From f93f9ac4d9da20749197abc5f272839da5519e1d Mon Sep 17 00:00:00 2001 From: Youness Alaoui Date: Fri, 2 Mar 2018 16:12:04 -0500 Subject: [PATCH 14/15] purism/librem_skl: Set TCC Activation at 95C Set the Thermal Control Circuit (TCC) activaction value to 95C even though FSP integration guide says to set it to 100C for SKL-U (offset at 0), because when the TCC activates at 100C, the CPU will have already shut itself down from overheating protection. This was tested on Purism Librem 13 v2. A bisect showed that the immediate shutdowns happened after commit [1] was merged which led to this solution. There is still a temperature ramping problem where a 'stress -c 4' command will bring the temperature up from 50 to 100C (95C after this patch) within a few milliseconds, instead of it taking many dozens of seconds to reach ~80C. A bisect shows this regression was introduced in commit [2] and still needs to be investigated. This change may not be necessary anymore once the temperature ramping problem is fixed, but it is still wise to keep it for preventing shutdowns in corner cases. [1] ec5a947b (soc/intel/skylake: make tcc_offset take effect) [2] fb1cd095 (purism/librem13v2: migrate from FSP 1.1 to 2.0) Change-Id: Idfc001c8e46ed3b07b24150c961c4b9bc9b71a62 Signed-off-by: Youness Alaoui --- src/mainboard/purism/librem_skl/variants/librem13v2/devicetree.cb | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/mainboard/purism/librem_skl/variants/librem13v2/devicetree.cb b/src/mainboard/purism/librem_skl/variants/librem13v2/devicetree.cb index 9ce1d91549..159d921046 100644 --- a/src/mainboard/purism/librem_skl/variants/librem13v2/devicetree.cb +++ b/src/mainboard/purism/librem_skl/variants/librem13v2/devicetree.cb @@ -10,6 +10,12 @@ chip soc/intel/skylake register "eist_enable" = "1" register "VmxEnable" = "1" + # Set the Thermal Control Circuit (TCC) activaction value to 95C + # even though FSP integration guide says to set it to 100C for SKL-U + # (offset at 0), because when the TCC activates at 100C, the CPU + # will have already shut itself down from overheating protection. + register "tcc_offset" = "5" # TCC of 95C + # GPE configuration # Note that GPE events called out in ASL code rely on this # route. i.e. If this route changes then the affected GPE -- 2.14.3