drm/nouveau/fifo/gf100-: provide notification to user if channel is killed

There are instances (such as non-recoverable GPU page faults) where
NVKM decides that a channel's context is no longer viable, and will
be removed from the runlist.

This commit notifies the owner of the channel when this happens, so
it has the opportunity to take some kind of recovery action instead
of hanging.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

11 files changed, 58 insertions, 5 deletions

diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl906f.h b/drivers/gpu/drm/nouveau/include/nvif/cl906f.h
index e9302ec5cbfa..a2d5410a491b 100644
--- a/drivers/gpu/drm/nouveau/include/nvif/cl906f.h
+++ b/drivers/gpu/drm/nouveau/include/nvif/cl906f.h
@@ -11,4 +11,5 @@ struct fermi_channel_gpfifo_v0 {
 };
 
 #define NV906F_V0_NTFY_NON_STALL_INTERRUPT                                 0x00
+#define NV906F_V0_NTFY_KILLED                                              0x01
 #endif
diff --git a/drivers/gpu/drm/nouveau/include/nvif/cla06f.h b/drivers/gpu/drm/nouveau/include/nvif/cla06f.h
index 567123a8f874..2efa3d048bb9 100644
--- a/drivers/gpu/drm/nouveau/include/nvif/cla06f.h
+++ b/drivers/gpu/drm/nouveau/include/nvif/cla06f.h
@@ -26,4 +26,5 @@ struct kepler_channel_gpfifo_a_v0 {
 };
 
 #define NVA06F_V0_NTFY_NON_STALL_INTERRUPT                                 0x00
+#define NVA06F_V0_NTFY_KILLED                                              0x01
 #endif
diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h
index ed92fec5292c..24efa900d8ca 100644
--- a/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h
+++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h
@@ -40,6 +40,7 @@ struct nvkm_fifo {
 
 	struct nvkm_event uevent; /* async user trigger */
 	struct nvkm_event cevent; /* channel creation event */
+	struct nvkm_event kevent; /* channel killed */
 };
 
 void nvkm_fifo_pause(struct nvkm_fifo *, unsigned long *);
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c
index 4a01afe91734..3567c9920a37 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c
@@ -90,6 +90,31 @@ nvkm_fifo_chan_chid(struct nvkm_fifo *fifo, int chid, unsigned long *rflags)
 	return NULL;
 }
 
+void
+nvkm_fifo_kevent(struct nvkm_fifo *fifo, int chid)
+{
+	nvkm_event_send(&fifo->kevent, 1, chid, NULL, 0);
+}
+
+static int
+nvkm_fifo_kevent_ctor(struct nvkm_object *object, void *data, u32 size,
+		      struct nvkm_notify *notify)
+{
+	struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object);
+	if (size == 0) {
+		notify->size  = 0;
+		notify->types = 1;
+		notify->index = chan->chid;
+		return 0;
+	}
+	return -ENOSYS;
+}
+
+static const struct nvkm_event_func
+nvkm_fifo_kevent_func = {
+	.ctor = nvkm_fifo_kevent_ctor,
+};
+
 static int
 nvkm_fifo_cevent_ctor(struct nvkm_object *object, void *data, u32 size,
 		      struct nvkm_notify *notify)
@@ -247,6 +272,7 @@ nvkm_fifo_dtor(struct nvkm_engine *engine)
 	void *data = fifo;
 	if (fifo->func->dtor)
 		data = fifo->func->dtor(fifo);
+	nvkm_event_fini(&fifo->kevent);
 	nvkm_event_fini(&fifo->cevent);
 	nvkm_event_fini(&fifo->uevent);
 	return data;
@@ -289,5 +315,9 @@ nvkm_fifo_ctor(const struct nvkm_fifo_func *func, struct nvkm_device *device,
 			return ret;
 	}
 
-	return nvkm_event_init(&nvkm_fifo_cevent_func, 1, 1, &fifo->cevent);
+	ret = nvkm_event_init(&nvkm_fifo_cevent_func, 1, 1, &fifo->cevent);
+	if (ret)
+		return ret;
+
+	return nvkm_event_init(&nvkm_fifo_kevent_func, 1, nr, &fifo->kevent);
 }
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h
index 55dc415c5c08..d8019bdacd61 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h
@@ -29,5 +29,5 @@ struct nvkm_fifo_chan_oclass {
 	struct nvkm_sclass base;
 };
 
-int g84_fifo_chan_ntfy(struct nvkm_fifo_chan *, u32, struct nvkm_event **);
+int gf100_fifo_chan_ntfy(struct nvkm_fifo_chan *, u32, struct nvkm_event **);
 #endif
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chang84.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chang84.c
index 66bd1cda5602..61797c4dd07a 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chang84.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chang84.c
@@ -30,7 +30,7 @@
 
 #include <nvif/cl826e.h>
 
-int
+static int
 g84_fifo_chan_ntfy(struct nvkm_fifo_chan *chan, u32 type,
 		   struct nvkm_event **pevent)
 {
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c
index 8772a995329a..cd468ab1db12 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c
@@ -190,6 +190,7 @@ gf100_fifo_recover(struct gf100_fifo *fifo, struct nvkm_engine *engine,
 	if (engine != &fifo->base.engine)
 		fifo->recover.mask |= 1ULL << engine->subdev.index;
 	schedule_work(&fifo->recover.work);
+	nvkm_fifo_kevent(&fifo->base, chid);
 }
 
 static const struct nvkm_enum
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c
index 73ae4ecc2e9a..67a6a514b5fa 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c
@@ -178,6 +178,7 @@ gk104_fifo_recover(struct gk104_fifo *fifo, struct nvkm_engine *engine,
 
 	fifo->recover.runm |= BIT(chan->runl);
 	schedule_work(&fifo->recover.work);
+	nvkm_fifo_kevent(&fifo->base, chid);
 }
 
 static const struct nvkm_enum
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogf100.c
index 12d964260a29..f9e0377d3d24 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogf100.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogf100.c
@@ -32,6 +32,23 @@
 #include <nvif/cl906f.h>
 #include <nvif/unpack.h>
 
+int
+gf100_fifo_chan_ntfy(struct nvkm_fifo_chan *chan, u32 type,
+		     struct nvkm_event **pevent)
+{
+	switch (type) {
+	case NV906F_V0_NTFY_NON_STALL_INTERRUPT:
+		*pevent = &chan->fifo->uevent;
+		return 0;
+	case NV906F_V0_NTFY_KILLED:
+		*pevent = &chan->fifo->kevent;
+		return 0;
+	default:
+		break;
+	}
+	return -EINVAL;
+}
+
 static u32
 gf100_fifo_gpfifo_engine_addr(struct nvkm_engine *engine)
 {
@@ -184,7 +201,7 @@ gf100_fifo_gpfifo_func = {
 	.dtor = gf100_fifo_gpfifo_dtor,
 	.init = gf100_fifo_gpfifo_init,
 	.fini = gf100_fifo_gpfifo_fini,
-	.ntfy = g84_fifo_chan_ntfy,
+	.ntfy = gf100_fifo_chan_ntfy,
 	.engine_ctor = gf100_fifo_gpfifo_engine_ctor,
 	.engine_dtor = gf100_fifo_gpfifo_engine_dtor,
 	.engine_init = gf100_fifo_gpfifo_engine_init,
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogk104.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogk104.c
index a2df4f3e7763..7f5386910162 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogk104.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogk104.c
@@ -213,7 +213,7 @@ gk104_fifo_gpfifo_func = {
 	.dtor = gk104_fifo_gpfifo_dtor,
 	.init = gk104_fifo_gpfifo_init,
 	.fini = gk104_fifo_gpfifo_fini,
-	.ntfy = g84_fifo_chan_ntfy,
+	.ntfy = gf100_fifo_chan_ntfy,
 	.engine_ctor = gk104_fifo_gpfifo_engine_ctor,
 	.engine_dtor = gk104_fifo_gpfifo_engine_dtor,
 	.engine_init = gk104_fifo_gpfifo_engine_init,
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h
index 9bf0251c05d0..1fc9d8bf12f2 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h
@@ -7,6 +7,7 @@ int nvkm_fifo_ctor(const struct nvkm_fifo_func *, struct nvkm_device *,
 		   int index, int nr, struct nvkm_fifo *);
 void nvkm_fifo_uevent(struct nvkm_fifo *);
 void nvkm_fifo_cevent(struct nvkm_fifo *);
+void nvkm_fifo_kevent(struct nvkm_fifo *, int chid);
 
 struct nvkm_fifo_chan_oclass;
 struct nvkm_fifo_func {