Hi Dave,
After merging the drm tree, today's linux-next build (x86_64 allmodconfig)
failed like this:
drivers/gpu/drm/i915/i915_irq.c: In function 'gen6_rps_reset_ei':
drivers/gpu/drm/i915/i915_irq.c:1115:38: error: 'struct intel_gen6_power_mgmt' has no member named 'down_ei'
vlv_c0_read(dev_priv, &dev_priv->rps.down_ei);
^
drivers/gpu/drm/i915/i915_irq.c:1116:15: error: 'struct intel_gen6_power_mgmt' has no member named 'up_ei'
dev_priv->rps.up_ei = dev_priv->rps.down_ei;
^
drivers/gpu/drm/i915/i915_irq.c:1116:37: error: 'struct intel_gen6_power_mgmt' has no member named 'down_ei'
dev_priv->rps.up_ei = dev_priv->rps.down_ei;
^
drivers/gpu/drm/i915/i915_irq.c: In function 'vlv_wa_c0_ei':
drivers/gpu/drm/i915/i915_irq.c:1133:21: error: 'struct intel_gen6_power_mgmt' has no member named 'down_ei'
&dev_priv->rps.down_ei, &now,
^
drivers/gpu/drm/i915/i915_irq.c:1136:16: error: 'struct intel_gen6_power_mgmt' has no member named 'down_ei'
dev_priv->rps.down_ei = now;
^
drivers/gpu/drm/i915/i915_irq.c:1141:20: error: 'struct intel_gen6_power_mgmt' has no member named 'up_ei'
&dev_priv->rps.up_ei, &now,
^
drivers/gpu/drm/i915/i915_irq.c:1144:16: error: 'struct intel_gen6_power_mgmt' has no member named 'up_ei'
dev_priv->rps.up_ei = now;
^
Caused by commit
8f68d591d476 ("drm/i915: Stop using RP_DOWN_EI on Baytrail")
from the drm-intel-fixes tree and a bad merge resolution on my part.
I reapplied that commit by hand. It would be nice if this conflict was
resolved in the drm tree.
This cherry picking of fixes from new development back to Linus' tree
can be a real pain when so many other changes happen in the same files.
--
Cheers,
Stephen Rothwell
On Mon, Mar 20, 2017 at 1:51 AM, Stephen Rothwell <[email protected]> wrote:
> This cherry picking of fixes from new development back to Linus' tree
> can be a real pain when so many other changes happen in the same files.
One possible fix for this would be if you reuse our rerere cache. The
only reason we don't go insane with all the drm conflicts is that we
completely distributed conflict resolution. Developers push a patch,
script tells them there's a conflict, they resolve it, maintainers
never even notice.We only notice when we double-check the merge
resolution when rerere re-applies it for the real backmerge :-) The
merge order in drm-tip should also match what you have in linux-next,
so you should be able to entirely reuse them.
Anyway, if you trust us enough to scoop up random git rerere caches
(or at least use them to double-check your own), they're all public:
git://anongit.freedesktop.org/drm-tip rerere-cache
Or
https://cgit.freedesktop.org/drm-tip/tree/rr-cache?h=rerere-cache
Yes we should probably gc them, but disk space is cheap.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch
On Mon, Mar 20, 2017 at 9:03 AM, Daniel Vetter <[email protected]> wrote:
> On Mon, Mar 20, 2017 at 1:51 AM, Stephen Rothwell <[email protected]> wrote:
>> This cherry picking of fixes from new development back to Linus' tree
>> can be a real pain when so many other changes happen in the same files.
>
> One possible fix for this would be if you reuse our rerere cache. The
> only reason we don't go insane with all the drm conflicts is that we
> completely distributed conflict resolution. Developers push a patch,
> script tells them there's a conflict, they resolve it, maintainers
> never even notice.We only notice when we double-check the merge
> resolution when rerere re-applies it for the real backmerge :-) The
> merge order in drm-tip should also match what you have in linux-next,
> so you should be able to entirely reuse them.
>
> Anyway, if you trust us enough to scoop up random git rerere caches
> (or at least use them to double-check your own), they're all public:
>
> git://anongit.freedesktop.org/drm-tip rerere-cache
>
> Or
>
> https://cgit.freedesktop.org/drm-tip/tree/rr-cache?h=rerere-cache
>
> Yes we should probably gc them, but disk space is cheap.
And the merge order, in case you want to check that. We can adjust it ofc:
https://cgit.freedesktop.org/drm-tip/tree/nightly.conf?h=rerere-cache
Cheers, Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch