On Tue, Mar 20, 2018 at 09:11:42AM -0600, Anthony J. Bentley wrote:
> Hi Bryan,
>
> Bryan Steele writes:
> > I think I had the same problem with:
> >
> > radeondrm0 at pci1 dev 5 function 0 "ATI Radeon HD 3000" rev 0x00
> >
> > I believe it's related to the radeon(4) driver switching to glamor by
> > default on older families instead of EXA. I believe I saw this before
> > on the older driver manually testing out glamor, so it may be our
> > kernel radeondrm isn't ready for it..
> >
> > I could also reproduce it by producing lots text in an xterm, for
> > example objdump -d on a large binary.
> >
> > Section "Device"
> > Identifier "Card0"
> > Driver "radeon"
> > Option "AccelMethod" "EXA" # or glamor
> > EndSection
> >
> > Worksaround it for me at least..
>
> Same here: after switching to EXA, I can no longer reproduce the hang.
Hi,
the patch below makes the radeon driver default to EXA accel, except
for newer chipsets where only glamor is supported.
Index: src/radeon_glamor.c
===================================================================
RCS file: /cvs/OpenBSD/xenocara/driver/xf86-video-ati/src/radeon_glamor.c,v
retrieving revision 1.7
diff -u -p -u -r1.7 radeon_glamor.c
--- src/radeon_glamor.c 13 Mar 2018 06:13:14 -0000 1.7
+++ src/radeon_glamor.c 20 Mar 2018 20:00:33 -0000
@@ -86,7 +86,7 @@ radeon_glamor_pre_init(ScrnInfoPtr scrn)
s = xf86GetOptValString(info->Options, OPTION_ACCELMETHOD);
if (!s) {
if (xorgGetVersion() >= XORG_VERSION_NUMERIC(1,18,3,0,0)) {
- if (info->ChipFamily < CHIP_FAMILY_R600)
+ if (info->ChipFamily < CHIP_FAMILY_TAHITI)
return FALSE;
} else {
if (info->ChipFamily < CHIP_FAMILY_TAHITI)
--
Matthieu Herrb
