Re: [ntp:questions] ATOM driver not working on Linux
Harlan Stenn wrote: Martin Burnicki writes: Harlan Stenn wrote: David Taylor writes: I have a newly installed 64-bit Linux Debian 7.7 system where I am trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get the expected assert messages. I have gpsd configured and working. Having configured and compiled NTP I get the error message: clock type 22 invalid That's the parse refclocks. Nope. The parse clocks are type 8. I caught that mistake already and posted about it. Yep. Sorry, I should have read the whole thread first. Did you configure with --enable-parse-clocks ? That should not be required to have support for the ATOM driver be compiled in. Agreed. On the other hand, I've just been in contact with someone else claiming that support for driver 28 was not been compiled in unless --enable-parse-clocks was passed to ./configure. I couldn't duplicate this, yet, but we should have a closer look at this before RC becomes the next release. I'm not sure how to do that yet, so if anybody has an idea about it please speak up or do some testing now. In the mean time I've run a few quick tests on different current Linux systems and everything worked as expected, i.e. if timepps.h is detected then PPS support is compiled in, but otherwise not. Martin ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
Martin Burnicki writes: > Harlan Stenn wrote: > > David Taylor writes: > >> I have a newly installed 64-bit Linux Debian 7.7 system where I am > >> trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get > >> the expected assert messages. I have gpsd configured and working. > >> Having configured and compiled NTP I get the error message: > >> > >> clock type 22 invalid > > > > That's the parse refclocks. > > Nope. The parse clocks are type 8. I caught that mistake already and posted about it. > > Did you configure with --enable-parse-clocks ? > > That should not be required to have support for the ATOM driver be > compiled in. Agreed. > On the other hand, I've just been in contact with someone else claiming > that support for driver 28 was not been compiled in unless > --enable-parse-clocks was passed to ./configure. > > I couldn't duplicate this, yet, but we should have a closer look at this > before RC becomes the next release. I'm not sure how to do that yet, so if anybody has an idea about it please speak up or do some testing now. I'm intending to release 4.2.8 on Thursday. Having said that, I'm also expecting that a few more currently-unresolved bugs will get fixed within the next month or so, which will result in 4.2.8p1. -- Harlan Stenn http://networktimefoundation.org - be a member! ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 16/12/2014 14:16, Martin Burnicki wrote: Harlan Stenn wrote: David Taylor writes: I have a newly installed 64-bit Linux Debian 7.7 system where I am trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get the expected assert messages. I have gpsd configured and working. Having configured and compiled NTP I get the error message: clock type 22 invalid That's the parse refclocks. Nope. The parse clocks are type 8. Did you configure with --enable-parse-clocks ? That should not be required to have support for the ATOM driver be compiled in. On the other hand, I've just been in contact with someone else claiming that support for driver 28 was not been compiled in unless --enable-parse-clocks was passed to ./configure. I couldn't duplicate this, yet, but we should have a closer look at this before RC becomes the next release. Martin The good news here is that with the current Raspberry Pi Linux (which as PPS built-in), and with pps-tools installed, it seems that I no longer need --enable-linuxcaps to get PPS support in NTP. -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
Harlan Stenn wrote: David Taylor writes: I have a newly installed 64-bit Linux Debian 7.7 system where I am trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get the expected assert messages. I have gpsd configured and working. Having configured and compiled NTP I get the error message: clock type 22 invalid That's the parse refclocks. Nope. The parse clocks are type 8. Did you configure with --enable-parse-clocks ? That should not be required to have support for the ATOM driver be compiled in. On the other hand, I've just been in contact with someone else claiming that support for driver 28 was not been compiled in unless --enable-parse-clocks was passed to ./configure. I couldn't duplicate this, yet, but we should have a closer look at this before RC becomes the next release. Martin ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 15/12/2014 14:53, Paul wrote: On Mon, Dec 15, 2014 at 3:13 AM, David Taylor < david-tay...@blueyonder.co.uk.invalid> wrote: ./configure --enable-linuxcaps Are you saying that config.h doesn't have #define HAVE_LINUX_CAPABILITIES if you don't explicitly --enable-linuxcaps? I think not, please see my later post. -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 15/12/2014 09:58, Harlan Stenn wrote: [] You don't need --enable-linuxcaps either - configure will do that for you automatically if it sees the right files. [] H Well, I recall being advised to add it in the early days of compiling on the Raspberry Pi, perhaps to get the "user" option. It's been there ever since. I've just tried on the Debian 7.7 system and, as you say, a version built without that flag compiles correctly and includes PPS support. I've updated my Web page accordingly. -- Thanks, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On Mon, Dec 15, 2014 at 3:13 AM, David Taylor < david-tay...@blueyonder.co.uk.invalid> wrote: > ./configure --enable-linuxcaps Are you saying that config.h doesn't have #define HAVE_LINUX_CAPABILITIES if you don't explicitly --enable-linuxcaps? ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
David Taylor writes: > On 14/12/2014 23:59, Harlan Stenn wrote: > [] > > You mean "with sys/timepps.h present", right? > > > > If clock type 22 is invalid then configure is not being passed the > > --enable-parse-clocks option. The above information from me was wrong - type 22 is the ATOM driver, and it should be automatically detected. > > You might also need to remove any top-level config.cache* files. > > > > H > > Harlan, > > Thanks for your suggestions. I could not find any config.cache* files, > but perhaps I wasn't looking in the right place? > >find / -name "config.cache*" > > So I deleted everything to do with compiling NTP, and re-ran the tar > command to build the source tree. I removed the /usr/include/timepps.h > file, leaving me with just /usr/include/sys/timepps.h, and the configure: > >./configure --enable-linuxcaps You don't need --enable-linuxcaps either - configure will do that for you automatically if it sees the right files. > This time, there were no error messages relating to timepps.h, and the > compiled ntp runs with PPS support. > > What /may/ have happened is that I did the first compile /before/ > installing pps-tools. > > Thanks to you, Brian, Paul and the others who helped. Glad to hear it's working now. H ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 14/12/2014 23:59, Harlan Stenn wrote: [] You mean "with sys/timepps.h present", right? If clock type 22 is invalid then configure is not being passed the --enable-parse-clocks option. You might also need to remove any top-level config.cache* files. H Harlan, Thanks for your suggestions. I could not find any config.cache* files, but perhaps I wasn't looking in the right place? find / -name "config.cache*" So I deleted everything to do with compiling NTP, and re-ran the tar command to build the source tree. I removed the /usr/include/timepps.h file, leaving me with just /usr/include/sys/timepps.h, and the configure: ./configure --enable-linuxcaps This time, there were no error messages relating to timepps.h, and the compiled ntp runs with PPS support. What /may/ have happened is that I did the first compile /before/ installing pps-tools. Thanks to you, Brian, Paul and the others who helped. -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
Paul writes: > On Sun, Dec 14, 2014 at 9:12 AM, David Taylor < > david-tay...@blueyonder.co.uk.invalid> wrote: > > > With sys/timepps.h present I got an ntp which didn't see PPS and produced > > a "clock type 22 invalid" error message, with timepps.h present the > > executable runs PPS correctly. > > Although these sorts of responses tend toward useless I built p485 with > sys/timepps.h on my 64-bit Ubuntu 12 S1 which is ATOM only. My bad - refclock 22 is the ATOM driver. If you look around line 1658 of configure.ac in ntp-dev you will see that in order to build the ATOM/PPSAPI we need: - 'inline' support - struct timespec - timepps.h or sys/timepps.h or be on SunOS, Solaris, SCO, or NetBSD and if the above 3 cases are true we #define HAVE_PPSAPI . H ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 2014-12-14 14:01, William Unruh wrote: On 2014-12-14, Brian Inglis wrote: On 2014-12-14 05:43, David Taylor wrote: I cheated: cp /usr/include/sys/timepps.h /user/include /usr/ I hope? and it compiles and runs with PPS support. Should I submit a bug report? Better cheat: rm -i /usr/include/timepps.h ln -rs /usr/include/sys/timepps.h /usr/include/ That way you don't have to remember to update /usr/include/ copy when you update .../sys/ copy. Has it changed in 10 years? In 2005 it became part of the Linux kernel; in 2007 it became part of pps-tools; in 2010 pps-tools adapted time_pps_kcbind from PHK BSD original as at 2005. In my experience, it will never happen, happens between 6 hours and 6 months after release! It's handy if you can just rebuild at that time with -DNEVER_HAPPEN to fix the problem. ;^> See other thread references to release engineering. -- Take care. Thanks, Brian Inglis ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On Sun, Dec 14, 2014 at 9:12 AM, David Taylor < david-tay...@blueyonder.co.uk.invalid> wrote: > With sys/timepps.h present I got an ntp which didn't see PPS and produced > a "clock type 22 invalid" error message, with timepps.h present the > executable runs PPS correctly. Although these sorts of responses tend toward useless I built p485 with sys/timepps.h on my 64-bit Ubuntu 12 S1 which is ATOM only. ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
David Taylor writes: > On 14/12/2014 13:23, Harlan Stenn wrote: > > David Taylor writes: > [] > >> I did try the ./configure with --enable-parse-clocks (as well as > >> --enable-linuxcaps) but I get the same error in config.log. > > > > You will always get the error that timepps.h doesn't exist because it > > doesn't exist. It doesn't need to exist if you have sys/timepps.h . > > > >> It confuses me that on Raspbian (a derivative of Debian) the same > >> source builds correctly, using the same commands, and both appear to > >> have timepps.h in the same place. > > > > Are you cross-compiling? If so, the headers you build with for > > cross-compiling are probably not in the same location, and you'll need > > to install the correct (appropriate) header in the location where the > > cross-compiler will see it. > > > > H > > No, this is a native compile, not cross. > > With sys/timepps.h present I got an ntp which didn't see PPS and > produced a "clock type 22 invalid" error message, with timepps.h > present the executable runs PPS correctly. You mean "with sys/timepps.h present", right? If clock type 22 is invalid then configure is not being passed the --enable-parse-clocks option. You might also need to remove any top-level config.cache* files. H ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 2014-12-14, Brian Inglis wrote: > On 2014-12-14 05:43, David Taylor wrote: >> I cheated: >> >>cp /usr/include/sys/timepps.h /user/include > /usr/ I hope? >> and it compiles and runs with PPS support. Should I submit a bug report? > > Better cheat: > > rm -i /usr/include/timepps.h > ln -rs /usr/include/sys/timepps.h /usr/include/ > > That way you don't have to remember to update /usr/include/ copy > when you update .../sys/ copy. Has it changed in 10 years? > ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 2014-12-14 05:43, David Taylor wrote: I cheated: cp /usr/include/sys/timepps.h /user/include /usr/ I hope? and it compiles and runs with PPS support. Should I submit a bug report? Better cheat: rm -i /usr/include/timepps.h ln -rs /usr/include/sys/timepps.h /usr/include/ That way you don't have to remember to update /usr/include/ copy when you update .../sys/ copy. -- Take care. Thanks, Brian Inglis ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 14/12/2014 13:23, Harlan Stenn wrote: David Taylor writes: [] I did try the ./configure with --enable-parse-clocks (as well as --enable-linuxcaps) but I get the same error in config.log. You will always get the error that timepps.h doesn't exist because it doesn't exist. It doesn't need to exist if you have sys/timepps.h . It confuses me that on Raspbian (a derivative of Debian) the same source builds correctly, using the same commands, and both appear to have timepps.h in the same place. Are you cross-compiling? If so, the headers you build with for cross-compiling are probably not in the same location, and you'll need to install the correct (appropriate) header in the location where the cross-compiler will see it. H No, this is a native compile, not cross. With sys/timepps.h present I got an ntp which didn't see PPS and produced a "clock type 22 invalid" error message, with timepps.h present the executable runs PPS correctly. -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 14/12/2014 13:25, Harlan Stenn wrote: David Taylor writes: I cheated: cp /usr/include/sys/timepps.h /user/include and it compiles and runs with PPS support. Should I submit a bug report? Not yet - it's still looking to me like a local config issue (ie, a problem with your local build configuration). H OK, Harlan. Now I have a working system I can run configure as you might with to debug this. -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
David Taylor writes: > I cheated: > >cp /usr/include/sys/timepps.h /user/include > > and it compiles and runs with PPS support. Should I submit a bug report? Not yet - it's still looking to me like a local config issue (ie, a problem with your local build configuration). H ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
David Taylor writes: > On 14/12/2014 11:56, Harlan Stenn wrote: >> David Taylor writes: >>> I have a newly installed 64-bit Linux Debian 7.7 system where I am >>> trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get >>> the expected assert messages. I have gpsd configured and working. >>> Having configured and compiled NTP I get the error message: >>> >>> clock type 22 invalid >> >> That's the parse refclocks. >> >> Did you configure with --enable-parse-clocks ? >> >>> On checking the config.log I see that timepps.h is missing: >>> >>> root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# cat config.log | grep timepps >>> configure:27520: checking for sys/timepps.h >>> configure:28812: checking timepps.h usability >>> conftest.c:230:21: fatal error: timepps.h: No such file or directory >>> | #include >>> configure:28812: checking timepps.h presence >>> conftest.c:197:21: fatal error: timepps.h: No such file or directory >>> | #include >>> configure:28812: checking for timepps.h >>> ac_cv_header_sys_timepps_h=yes >>> ac_cv_header_timepps_h=no >>> >>> but on searching the disk timepps.h does appear to be present, and in >>> the correct directory: >>> >>> root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# find / -name 'timepps.h' >>> /usr/include/sys/timepps.h >>> /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/ppsapi/loopback/src/timepps.h >>> /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/include/timepps.h >>> >>> What might be happening, please? What's wrong? >> >> Some systems provide and some . >> >> We look for each. >> >> You have the latter. My FreeBSD systems also only have sys/timepps.h . >> >> It should be OK that you only have sys/timepps.h . >> >> I suspect your problem is that you built without --enable-parse-clocks . >> >> H > Thanks for that, Harlan. I'm at the limit of my knowledge here, so any > suggestions are appreciated. > > I did try the ./configure with --enable-parse-clocks (as well as > --enable-linuxcaps) but I get the same error in config.log. You will always get the error that timepps.h doesn't exist because it doesn't exist. It doesn't need to exist if you have sys/timepps.h . > It confuses me that on Raspbian (a derivative of Debian) the same > source builds correctly, using the same commands, and both appear to > have timepps.h in the same place. Are you cross-compiling? If so, the headers you build with for cross-compiling are probably not in the same location, and you'll need to install the correct (appropriate) header in the location where the cross-compiler will see it. H ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
I cheated: cp /usr/include/sys/timepps.h /user/include and it compiles and runs with PPS support. Should I submit a bug report? Thanks for your help! -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 14/12/2014 11:56, Harlan Stenn wrote: David Taylor writes: I have a newly installed 64-bit Linux Debian 7.7 system where I am trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get the expected assert messages. I have gpsd configured and working. Having configured and compiled NTP I get the error message: clock type 22 invalid That's the parse refclocks. Did you configure with --enable-parse-clocks ? On checking the config.log I see that timepps.h is missing: root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# cat config.log | grep timepps configure:27520: checking for sys/timepps.h configure:28812: checking timepps.h usability conftest.c:230:21: fatal error: timepps.h: No such file or directory | #include configure:28812: checking timepps.h presence conftest.c:197:21: fatal error: timepps.h: No such file or directory | #include configure:28812: checking for timepps.h ac_cv_header_sys_timepps_h=yes ac_cv_header_timepps_h=no but on searching the disk timepps.h does appear to be present, and in the correct directory: root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# find / -name 'timepps.h' /usr/include/sys/timepps.h /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/ppsapi/loopback/src/timepps.h /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/include/timepps.h What might be happening, please? What's wrong? Some systems provide and some . We look for each. You have the latter. My FreeBSD systems also only have sys/timepps.h . It should be OK that you only have sys/timepps.h . I suspect your problem is that you built without --enable-parse-clocks . H Thanks for that, Harlan. I'm at the limit of my knowledge here, so any suggestions are appreciated. I did try the ./configure with --enable-parse-clocks (as well as --enable-linuxcaps) but I get the same error in config.log. It confuses me that on Raspbian (a derivative of Debian) the same source builds correctly, using the same commands, and both appear to have timepps.h in the same place. -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
David Taylor writes: > I have a newly installed 64-bit Linux Debian 7.7 system where I am > trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get > the expected assert messages. I have gpsd configured and working. > Having configured and compiled NTP I get the error message: > >clock type 22 invalid That's the parse refclocks. Did you configure with --enable-parse-clocks ? > On checking the config.log I see that timepps.h is missing: > > root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# cat config.log | grep timepps > configure:27520: checking for sys/timepps.h > configure:28812: checking timepps.h usability > conftest.c:230:21: fatal error: timepps.h: No such file or directory > | #include > configure:28812: checking timepps.h presence > conftest.c:197:21: fatal error: timepps.h: No such file or directory > | #include > configure:28812: checking for timepps.h > ac_cv_header_sys_timepps_h=yes > ac_cv_header_timepps_h=no > > but on searching the disk timepps.h does appear to be present, and in > the correct directory: > > root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# find / -name 'timepps.h' > /usr/include/sys/timepps.h > /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/ppsapi/loopback/src/timepps.h > /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/include/timepps.h > > What might be happening, please? What's wrong? Some systems provide and some . We look for each. You have the latter. My FreeBSD systems also only have sys/timepps.h . It should be OK that you only have sys/timepps.h . I suspect your problem is that you built without --enable-parse-clocks . H ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 14/12/2014 09:40, David Woolley wrote: On 14/12/14 09:33, David Taylor wrote: On checking the config.log I see that timepps.h is missing: Please read before posting. This has been extensively discussed over the last week. Thanks, David. Yes, I have seen the discussion, but as far as I can see the file is actually present, and not missing. This is comparing two variants of Debian Linux, one for Intel 64-bit, and the other Raspberry Pi 32-bit. One it compiles (RPi) and the other it doesn't (64-bit Debian). I am unclear how to resolve this issue. -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
Re: [ntp:questions] ATOM driver not working on Linux
On 14/12/14 09:33, David Taylor wrote: On checking the config.log I see that timepps.h is missing: Please read before posting. This has been extensively discussed over the last week. ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions
[ntp:questions] ATOM driver not working on Linux
I have a newly installed 64-bit Linux Debian 7.7 system where I am trying to bring up NTP with PPS support. Using ppstest /dev/pps0 I get the expected assert messages. I have gpsd configured and working. Having configured and compiled NTP I get the error message: clock type 22 invalid On checking the config.log I see that timepps.h is missing: root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# cat config.log | grep timepps configure:27520: checking for sys/timepps.h configure:28812: checking timepps.h usability conftest.c:230:21: fatal error: timepps.h: No such file or directory | #include configure:28812: checking timepps.h presence conftest.c:197:21: fatal error: timepps.h: No such file or directory | #include configure:28812: checking for timepps.h ac_cv_header_sys_timepps_h=yes ac_cv_header_timepps_h=no but on searching the disk timepps.h does appear to be present, and in the correct directory: root@Pixie:~/ntp/ntp-dev-4.2.7p485-RC# find / -name 'timepps.h' /usr/include/sys/timepps.h /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/ppsapi/loopback/src/timepps.h /root/ntp/ntp-dev-4.2.7p485-RC/ports/winnt/include/timepps.h What might be happening, please? What's wrong? Thanks, David -- Cheers, David Web: http://www.satsignal.eu ___ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions