| Message ID | 1462876002-6462-1-git-send-email-kwizart@gmail.com |
|---|---|
| State | Accepted |
| Headers | show |
On 05/10/2016 04:26 AM, Nicolas Chauvet wrote: > This module is loaded by the related mfd driver which has > the needed MODULE_DEVICE_TABLE(i2c,...). > > This patch fix the modalias when the rtc driver is built > as a module, so the right name is used. > Everything operates correctly when this module is builtin. I'm sure this used to work. Do you know when/why it broke? Perhaps a Fixes: tag would be useful, and perhaps a Cc: stable?
2016-05-10 17:43 GMT+02:00 Stephen Warren <swarren@wwwdotorg.org>: > On 05/10/2016 04:26 AM, Nicolas Chauvet wrote: > >> This module is loaded by the related mfd driver which has >> the needed MODULE_DEVICE_TABLE(i2c,...). >> >> This patch fix the modalias when the rtc driver is built >> as a module, so the right name is used. >> Everything operates correctly when this module is builtin. >> > > I'm sure this used to work. Do you know when/why it broke? Perhaps a > Fixes: tag would be useful, and perhaps a Cc: stable? > I don't remember autoloading of rtc-tps6586x.ko has ever worked. Then there is still the issue of rtc modules loaded too late for the kernel (or the kernel not capable to defer the read of hw clock), but that's another issue. I first tried to report the issue on this bug tracker and at that time, autoloading wasn't working (was in 2014). https://bugzilla.redhat.com/show_bug.cgi?id=1074002 I'm sure current distros using an arm generic kernel (with all modules) will rebase on kernel 4.4, but technically the bug is present in kernel 3.10 from the oldest current maintained branch (since the driver introduction).
2016-05-10 17:43 GMT+02:00 Stephen Warren <swarren@wwwdotorg.org>: > On 05/10/2016 04:26 AM, Nicolas Chauvet wrote: >> >> This module is loaded by the related mfd driver which has >> the needed MODULE_DEVICE_TABLE(i2c,...). >> >> This patch fix the modalias when the rtc driver is built >> as a module, so the right name is used. >> Everything operates correctly when this module is builtin. > > > I'm sure this used to work. Do you know when/why it broke? Perhaps a Fixes: > tag would be useful, and perhaps a Cc: stable? I don't remember autoloading of rtc-tps6586x.ko has ever worked. Then there is still the issue of rtc modules loaded too late for the kernel (or the kernel not capable to defer the read of hw clock), but that's another issue. I first tried to report the issue on this bug tracker and at that time, autoloading wasn't working (was in 2014). https://bugzilla.redhat.com/show_bug.cgi?id=1074002 I'm sure current distros using an arm generic kernel (with all modules) will rebase on kernel 4.4, but technically the bug is present in kernel 3.10 from the oldest current maintained branch (since the driver introduction).
On 05/10/2016 03:44 PM, Nicolas Chauvet wrote: > 2016-05-10 17:43 GMT+02:00 Stephen Warren <swarren@wwwdotorg.org>: >> On 05/10/2016 04:26 AM, Nicolas Chauvet wrote: >>> >>> This module is loaded by the related mfd driver which has >>> the needed MODULE_DEVICE_TABLE(i2c,...). >>> >>> This patch fix the modalias when the rtc driver is built >>> as a module, so the right name is used. >>> Everything operates correctly when this module is builtin. >> >> >> I'm sure this used to work. Do you know when/why it broke? Perhaps a Fixes: >> tag would be useful, and perhaps a Cc: stable? > > I don't remember autoloading of rtc-tps6586x.ko has ever worked. Then > there is still the issue of rtc modules loaded too late for the kernel > (or the kernel not capable to defer the read of hw clock), but that's > another issue. > > I first tried to report the issue on this bug tracker and at that > time, autoloading wasn't working (was in 2014). > https://bugzilla.redhat.com/show_bug.cgi?id=1074002 > > I'm sure current distros using an arm generic kernel (with all > modules) will rebase on kernel 4.4, but technically the bug is present > in kernel 3.10 from the oldest current maintained branch (since the > driver introduction). Ah, all the cases I used previous did indeed have the RTC driver built-in not a module, so that explains it.
On 10/05/2016 at 12:26:42 +0200, Nicolas Chauvet wrote : > This module is loaded by the related mfd driver which has > the needed MODULE_DEVICE_TABLE(i2c,...). > > This patch fix the modalias when the rtc driver is built > as a module, so the right name is used. > Everything operates correctly when this module is builtin. > > Signed-off-by: Nicolas Chauvet <kwizart@gmail.com> > --- > drivers/rtc/rtc-tps6586x.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > Applied, thanks.
diff --git a/drivers/rtc/rtc-tps6586x.c b/drivers/rtc/rtc-tps6586x.c index e404faa..a3418a8 100644 --- a/drivers/rtc/rtc-tps6586x.c +++ b/drivers/rtc/rtc-tps6586x.c @@ -344,7 +344,7 @@ static struct platform_driver tps6586x_rtc_driver = { }; module_platform_driver(tps6586x_rtc_driver); -MODULE_ALIAS("platform:rtc-tps6586x"); +MODULE_ALIAS("platform:tps6586x-rtc"); MODULE_DESCRIPTION("TI TPS6586x RTC driver"); MODULE_AUTHOR("Laxman dewangan <ldewangan@nvidia.com>"); MODULE_LICENSE("GPL v2");
This module is loaded by the related mfd driver which has the needed MODULE_DEVICE_TABLE(i2c,...). This patch fix the modalias when the rtc driver is built as a module, so the right name is used. Everything operates correctly when this module is builtin. Signed-off-by: Nicolas Chauvet <kwizart@gmail.com> --- drivers/rtc/rtc-tps6586x.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)