| Message ID | 20240408231727.396452-1-andriy.shevchenko@linux.intel.com |
|---|---|
| Headers | show |
| Series | gpiolib: Fix gpio_lookup_flags mess and add Return sections | expand |
On Tue, Apr 9, 2024 at 1:17 AM Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote: > > The GPIO_* flag definitions are *almost* duplicated in two files > (with unmatches OPEN_SOURCE / OPEN_DRAIN). Moreover, some code relies > on one set of definitions while the rest is on the other. Clean up > this mess by providing only one source of the definitions to all. > > Fixes: b424808115cb ("brcm80211: brcmsmac: Move LEDs to GPIO descriptors") > Fixes: 5923ea6c2ce6 ("gpio: pass lookup and descriptor flags to request_own") > Fixes: fed7026adc7c ("gpiolib: Make use of enum gpio_lookup_flags consistent") > Fixes: 4c0facddb7d8 ("gpio: core: Decouple open drain/source flag with active low/high") > Fixes: 69d301fdd196 ("gpio: add DT bindings for existing consumer flags") > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > --- > drivers/gpio/gpiolib-of.c | 5 ++--- > drivers/gpio/gpiolib.c | 8 +++----- > .../broadcom/brcm80211/brcmsmac/led.c | 2 +- > include/linux/gpio/driver.h | 3 +-- > include/linux/gpio/machine.h | 20 +++++-------------- > 5 files changed, 12 insertions(+), 26 deletions(-) > I don't think ./dt-bindings/gpio/gpio.h is the right source of these defines for everyone - including non-OF systems. I would prefer the ones in include/linux/gpio/machine.h be the upstream source but then headers in include/dt-bindings/ cannot include them so my second-best suggestion is to rename the ones in include/linux/gpio/machine.h and treewide too. In general values from ./dt-bindings/gpio/gpio.h should only be used in DTS sources and gpiolib-of code. Bart
On Tue, Apr 09, 2024 at 11:42:37AM +0200, Bartosz Golaszewski wrote: > On Tue, Apr 9, 2024 at 1:17 AM Andy Shevchenko > <andriy.shevchenko@linux.intel.com> wrote: > > > > The GPIO_* flag definitions are *almost* duplicated in two files > > (with unmatches OPEN_SOURCE / OPEN_DRAIN). Moreover, some code relies > > on one set of definitions while the rest is on the other. Clean up > > this mess by providing only one source of the definitions to all. > > > > Fixes: b424808115cb ("brcm80211: brcmsmac: Move LEDs to GPIO descriptors") > > Fixes: 5923ea6c2ce6 ("gpio: pass lookup and descriptor flags to request_own") > > Fixes: fed7026adc7c ("gpiolib: Make use of enum gpio_lookup_flags consistent") > > Fixes: 4c0facddb7d8 ("gpio: core: Decouple open drain/source flag with active low/high") > > Fixes: 69d301fdd196 ("gpio: add DT bindings for existing consumer flags") > > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > > --- > > drivers/gpio/gpiolib-of.c | 5 ++--- > > drivers/gpio/gpiolib.c | 8 +++----- > > .../broadcom/brcm80211/brcmsmac/led.c | 2 +- > > include/linux/gpio/driver.h | 3 +-- > > include/linux/gpio/machine.h | 20 +++++-------------- > > 5 files changed, 12 insertions(+), 26 deletions(-) > > I don't think ./dt-bindings/gpio/gpio.h is the right source of these > defines for everyone - including non-OF systems. I would prefer the > ones in include/linux/gpio/machine.h be the upstream source but then > headers in include/dt-bindings/ cannot include them so my second-best > suggestion is to rename the ones in include/linux/gpio/machine.h and > treewide too. In general values from ./dt-bindings/gpio/gpio.h should > only be used in DTS sources and gpiolib-of code. Then, please fix that your way. It's quite annoying issue.
On Tue, Apr 9, 2024 at 2:51 PM Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote: > > On Tue, Apr 09, 2024 at 11:42:37AM +0200, Bartosz Golaszewski wrote: > > On Tue, Apr 9, 2024 at 1:17 AM Andy Shevchenko > > <andriy.shevchenko@linux.intel.com> wrote: > > > > > > The GPIO_* flag definitions are *almost* duplicated in two files > > > (with unmatches OPEN_SOURCE / OPEN_DRAIN). Moreover, some code relies > > > on one set of definitions while the rest is on the other. Clean up > > > this mess by providing only one source of the definitions to all. > > > > > > Fixes: b424808115cb ("brcm80211: brcmsmac: Move LEDs to GPIO descriptors") > > > Fixes: 5923ea6c2ce6 ("gpio: pass lookup and descriptor flags to request_own") > > > Fixes: fed7026adc7c ("gpiolib: Make use of enum gpio_lookup_flags consistent") > > > Fixes: 4c0facddb7d8 ("gpio: core: Decouple open drain/source flag with active low/high") > > > Fixes: 69d301fdd196 ("gpio: add DT bindings for existing consumer flags") > > > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > > > --- > > > drivers/gpio/gpiolib-of.c | 5 ++--- > > > drivers/gpio/gpiolib.c | 8 +++----- > > > .../broadcom/brcm80211/brcmsmac/led.c | 2 +- > > > include/linux/gpio/driver.h | 3 +-- > > > include/linux/gpio/machine.h | 20 +++++-------------- > > > 5 files changed, 12 insertions(+), 26 deletions(-) > > > > I don't think ./dt-bindings/gpio/gpio.h is the right source of these > > defines for everyone - including non-OF systems. I would prefer the > > ones in include/linux/gpio/machine.h be the upstream source but then > > headers in include/dt-bindings/ cannot include them so my second-best > > suggestion is to rename the ones in include/linux/gpio/machine.h and > > treewide too. In general values from ./dt-bindings/gpio/gpio.h should > > only be used in DTS sources and gpiolib-of code. > > Then, please fix that your way. It's quite annoying issue. > This is not difficult in itself but it's a tree-wide change so we will probably have to send it to Torvalds at the end of the merge window in a separate pull-request. I don't really have time now, I'll be travelling for 5 weeks in a row. I'll see closer to the merge window. Or next release cycle. Bart
On Tue, Apr 09, 2024 at 02:55:20PM +0200, Bartosz Golaszewski wrote: > On Tue, Apr 9, 2024 at 2:51 PM Andy Shevchenko > <andriy.shevchenko@linux.intel.com> wrote: > > On Tue, Apr 09, 2024 at 11:42:37AM +0200, Bartosz Golaszewski wrote: > > > On Tue, Apr 9, 2024 at 1:17 AM Andy Shevchenko > > > <andriy.shevchenko@linux.intel.com> wrote: > > > > > > > > The GPIO_* flag definitions are *almost* duplicated in two files > > > > (with unmatches OPEN_SOURCE / OPEN_DRAIN). Moreover, some code relies > > > > on one set of definitions while the rest is on the other. Clean up > > > > this mess by providing only one source of the definitions to all. > > > > > > > > Fixes: b424808115cb ("brcm80211: brcmsmac: Move LEDs to GPIO descriptors") > > > > Fixes: 5923ea6c2ce6 ("gpio: pass lookup and descriptor flags to request_own") > > > > Fixes: fed7026adc7c ("gpiolib: Make use of enum gpio_lookup_flags consistent") > > > > Fixes: 4c0facddb7d8 ("gpio: core: Decouple open drain/source flag with active low/high") > > > > Fixes: 69d301fdd196 ("gpio: add DT bindings for existing consumer flags") > > > > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> > > > > --- > > > > drivers/gpio/gpiolib-of.c | 5 ++--- > > > > drivers/gpio/gpiolib.c | 8 +++----- > > > > .../broadcom/brcm80211/brcmsmac/led.c | 2 +- > > > > include/linux/gpio/driver.h | 3 +-- > > > > include/linux/gpio/machine.h | 20 +++++-------------- > > > > 5 files changed, 12 insertions(+), 26 deletions(-) > > > > > > I don't think ./dt-bindings/gpio/gpio.h is the right source of these > > > defines for everyone - including non-OF systems. I would prefer the > > > ones in include/linux/gpio/machine.h be the upstream source but then > > > headers in include/dt-bindings/ cannot include them so my second-best > > > suggestion is to rename the ones in include/linux/gpio/machine.h and > > > treewide too. In general values from ./dt-bindings/gpio/gpio.h should > > > only be used in DTS sources and gpiolib-of code. > > > > Then, please fix that your way. It's quite annoying issue. > > This is not difficult in itself I'm not sure, what about enum gpio_lookup_flags? Shall we resurrect it? I see that you have better vision anyway. Consider my patch as a problem report (and as bonus you have already list of Fixes tags :-). > but it's a tree-wide change so we will > probably have to send it to Torvalds at the end of the merge window in > a separate pull-request. WFM! > I don't really have time now, I'll be travelling for 5 weeks in a row. > I'll see closer to the merge window. Or next release cycle. But can you prioritize this? It's a carefully planted minefield with already a bug and confusion here.