From cd86612112b2156caf0576f2af0b80de8b92fb3e Mon Sep 17 00:00:00 2001
From: Ben Skeggs <bskeggs@redhat.com>
Date: Wed, 2 Sep 2020 15:30:33 +1000
Subject: drm/nouveau/kms/nv50-gp1xx: add WAR for EVO push buffer HW bug
Git-commit: ca386aa7155a5467fa7b2b8376f4da8f8e59be4d
Patch-mainline: v5.9-rc4
References: jsc#SLE-12680, jsc#SLE-12880, jsc#SLE-12882, jsc#SLE-12883, jsc#SLE-13496, jsc#SLE-15322
Thanks to NVIDIA for confirming this workaround, and clarifying which HW
is affected.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Tested-by: Alexander Kapshuk <alexander.kapshuk@gmail.com>
Signed-off-by: Patrik Jakobsson <pjakobsson@suse.de>
---
drivers/gpu/drm/nouveau/dispnv50/disp.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index e7874877da85..1ed242070001 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -257,6 +257,12 @@ nv50_dmac_create(struct nvif_device *device, struct nvif_object *disp,
dmac->push->end = dmac->push->bgn;
dmac->max = 0x1000/4 - 1;
+ /* EVO channels are affected by a HW bug where the last 12 DWORDs
+ * of the push buffer aren't able to be used safely.
+ */
+ if (disp->oclass < GV100_DISP)
+ dmac->max -= 12;
+
args->pushbuf = nvif_handle(&dmac->_push.mem.object);
ret = nv50_chan_create(device, disp, oclass, head, data, size,
--
2.29.2