Re: [PATCH 0/3] regulator: unexport regulator_lock/unlock()
On Tue, Sep 15, 2020 at 08:55:06PM +0200, Michał Mirosław wrote: > On Mon, Sep 07, 2020 at 07:05:49PM +0100, Mark Brown wrote: > > [3/3] regulator: unexport regulator_lock/unlock() > > (no commit info) > It looks like the third one didn't get in? (Can't see it in the > for-next branch). It probably didn't apply, please check and resend. signature.asc Description: PGP signature
Re: [PATCH 0/3] regulator: unexport regulator_lock/unlock()
On Mon, Sep 07, 2020 at 07:05:49PM +0100, Mark Brown wrote: > On Mon, 10 Aug 2020 06:33:30 +0200, Michał Mirosław wrote: > > This removes regulator_lock/unlock() calls around > > regulator_notifier_call_chain() as they are redundant - drivers > > already have to guarantee regulator_dev's existence during the call. > > > > This should make reasoing about the lock easier, as this was the only > > use outside regulator core code. > > > > [...] > > Applied to > >https://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator.git > for-next > > Thanks! > > [1/3] regulator: don't require mutex for regulator_notifier_call_chain() > commit: 3bca239d6184df61a619d78764e0481242d844b4 > [2/3] regulator: remove locking around regulator_notifier_call_chain() > commit: e9c142b0d2c08178a9e146d47d8fe397373bda3e > [3/3] regulator: unexport regulator_lock/unlock() > (no commit info) [...] It looks like the third one didn't get in? (Can't see it in the for-next branch). Best Regards Michał Mirosław
Re: [PATCH 0/3] regulator: unexport regulator_lock/unlock()
On Mon, Aug 10, 2020 at 06:33:30AM +0200, Michał Mirosław wrote: > This removes regulator_lock/unlock() calls around > regulator_notifier_call_chain() as they are redundant - drivers > already have to guarantee regulator_dev's existence during the call. > > This should make reasoing about the lock easier, as this was the only > use outside regulator core code. > > The only client that needed recursive locking from the notifier chain > was drivers/usb/host/ohci-da8xx.c, which responds to over-current > notification by calling regulator_disable(). I can't see the series in regulator/for-next and got no comments. Should I resend? As a side note: this is a step towards fixing regulator-coupling-related locking issues. Best Regards, Michał Mirosław
[PATCH 0/3] regulator: unexport regulator_lock/unlock()
This removes regulator_lock/unlock() calls around regulator_notifier_call_chain() as they are redundant - drivers already have to guarantee regulator_dev's existence during the call. This should make reasoing about the lock easier, as this was the only use outside regulator core code. The only client that needed recursive locking from the notifier chain was drivers/usb/host/ohci-da8xx.c, which responds to over-current notification by calling regulator_disable(). Michał Mirosław (3): regulator: don't require mutex for regulator_notifier_call_chain() regulator: remove locking around regulator_notifier_call_chain() regulator: unexport regulator_lock/unlock() drivers/regulator/core.c | 11 +++ drivers/regulator/da9055-regulator.c | 2 -- drivers/regulator/da9062-regulator.c | 2 -- drivers/regulator/da9063-regulator.c | 2 -- drivers/regulator/da9210-regulator.c | 4 drivers/regulator/da9211-regulator.c | 4 drivers/regulator/lp8755.c | 6 -- drivers/regulator/ltc3589.c| 10 ++ drivers/regulator/ltc3676.c| 10 ++ drivers/regulator/pv88060-regulator.c | 10 ++ drivers/regulator/pv88080-regulator.c | 10 ++ drivers/regulator/pv88090-regulator.c | 10 ++ drivers/regulator/slg51000-regulator.c | 4 drivers/regulator/stpmic1_regulator.c | 4 drivers/regulator/wm831x-dcdc.c| 4 drivers/regulator/wm831x-isink.c | 2 -- drivers/regulator/wm831x-ldo.c | 2 -- drivers/regulator/wm8350-regulator.c | 2 -- include/linux/regulator/driver.h | 3 --- 19 files changed, 13 insertions(+), 89 deletions(-) -- 2.20.1