Quoting Daniele Ceraolo Spurio (2019-05-31 23:24:07) > Separate the display PM from the PCI-level runtime PM. > I'll follow this up with v2 of the rpm encapsulation series [1], but > I'd like to get this in before that to avoid having to carry this > big dumb diff in that series. With RUNTIME_PM_DEBUG disabled, add/remove: 3/1 grow/shrink: 6/8 up/down: 396/-393 (3) Function old new delta intel_runtime_pm_release - 274 +274 intel_runtime_pm_put_raw - 45 +45 intel_runtime_pm_put_unchecked 10 48 +38 intel_display_power_put_async_work 179 192 +13 intel_display_power_flush_work 117 126 +9 __intel_display_power_put_async 193 199 +6 intel_runtime_pm_get_raw - 4 +4 intel_display_power_grab_async_put_ref 117 121 +4 __warned 469 472 +3 intel_runtime_pm_get 10 7 -3 intel_power_domains_enable 38 33 -5 intel_display_power_put_unchecked 23 18 -5 intel_display_power_get_if_enabled 143 138 -5 intel_display_power_get 84 79 -5 intel_power_domains_suspend 490 480 -10 intel_power_domains_fini_hw 116 106 -10 release_async_put_domains 220 203 -17 __intel_runtime_pm_put.constprop 333 - -333 Total: Before=23394388, After=23394391, chg +0.00% which is my biggest worry when meddling with these, that we accidentally explode production code with unused debugging (all those wakerefs). Lgtm, I would like Jani to indicate that he's happy with this split as well since he has been looking at very similar ideas. -Chris _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
