mesa/src/freedreno/drm/freedreno_pipe.c

216 lines
5.8 KiB
C
Raw Normal View History

freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
/*
* Copyright (C) 2012-2018 Rob Clark <robclark@freedesktop.org>
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* Authors:
* Rob Clark <robclark@freedesktop.org>
*/
#include "freedreno_drmif.h"
#include "freedreno_priv.h"
/**
* priority of zero is highest priority, and higher numeric values are
* lower priorities
*/
struct fd_pipe *
fd_pipe_new2(struct fd_device *dev, enum fd_pipe_id id, uint32_t prio)
{
struct fd_pipe *pipe;
uint64_t val;
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
if (id > FD_PIPE_MAX) {
ERROR_MSG("invalid pipe id: %d", id);
return NULL;
}
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
if ((prio != 1) && (fd_device_version(dev) < FD_VERSION_SUBMIT_QUEUES)) {
ERROR_MSG("invalid priority!");
return NULL;
}
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
pipe = dev->funcs->pipe_new(dev, id, prio);
if (!pipe) {
ERROR_MSG("allocation failed");
return NULL;
}
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
pipe->dev = fd_device_ref(dev);
pipe->id = id;
p_atomic_set(&pipe->refcnt, 1);
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
fd_pipe_get_param(pipe, FD_GPU_ID, &val);
pipe->dev_id.gpu_id = val;
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
fd_pipe_get_param(pipe, FD_CHIP_ID, &val);
pipe->dev_id.chip_id = val;
pipe->control_mem = fd_bo_new(dev, sizeof(*pipe->control),
FD_BO_CACHED_COHERENT,
"pipe-control");
pipe->control = fd_bo_map(pipe->control_mem);
/* We could be getting a bo from the bo-cache, make sure the fence value
* is not garbage:
*/
pipe->control->fence = 0;
/* We don't want the control_mem bo to hold a reference to the ourself,
* so disable userspace fencing. This also means that we won't be able
* to determine if the buffer is idle which is needed by bo-cache. But
* pipe creation/destroy is not a high frequency event so just disable
* the bo-cache as well:
*/
pipe->control_mem->nosync = true;
pipe->control_mem->bo_reuse = NO_CACHE;
return pipe;
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
}
struct fd_pipe *
fd_pipe_new(struct fd_device *dev, enum fd_pipe_id id)
{
return fd_pipe_new2(dev, id, 1);
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
}
struct fd_pipe *
fd_pipe_ref(struct fd_pipe *pipe)
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
{
simple_mtx_lock(&table_lock);
fd_pipe_ref_locked(pipe);
simple_mtx_unlock(&table_lock);
return pipe;
}
struct fd_pipe *
fd_pipe_ref_locked(struct fd_pipe *pipe)
{
simple_mtx_assert_locked(&table_lock);
pipe->refcnt++;
return pipe;
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
}
void
fd_pipe_del(struct fd_pipe *pipe)
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
{
simple_mtx_lock(&table_lock);
fd_pipe_del_locked(pipe);
simple_mtx_unlock(&table_lock);
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
}
void
fd_pipe_del_locked(struct fd_pipe *pipe)
{
simple_mtx_assert_locked(&table_lock);
if (!p_atomic_dec_zero(&pipe->refcnt))
return;
fd_bo_del_locked(pipe->control_mem);
fd_device_del_locked(pipe->dev);
pipe->funcs->destroy(pipe);
}
/**
* Discard any unflushed deferred submits. This is called at context-
* destroy to make sure we don't leak unflushed submits.
*/
void
fd_pipe_purge(struct fd_pipe *pipe)
{
struct fd_device *dev = pipe->dev;
struct list_head deferred_submits;
list_inithead(&deferred_submits);
simple_mtx_lock(&dev->submit_lock);
foreach_submit_safe (deferred_submit, &dev->deferred_submits) {
if (deferred_submit->pipe != pipe)
continue;
list_del(&deferred_submit->node);
list_addtail(&deferred_submit->node, &deferred_submits);
dev->deferred_cmds -= fd_ringbuffer_cmd_count(deferred_submit->primary);
}
simple_mtx_unlock(&dev->submit_lock);
foreach_submit_safe (deferred_submit, &deferred_submits) {
list_del(&deferred_submit->node);
fd_submit_del(deferred_submit);
}
}
int
fd_pipe_get_param(struct fd_pipe *pipe, enum fd_param_id param, uint64_t *value)
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
{
return pipe->funcs->get_param(pipe, param, value);
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
}
int
fd_pipe_set_param(struct fd_pipe *pipe, enum fd_param_id param, uint64_t value)
{
return pipe->funcs->set_param(pipe, param, value);
}
const struct fd_dev_id *
fd_pipe_dev_id(struct fd_pipe *pipe)
{
return &pipe->dev_id;
}
int
fd_pipe_wait(struct fd_pipe *pipe, const struct fd_fence *fence)
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
{
return fd_pipe_wait_timeout(pipe, fence, ~0);
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
}
int
fd_pipe_wait_timeout(struct fd_pipe *pipe, const struct fd_fence *fence,
uint64_t timeout)
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
{
if (!fd_fence_after(fence->ufence, pipe->control->fence))
return 0;
fd_pipe_flush(pipe, fence->ufence);
return pipe->funcs->wait(pipe, fence, timeout);
freedreno: import libdrm_freedreno + redesign submit In the pursuit of lowering driver overhead, it became clear that some amount of redesign of how libdrm_freedreno constructs the submit ioctl would be needed. In particular, as the gallium driver is starting to make heavier use of CP_SET_DRAW_STATE state groups/objects, the over- head of tracking cmd buffers and relocs becomes too much. And for "streaming" state, which isn't ever reused (like uniform uploads) the overhead of allocating/freeing ringbuffer[1] objects is too high. This redesign makes two main changes: 1) Introduces a fd_submit object for tracking bos and cmds table for the submit ioctl, making ringbuffer objects more light- weight. This was previously done in the ringbuffer. But we have many ringbuffer instances involved in a submit (gmem + draw + potentially 1000's of state-group rbs), and only need a single bos and cmds table. (Reloc table is still per-rb) The submit is also a convenient place for a slab allocator for ringbuffer objects. Other options would have required locking because, while we can guarantee allocations will only happen on a single thread, free's could happen either on the application thread or the flush_queue thread. With the slab allocator in the submit object, any frees that happen on the flush_queue thread happen after we know that the application thread is done with the submit. 2) Introduce a new "softpin" msm_ringbuffer_sp implementation that does not use relocs and only has cmds table entries for IB1 (ie. the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER to from the RB). To do this properly will require some updates on the kernel side, so whether you get the softpin or legacy submit/ringbuffer implementation at runtime depends on your kernel version. To make all these changes in libdrm would basically require adding a libdrm_freedreno2, so this is a good point to just pull the libdrm code into mesa. Plus it allows for using mesa's hashtable, slab allocator, etc. And it lets us have asserts enabled for debug mesa buids but omitted for release builds. And it makes life easier if further API changes become necessary. At this point I haven't tried to pull in the kgsl backend. Although I left the level of vfunc indirection which would make it possible to have other backends. (And this was convenient to keep to allow for the "softpin" ringbuffer to coexist.) NOTE: if bisecting a build error takes you here, try a clean build. There are a bunch of ways things can go wrong if you still have libdrm_freedreno cflags. [1] "ringbuffer" is probably a bad name, the only level of cmdstream buffer that is actually a ring is RB managed by kernel. User- space cmdstream is all IB1/IB2 and state-groups. Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org> Reviewed-by: Eric Engestrom <eric.engestrom@intel.com> Signed-off-by: Rob Clark <robdclark@gmail.com>
2018-10-21 15:22:11 +01:00
}
uint32_t
fd_pipe_emit_fence(struct fd_pipe *pipe, struct fd_ringbuffer *ring)
{
uint32_t fence = ++pipe->last_fence;
if (fd_dev_64b(&pipe->dev_id)) {
OUT_PKT7(ring, CP_EVENT_WRITE, 4);
OUT_RING(ring, CP_EVENT_WRITE_0_EVENT(CACHE_FLUSH_TS));
OUT_RELOC(ring, control_ptr(pipe, fence)); /* ADDR_LO/HI */
OUT_RING(ring, fence);
} else {
OUT_PKT3(ring, CP_EVENT_WRITE, 3);
OUT_RING(ring, CP_EVENT_WRITE_0_EVENT(CACHE_FLUSH_TS));
OUT_RELOC(ring, control_ptr(pipe, fence)); /* ADDR */
OUT_RING(ring, fence);
}
return fence;
}