Hi Linus,
Please pull from the tag
git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git \
pm-4.15-rc5
with top-most commit 63d15e8c2ac134b3fd5460d19043ffde6f6eeb61
Merge branch 'pm-pci'
on top of commit 1291a0d5049dbc06baaaf66a9ff3f53db493b19b
Linux 4.15-rc4
to receive power management fixes for 4.15-rc5.
These fix a regression in the ondemand and conservative cpufreq
governors that was introduced during the 4.13 cycle, a recent
regression in the imx6q cpufreq driver and a regression in the PCI
handling of hibernation from the 4.14 cycle.
Specifics:
- Fix an issue in the PCI handling of the "thaw" transition during
hibernation (after creating an image), introduced by a bug fix
from the 4.13 cycle and exposed by recent changes in the IRQ
subsystem, that caused pci_restore_state() to be called for
devices in low-power states in some cases which is incorrect
and breaks MSI management on some systems (Rafael Wysocki).
- Fix a recent regression in the imx6q cpufreq driver that broke
speed grading on i.MX6 QuadPlus by omitting checks causing
invalid operating performance points (OPPs) to be disabled on
that SoC as appropriate (Lucas Stach).
- Fix a regression introduced during the 4.14 cycle in the ondemand
and conservative cpufreq governors that causes the sampling
interval used by them to be shorter than the tick period in some
cases which leads to incorrect decisions (Rafael Wysocki).
Thanks!
---------------
Lucas Stach (1):
cpufreq: imx6q: fix speed grading regression on i.MX6 QuadPlus
Rafael J. Wysocki (2):
PCI / PM: Force devices to D0 in pci_pm_thaw_noirq()
cpufreq: governor: Ensure sufficiently large sampling intervals
---------------
drivers/cpufreq/cpufreq_governor.c | 19 ++++++++++++++++---
drivers/cpufreq/imx6q-cpufreq.c | 11 ++++++-----
drivers/pci/pci-driver.c | 7 ++++++-
3 files changed, 28 insertions(+), 9 deletions(-)
