aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2017-08-15drm: omapdrm: Move reg_fields to dispc_features structureLaurent Pinchart3-112/+76
The reg_fields feature describes DISPC registers only. Move it from the omap_dss_features structure to the dispc_features structure. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Move DISPC_CLK_SWITCH reg feature to struct dss_featuresLaurent Pinchart4-15/+16
The register belongs to the DSS, move the feature to the dss_features structure. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Move num_ovls and num_mgrs to dispc_features structureLaurent Pinchart3-58/+66
The num_ovls and num_mgrs are dispc features. Move them from the omap_dss_features structure to the dispc_features structure. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Move overlay caps features to dispc_features structureLaurent Pinchart3-84/+102
The overlay_caps is a dispc feature. Move it from the omap_dss_features structure to the dispc_features structure. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Move color modes feature to dispc_features structureLaurent Pinchart3-136/+124
The supported_color_modes is a dispc feature. Move it from the omap_dss_features structure to the dispc_features structure. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Move size unit features to dispc_features structureLaurent Pinchart3-41/+18
The buffer_size_unit and burst_size_unit are dispc features. Move them from the omap_dss_features structure to the dispc_features structure. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Move shutdown() handler from core to dssLaurent Pinchart2-20/+16
In preparation for removal of the core module, move the shutdown() handler from core to dss. Signed-off-by: Laurent Pinchart <[email protected]> Reviewed-by: Tomi Valkeinen <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Move all debugfs code from core to dssLaurent Pinchart3-89/+84
debugfs code is spread between the core and dss drivers. In preparation for removal of the core driver, move it all to the dss driver. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dss: Initialize DSS internal features at probe timeLaurent Pinchart1-13/+13
The DSS internal features are derived from the platform device compatible string which is available at probe time. Don't delay initialization until bind time. This prepares for the merge of the two DSS features structures that will be needed before the DSS is bound. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dss: Use supported outputs instead of display typesLaurent Pinchart3-87/+17
The dss driver uses the supported display types to check whether the DSS has SDI, VENC or HDMI outputs. We can instead use the supported outputs the provide the same information. This removes the last use of the supported display types, that we can then remove as well. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dss: Select features based on compatible stringLaurent Pinchart1-68/+29
Use the compatible string instead of the OMAP SoC revision to determine device features. The various OMAP3-based SoCs can't be told apart using the compatible string, use soc_device_match() to tell them apart. Signed-off-by: Laurent Pinchart <[email protected]> Reviewed-by: Tomi Valkeinen <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dpi: Replace OMAP SoC model checks with DSS modelLaurent Pinchart3-37/+45
The DPI code only needs to differentiate between major OMAP revisions, which can be obtained from the DSS compatible string. Replace the OMAP SoC model checks to prepare for removal of the OMAP SoC version platform data. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dispc: Select features based on compatible stringLaurent Pinchart1-60/+27
Use the compatible string instead of the OMAP SoC revision to determine device features. On OMAP34xx the features depend on the ES revision that can not be determined from the compatible string. Use soc_device_match() in that case. Signed-off-by: Laurent Pinchart <[email protected]> Reviewed-by: Tomi Valkeinen <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: Don't forward set_min_bus_tput() to no-op platform codeLaurent Pinchart2-11/+6
The OMAP implementation of the set_min_bus_tput() API is a no-op. There's no point in forwarding the driver calls to the platform code. Remove the use of the related platform data callback, but keep the internal function as a reminder that the feature will need to be implemented when the OMAP platform will provide support. Signed-off-by: Laurent Pinchart <[email protected]> Reviewed-by: Tomi Valkeinen <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dsi: Handle pin muxing internallyLaurent Pinchart3-25/+79
Don't rely on callback functions provided by the platform, but access the syscon internally to mux the DSI pins. Signed-off-by: Laurent Pinchart <[email protected]> Reviewed-by: Tomi Valkeinen <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dsi: Store DSI model and PLL hardware data in OF dataLaurent Pinchart1-61/+55
The DSI PLL hardware data and DSS channels are selected based on the OMAP SoC model. There's no need for fine-grained model information, as the driver only needs to differentiate between OMAP3, OMAP4 and OMAP5. As this can be done through the DSI compatible string, store the corresponding information in OF match data instead to avoid accessing the OMAP SoC model. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dss: Split operations out of dss_features structureLaurent Pinchart1-15/+35
Move the two function pointers to a new dss_ops structure. This will allow merging the dss_features and omap_dss_features structures without having to expose the DPI source selection and LCD clock muxing functions in header files. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: hdmi: Store PHY features in PHY data structureLaurent Pinchart2-15/+15
PHY features are stored in a global variable, while they should be properties of the PHY object. As existing OMAP platforms have a single HDMI PHY this doesn't cause any issue, but doesn't follow the driver model. Move the PHY features to the HDMI PHY data structure to follow the driver model and pave the road for multiple HDMI PHYs support. Signed-off-by: Laurent Pinchart <[email protected]> Reviewed-by: Tomi Valkeinen <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: venc: Don't export omap_dss_pal_vm and omap_dss_ntsc_vmLaurent Pinchart2-11/+2
The two variables are never used outside of their compilation unit, make them static. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: dpi: Remove unneeded regulator checkLaurent Pinchart1-10/+3
The dpi_display_enable() function ensures that a VDDS_DSI regulator is available if the DSI uses the VDDS_DSI supply. This is not needed, as a failure to get the VDDS_DSI supply will result in a probe failure, dpi_display_enable() will thus never be called in that case. Remove the check, and replace tests for the FEAT_DPI_USES_VDDS_DSI feature with a test for the regulator object. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: panel-dpi: Remove unneeded check for OF nodeLaurent Pinchart1-0/+3
As non-DT booting is no longer supported, the only way to instantiate the device is through an OF node, which is guaranteed to be present. Remove the unneeded check. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: connector-analog-tv: Remove unneeded check for OF nodeLaurent Pinchart1-0/+3
As non-DT booting is no longer supported, the only way to instantiate the device is through an OF node, which is guaranteed to be present. Remove the unneeded check. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm: omapdrm: acx565akm: Remove unneeded check for OF nodeLaurent Pinchart1-0/+3
As non-DT booting is no longer supported, the only way to instantiate the device is through an OF node, which is guaranteed to be present. Remove the unneeded check. Signed-off-by: Laurent Pinchart <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15ARM: OMAP2+: Register SoC device attributes from machine .init()Laurent Pinchart2-1/+1
SoC device attributes are registered with a call to soc_device_register() from the machine .init_late() operation, which is called from the late initcall, after all drivers built-in drivers have been probed. This results in the impossibility for drivers to use SoC device matching in their probe function. The omap_soc_device_init() function is safe to call from the machine .init() operation, as all data it depends on is initialized from the .init_early() operation. Move SoC device attribute registration to machine .init() like on all other ARM platforms. Signed-off-by: Laurent Pinchart <[email protected]> Acked-by: Tony Lindgren <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: panel-dsi-cm: constify attribute_group structures.Arvind Yadav1-1/+1
attribute_groups are not supposed to change at runtime. All functions working with attribute_groups provided by <linux/sysfs.h> work with const attribute_group. So mark the non-const structs as const. File size before: text data bss dec hex filename 4669 464 0 5133 140d panel-dsi-cm.o File size After adding 'const': text data bss dec hex filename 4725 400 0 5125 1405 panel-dsi-cm.o Signed-off-by: Arvind Yadav <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: panel-sony-acx565akm: constify attribute_group structures.Arvind Yadav1-1/+1
attribute_groups are not supposed to change at runtime. All functions working with attribute_groups provided by <linux/sysfs.h> work with const attribute_group. So mark the non-const structs as const. File size before: text data bss dec hex filename 5147 560 0 5707 164b panel-sony-acx565akm.o File size After adding 'const': text data bss dec hex filename 5211 496 0 5707 164b panel-sony-acx565akm.o Signed-off-by: Arvind Yadav <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: constify attribute_group structures.Arvind Yadav1-1/+1
attribute_groups are not supposed to change at runtime. All functions working with attribute_groups provided by <linux/sysfs.h> work with const attribute_group. So mark the non-const structs as const. File size before: text data bss dec hex filename 2673 368 0 3041 be1 panel-tpo-td043mtea1.o File size After adding 'const': text data bss dec hex filename 2721 304 0 3025 bd1 panel-tpo-td043mtea1.o Signed-off-by: Arvind Yadav <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: dma-buf: Constify dma_buf_ops structures.Arvind Yadav1-1/+1
dma_buf_ops are not supposed to change at runtime. All functions working with dma_buf_ops provided by <linux/dma-buf.h> work with const dma_buf_ops. So mark the non-const structs as const. File size before: text data bss dec hex filename 1240 112 0 1352 548 drivers/gpu/drm/omapdrm/omap_gem_dmabuf.o File size After adding 'const': text data bss dec hex filename 1352 0 0 1352 548 drivers/gpu/drm/omapdrm/omap_gem_dmabuf.o Signed-off-by: Arvind Yadav <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: omap_display_timings: constify videomode structuresJulia Lawall2-2/+2
The videomode structures are only copied into the vm field of a panel_drv_data structure, so they can be const. Done with the help of Coccinelle. Signed-off-by: Julia Lawall <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: displays: encoder-tpd12s015: Support for hot plug detectionPeter Ujfalusi1-0/+81
Use interrupt handler for hpd GPIO to react to HPD changes. Signed-off-by: Peter Ujfalusi <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: displays: connector-hdmi: Support for hot plug detectionPeter Ujfalusi1-0/+104
If the hpd_gpio is valid, use interrupt handler to react to HPD changes. In case the hpd_gpio is not valid, try to enable hpd detection on the encoder if it supports it. Signed-off-by: Peter Ujfalusi <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: Support for HDMI hot plug detectionPeter Ujfalusi3-1/+82
The HPD signal can be used for detecting HDMI cable plug and unplug event without the need for polling the status of the line. This will speed up detecting such event because we do not need to wait for the next poll event to notice the state change. Signed-off-by: Peter Ujfalusi <[email protected]> Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/omap: fix memory leak when FB init failsTomi Valkeinen1-1/+1
omap_framebuffer_create() fails to unref all the gem objects if creating the FB fails, leading to a memory leak. Fix the loop so that it goes through all the reffed gem objects. Signed-off-by: Tomi Valkeinen <[email protected]>
2017-08-15drm/i915: Work around GCC anonymous union initialization bugImre Deak1-24/+54
GCC 4.4 can't cope with anonymous union initializers which seems to be a bug in that version (see the Reference) and is fixed since GCC version 4.6. A workaround which is also used elsewhere in the kernel for the same purpose is to wrap the initialization in curly braces, so do the same here. Fixes: b5565a2efc12 ("drm/i915/bxt, glk: Give a proper name to the power well struct phy field") Reference: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10676 Reported-by: Fengguang Wu <[email protected]> Cc: Arkadiusz Hiler <[email protected]> Signed-off-by: Imre Deak <[email protected]> Reviewed-by: Chris Wilson <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-15Merge tag 'gvt-next-2017-08-15' of https://github.com/01org/gvt-linux into ↵Daniel Vetter18-133/+297
drm-intel-next-queued gvt-next-2017-08-15 gvt update for 4.14 - MMIO save/restore optimization (Changbin) - Split workload scan vs. dispatch for more parallel exec (Ping) - vGPU full 48bit ppgtt support (Joonas, Tina) - vGPU hw id expose for perf (Zhenyu) - other misc cleanup and fixes Signed-off-by: Daniel Vetter <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-15drm/i915: Initialize 'data' in intel_dsi_dcs_backlight.cBalasubramaniam, Hari Chand1-1/+1
variable 'data' may be used uninitialized in this function. thus, 'function dcs_get_backlight' will return unwanted value/fail. Thus, adding NULL initialized to 'data' variable will solve the return failure happening. v2: Change commit message to reflect upstream with proper message Fixes: 90198355b83c ("drm/i915/dsi: Add DCS control for Panel PWM") Cc: Jani Nikula <[email protected]> Cc: Daniel Vetter <[email protected]> Cc: Yetunde Adebisi <[email protected]> Cc: Deepak M <[email protected]> Cc: Jani Nikula <[email protected]> Signed-off-by: Balasubramaniam, Hari Chand <[email protected]> Reviewed-by: Jani Nikula <[email protected]> Signed-off-by: Jani Nikula <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/1502762746-191826-1-git-send-email-hari.chand.balasubramaniam@intel.com
2017-08-15drm/i915/dp: Validate the compliance test link parametersManasi Navare1-17/+17
Validate the compliance test link parameters when the compliance test dpcd registers are read. Also validate them in compute_config before using them since the max values might have been reduced due to link training fallback. If either the link rate or lane count is invalid, we still bail from using the test parameters since the combination would not work and instead use the fallback values. v2: * Added commit message to explain why we still bail when either of of the params is invalid (Ville Syrjala) * Add reason for validating in the comment (Jani Nikula) * Also check if index >= 0 after validating (Jani Nikula) Signed-off-by: Manasi Navare <[email protected]> Cc: Jani Nikula <[email protected]> Cc: Ville Syrjala <[email protected]> Reviewed-by: Jani Nikula <[email protected]> Tested-by: Nathan Ciobanu <[email protected]> Signed-off-by: Jani Nikula <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-15drm/i915/dp: Generalize intel_dp_link_params function to accept arguments to ↵Manasi Navare1-6/+8
be validated This function now takes the link rate and lane ocunt to be validated as an argument so that this can be used for validating even the compliance test link parameters. Signed-off-by: Manasi Navare <[email protected]> Cc: Ville Syrjala <[email protected]> Cc: Jani Nikula <[email protected]> Reviewed-by: Jani Nikula <[email protected]> Tested-by: Nathan Ciobanu <[email protected]> Signed-off-by: Jani Nikula <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-15Backmerge tag 'v4.13-rc5' into drm-nextDave Airlie866-18214/+14179
Linux 4.13-rc5 There's a really nasty nouveau collision, hopefully someone can take a look once I pushed this out.
2017-08-15drm/i915/gvt: Fix guest i915 full ppgtt blocking issueTina Zhang2-17/+29
Guest i915 full ppgtt functionality was blocking by an issue, which would lead to gpu hardware hang. Guest i915 driver may update the ppgtt table just before this workload is going to be submitted to the hardware by device model. This case wasn't handled well by device model before, due to the small time window between removing old ppgtt entry and adding the new one. Errors occur when the workload is executed by hardware during that small time window. This patch is to remove this time window by adding the new ppgtt entry first and then remove the old one. Changes in v2: - Move VGT_CAPS_FULL_PPGTT introduction to patch 2/4. (Joonas) Changes since v2: - Divide the whole patch set into two separate patch series, with one patch in i915 side to check guest i915 full ppgtt capability and enable it when this capability is supported by the device model, and the other one in gvt side which fixs the blocking issue and enables the device model to provide the capability to guest. And this patch focuses on gvt side. (Joonas) - Change the title from "reorder the shadow ppgtt update process by adding entry first" to "Fix guest i915 full ppgtt blocking issue". (Tina) Changes since v3: - Rebase to the latest branch. Changes since v4: - Tested by Tina Zhang. Changes since v5: - Rebase to the latest branch. v6: - Update full 48bit ppgtt definition Cc: Tina Zhang <[email protected]> Signed-off-by: Tina Zhang <[email protected]> Signed-off-by: Zhenyu Wang <[email protected]>
2017-08-15drm/i915: Enable guest i915 full ppgtt functionalityTina Zhang5-3/+20
Enable the guest i915 full ppgtt functionality when host can provide this capability. vgt_caps is introduced to guest i915 driver to get the vgpu capabilities from the device model. VGT_CPAS_FULL_PPGTT is one of the capabilities type to let guest i915 dirver know that the guest i915 full ppgtt is supported by device model. Notice that the minor version of pvinfo isn't bumped because of this vgt_caps introduction, due to older guest would be broken by simply increasing the pvinfo version. Although the pvinfo minor version doesn't increase, the compatibility won't be blocked. The compatibility is ensured by checking the value of caps field in pvinfo. Zero means no full ppgtt support and BIT(2) means this feature is provided. Changes since v1: - Use u32 instead of uint32_t (Joonas) - Move VGT_CAPS_FULL_PPGTT introduction to this patch and use #define instead of enum (Joonas) - Rewrite the vgpu full ppgtt capability checking logic. (Joonas) - Some coding style refine. (Joonas) Changes since v2: - Divide the whole patch set into two separate patch series, with one patch in i915 side to check guest i915 full ppgtt capability and enable it when this capability is supported by the device model, and the other one in gvt side which fixs the blocking issue and enables the device model to provide the capability to guest. And this patch focuses on guest i915 side. (Joonas) - Change the title from "introduce vgt_caps to pvinfo" to "Enable guest i915 full ppgtt functionality". (Tina) Change since v3: - Add some comments about pvinfo caps and version. (Joonas) Change since v4: - Tested by Tina Zhang. Change since v5: - Add limitation about supporting 32bit full ppgtt. Change since v6: - Change the fallback to 48bit full ppgtt if i915.ppgtt_enable=2. (Zhenyu) Change in v9: - Remove the fixme comment due to no plan for 32bit full ppgtt support. (Zhenyu) - Reorder the patch-set to fix compiling issue with git-bisect. (Zhenyu) - Add print log when forcing guest 48bit full ppgtt. (Zhenyu) v10: - Update against Joonas's has_full_ppgtt and has_full_48bit_ppgtt disconnect change. (Zhenyu) Reviewed-by: Joonas Lahtinen <[email protected]> # in v2 Cc: Joonas Lahtinen <[email protected]> Cc: Tina Zhang <[email protected]> Signed-off-by: Tina Zhang <[email protected]> Signed-off-by: Zhenyu Wang <[email protected]>
2017-08-15drm/i915: Disconnect 32 and 48 bit ppGTT supportJoonas Lahtinen1-4/+9
Configurations like virtualized environments may support only 48 bit ppGTT without supporting 32 bit ppGTT. Support this by disconnecting the relationship of the two feature bits. Cc: Tina Zhang <[email protected]> Cc: Chris Wilson <[email protected]> Cc: Zhi Wang <[email protected]> Reviewed-by: Chris Wilson <[email protected]> Signed-off-by: Joonas Lahtinen <[email protected]> Signed-off-by: Zhenyu Wang <[email protected]>
2017-08-14drm/vc4: Continue the switch to drm_*_put() helpersCihangir Akturk1-1/+1
Use drm_*_get() and drm_*_put() helpers instead of drm_*_reference() and drm_*_unreference() helpers. drm_*_reference() and drm_*_unreference() functions are just compatibility alias for drm_*_get() and drm_*_put() and should not be used by new code. So convert all users of compatibility functions to use the new APIs. Generated by: scripts/coccinelle/api/drm-get-put.cocci v2: Tweak commit summary to distinguish it from the previous commit covering everything but the new bo_label ioctl (by anholt). Signed-off-by: Cihangir Akturk <[email protected]> Signed-off-by: Eric Anholt <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected] Reviewed-by: Eric Anholt <[email protected]>
2017-08-14drm/vc4: Fix leak of HDMI EDIDEric Anholt1-0/+1
We don't keep a pointer to it around anywhere, so it's our job to free it. Cc: Stefan Wahren <[email protected]> Link: https://github.com/anholt/linux/issues/101 Fixes: c8b75bca92cb ("drm/vc4: Add KMS support for Raspberry Pi.") Signed-off-by: Eric Anholt <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected] Reviewed-by: Daniel Vetter <[email protected]> Tested-by: Stefan Wahren <[email protected]>
2017-08-14dma-buf: fix reservation_object_wait_timeout_rcu to wait correctly v2Christian König1-18/+15
With hardware resets in mind it is possible that all shared fences are signaled, but the exlusive isn't. Fix waiting for everything in this situation. v2: make sure we always wait for the exclusive fence Acked-by: Sumit Semwal <[email protected]> Signed-off-by: Christian König <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Reviewed-by: Chunming Zhou <[email protected]> Signed-off-by: Alex Deucher <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-14dma-buf: add reservation_object_copy_fences (v2)Christian König2-0/+63
Allows us to copy all the fences in a reservation object to another one. v2: handle NULL src_list Signed-off-by: Christian König <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Alex Deucher <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-14drm/i915: More surgically unbreak the modeset vs reset deadlockDaniel Vetter4-10/+52
There's no reason to entirely wedge the gpu, for the minimal deadlock bugfix we only need to unbreak/decouple the atomic commit from the gpu reset. The simplest way to fix that is by replacing the unconditional fence wait a the top of commit_tail by a wait which completes either when the fences are done (normal case, or when a reset doesn't need to touch the display state). Or when the gpu reset needs to force-unblock all pending modeset states. The lesser source of deadlocks is when we try to pin a new framebuffer and run into a stall. There's a bunch of places this can happen, like eviction, changing the caching mode, acquiring a fence on older platforms. And we can't just break the depency loop and keep going, the only way would be to break out and restart. But the problem with that approach is that we must stall for the reset to complete before we grab any locks, and with the atomic infrastructure that's a bit tricky. The only place is the ioctl code, and we don't want to insert code into e.g. the BUSY ioctl. Hence for that problem just create a critical section, and if any code is in there, wedge the GPU. For the steady-state this should never be a problem. Note that in both cases TDR itself keeps working, so from a userspace pov this trickery isn't observable. Users themselvs might spot a short glitch while the rendering is catching up again, but that's still better than pre-TDR where we've thrown away all the rendering, including innocent batches. Also, this fixes the regression TDR introduced of making gpu resets deadlock-prone when we do need to touch the display. One thing I noticed is that gpu_error.flags seems to use both our own wait-queue in gpu_error.wait_queue, and the generic wait_on_bit facilities. Not entirely sure why this inconsistency exists, I just picked one style. A possible future avenue could be to insert the gpu reset in-between ongoing modeset changes, which would avoid the momentary glitch. But that's a lot more work to implement in the atomic commit machinery, and given that we only need this for pre-g4x hw, of questionable utility just for the sake of polishing gpu reset even more on those old boxes. It might be useful for other features though. v2: Rebase onto 4.13 with a s/wait_queue_t/struct wait_queue_entry/. v3: Really emabarrassing fixup, I checked the wrong bit and broke the unbreak/wakeup logic. v4: Also handle deadlocks in pin_to_display. v5: Review from Michel: - Fixup the BUILD_BUG_ON - Don't forget about the overlay Cc: Michel Thierry <[email protected]> Cc: Chris Wilson <[email protected]> Cc: Mika Kuoppala <[email protected]> Cc: Joonas Lahtinen <[email protected]> Cc: Tvrtko Ursulin <[email protected]> (v2) Cc: Michel Thierry <[email protected]> Signed-off-by: Daniel Vetter <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected] Reviewed-by: Michel Thierry <[email protected]>
2017-08-14drm/i915: Push i915_sw_fence_wait into the nonblocking atomic commitDaniel Vetter1-8/+8
Blocking in a worker is ok, that's what the unbound_wq is for. And it unifies the paths between the blocking and nonblocking commit, giving me just one path where I have to implement the deadlock avoidance trickery in the next patch. I first tried to implement the following patch without this rework, but force-completing i915_sw_fence creates some serious challenges around properly cleaning things up. So wasn't a feasible short-term approach. Another approach would be to simple keep track of all pending atomic commit work items and manually queue them from the reset code. With the caveat that double-queue in case we race with the i915_sw_fence must be avoided. Given all that, taking the cost of a double schedule in atomic for the short-term fix is the best approach, but can be changed in the future of course. v2: Amend commit message (Chris). v3: Add comment explaining why we do nothing in the sw_fence complete callback (Michel). Reviewed-by: Maarten Lankhorst <[email protected]> Cc: Chris Wilson <[email protected]> Cc: Mika Kuoppala <[email protected]> Cc: Joonas Lahtinen <[email protected]> Cc: Tvrtko Ursulin <[email protected]> (v2) Cc: Michel Thierry <[email protected]> Reviewed-by: Michel Thierry <[email protected]> Signed-off-by: Daniel Vetter <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-14drm/i915: Avoid the gpu reset vs. modeset deadlockDaniel Vetter1-0/+7
... using the biggest hammer we have. This is essentially a weaponized version of the timeout-based wedging Chris added in commit 36703e79a982c8ce5a8e43833291f2719e92d0d1 Author: Chris Wilson <[email protected]> Date: Thu Jun 22 11:56:25 2017 +0100 drm/i915: Break modeset deadlocks on reset Because defense-in-depth is good it's good to still have both. Also note that with the locking change we can now restrict this a lot (old gpus and special testing only), so this doesn't kill the TDR benefits on at least anything remotely modern. And futuremore with a few tricks it should be possible to make a much more educated guess about whether an atomic commit is stuck waiting on the gpu (atomic_t counting the pending i915_sw_fence used by the atomic modeset code should do it), so we can improve this. But for now just start with something that is guaranteed to recover faster, for much better CI througput. This defacto reverts TDR on these platforms, but there's not really a single commit to specify as the sole offender. v2: Add a debug message to explain what's going on. We can't DRM_ERROR because that spams CI. And the timeout based fallback still prints a DRM_ERROR, in case something goes wrong. v3: Fix comment layout (Michel) Fixes: 4680816be336 ("drm/i915: Wait first for submission, before waiting for request completion") Fixes: 221fe7994554 ("drm/i915: Perform a direct reset of the GPU from the waiter") Cc: Chris Wilson <[email protected]> Cc: Mika Kuoppala <[email protected]> Cc: Joonas Lahtinen <[email protected]> Cc: Tvrtko Ursulin <[email protected]> (v2) Cc: Michel Thierry <[email protected]> Reviewed-by: Tvrtko Ursulin <[email protected]> (v2) Reviewed-by: Michel Thierry <[email protected]> Signed-off-by: Daniel Vetter <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
2017-08-14drm/i915/gen9: Send all components in VF stateMika Kuoppala1-2/+2
Update gen9 renderstate to account the, long overdue, changes for igt commit 5c07135b7bd2 ("tools/null_state/gen9: Send all components in VF state"). Signed-off-by: Mika Kuoppala <[email protected]> Reviewed-by: Chris Wilson <[email protected]> Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
Powered by cgit, Gruvboxed by Blaster4385.