drm/linux: Rewrite the workqueue implementation

Most workqueues now use one runner thread per logical CPU, except for
WQ_UNBOUND ones, which only have a single kernel thread.

This commit contains c

drm/linux: Rewrite the workqueue implementation

Most workqueues now use one runner thread per logical CPU, except for
WQ_UNBOUND ones, which only have a single kernel thread.

This commit contains critical fixes from Matthew Dillon for multithread
interaction issues in queue_work() and the cancel functions.

show more ...

drm: Add headers and infrastructure required by Linux 4.13+ code

drm/linux: Add work_pending()

drm/linux: Add drain_workqueue()

drm - Fix workqueue bugs w/recent drm pushes

* The linux workqueue code uses pcpu queues in many situations.
At least one use-case in the recent drm code push can result in
taskqueue_enqueue() c

drm - Fix workqueue bugs w/recent drm pushes

* The linux workqueue code uses pcpu queues in many situations.
At least one use-case in the recent drm code push can result in
taskqueue_enqueue() calls being issued on active tasks to a different
queue, causing (typically) overnight or idle related panics.

* Adjust the linux workqueue code to use the new taskqueue API
functions that supports this behavior.

show more ...

drm/linux: Add flush_delayed_work()

drm: Do not use PHOLD/PRELE() in flush_taskqueue()

kernel - Rewrite the callout_*() API

* Rewrite the entire API from scratch and improve compatibility
with FreeBSD. This is not an attempt to achieve full API compatibility,
as FreeBSD's API has

kernel - Rewrite the callout_*() API

* Rewrite the entire API from scratch and improve compatibility
with FreeBSD. This is not an attempt to achieve full API compatibility,
as FreeBSD's API has unnecessary complexity that coders would frequently
make mistakes interpreting.

* Remove the IPI mechanisms in favor of fine-grained spin-locks instead.

* Add some robustness features in an attempt to track down corrupted
callwheel lists due to originating subsystems freeing structures out
from under an active callout.

* The code supports a full-blown type-stable/adhoc-reuse structural
separation between the front-end and the back-end, but this feature
is currently not operational and may be removed at some future point.
Instead we currently just embed the struct _callout inside the
struct callout.

* Replace callout_stop_sync() with callout_cancel().

* callout_drain() is now implemented as a synchronous cancel instead
of an asynchronous stop, which is closer to the FreeBSD API and
expected operation for ported code (usb stack in particular). We
will just have to fix any deadlocks which we come across.

* Retain our callout_terminate() function as the 'better' way to
stop using a callout, as it will not only cancel the callout but
also de-flag the structure so it can no longer be used.

show more ...

drm/linux/workqueue: the internal function is named func

Some driver code expects to be able to access it.

drm - No need to protect a callout struct from MP accesses.

drm: Sync include directives with Linux

* Add a few key include/asm or include/linux headers

* Move some code from .h to .c files in order to avoid clashes
between the DragonFly and Linux variant

drm: Sync include directives with Linux

* Add a few key include/asm or include/linux headers

* Move some code from .h to .c files in order to avoid clashes
between the DragonFly and Linux variants of kmalloc() and kfree()

show more ...

drm/linux: Add system_unbound_wq

drm/linux: Add work_busy()

drm/linux: make flush_work() more robust

drm/linux: Improve linux/workqueue.h

drm - Fix major stalls by fixing an improper taskqueue priority

* drm was creating task queues with a LWKT priority of 0, which is
lower than the priority of a running user thread.

* Fix all case

drm - Fix major stalls by fixing an improper taskqueue priority

* drm was creating task queues with a LWKT priority of 0, which is
lower than the priority of a running user thread.

* Fix all cases where improper priorities are passed to
taskqueue_start_threads(). This fixes major video stalls and glitches
that occur when other things might be running on the system cpu-bound.

* taskqueue_start_threads() now asserts if the priority passed to it is
illegal.

show more ...

drm/linux: Implement flush_work()

drm/radeon: Update to Linux 3.18

Mostly bugfixes and audio refactoring, no new functionalities.

This update has been prepared by Rimvydas Jasinskas.

drm: Add linux/lockdep.h

drm/linux: Improve the workqueue subsystem

Adding some APIs used by the drm/i915 code from Linux 3.17

drm/i915: Update to Linux 3.16

* Much improved support for Broadwell GPUs. Acceleration should now be fully
operational and the giant L4 eDRAM cache is now enabled when present.

* Baytrail/Valley

drm/i915: Update to Linux 3.16

* Much improved support for Broadwell GPUs. Acceleration should now be fully
operational and the giant L4 eDRAM cache is now enabled when present.

* Baytrail/Valleyview support improvements

* prelimary support for Cherryview (14nm Atom SOCs)

* Various fixes and performance improvements on most other GPU generations

* Improved runtime power management

* 5.4GHz DisplayPort support

* Large cursor support (up to 256x256 pixels), useful for high-dpi displays.

* Mapping of user pages into video memory (userptr)
This allows zero-copy downloads and efficient readback to/from the GPU,
allowing faster rendering of client-side software rasterisers, mitigation
of stalls due to read back and faster pipelining of texture data (such as
pixel buffer objects in GL or data blobs in CL). Mixed CPU/GPU operations
become more efficient in general.

show more ...

drm: Update linux/workqueue.h

Obtained-from: FreeBSD's OFED subsystem

drm: Implement and use the Linux version of kfree()

drm: Consolidate memory allocation types

* The DRM subsystem was using many different kmalloc types for no
good reason

* Some types were only used twice (1x kmalloc() and 1x kfree() calls)

* Rem

drm: Consolidate memory allocation types

* The DRM subsystem was using many different kmalloc types for no
good reason

* Some types were only used twice (1x kmalloc() and 1x kfree() calls)

* Remove that useless complexity and only keep one kmalloc type for
the whole DRM subsystem and its drivers

Inspired-by: OpenBSD

show more ...

linux/workqueue.h: Adapt to DragonFly

* Protect callout structures from concurrent operations on MP machines

* Add some functions used by the drm/i915 code from Linux 3.8.x