| Message ID | 20181005153638.1886-1-digetx@gmail.com |
|---|---|
| Headers | show |
| Series | Continuing the work on coupled regulators | expand |
* Dmitry Osipenko <digetx@gmail.com> [181005 15:43]: > 1. Re-worked the original "Add voltage balancing mechanism" patch from > Maciej by: > 1) Fixing infinite loop within regulator_balance_voltage(). > 2) Handling suspend_state_t properly. > 3) Fixing broken compilation of the patch. Thanks for fixing these, this series no longer hangs for me on beagle-x15. Regards, Tony
On 10/8/18 8:58 PM, Tony Lindgren wrote: > * Dmitry Osipenko <digetx@gmail.com> [181005 15:43]: >> 1. Re-worked the original "Add voltage balancing mechanism" patch from >> Maciej by: >> 1) Fixing infinite loop within regulator_balance_voltage(). >> 2) Handling suspend_state_t properly. >> 3) Fixing broken compilation of the patch. > > Thanks for fixing these, this series no longer hangs for me > on beagle-x15. Thank you very much for testing once again!
On Fri, Oct 05, 2018 at 06:36:37PM +0300, Dmitry Osipenko wrote: > Wait/wound mutex shall be used in order to avoid lockups on locking of > coupled regulators. This breaks the build due to a few of the drivers (wm8350 and da9210 at least) taking rdev locks in their interrupt handlers. I'd suggest adding a function for those drivers to use - exposing the ww context seems like the wrong thing to do there. Otherwise the series looks good, I've applied everything up to here. Thanks for taking the time to pick this up, and especially for figuring out the ww_mutex conversion.
On 08.11.2018 16:07, Mark Brown wrote: > On Fri, Oct 05, 2018 at 06:36:37PM +0300, Dmitry Osipenko wrote: > >> Wait/wound mutex shall be used in order to avoid lockups on locking of >> coupled regulators. > > This breaks the build due to a few of the drivers (wm8350 and da9210 at > least) taking rdev locks in their interrupt handlers. I'd suggest > adding a function for those drivers to use - exposing the ww context > seems like the wrong thing to do there. I totally missed those drivers, will fix them in v2. > Otherwise the series looks good, I've applied everything up to here. > Thanks for taking the time to pick this up, and especially for figuring > out the ww_mutex conversion. > Thank you.
Hello, Maciej moved on to work on something else and kindly allowed me to take over the coupled regulators patches [0] which I'll try to finalize. I recently started to work on adding DVFS support to the CPUFreq driver of NVIDIA Tegra20/30 which require the coupled regulators functionality. Both Tegra20 and Tegra30 have the voltage "max-spread" constraint. Tegra30 has an additional constraint, the "max-step" voltage. What is done in this series: 1. Re-worked the original "Add voltage balancing mechanism" patch from Maciej by: 1) Fixing infinite loop within regulator_balance_voltage(). 2) Handling suspend_state_t properly. 3) Fixing broken compilation of the patch. 2. Added pair of patches to clean up regulator-couple resolving. 3. Changed device tree binding of the regulator-coupled-max-spread property to make it more flexible. 4. Added new device tree property, the regulator-max-step-microvolt. 5. Implemented wait/wound mutex which was suggested by Lucas Stach in the review comment to [0]. I've tested w/w locking extensively (including CONFIG_DEBUG_WW_MUTEX_SLOWPATH) and quite confident that everything is okay. 6. Patches were tested with KASAN on Tegra20 / Tegra30, I tried to cover various cases and verified that voltage constraints not violated under different circumstances. Please review, thanks. [0] https://lkml.org/lkml/2018/4/23/619 Dmitry Osipenko (9): regulator: core: Mutually resolve regulators coupling regulator: core: Don't allow to get regulator until all couples resolved dt-bindings: regulator: Change regulator-coupled-max-spread property regulator: core: Limit regulators coupling to a single couple dt-bindings: regulator: Document new regulator-max-step-microvolt property regulator: core: Add new max_uV_step constraint regulator: core: Decouple regulators on regulator_unregister() regulator: core: Use ww_mutex for regulators locking regulator: core: Properly handle case where supply is the couple Maciej Purski (2): regulator: core: Add voltage balancing mechanism regulator: core: Change voltage setting path .../bindings/regulator/regulator.txt | 7 +- drivers/regulator/core.c | 821 +++++++++++++++--- drivers/regulator/of_regulator.c | 4 + include/linux/regulator/driver.h | 5 +- include/linux/regulator/machine.h | 3 + 5 files changed, 713 insertions(+), 127 deletions(-)