| Message ID | 1503531988-15429-1-git-send-email-jasmin@anw.at (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi! Just some notes on that patch. I have *not* tested it due to the lack of an ir remote control. So someone needs to test this on an <= 4.9 Kernel, if the ir core is still working as expected. Even if I fixed that in media_build, it may be better to apply this code change in media_tree. This because the involved variables are all of type ktime_t and there are accessor and converter functions available for that type, which should have been used by the original author of 86fe1ac0d and 48b2de197 in my opinion. BR, Jasmin
On 08/24/2017 02:07 AM, Jasmin J. wrote: > Hi! > > Just some notes on that patch. > > I have *not* tested it due to the lack of an ir remote control. So someone > needs to test this on an <= 4.9 Kernel, if the ir core is still working as > expected. > > Even if I fixed that in media_build, it may be better to apply this code change > in media_tree. This because the involved variables are all of type ktime_t and > there are accessor and converter functions available for that type, which > should have been used by the original author of 86fe1ac0d and 48b2de197 in my > opinion. Sean, I agree with Jasmin here. I noticed the same errors in the daily build and it is really caused by not using the correct functions. I just didn't have the time to follow up on it. Can you take a look at Jasmin's patch and, if OK, make a pull request for it? Regards, Hans
Hello Sean! > I agree with Jasmin here. I noticed the same errors in the daily build and it > is really caused by not using the correct functions. I just didn't have the > time to follow up on it. I started to fix also gpio-ir-tx.c, but stopped that because it was too late. So I simply deactivated the driver: https://www.mail-archive.com/linux-media@vger.kernel.org/msg117607.html Maybe you can fix gpio-ir-tx.c also by using the right functions to access ktime_t, so that this driver would be available for older Kernels, too. But there was another problem beside the ktime_t accessors, which I didn't analyze (symbol missing). BR, Jasmin
On Thu, Aug 24, 2017 at 08:47:54AM +0200, Hans Verkuil wrote: > On 08/24/2017 02:07 AM, Jasmin J. wrote: > > Hi! > > > > Just some notes on that patch. > > > > I have *not* tested it due to the lack of an ir remote control. So someone > > needs to test this on an <= 4.9 Kernel, if the ir core is still working as > > expected. > > > > Even if I fixed that in media_build, it may be better to apply this code change > > in media_tree. This because the involved variables are all of type ktime_t and > > there are accessor and converter functions available for that type, which > > should have been used by the original author of 86fe1ac0d and 48b2de197 in my > > opinion. > > Sean, > > I agree with Jasmin here. I noticed the same errors in the daily build and it > is really caused by not using the correct functions. I just didn't have the > time to follow up on it. > > Can you take a look at Jasmin's patch and, if OK, make a pull request for > it? I hadn't taken media_build into account when I wrote this, so yes I'll have a look and create a pull request. Sean
diff --git a/backports/backports.txt b/backports/backports.txt index 87b9ee8..5691a3e 100644 --- a/backports/backports.txt +++ b/backports/backports.txt @@ -38,6 +38,7 @@ add v4.10_refcount.patch add v4.9_mm_address.patch add v4.9_dvb_net_max_mtu.patch add v4.9_ktime_cleanups.patch +add v4.9_rc_ir_raw_ktime.patch [4.8.255] add v4.8_user_pages_flag.patch diff --git a/backports/v4.9_rc_ir_raw_ktime.patch b/backports/v4.9_rc_ir_raw_ktime.patch new file mode 100644 index 0000000..407128d --- /dev/null +++ b/backports/v4.9_rc_ir_raw_ktime.patch @@ -0,0 +1,36 @@ +diff --git a/drivers/media/rc/rc-ir-raw.c b/drivers/media/rc/rc-ir-raw.c +index f495709..f1259d1 100644 +--- a/drivers/media/rc/rc-ir-raw.c ++++ b/drivers/media/rc/rc-ir-raw.c +@@ -106,7 +107,7 @@ int ir_raw_event_store_edge(struct rc_dev *dev, bool pulse) + return -EINVAL; + + now = ktime_get(); +- ev.duration = ktime_sub(now, dev->raw->last_event); ++ ev.duration = ktime_to_ns(ktime_sub(now, dev->raw->last_event)); + ev.pulse = !pulse; + + rc = ir_raw_event_store(dev, &ev); +@@ -474,18 +479,18 @@ EXPORT_SYMBOL(ir_raw_encode_scancode); + static void edge_handle(unsigned long arg) + { + struct rc_dev *dev = (struct rc_dev *)arg; +- ktime_t interval = ktime_get() - dev->raw->last_event; ++ ktime_t interval = ktime_sub(ktime_get(), dev->raw->last_event); + +- if (interval >= dev->timeout) { ++ if (ktime_to_ns(interval) >= dev->timeout) { + DEFINE_IR_RAW_EVENT(ev); + + ev.timeout = true; +- ev.duration = interval; ++ ev.duration = ktime_to_ns(interval); + + ir_raw_event_store(dev, &ev); + } else { + mod_timer(&dev->raw->edge_handle, +- jiffies + nsecs_to_jiffies(dev->timeout - interval)); ++ jiffies + nsecs_to_jiffies(dev->timeout - ktime_to_ns(interval))); + } + + ir_raw_event_handle(dev);