The req->fence.error will be set if this request caused GPU hang so
we can use this value to workload->status to indicate whether this
GVT request caused any problem. If it caused GPU hang, we shouldn't
trigger any context switch back to the guest.
v2:
- only take -EIO from fence->error. (Zhenyu)
Fixes: 8f1117abb4 (drm/i915/gvt: handle workload lifecycle properly)
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
For the vGPU workloads, now GVT-g use per vGPU scheduler, the per-ring
work_thread only pick workload belongs to the current vGPU. And with time
slice based scheduler, it waits all the engines become idle before do vGPU
switch. So we can run free dispatch in per-ring work_thread, different ring
running in different 'vGPU' won't happen.
For the workloads between vGPU and Host, this scheduler_mutex can't block
host to dispatch workload into other ring engines.
Here remove this mutex since it impacts the performance when applications
use more than 1 ring engines in 1 vgpu.
ring0 running in vGPU1, ring1 running in Host. Will happen.
ring0 running in vGPU1, ring1 running in vGPU2. Won't happen.
Signed-off-by: Weinan Li <weinan.z.li@intel.com>
Signed-off-by: Ping Gao <ping.a.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
The time based scheduler poll context busy status at every
micro-second during vGPU switch, it will make GPU idle for a while
when the context is very small and completed before the next
micro-second arrival. Trigger scheduling immediately after context
complete will eliminate GPU idle and improve performance.
Create two vGPU with same type, run Heaven simultaneously:
Before this patch:
+---------+----------+----------+
| | vGPU1 | vGPU2 |
+---------+----------+----------+
| Heaven | 357 | 354 |
+-------------------------------+
After this patch:
+---------+----------+----------+
| | vGPU1 | vGPU2 |
+---------+----------+----------+
| Heaven | 397 | 398 |
+-------------------------------+
v2: Let need_reschedule protect by gvt-lock.
Signed-off-by: Ping Gao <ping.a.gao@intel.com>
Signed-off-by: Weinan Li <weinan.z.li@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Commit ab9da627906a ("drm/i915: make context status notifier head be
per engine") gives us a chance to inspect every single request. Then
we can eliminate unnecessary mmio switching for same vGPU. We only
need mmio switching for different VMs (including host).
This patch introduced a new general API intel_gvt_switch_mmio() to
replace the old intel_gvt_load/restore_render_mmio(). This function
can be further optimized for vGPU to vGPU switching.
To support individual ring switch, we track the owner who occupy
each ring. When another VM or host request a ring we do the mmio
context switching. Otherwise no need to switch the ring.
This optimization is very useful if only one guest has plenty of
workloads and the host is mostly idle. The best case is no mmio
switching will happen.
v2:
o fix missing ring switch issue. (chuanxiao)
o support individual ring switch.
Signed-off-by: Changbin Du <changbin.du@intel.com>
Reviewed-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Since unifying ringbuffer/execlist submission to use
engine->pin_context, we ensure that the intel_ring is available before
we start constructing the request. We can therefore move the assignment
of the request->ring to the central i915_gem_request_alloc() and not
require it in every engine->request_alloc() callback. Another small step
towards simplification (of the core, but at a cost of handling error
pointers in less important callers of engine->pin_context).
v2: Rearrange a few branches to reduce impact of PTR_ERR() on gcc's code
generation.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Oscar Mateo <oscar.mateo@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Oscar Mateo <oscar.mateo@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170504093308.4137-1-chris@chris-wilson.co.uk
Backmerge the main drm-next pull to sync up.
Chris also pointed out that
commit ade0b0c965
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Sat Apr 22 09:15:37 2017 +0100
drm/i915: Confirm the request is still active before adding it to the await
is double-applied in the git merge, so make sure we get this right.
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
gvt-next-fixes-2017-04-20
- some code optimization from Changbin
- debug message cleanup after QoS merge
- misc fixes for display mmio init, reset vgpu warning, etc.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
As those debug messages might appear in every timer call for scheduler,
it's too noisy, eat too much log and aren't meaningful. So remove them.
Cc: Ping Gao <ping.a.gao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
In my previous Commit ab9da627906a ("drm/i915: make context status
notifier head be per engine") rely on scheduler->current_workload[x]
to distinguish gvt spacial request from i915 request. But this is
not always true since no synchronization between workload_thread and
lrc irq handler.
lrc irq handler workload_thread
---- ----
pick i915 requests;
intel_vgpu_submit_execlist();
current_workload[x] = xxx;
shadow_context_status_change();
Then current_workload[x] is not null but current request is of i915 self.
So instead we check ctx flag CONTEXT_FORCE_SINGLE_SUBMISSION. Only gvt
request set this flag and always set.
v2: Reverse the order of multi-condition 'if' statement.
Fixes: ab9da6279 ("drm/i915: make context status notifier head be per engine")
Signed-off-by: Changbin Du <changbin.du@intel.com>
Reviewed-by: Yulei Zhang <yulei.zhang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
GVTg has introduced the context status notifier to schedule the GVTg
workload. At that time, the notifier is bound to GVTg context only,
so GVTg is not aware of host workloads.
Now we are going to improve GVTg's guest workload scheduler policy,
and add Guc emulation support for new Gen graphics. Both these two
features require acknowledgment for all contexts running on hardware.
(But will not alter host workload.) So here try to make some change.
The change is simple:
1. Move the context status notifier head from i915_gem_context to
intel_engine_cs. Which means there is a notifier head per engine
instead of per context. Execlist driver still call notifier for
each context sched-in/out events of current engine.
2. At GVTg side, it binds a notifier_block for each physical engine
at GVTg initialization period. Then GVTg can hear all context
status events.
In this patch, GVTg do nothing for host context event, but later
will add a function there. But in any case, the notifier callback is
a noop if this is no active vGPU.
Since intel_gvt_init() is called at early initialization stage and
require the status notifier head has been initiated, I initiate it in
intel_engine_setup().
v2: remove a redundant newline. (chris)
Fixes: 3c7ba6359d ("drm/i915: Introduce execlist context status change notification")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=100232
Signed-off-by: Changbin Du <changbin.du@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Zhi Wang <zhi.a.wang@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170313024711.28591-1-changbin.du@intel.com
Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
(cherry picked from commit 3fc03069bc)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170321144720.17020-1-chris@chris-wilson.co.uk
When handling guest request, GVT needs to populate/update shadow_ctx
with guest context. This behavior needs to make sure the shadow_ctx
is pinned. The current implementation is relying on i195 allocate request
to pin but this way cannot guarantee the i915 not to unpin the shadow_ctx
when GVT update the guest context from shadow_ctx. So GVT should pin/unpin
the shadow_ctx by itself.
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
The shadow indirect context image should be only scanned when valid.
So far, Only RCS ring has the shadow indirect context image. This patch
limits the scan logic only for RCS ring.
v2. refine description of the subject
v3. fix alignment. (Zhenyu)
Signed-off-by: Tina Zhang <tina.zhang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
gvt_err should be used only for the very few critical error message
during host i915 drvier initialization. This patch
1. removes the redundant gvt_err;
2. creates a new gvt_vgpu_err to show errors caused by vgpu;
3. replaces the most gvt_err with gvt_vgpu_err;
4. leaves very few gvt_err for dumping gvt error during host gvt
initialization.
v2. change name to gvt_vgpu_err and add vgpu id to the message. (Kevin)
add gpu id to gvt_vgpu_err. (Zhi)
v3. remove gpu id from gvt_vgpu_err caller. (Zhi)
v4. add vgpu check to the gvt_vgpu_err macro. (Zhiyuan)
v5. add comments for v3 and v4.
v6. split the big patch into two, with this patch only for checking
gvt_vgpu_err. (Zhenyu)
v7. rebase to staging branch
v8. rebase to fix branch
Signed-off-by: Tina Zhang <tina.zhang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
GVTg has introduced the context status notifier to schedule the GVTg
workload. At that time, the notifier is bound to GVTg context only,
so GVTg is not aware of host workloads.
Now we are going to improve GVTg's guest workload scheduler policy,
and add Guc emulation support for new Gen graphics. Both these two
features require acknowledgment for all contexts running on hardware.
(But will not alter host workload.) So here try to make some change.
The change is simple:
1. Move the context status notifier head from i915_gem_context to
intel_engine_cs. Which means there is a notifier head per engine
instead of per context. Execlist driver still call notifier for
each context sched-in/out events of current engine.
2. At GVTg side, it binds a notifier_block for each physical engine
at GVTg initialization period. Then GVTg can hear all context
status events.
In this patch, GVTg do nothing for host context event, but later
will add a function there. But in any case, the notifier callback is
a noop if this is no active vGPU.
Since intel_gvt_init() is called at early initialization stage and
require the status notifier head has been initiated, I initiate it in
intel_engine_setup().
v2: remove a redundant newline. (chris)
Fixes: 3c7ba6359d ("drm/i915: Introduce execlist context status change notification")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=100232
Signed-off-by: Changbin Du <changbin.du@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Zhi Wang <zhi.a.wang@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170313024711.28591-1-changbin.du@intel.com
Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Currently i915 has a request replay mechanism which can make sure
the request can be replayed after a GPU reset. With this mechanism,
gvt should wait until the GVT request seqno passed before complete
the current workload. So that there should be a context switch interrupt
come before gvt free the workload. In this way, workload lifecylce
matches with the i915 request lifecycle. The workload can only be freed
after the request is completed.
v2: use gvt_dbg_sched instead of gvt_err to print when wait again
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Due to the request replay, context switch interrupt may come after
gvt free the workload thus can cause a kernel NULL pointer kernel
panic. This patch will add a simple check to avoid this for a short
term.
From long term, gvt workload lifecycle doesn't match with i915 request
and need to find a proper way to manage this.
v4: simplify the NULL pointer check.
v5: add unlikely to optimize.
Signed-off-by: Chuanxiao Dong <chuanxiao.dong@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
We need to be careful to only update addr mode for gvt shadow context
descriptor but keep other valid config. This fixes GPU hang caused by
invalid descriptor submitted for gvt workload.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Remove a redundant end of line in below log.
'will complete workload %p\n, status: %d\n'
Signed-off-by: Changbin Du <changbin.du@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Backmerge Linus master to get the connector locking revert.
* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux: (645 commits)
sysctl: fix proc_doulongvec_ms_jiffies_minmax()
Revert "drm/probe-helpers: Drop locking from poll_enable"
MAINTAINERS: add Dan Streetman to zbud maintainers
MAINTAINERS: add Dan Streetman to zswap maintainers
mm: do not export ioremap_page_range symbol for external module
mn10300: fix build error of missing fpu_save()
romfs: use different way to generate fsid for BLOCK or MTD
frv: add missing atomic64 operations
mm, page_alloc: fix premature OOM when racing with cpuset mems update
mm, page_alloc: move cpuset seqcount checking to slowpath
mm, page_alloc: fix fast-path race with cpuset update or removal
mm, page_alloc: fix check for NULL preferred_zone
kernel/panic.c: add missing \n
fbdev: color map copying bounds checking
frv: add atomic64_add_unless()
mm/mempolicy.c: do not put mempolicy before using its nodemask
radix-tree: fix private list warnings
Documentation/filesystems/proc.txt: add VmPin
mm, memcg: do not retry precharge charges
proc: add a schedule point in proc_pid_readdir()
...
The vgpu->running_workload_num is used to determine whether a vgpu has
any workload running or not. So we should make sure the workload is
really done before we dec running_workload_num. Function
complete_current_workload is not the right place to do it, since this
function is still processing the workload. This patch move the dec op
afterward.
v2: move dec op before wake_up(&scheduler->workload_complete_wq) (Min He)
Signed-off-by: Changbin Du <changbin.du@intel.com>
Reviewed-by: Min He <min.he@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
In the function workload_thread(), we invoke complete_current_workload()
to cleanup the just processed workload (workload will be freed there).
So we cannot access workload->req after that. This patch move
complete_current_workload() afterward.
Signed-off-by: Changbin Du <changbin.du@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Backmerge Linux 4.10-rc2 to resync with our -fixes cherry-picks. I've
done the backmerge directly because Dave is on vacation.
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Need to be careful to release struct_mutext when request alloc
failed and take consistent handling for return status as with
normal go out path. Ensure to check correct workload request in
complete path too.
v2: Add Fixes note
Fixes: 90d27a1b18 ("drm/i915/gvt: fix deadlock in workload_thread")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Pei Zhang <pei.zhang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
It's a classical abba type deadlock when using 2 mutex objects, which
are gvt.lock(a) and drm.struct_mutex(b). Deadlock happens in threads:
1. intel_gvt_create/destroy_vgpu: P(a)->P(b)
2. workload_thread: P(b)->P(a)
Fix solution is align the lock acquire sequence in both threads. This
patch choose to adjust the sequence in workload_thread function.
This fixed lockup symptom for guest-reboot stress test.
v2: adjust sequence in workload_thread based on zhenyu's suggestion.
adjust sequence in create/destroy_vgpu function.
v3: fix to still require struct_mutex for dispatch_workload()
Signed-off-by: Pei Zhang <pei.zhang@intel.com>
[zhenyuw: fix unused variables warnings.]
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
kmap_atomic doesn't allow sleep until unmapped. However,
it's necessary to allow sleep during reading/writing guest
memory, so use kmap instead.
Signed-off-by: Bing Niu <bing.niu@intel.com>
Signed-off-by: Xiaoguang Chen <xiaoguang.chen@intel.com>
Signed-off-by: Jike Song <jike.song@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
From commit e95433c73a, workload status setting
was changed to only capture on error path, but we need to set it properly in
normal path too, otherwise we'll fail to complete workload which could lead
guest VM vGPU reset.
v2: uses braces and add Fixes tag.
Fixes: e95433c73a ("drm/i915: Rearrange i915_wait_request() accounting with callers")
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Our low-level wait routine has evolved from our generic wait interface
that handled unlocked, RPS boosting, waits with time tracking. If we
push our GEM fence tracking to use reservation_objects (required for
handling multiple timelines), we lose the ability to pass the required
information down to i915_wait_request(). However, if we push the extra
functionality from i915_wait_request() to the individual callsites
(i915_gem_object_wait_rendering and i915_gem_wait_ioctl) that make use
of those extras, we can both simplify our low level wait and prepare for
extending the GEM interface for use of reservation_objects.
v2: Rewrite i915_wait_request() kerneldocs
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20161028125858.23563-4-chris@chris-wilson.co.uk
We cannot use blocking method mutex_lock inside a wait loop.
Here we invoke pick_next_workload() which needs acquire a
mutex in our "condition" experssion. Then we go into a another
of the going-to-sleep sequence and changing the task state.
This is a dangerous. Let's rewrite the wait sequence to avoid
nested sleeping.
v2: fix do...while loop exit condition (zhenyu)
v3: rebase to gvt-staging branch
Signed-off-by: Du, Changbin <changbin.du@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Switch to use new for_each_engine() helper to properly access
enabled intel_engine_cs as i915 core has changed that to be
dynamic managed. At GVT-g init time would still depend on ring
mask to determine engine list as it's earlier.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
For whatever reason, the gvt scheduler runs synchronously. At the very
least, lets run synchronously without holding the struct_mutex.
v2: cut'n'paste mutex_lock instead of unlock.
Replace long hold of struct_mutex with a mutex to serialise the worker
threads.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
The workload took a pointer to the request, and even waited upon,
without holding a reference on the request. Take that reference
explicitly and fix up the error path following request allocation that
missed flushing the request.
v2: [zhenyuw]
- drop request put in error path for dispatch, as main thread
caller will handle it identically to a real request.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
i915 core should only call functions and structures exposed through
intel_gvt.h. Remove internal gvt.h and i915_pvinfo.h.
Change for internal intel_gvt structure as private handler which
not requires to expose gvt internal structure for i915 core.
v2: Fix per Chris's comment
- carefully handle dev_priv->gvt assignment
- add necessary bracket for macro helper
- forward declartion struct intel_gvt
- keep free operation within same file handling alloc
v3: fix use after free and remove intel_gvt.initialized
v4: change to_gvt() to an inline
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
As different VM may configure different render MMIOs when executing
workload, to schedule workloads between different VM, the render MMIOs
have to be switched.
Signed-off-by: Zhi Wang <zhi.a.wang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
This patch introduces the vGPU workload scheduler routines.
GVT workload scheduler is responsible for picking and executing GVT workload
from current scheduled vGPU. Before the workload is submitted to host i915,
the guest execlist context will be shadowed in the host GVT shadow context.
the instructions in guest ring buffer will be copied into GVT shadow ring
buffer. Then GVT-g workload scheduler will scan the instructions in guest
ring buffer and submit it to host i915.
Signed-off-by: Zhi Wang <zhi.a.wang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
