lirc git 0.10.0-devel.git+e3225584 stops working, uses 100% CPU

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

lirc git 0.10.0-devel.git+e3225584 stops working, uses 100% CPU

VDR User
Hi. I just updated my previously compiled lirc git
(0.9.5-devel.git+8c87a28), which worked fine, to the current lirc git
(0.10.0-devel.git+e3225584). I have not changed any config files.
Using irw, I can see the new lirc does work with the first button
press (both single press or press & hold/repeat) on my remote control.
However, when I attempt another button press, nothing happens. When I
look in top I see:

  PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
 9884 vdr       20   0    7864    376      0 R 100.0  0.0   0:59.17 lircd

So for some reason lircd is taking up 100% CPU. How I start lircd is:

sudo modprobe -a serial_ir 2>&1 &
sudo lircd --effective-user=vdr -O
/ldvb/lirc.config/lirc_options_serial.conf
/ldvb/lirc.config/lircd.20.1_IR.serial.conf

My lirc_options_serial.conf is:
[lircd]
nodaemon        = False
driver          = default
device          = /dev/lirc0
output          = /var/run/lirc/lircd
pidfile         = /var/run/lirc/lircd.pid
permission      = 666
allow-simulate  = No
repeat-max      = 600
logfile         = /logs/lircd.log

[lircmd]
uinput          = False
nodaemon        = False

Everything works fine if I reinstall my previous
lirc_0.9.5-devel-git+8c87a28_i386.deb package.. I don't know how to
debug this further. Any help is greatly appreciated!

Thanks,
Derek

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lirc git 0.10.0-devel.git+e3225584 stops working, uses 100% CPU

Bengt Martensson-2
On 06/25/17 05:21, VDR User wrote:

> Hi. I just updated my previously compiled lirc git
> (0.9.5-devel.git+8c87a28), which worked fine, to the current lirc git
> (0.10.0-devel.git+e3225584). I have not changed any config files.
> Using irw, I can see the new lirc does work with the first button
> press (both single press or press & hold/repeat) on my remote control.
> However, when I attempt another button press, nothing happens. When I
> look in top I see:
>
>    PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
>   9884 vdr       20   0    7864    376      0 R 100.0  0.0   0:59.17 lircd
>
> So for some reason lircd is taking up 100% CPU. How I start lircd is:

I have seen something like this a few years ago, but I was not able to
reproduce it. Can you deterministically reproduce? Please raise an issue
at https://sourceforge.net/p/lirc/tickets/ . I think some more details
are required, kernel, distro, the lircd.conf.


> I don't know how to
> debug this further.

The canonical suggestion is to raise the log level to max, and look in
the log file.

Alec may have some more suggestions.

Greetz,

Bengt

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lirc git 0.10.0-devel.git+e3225584 stops working, uses 100% CPU

Alec Leamas
In reply to this post by VDR User

On 25/06/17 11:09, Bengt Martensson wrote:
> On 06/25/17 05:21, VDR User wrote:

>> I don't know how to
>> debug this further.
>
> The canonical suggestion is to raise the log level to max, and look in
> the log file.
>
> Alec may have some more suggestions.

Well, since it works for you in 9.9.4, the obvious debug path is to make
a git bisect. However, it seems important to know if this affects also
the release branch before we go into this.

Cheers!

--alec

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lirc git 0.10.0-devel.git+e3225584 stops working, uses 100% CPU

VDR User
In reply to this post by Bengt Martensson-2
> I have seen something like this a few years ago, but I was not able to
> reproduce it. Can you deterministically reproduce? Please raise an issue at
> https://sourceforge.net/p/lirc/tickets/ . I think some more details are
> required, kernel, distro, the lircd.conf.

Following Alec's suggestion to try the release branch rather than
master branch, I've recompiled (0.10.0-rc3.git+5259c456) and haven't
had a problem so far. However, I can easily reproduce the 100% CPU
issue with the master branch by simply starting lircd and doing two
button presses. Should I still raise an issue although the release
branch works?

> The canonical suggestion is to raise the log level to max, and look in the
> log file.

In the event I should still raise an issue, should the logging level
be 10, trace2, or? I see when I was trying the girs driver I was using
"trace2" but I don't know if that's useful here, if this is even
pursued further.

Thanks,
Derek

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lirc git 0.10.0-devel.git+e3225584 stops working, uses 100% CPU

Alec Leamas
In reply to this post by VDR User


On 25/06/17 19:30, VDR User wrote:

> After adding `-b release` to the git clone, I've now compiled lirc
> 0.10.0-rc3.git+5259c456, which does seem to be working.

OK, that's a relief. So we can still ship the release branch.

> I don't know
> what differences there are between the release and master branches but
> with the master branch the 100% CPU is easily reproducible here for
> future reference. For the record, I'm using Debian Testing with stable
> kernel (4.11.7), using a homebrew serial ir (serial_ir driver).
>


OK, so we have an offending commit in the master branch. Have you ever
done a git bisect?

In any case: could you test the commit 9a25a86? I. e.

    $ git checkout 9a25a86
    $ ./autogen.sh
    $ ./configure...
    [build and test]


and report back?


Cheers!

--alec

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lirc git 0.10.0-devel.git+e3225584 stops working, uses 100% CPU

VDR User
>> I don't know
>> what differences there are between the release and master branches but
>> with the master branch the 100% CPU is easily reproducible here for
>> future reference. For the record, I'm using Debian Testing with stable
>> kernel (4.11.7), using a homebrew serial ir (serial_ir driver).
>
> OK, so we have an offending commit in the master branch. Have you ever done
> a git bisect?
>
> In any case: could you test the commit 9a25a86? I. e.

Just did a git bisect on the master branch. Results are:

beeb4adb1d114bcf5faf060b9543cc504e24a221 is the first bad commit
commit beeb4adb1d114bcf5faf060b9543cc504e24a221
Author: Alec Leamas <[hidden email]>
Date:   Fri Jun 16 10:13:31 2017 +0200

    lib, lircd: Remove --release support.

:040000 040000 43d58ad2598be91058bd46774b6dd6fb876f26b8
9a8211ed85578a22a8c9508d1cf1517300071d4c M      daemons
:040000 040000 6cbd3993d9ae4338fefacbe7c0cf6f1b8bba0177
e83e8419964b8e9dd10204115012fa5a0d3ce641 M      doc
:040000 040000 0537275592a0b8000440d64c4e713842ce77488c
f6e4e7dbd90c5ef5ed260d09e66ffefc60efb69b M      lib

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Loading...