Nouveau - Aug 2015 - [PATCH] Revert "drm/nouveau/fifo/gk104: kick channels when deactivating them"

This reverts commit 1addc1264852

This commit seems to cause crashes in gk104_fifo_intr_runlist() by
returning 0xbad0da00 when register 0x2a00 is read. Since this commit was
intended for GM20B which is not completely supported yet, let's revert
it for the time being.

Reported-by: Eric Biggers <ebiggers3 at gmail.com>
Signed-off-by: Alexandre Courbot <acourbot at nvidia.com>
---
David, it would be great if this could be merged for 4.2 since lots of
users could potentially experience this issue. Thanks!

 drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c | 29 +++++++-----------------
 1 file changed, 8 insertions(+), 21 deletions(-)

diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c
b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c
index 52c22b026005..e10f9644140f 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c
@@ -166,30 +166,14 @@ gk104_fifo_context_attach(struct nvkm_object *parent,
 }
 
 static int
-gk104_fifo_chan_kick(struct gk104_fifo_chan *chan)
-{
-	struct nvkm_object *obj = (void *)chan;
-	struct gk104_fifo_priv *priv = (void *)obj->engine;
-
-	nv_wr32(priv, 0x002634, chan->base.chid);
-	if (!nv_wait(priv, 0x002634, 0x100000, 0x000000)) {
-		nv_error(priv, "channel %d [%s] kick timeout\n",
-			 chan->base.chid, nvkm_client_name(chan));
-		return -EBUSY;
-	}
-
-	return 0;
-}
-
-static int
 gk104_fifo_context_detach(struct nvkm_object *parent, bool suspend,
 			  struct nvkm_object *object)
 {
 	struct nvkm_bar *bar = nvkm_bar(parent);
+	struct gk104_fifo_priv *priv = (void *)parent->engine;
 	struct gk104_fifo_base *base = (void *)parent->parent;
 	struct gk104_fifo_chan *chan = (void *)parent;
 	u32 addr;
-	int ret;
 
 	switch (nv_engidx(object->engine)) {
 	case NVDEV_ENGINE_SW    : return 0;
@@ -204,9 +188,13 @@ gk104_fifo_context_detach(struct nvkm_object *parent, bool
suspend,
 		return -EINVAL;
 	}
 
-	ret = gk104_fifo_chan_kick(chan);
-	if (ret && suspend)
-		return ret;
+	nv_wr32(priv, 0x002634, chan->base.chid);
+	if (!nv_wait(priv, 0x002634, 0xffffffff, chan->base.chid)) {
+		nv_error(priv, "channel %d [%s] kick timeout\n",
+			 chan->base.chid, nvkm_client_name(chan));
+		if (suspend)
+			return -EBUSY;
+	}
 
 	if (addr) {
 		nv_wo32(base, addr + 0x00, 0x00000000);
@@ -331,7 +319,6 @@ gk104_fifo_chan_fini(struct nvkm_object *object, bool
suspend)
 		gk104_fifo_runlist_update(priv, chan->engine);
 	}
 
-	gk104_fifo_chan_kick(chan);
 	nv_wr32(priv, 0x800000 + (chid * 8), 0x00000000);
 	return nvkm_fifo_channel_fini(&chan->base, suspend);
 }
-- 
2.5.0

Hi,

On Wed, Aug 12, 2015 at 01:17:38PM +0900, Alexandre Courbot
wrote:
> This reverts commit 1addc1264852
> 
> This commit seems to cause crashes in gk104_fifo_intr_runlist() by
> returning 0xbad0da00 when register 0x2a00 is read. Since this commit was
> intended for GM20B which is not completely supported yet, let's revert
> it for the time being.
> 
> Reported-by: Eric Biggers <ebiggers3 at gmail.com>
> Signed-off-by: Alexandre Courbot <acourbot at nvidia.com>
> ---
> David, it would be great if this could be merged for 4.2 since lots of
> users could potentially experience this issue. Thanks!
Tested-by: Afzal Mohammed <afzal.mohd.ma at gmail.com>

Please help $subject reach mainline for 4.2, w/o this revert, the
system here hangs most (>90%) of the time at boot time.

As an aside, yesterday after a marathon git bisect, came to the same
solution (though I don't understand what that change means). Was
about to report it and saw this one. Thanks Alexandre.

W/o the revert, in the rare case where it boots, below is observed in
addition to as compared to w/ revert,

[    9.826010] nouveau E[    PBUS][0000:01:00.0] MMIO read of 0x00000000 FAULT
at 0x122130 [ !ENGINE ]

Regards
Afzal

On Wed, Aug 12, 2015 at 3:00 PM, Afzal Mohammed <afzal.mohd.ma at
gmail.com> wrote:
> Hi,
>
> On Wed, Aug 12, 2015 at 01:17:38PM +0900, Alexandre Courbot wrote:
>> This reverts commit 1addc1264852
>>
>> This commit seems to cause crashes in gk104_fifo_intr_runlist() by
>> returning 0xbad0da00 when register 0x2a00 is read. Since this commit
was
>> intended for GM20B which is not completely supported yet, let's
revert
>> it for the time being.
>>
>> Reported-by: Eric Biggers <ebiggers3 at gmail.com>
>> Signed-off-by: Alexandre Courbot <acourbot at nvidia.com>
>> ---
>> David, it would be great if this could be merged for 4.2 since lots of
>> users could potentially experience this issue. Thanks!
>
> Tested-by: Afzal Mohammed <afzal.mohd.ma at gmail.com>
Thanks!
>
> Please help $subject reach mainline for 4.2, w/o this revert, the
> system here hangs most (>90%) of the time at boot time.
>
> As an aside, yesterday after a marathon git bisect, came to the same
> solution (though I don't understand what that change means). Was
> about to report it and saw this one. Thanks Alexandre.
All credit goes to Eric for bisecting and reporting this issue.
>
> W/o the revert, in the rare case where it boots, below is observed in
> addition to as compared to w/ revert,
>
> [    9.826010] nouveau E[    PBUS][0000:01:00.0] MMIO read of 0x00000000
FAULT at 0x122130 [ !ENGINE ]
Could you let me know what your card is? It may be useful to know the
range of affected cards when trying to fix this.

Thanks,
Alex.