Commit Graph

2 Commits

Author SHA1 Message Date
Shane Synan
9baca41064 ipq806x: fix min<>target opp-microvolt DTS mixup
Rearrange all voltage triplets for "opp_table0" to match the
specifications.  "opp-microvolt" and "opp-microvolt-<name>" triplets
are in order of <target min max>, and NOT <min target max>.

Previously, the CPU would *always* spend its time at the "minimum"
voltage, ignoring the actual intended target.  This is a regression
from previous behavior.

On an NBG6817 with a Qualcomm CPU of PVS bin #2...
(see &opp_table0 -> opp-1725000000 -> opp-microvolt-speed0-pvs2-v0)

* Before:
/usr/bin/tail -n +1 /sys/kernel/debug/opp/cpu0/opp\:1725000000/supply-0/u_volt_*
==> /sys/kernel/debug/opp/cpu0/opp:1725000000/supply-0/u_volt_max <==
1260000

==> /sys/kernel/debug/opp/cpu0/opp:1725000000/supply-0/u_volt_min <==
1200000

==> /sys/kernel/debug/opp/cpu0/opp:1725000000/supply-0/u_volt_target <==
1140000

* After:
/usr/bin/tail -n +1 /sys/kernel/debug/opp/cpu0/opp\:1725000000/supply-0/u_volt_*
==> /sys/kernel/debug/opp/cpu0/opp:1725000000/supply-0/u_volt_max <==
1260000

==> /sys/kernel/debug/opp/cpu0/opp:1725000000/supply-0/u_volt_min <==
1140000

==> /sys/kernel/debug/opp/cpu0/opp:1725000000/supply-0/u_volt_target <==
1200000

To check voltages and frequencies at run time, use...
/bin/cat /sys/kernel/debug/regulator/regulator_summary &&
  /bin/cat /sys/kernel/debug/clk/clk_summary | grep "hfpll"

See
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/Documentation/devicetree/bindings/opp/opp.txt?h=v5.4.142#n91

Fixes: 1e25423be8 ("ipq806x: refresh dtsi patches")

Signed-off-by: Shane Synan <digitalcircuit36939@gmail.com>
Reviewed-by: Ansuel Smith <ansuelsmth@gmail.com>
[commit message style cleanup, another kernel refresh]
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
2021-08-21 23:54:31 +02:00
Ansuel Smith
e5d50f69e8 ipq806x: reduce pci IO space to 64k
With some talk with the ARM maintainer, it was notice that enlarging the limit
to the current value is VERY wrong and clash with other memory.
A better solution would be to reduce the IO space from 1MB to 64K as probably
it's a long lasting typo and even x86 arch doesn't have a IO space that big.

Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
2021-05-23 15:09:06 +02:00