> dmesg:[ 68.627296] nouveau E[ PBUS][0000:00:0d.0] MMIO write of
> x00820001 FAULT at 0x00b000
I've had the same here, with a C61 onboard chip.
The next message, caught with a serial console just before freezing
with a sheared pattern on the screen, is:
nouveau E[ PBUS][0000:00:0d.0] MMIO write of 0x010c0001 FAULT at 0x00b010
They're caused the following line in core/engine/mpeg/nv31.c:194:
nv_wr32(priv, 0x00b008 + (i * 0x10), tile->pitch);
So, disabling just the mpeg engine in core/engine/device/nv40.c did the trick
for me: no more freezing or fault messages in dmesg.
For now, I think that's better than going full nouveau.noaccel=1.
Patch is just for the 0x4c/C61 chipset; the same could be done for other
chipsets.
--- linux-source-3.16/drivers/gpu/drm/nouveau/core/engine/device/nv40.c~
2015-05-29 09:12:37.876649108 +0300
+++ linux-source-3.16/drivers/gpu/drm/nouveau/core/engine/device/nv40.c
2015-05-29 09:17:10.716649906 +0300
@@ -322,7 +322,9 @@ nv40_identify(struct nouveau_device *dev
device->oclass[NVDEV_ENGINE_FIFO ] = nv40_fifo_oclass;
device->oclass[NVDEV_ENGINE_SW ] = nv10_software_oclass;
device->oclass[NVDEV_ENGINE_GR ] = &nv40_graph_oclass;
+ /* XXX
device->oclass[NVDEV_ENGINE_MPEG ] = &nv44_mpeg_oclass;
+ */
device->oclass[NVDEV_ENGINE_DISP ] = nv04_disp_oclass;
device->oclass[NVDEV_ENGINE_PERFMON] = nv40_perfmon_oclass;
break;
--
To UNSUBSCRIBE, email to debian-x-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive:
https://lists.debian.org/1e533c87ca19074f531ad85ceda35e08.6b0b1753@showman
