /* If everything else has failed, fall back to safe default mode */ - if (default_vmode == NULL) { - - struct fb_videomode fb_vmode = {0}; - - /* - * Add the standard VESA modes to our modelist - * Since we don't have EDID, there may be modes that - * overspec monitor and/or are incorrect aspect ratio, etc. - * But at least the user has a chance to choose - */ - for (i = 0; i < VESA_MODEDB_SIZE; i++) { - mode = (struct fb_videomode *)&vesa_modes[i]; - if (dlfb_is_valid_mode(mode, dlfb)) - fb_add_videomode(mode, &info->modelist); - else - dev_dbg(dev, "VESA mode %dx%d too big\n", - mode->xres, mode->yres); - } - - /* - * default to resolution safe for projectors - * (since they are most common case without EDID) - */ - fb_vmode.xres = 800; - fb_vmode.yres = 600; - fb_vmode.refresh = 60; - default_vmode = fb_find_nearest_mode(&fb_vmode, - &info->modelist); - } - - /* If we have good mode and no active clients*/ - if ((default_vmode != NULL) && (dlfb->fb_count == 0)) { - - fb_videomode_to_var(&info->var, default_vmode); - dlfb_var_color_format(&info->var); - - /* - * with mode size info, we can now alloc our framebuffer. - */ - memcpy(&info->fix, &dlfb_fix, sizeof(dlfb_fix)); - } else - result = -EINVAL; - -error: - if (edid && (dlfb->edid != edid)) - kfree(edid); - - if (info->dev) - mutex_unlock(&info->lock); - - return result; -} - -static ssize_t metrics_bytes_rendered_show(struct device *fbdev, - struct device_attribute *a, char *buf) { - struct fb_info *fb_info = dev_get_drvdata(fbdev); - struct dlfb_data *dlfb = fb_info->par; - return snprintf(buf, PAGE_SIZE, "%u\n", - atomic_read(&dlfb->bytes_rendered)); -} - -static ssize_t metrics_bytes_identical_show(struct device *fbdev, - struct device_attribute *a, char *buf) { - struct fb_info *fb_info = dev_get_drvdata(fbdev); - struct dlfb_data *dlfb = fb_info->par; - return snprintf(buf, PAGE_SIZE, "%u\n", - atomic_read(&dlfb->bytes_identical)); -} - -static ssize_t metrics_bytes_sent_show(struct device *fbdev, - struct device_attribute *a, char *buf) { - struct fb_info *fb_info = dev_get_drvdata(fbdev); - struct dlfb_data *dlfb = fb_info->par; - return snprintf(buf, PAGE_SIZE, "%u\n", - atomic_read(&dlfb->bytes_sent)); -} - -static ssize_t metrics_cpu_kcycles_used_show(struct device *fbdev, - struct device_attribute *a, char *buf) { - struct fb_info *fb_info = dev_get_drvdata(fbdev); - struct dlfb_data *dlfb = fb_info->par; - return snprintf(buf, PAGE_SIZE, "%u\n", - atomic_read(&dlfb->cpu_kcycles_used)); -} - -static ssize_t edid_show( - struct file *filp, - struct kobject *kobj, struct bin_attribute *a, - char *buf, loff_t off, size_t count) { - struct device *fbdev = kobj_to_dev(kobj); - struct fb_info *fb_info = dev_get_drvdata(fbdev); - struct dlfb_data *dlfb = fb_info->par; - - if (dlfb->edid == NULL) - return 0; - - if ((off >= dlfb->edid_size) || (count > dlfb->edid_size)) - return 0; - - if (off + count > dlfb->edid_size) - count = dlfb->edid_size - off; - - memcpy(buf, dlfb->edid, count); - - return count; -} - -static ssize_t edid_store( - struct file *filp, - struct kobject *kobj, struct bin_attribute *a, - char *src, loff_t src_off, size_t src_size) { - struct device *fbdev = kobj_to_dev(kobj); - struct fb_info *fb_info = dev_get_drvdata(fbdev); - struct dlfb_data *dlfb = fb_info->par; - int ret; - - /* We only support write of entire EDID at once, no offset*/ - if ((src_size != EDID_LENGTH) || (src_off != 0)) - return -EINVAL; - - ret = dlfb_setup_modes(dlfb, fb_info, src, src_size); - if (ret) - return ret; - - if (!dlfb->edid || memcmp(src, dlfb->edid, src_size)) - return -EINVAL; - - ret = dlfb_ops_set_par(fb_info); - if (ret) - return ret; - - return src_size; -} - -static ssize_t metrics_reset_store(struct device *fbdev, - struct device_attribute *attr, - const char *buf, size_t count) -{ - struct fb_info *fb_info = dev_get_drvdata(fbdev); - struct dlfb_data *dlfb = fb_info->par; - - atomic_set(&dlfb->bytes_rendered, 0); - atomic_set(&dlfb->bytes_identical, 0); - atomic_set(&dlfb->bytes_sent, 0); - atomic_set(&dlfb->cpu_kcycles_used, 0); - - return count; -} - -static const struct bin_attribute edid_attr = { - .attr.name = "edid", - .attr.mode = 0666, - .size = EDID_LENGTH, - .read = edid_show, - .write = edid_store -}; - -static const struct device_attribute fb_device_attrs[] = { - __ATTR_RO(metrics_bytes_rendered), - __ATTR_RO(metrics_bytes_identical), - __ATTR_RO(metrics_bytes_sent), - __ATTR_RO(metrics_cpu_kcycles_used), - __ATTR(metrics_reset, S_IWUSR, NULL, metrics_reset_store), -}; - -/* - * This is necessary before we can communicate with the display controller. - */ -static int dlfb_select_std_channel(struct dlfb_data *dlfb) -{ - int ret; - void *buf; - static const u8 set_def_chn[] = { - 0x57, 0xCD, 0xDC, 0xA7, - 0x1C, 0x88, 0x5E, 0x15, - 0x60, 0xFE, 0xC6, 0x97, - 0x16, 0x3D, 0x47, 0xF2 }; - - buf = kmemdup(set_def_chn, sizeof(set_def_chn), GFP_KERNEL); - - if (!buf) - return -ENOMEM; - - ret = usb_control_msg(dlfb->udev, usb_sndctrlpipe(dlfb->udev, 0), - NR_USB_REQUEST_CHANNEL, - (USB_DIR_OUT | USB_TYPE_VENDOR), 0, 0, - buf, sizeof(set_def_chn), USB_CTRL_SET_TIMEOUT); - - kfree(buf); - - return ret; -} - -static int dlfb_parse_vendor_descriptor(struct dlfb_data *dlfb, - struct usb_interface *intf) -{ - char *desc; - char *buf; - char *desc_end; - int total_len; - - buf = kzalloc(MAX_VENDOR_DESCRIPTOR_SIZE, GFP_KERNEL); - if (!buf) - return false; - desc = buf; - - total_len = usb_get_descriptor(interface_to_usbdev(intf), - 0x5f, /* vendor specific */ - 0, desc, MAX_VENDOR_DESCRIPTOR_SIZE); - - /* if not found, look in configuration descriptor */ - if (total_len < 0) { - if (0 == usb_get_extra_descriptor(intf->cur_altsetting, - 0x5f, &desc)) - total_len = (int) desc[0]; - } - - if (total_len > 5) { - dev_info(&intf->dev, - "vendor descriptor length: %d data: %11ph\n", - total_len, desc); - - if ((desc[0] != total_len) || /* descriptor length */ - (desc[1] != 0x5f) || /* vendor descriptor type */ - (desc[2] != 0x01) || /* version (2 bytes) */ - (desc[3] != 0x00) || - (desc[4] != total_len - 2)) /* length after type */ - goto unrecognized; - - desc_end = desc + total_len; - desc += 5; /* the fixed header we've already parsed */ - - while (desc < desc_end) { - u8 length; - u16 key; - - key = *desc++; - key |= (u16)*desc++ << 8; - length = *desc++; - - switch (key) { - case 0x0200: { /* max_area */ - u32 max_area = *desc++; - max_area |= (u32)*desc++ << 8; - max_area |= (u32)*desc++ << 16; - max_area |= (u32)*desc++ << 24; - dev_warn(&intf->dev, - "DL chip limited to %d pixel modes\n", - max_area); - dlfb->sku_pixel_limit = max_area; - break; - } - default: - break; - } - desc += length; - } - } else { - dev_info(&intf->dev, "vendor descriptor not available (%d)\n", - total_len); - } - - goto success; - -unrecognized: - /* allow udlfb to load for now even if firmware unrecognized */ - dev_err(&intf->dev, "Unrecognized vendor firmware descriptor\n"); - -success: - kfree(buf); - return true; -} - -static int dlfb_usb_probe(struct usb_interface *intf, - const struct usb_device_id *id) -{ - int i; - const struct device_attribute *attr; - struct dlfb_data *dlfb; - struct fb_info *info; - int retval = -ENOMEM; - struct usb_device *usbdev = interface_to_usbdev(intf); - - /* usb initialization */ - dlfb = kzalloc(sizeof(*dlfb), GFP_KERNEL); - if (!dlfb) { - dev_err(&intf->dev, "%s: failed to allocate dlfb\n", __func__); - return -ENOMEM; - } - - INIT_LIST_HEAD(&dlfb->deferred_free); - - dlfb->udev = usb_get_dev(usbdev); - usb_set_intfdata(intf, dlfb); - - dev_dbg(&intf->dev, "console enable=%d\n", console); - dev_dbg(&intf->dev, "fb_defio enable=%d\n", fb_defio); - dev_dbg(&intf->dev, "shadow enable=%d\n", shadow); - - dlfb->sku_pixel_limit = 2048 * 1152; /* default to maximum */ - - if (!dlfb_parse_vendor_descriptor(dlfb, intf)) { - dev_err(&intf->dev, - "firmware not recognized, incompatible device?\n"); - goto error; - } - - if (pixel_limit) { - dev_warn(&intf->dev, - "DL chip limit of %d overridden to %d\n", - dlfb->sku_pixel_limit, pixel_limit); - dlfb->sku_pixel_limit = pixel_limit; - } - - - /* allocates framebuffer driver structure, not framebuffer memory */ - info = framebuffer_alloc(0, &dlfb->udev->dev); - if (!info) - goto error; - - dlfb->info = info; - info->par = dlfb; - info->pseudo_palette = dlfb->pseudo_palette; - dlfb->ops = dlfb_ops; - info->fbops = &dlfb->ops; - - mutex_init(&dlfb->render_mutex); - dlfb_init_damage(dlfb); - spin_lock_init(&dlfb->damage_lock); - INIT_WORK(&dlfb->damage_work, dlfb_damage_work); - - INIT_LIST_HEAD(&info->modelist); - - if (!dlfb_alloc_urb_list(dlfb, WRITES_IN_FLIGHT, MAX_TRANSFER)) { - retval = -ENOMEM; - dev_err(&intf->dev, "unable to allocate urb list\n"); - goto error; - } - - /* We don't register a new USB class. Our client interface is dlfbev */ - - retval = fb_alloc_cmap(&info->cmap, 256, 0); - if (retval < 0) { - dev_err(info->device, "cmap allocation failed: %d\n", retval); - goto error; - } - - retval = dlfb_setup_modes(dlfb, info, NULL, 0); - if (retval != 0) { - dev_err(info->device, - "unable to find common mode for display and adapter\n"); - goto error; - } - - /* ready to begin using device */ - - atomic_set(&dlfb->usb_active, 1); - dlfb_select_std_channel(dlfb); - - dlfb_ops_check_var(&info->var, info); - retval = dlfb_ops_set_par(info); - if (retval) - goto error; - - retval = register_framebuffer(info); - if (retval < 0) { - dev_err(info->device, "unable to register framebuffer: %d\n", - retval); - goto error; - } - - for (i = 0; i < ARRAY_SIZE(fb_device_attrs); i++) { - attr = &fb_device_attrs[i]; - retval = device_create_file(info->dev, attr); - if (retval) - dev_warn(info->device, - "failed to create '%s' attribute: %d\n", - attr->attr.name, retval); - } - - retval = device_create_bin_file(info->dev, &edid_attr); - if (retval) - dev_warn(info->device, "failed to create '%s' attribute: %d\n", - edid_attr.attr.name, retval); - - dev_info(info->device, - "%s is DisplayLink USB device (%dx%d, %dK framebuffer memory)\n", - dev_name(info->dev), info->var.xres, info->var.yres, - ((dlfb->backing_buffer) ? - info->fix.smem_len * 2 : info->fix.smem_len) >> 10); - return 0; - -error: - if (dlfb->info) { - dlfb_ops_destroy(dlfb->info); - } else { - usb_put_dev(dlfb->udev); - kfree(dlfb); - } - return retval; -} - -static void dlfb_usb_disconnect(struct usb_interface *intf) -{ - struct dlfb_data *dlfb; - struct fb_info *info; - int i; - - dlfb = usb_get_intfdata(intf); - info = dlfb->info; - - dev_dbg(&intf->dev, "USB disconnect starting\n"); - - /* we virtualize until all fb clients release. Then we free */ - dlfb->virtualized = true; - - /* When non-active we'll update virtual framebuffer, but no new urbs */ - atomic_set(&dlfb->usb_active, 0); - - /* this function will wait for all in-flight urbs to complete */ - dlfb_free_urb_list(dlfb); - - /* remove udlfb's sysfs interfaces */ - for (i = 0; i < ARRAY_SIZE(fb_device_attrs); i++) - device_remove_file(info->dev, &fb_device_attrs[i]); - device_remove_bin_file(info->dev, &edid_attr); - - unregister_framebuffer(info); -} - -static struct usb_driver dlfb_driver = { - .name = "udlfb", - .probe = dlfb_usb_probe, - .disconnect = dlfb_usb_disconnect, - .id_table = id_table, -}; - -module_usb_driver(dlfb_driver); - -static void dlfb_urb_completion(struct urb *urb) -{ - struct urb_node *unode = urb->context; - struct dlfb_data *dlfb = unode->dlfb; - unsigned long flags; - - switch (urb->status) { - case 0: - /* success */ - break; - case -ECONNRESET: - case -ENOENT: - case -ESHUTDOWN: - /* sync/async unlink faults aren't errors */ - break; - default: - dev_err(&dlfb->udev->dev, - "%s - nonzero write bulk status received: %d\n", - __func__, urb->status); - atomic_set(&dlfb->lost_pixels, 1); - break; - } - - urb->transfer_buffer_length = dlfb->urbs.size; /* reset to actual */ - - spin_lock_irqsave(&dlfb->urbs.lock, flags); - list_add_tail(&unode->entry, &dlfb->urbs.list); - dlfb->urbs.available++; - spin_unlock_irqrestore(&dlfb->urbs.lock, flags); - - up(&dlfb->urbs.limit_sem); -} - -static void dlfb_free_urb_list(struct dlfb_data *dlfb) -{ - int count = dlfb->urbs.count; - struct list_head *node; - struct urb_node *unode; - struct urb *urb; - - /* keep waiting and freeing, until we've got 'em all */ - while (count--) { - down(&dlfb->urbs.limit_sem); - - spin_lock_irq(&dlfb->urbs.lock); - - node = dlfb->urbs.list.next; /* have reserved one with sem */ - list_del_init(node); - - spin_unlock_irq(&dlfb->urbs.lock); - - unode = list_entry(node, struct urb_node, entry); - urb = unode->urb; - - /* Free each separately allocated piece */ - usb_free_coherent(urb->dev, dlfb->urbs.size, - urb->transfer_buffer, urb->transfer_dma); - usb_free_urb(urb); - kfree(node); - } - - dlfb->urbs.count = 0; -} - -static int dlfb_alloc_urb_list(struct dlfb_data *dlfb, int count, size_t size) -{ - struct urb *urb; - struct urb_node *unode; - char *buf; - size_t wanted_size = count * size; - - spin_lock_init(&dlfb->urbs.lock); - -retry: - dlfb->urbs.size = size; - INIT_LIST_HEAD(&dlfb->urbs.list); - - sema_init(&dlfb->urbs.limit_sem, 0); - dlfb->urbs.count = 0; - dlfb->urbs.available = 0; - - while (dlfb->urbs.count * size < wanted_size) { - unode = kzalloc(sizeof(*unode), GFP_KERNEL); - if (!unode) - break; - unode->dlfb = dlfb; - - urb = usb_alloc_urb(0, GFP_KERNEL); - if (!urb) { - kfree(unode); - break; - } - unode->urb = urb; - - buf = usb_alloc_coherent(dlfb->udev, size, GFP_KERNEL, - &urb->transfer_dma); - if (!buf) { - kfree(unode); - usb_free_urb(urb); - if (size > PAGE_SIZE) { - size /= 2; - dlfb_free_urb_list(dlfb); - goto retry; - } - break; - } - - /* urb->transfer_buffer_length set to actual before submit */ - usb_fill_bulk_urb(urb, dlfb->udev, usb_sndbulkpipe(dlfb->udev, 1), - buf, size, dlfb_urb_completion, unode); - urb->transfer_flags |= URB_NO_TRANSFER_DMA_MAP; - - list_add_tail(&unode->entry, &dlfb->urbs.list); - - up(&dlfb->urbs.limit_sem); - dlfb->urbs.count++; - dlfb->urbs.available++; - } - - return dlfb->urbs.count; -} - -static struct urb *dlfb_get_urb(struct dlfb_data *dlfb) -{ - int ret; - struct list_head *entry; - struct urb_node *unode; - - /* Wait for an in-flight buffer to complete and get re-queued */ - ret = down_timeout(&dlfb->urbs.limit_sem, GET_URB_TIMEOUT); - if (ret) { - atomic_set(&dlfb->lost_pixels, 1); - dev_warn(&dlfb->udev->dev, - "wait for urb interrupted: %d available: %d\n", - ret, dlfb->urbs.available); - return NULL; - } - - spin_lock_irq(&dlfb->urbs.lock); - - BUG_ON(list_empty(&dlfb->urbs.list)); /* reserved one with limit_sem */ - entry = dlfb->urbs.list.next; - list_del_init(entry); - dlfb->urbs.available--; - - spin_unlock_irq(&dlfb->urbs.lock); - - unode = list_entry(entry, struct urb_node, entry); - return unode->urb; -} - -static int dlfb_submit_urb(struct dlfb_data *dlfb, struct urb *urb, size_t len) -{ - int ret; - - BUG_ON(len > dlfb->urbs.size); - - urb->transfer_buffer_length = len; /* set to actual payload len */ - ret = usb_submit_urb(urb, GFP_KERNEL); - if (ret) { - dlfb_urb_completion(urb); /* because no one else will */ - atomic_set(&dlfb->lost_pixels, 1); - dev_err(&dlfb->udev->dev, "submit urb error: %d\n", ret); - } - return ret; -} - -module_param(console, bool, S_IWUSR | S_IRUSR | S_IWGRP | S_IRGRP); -MODULE_PARM_DESC(console, "Allow fbcon to open framebuffer"); - -module_param(fb_defio, bool, S_IWUSR | S_IRUSR | S_IWGRP | S_IRGRP); -MODULE_PARM_DESC(fb_defio, "Page fault detection of mmap writes"); - -module_param(shadow, bool, S_IWUSR | S_IRUSR | S_IWGRP | S_IRGRP); -MODULE_PARM_DESC(shadow, "Shadow vid mem. Disable to save mem but lose perf"); - -module_param(pixel_limit, int, S_IWUSR | S_IRUSR | S_IWGRP | S_IRGRP); -MODULE_PARM_DESC(pixel_limit, "Force limit on max mode (in x*y pixels)"); - -MODULE_AUTHOR("Roberto De Ioris <roberto(a)unbit.it>, " - "Jaya Kumar <jayakumar.lkml(a)gmail.com>, " - "Bernie Thompson <bernie(a)plugable.com>"); -MODULE_DESCRIPTION("DisplayLink kernel framebuffer driver"); -MODULE_LICENSE("GPL"); - -- 2.29.2 [View Less]

5 14

[PATCH rdma-core v3 0/6] Add user space dma-buf support
by Jianxin Xiong 02 Dec '20

02 Dec '20

This is the third version of the patch series. Change log: v3: * Add parameter 'iova' to the new ibv_reg_dmabuf_mr() API * Change the way of allocating dma-buf object - use /dev/dri/renderD* instead of /dev/dri/card* and use GEM object instead of dumb buffer * Add cmake function to allow building modules with mixed cython and C source files * Add new tests that use dma-buf MRs for send/recv and rdma traffic * Skip dma-buf tests on unsupported systems * Remove some use of random values in … [View More]the new tests * Add dealloc() and close() methods to the new classes * Replace string.format with f-string in python code * Fix some coding style issues: spacing, indentation, typo, comments v2: https://www.spinics.net/lists/linux-rdma/msg97936.html * Put the kernel header updates into a separate commit * Add comments for the data structure used in python ioctl calls * Fix issues related to symbol versioning * Fix styling issues: extra spaces, unncecessary variable, typo * Fix an inproper error code usage * Put the new op into ibv_context_ops instead if verbs_context v1: https://www.spinics.net/lists/linux-rdma/msg97865.html * Add user space API for registering dma-buf based memory regions * Update pyverbs with the new API * Add new tests This is the user space counter-part of the kernel patch set to add dma-buf support to the RDMA subsystem. This series consists of six patches. The first patch updates the kernel headers for dma-buf support. Patch 2 adds the new API function and updates the man pages. Patch 3 implements the new API in the mlx5 provider. Patch 4 adds new class definitions to pyverbs for the new API. Patch 5 adds a set of new tests for the new API. Patch 6 fixes bug in the utility code of the tests. Pull request at github: https://github.com/linux-rdma/rdma-core/pull/895 Jianxin Xiong (6): Update kernel headers verbs: Support dma-buf based memory region mlx5: Support dma-buf based memory region pyverbs: Add dma-buf based MR support tests: Add tests for dma-buf based memory regions tests: Bug fix for get_access_flags() buildlib/pyverbs_functions.cmake | 52 ++++++ debian/libibverbs1.symbols | 2 + kernel-headers/rdma/ib_user_ioctl_cmds.h | 14 ++ kernel-headers/rdma/ib_user_verbs.h | 14 -- libibverbs/CMakeLists.txt | 2 +- libibverbs/cmd_mr.c | 38 ++++ libibverbs/driver.h | 7 + libibverbs/dummy_ops.c | 11 ++ libibverbs/libibverbs.map.in | 6 + libibverbs/man/ibv_reg_mr.3 | 27 ++- libibverbs/verbs.c | 18 ++ libibverbs/verbs.h | 11 ++ providers/mlx5/mlx5.c | 2 + providers/mlx5/mlx5.h | 3 + providers/mlx5/verbs.c | 22 +++ pyverbs/CMakeLists.txt | 7 + pyverbs/dmabuf.pxd | 15 ++ pyverbs/dmabuf.pyx | 72 ++++++++ pyverbs/dmabuf_alloc.c | 296 +++++++++++++++++++++++++++++++ pyverbs/dmabuf_alloc.h | 19 ++ pyverbs/libibverbs.pxd | 2 + pyverbs/mr.pxd | 6 + pyverbs/mr.pyx | 103 ++++++++++- tests/test_mr.py | 239 ++++++++++++++++++++++++- tests/utils.py | 30 +++- 25 files changed, 996 insertions(+), 22 deletions(-) create mode 100644 pyverbs/dmabuf.pxd create mode 100644 pyverbs/dmabuf.pyx create mode 100644 pyverbs/dmabuf_alloc.c create mode 100644 pyverbs/dmabuf_alloc.h -- 1.8.3.1 [View Less]

5 21

[PATCH v2 0/3] drm/mipi-dbi: Type B bus support, drm/tiny: MRB2801
by mdurnev＠gmail.com 02 Dec '20

02 Dec '20

From: Mikhail Durnev <mikhail_durnev(a)mentor.com> Hi All, This patch series is aiming at extending the mipi-dbi bus driver to support Intel 8080 type parallel bus (Type B) over GPIO and adding a new driver for ILI9341 display panels with 8- or 16-bit parallel interface. It was tested with the MRB2801 display module [1] that had a connector compatible with the ALIENTEK STM32 development board. The module was connected to Raspberry Pi 3 via GPIO pins. The parallel bus is implemented … [View More]partially. It supports only write operations from the host to the display. Read operations would require switching GPIO mode between input and output back and forth. But this implementation is very simple, and GPIO mode can be set for all used pins to output once at initialization. The RD pin of the display has to always receive the logic high signal to make sure the data bus pins from the dislay side are always in the high impedance state. Otherwise the display module as well as the GPIO controller of the host can be damaged. To be on the safe side I recommend using protective resistors for all GPIO pins conneced to DB pins of the display. Resistors of 10 kOhms are just fine for RPi 3. The WR and DC pins may not work well with 10K resistors. Although there is no need to protect them, you can try using 1K resistors if you want. Bit banging is used to transmit data over the parallel bus from host to display. There are two numbers that contol timings. They should be defined in the device tree via the wr-up-down-delays property. The first number is related to the write control pulse duration, and the second number is related to the write cycle duration. For ILI9341, the write pulse cannot be shorter than 15 ns, and the write duration cannot be shorter than 66 ns. Delays of 10 and 51 ns respectively allow to meet the specifications on RPi 3. Faster machines may need values closer to 15 and 66. [1] http://www.lcdwiki.com/2.8inch_16BIT_Module_ILI9341_SKU:MRB2801 Signed-off-by: Mikhail Durnev <mikhail_durnev(a)mentor.com> v1 -> v2: - Moved the definition of mipi_dbi_machine_little_endian() out of the "#if IS_ENABLED(CONFIG_SPI)" clause. That static function is used in mipi_dbi_gpio_init which does not need CONFIG_SPI enabled v0 -> v1: - Rebased on v5.10-rc6 - Replaced "dbi->spi = 0;" with "dbi->spi = NULL;" in mipi_dbi_gpio_init v0: - Based on branch rpi-5.10.y - Tested on Raspberry Pi 3 Model B V 1.2 Mikhail Durnev (3): drm/mipi-dbi: Add support for Type B drm/tiny: Add driver for ili9341 with parallel bus dt-bindings: panel: Add bindings for MRB2801 .../devicetree/bindings/display/ronbo,mrb2801.txt | 42 +++ drivers/gpu/drm/drm_mipi_dbi.c | 134 +++++++++- drivers/gpu/drm/tiny/Kconfig | 13 + drivers/gpu/drm/tiny/Makefile | 1 + drivers/gpu/drm/tiny/ili9341_gpio.c | 290 +++++++++++++++++++++ include/drm/drm_mipi_dbi.h | 30 ++- 6 files changed, 499 insertions(+), 11 deletions(-) create mode 100644 Documentation/devicetree/bindings/display/ronbo,mrb2801.txt create mode 100644 drivers/gpu/drm/tiny/ili9341_gpio.c -- 2.7.4 [View Less]

2 4

← Newer
1
...
14
15
16
17
18
19
20
...
82
Older →

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel November 2020