Loading ...
Sorry, an error occurred while loading the content.
 

Re: Solution to time/clock problems with Inspiron laptops in linux

Expand Messages
  • ddhummel
    ... This is a linux list. ... You can pretty much count on this from Dell these days for anything having to do with Linux. Don t bother. ... If you have
    Message 1 of 22 , Dec 5, 2002
      --- In linux-dell-laptops@y..., "ya_hoo_com" <ya_hoo_com@y...> wrote:
      > All this was whilst Win2k kept perfect time for months on end.

      This is a linux list.

      > The last time I was receiving emails from this group, it seemed Dell
      > were offering no help, and no-one quite knew the problem.

      You can pretty much count on this from Dell these days for anything
      having to do with Linux. Don't bother.

      > I have since discovered (after lots of experimentation) that the
      > problem (for me at least) is due to bugs in the Linux kernel. 2.4.18
      > (and possibly many other versions) have big problems with
      > variable-speed laptop CPUs (allegedly).

      If you have identified bugs in the kernel regarding system clock
      accuracy, perhaps you should provide the names of the kernel source
      files containing the bugs to the kernel maintainers. Also, please
      provide the source of your information specifically regarding laptop
      CPUs. A URL is a wonderful thing.

      > The "hardware clock" is perfect,

      Assuming you've set it perfectly.

      > and win2k just gets its time
      > from this hardware clock all the time, instead of trying to
      > do it itself.

      Nope, Windows uses a software clock as well, but this is, erh, a linux
      list.

      > Even more unfortunately, (probably because of linux's background of
      > being developed to run on crappy old hardware, and broken PC-clones
      > with bug-ridden BIOS's and hardware that the manufacturers only
      > support under windows)

      Damn I guess we should all be using Windows then.

      > all linux distributions and most software
      > assume that the "hardware clock" is not to be trusted, and that the
      > linux "system clock" is always perfect.

      Uh, the system clock is set from the hardware clock at boot time, and
      the kernel doesn't know or care if it is inaccurate. I suspect most
      software does not poll the hardware clock because this would be
      impractical/inefficient when the current time is available from the
      running kernel.

      As you've since found out, hwclock's hctosys option is your friend.

      This is all redundant anyhow. Paul Waserbrot's post of 11/21/02 gives
      a perfectly adequate solution that will work on most systems.

      -DDH
    • jsc3@gtf.org
      ... Really? I guess that must be why they call it linux-dell-laptops . What s your point? He s talking about a time issue with Linux. I too have noticed my
      Message 2 of 22 , Dec 5, 2002
        ddhummel wrote:
        >
        > --- In linux-dell-laptops@y..., "ya_hoo_com" <ya_hoo_com@y...> wrote:
        > > All this was whilst Win2k kept perfect time for months on end.
        >
        > This is a linux list.

        Really? I guess that must be why they call it "linux-dell-laptops".

        What's your point? He's talking about a time issue with Linux. I
        too have noticed my Dell i8k does not keep time very well when running
        Linux unless I use ntp. I don't run Windows enough to have any valid
        data points, but a comparison with Windows performance on the same
        hardware certainly seems reasonable for this forum. Are you such
        a Linux zealot that any mention of a Linux flaw or any comparison with
        how some other OS (such as Windows) performs a similar task is not
        acceptable to you?

        > > The last time I was receiving emails from this group, it seemed Dell
        > > were offering no help, and no-one quite knew the problem.
        >
        > You can pretty much count on this from Dell these days for anything
        > having to do with Linux. Don't bother.

        The same is true enough for other PC makers as well. I am sure Sony
        will give you loads of Linux help.

        > > I have since discovered (after lots of experimentation) that the
        > > problem (for me at least) is due to bugs in the Linux kernel. 2.4.18
        > > (and possibly many other versions) have big problems with
        > > variable-speed laptop CPUs (allegedly).
        >
        > If you have identified bugs in the kernel regarding system clock
        > accuracy, perhaps you should provide the names of the kernel source
        > files containing the bugs to the kernel maintainers. Also, please
        > provide the source of your information specifically regarding laptop
        > CPUs. A URL is a wonderful thing.

        His experimentation and level of experience may not enable him to point
        to specific kernel source files. It is perfectly possible to find flaws
        without examining the source - people do it with Windows all the time.

        A URL or some other solid reference for the laptop CPU part of this
        issues would be quite useful, and it would be nice if the original
        poster could direct us to more details on this.

        > > The "hardware clock" is perfect,
        >
        > Assuming you've set it perfectly.

        And nothing's perfect - the hardware clock may keep time acceptably
        but I seriously doubt it is perfect. For the sake of this argument,
        all it has to do is work better than Linux.

        > > and win2k just gets its time
        > > from this hardware clock all the time, instead of trying to
        > > do it itself.
        >
        > Nope, Windows uses a software clock as well, but this is, erh, a linux
        > list.

        You keep saying that like some kind of mantra. It is a Linux list, but
        comparing Linux with Windows is not out of the scope of the list
        charter, if I remember correctly.

        > > Even more unfortunately, (probably because of linux's background of
        > > being developed to run on crappy old hardware, and broken PC-clones
        > > with bug-ridden BIOS's and hardware that the manufacturers only
        > > support under windows)
        >
        > Damn I guess we should all be using Windows then.

        No, maybe we should try to figure out how to make Linux keep time at
        least as well as Windows does on the same hardware. I think the point
        that was being made may be relevant. I think your attitude is counter
        productive. The original poster was not slamming Linux as far as I
        could tell, he was just stating his case and doing a reasonable job of
        it (ie time is kept well when running Windows, time is not kept well
        running Linux on the same hardware, perhaps Linux could be improved
        somehow to keep time as well as Windows). There may be a few flaws in
        the argument (ie Windows uses the hardware clock all the time - I am
        pretty certain that is not true) but the flaws I see don't discredit the
        basic argument. And the original poster NEVER seemed to suggest that we
        should all be using Windows.

        > > all linux distributions and most software
        > > assume that the "hardware clock" is not to be trusted, and that the
        > > linux "system clock" is always perfect.
        >
        > Uh, the system clock is set from the hardware clock at boot time, and
        > the kernel doesn't know or care if it is inaccurate.

        And then some Linux distributions set the hardware clock from the system
        clock when shutting down, which means that any inaccuracy in the Linux
        time will now be passed on to the hardware.

        > I suspect most
        > software does not poll the hardware clock because this would be
        > impractical/inefficient when the current time is available from the
        > running kernel.

        This is reasonable - one should only poll the hardware time when
        absolutely necessary, such as at boot time and perhaps after a CPU
        speed transition.

        Now that I think about it, it does seem reasonable that the time gets
        messed up when the CPU switches speeds. Sometimes my clock gets way
        of synch when running Linux, and now that I think about it, it could be
        at times when I plug in or unplug my laptop from the AC power grid. It
        is certainly something that could be tested (without inspecting the
        kernel source, or even using strace to monitor system calls) quite
        easily.

        > As you've since found out, hwclock's hctosys option is your friend.

        Yes, the original poster pointed out a Linux problem and indicated a
        Linux work-around for it. This is not at all the same as saying "Linux
        is broken and 'Damn, I guess we should all be using Windows then.'"

        > This is all redundant anyhow. Paul Waserbrot's post of 11/21/02 gives
        > a perfectly adequate solution that will work on most systems.

        This is correct - it is also similar to what was proposed by the
        original post you were responding to. Admittedly, Paul's post
        was more complete and more detailed, but the basic idea was the same in
        the original post in this thread.

        --
        John Cronin
        mailto: `echo NjsOc3@... | sed 's/[NOSPAM]//g'`
      • James D Strandboge
        ... yep ... Partially right. Are you using a battery monitor? If so, this is the source of your time problems. Win2k uses acpi to monitor the battery, which
        Message 3 of 22 , Dec 5, 2002
          On Thu, 2002-12-05 at 08:52, ya_hoo_com wrote:
          > Like apparently many others in this group, I had massive problems with
          > linux + dell laptop being totally rubbish at keeping time. At one
          > point, my laptop was losing 2 minutes every hour. Two days was enough
          > to be almost an hour wrong!
          >
          yep

          > All this was whilst Win2k kept perfect time for months on end.

          > The last time I was receiving emails from this group, it seemed Dell
          > were offering no help, and no-one quite knew the problem.
          >
          > I have since discovered (after lots of experimentation) that the
          > problem (for me at least) is due to bugs in the Linux kernel. 2.4.18
          > (and possibly many other versions) have big problems with
          > variable-speed laptop CPUs (allegedly).
          >
          Partially right. Are you using a battery monitor? If so, this is the
          source of your time problems. Win2k uses acpi to monitor the battery,
          which does not require interrupts to query the battery status. Linux,
          by default, uses apm, which DOES require interrupts to poll the battery
          status. Poll every couple of seconds, and you get a lot of drift.
          After removing my battery monitor, my i8200 keeps good time (2 sec drift
          every 8 hours).

          What to do?

          1) Poll battery status far less frequently and/or use ntp to reset it.
          Not great, but as you will see, the only viable 'solution' at present.

          2) use ACPI. Unfortunately, though ACPI does not cause clock drift
          when polling the battery, polling the battery repeatedly causes a system
          hang. Not much better, but there is talk of this issue on the ACPI
          development lists. It appears to be Dell specific-- I will post to the
          list my findings.

          So the only real option is to use ntp to set the clock or not use a
          battery monitor. You should know that using apm also causes other
          problems like usb hiccups. The acpi implementation on linux is coming
          along, and is quite solid in my experience, other than this battery
          issue-- hopefully it will be resolved soon.

          Jamie Strandboge

          --
          Email: jstrand1@...
          GPG/PGP ID: 26384A3A
          Fingerprint: D9FF DF4A 2D46 A353 A289 E8F5 AA75 DCBE 2638 4A3A
        • ddhummel
          ... Right, but Windows has nothing to do with the solution. ... What purpose does it serve for a Linux list? I ve been on lots of Linux lists for a long time.
          Message 4 of 22 , Dec 5, 2002
            --- In linux-dell-laptops@y..., jsc3@g... wrote:
            > ddhummel wrote:
            > > --- "ya_hoo_com" <ya_hoo_com@y...> wrote:
            > > > All this was whilst Win2k kept perfect time for months on end.
            > >
            > > This is a linux list.
            >
            > Really? I guess that must be why they call it "linux-dell-laptops".

            ???

            > What's your point? He's talking about a time issue with Linux.

            Right, but Windows has nothing to do with the solution.

            > but a comparison with Windows performance on the same hardware
            > certainly seems reasonable for this forum.

            What purpose does it serve for a Linux list? I've been on lots of
            Linux lists for a long time. These kinds of comparisons are useless
            for arriving at solutions to specific linux problems.

            > Are you such a Linux zealot that any mention
            > of a Linux flaw or any comparison with how some other OS
            > (such as Windows) performs a similar task is not acceptable to you?

            Zealot? Sometimes ;->. I was just asking him to clarify or provide
            proof of the kernel flaws or bugs he claimed to have discovered, for
            the benefit of everyone that might be experiencing this problem, and
            so that more permanent fixes could be found.

            > His experimentation and level of experience may not enable him
            > to point to specific kernel source files.

            Then he shouldn't have claimed to have found bugs in the kernel.

            > It is perfectly possible to find flaws without examining the source
            > - people do it with Windows all the time.

            Yep, and these flaws are usually due to misconfigurations...not kernel
            bugs. One of the strong points of Linux is that is almost infinitely
            configurable. The trick is finding the exact configuration that works
            best with your system.

            > A URL or some other solid reference for the laptop CPU part of this
            > issues would be quite useful, and it would be nice if the original
            > poster could direct us to more details on this.

            Agreed.

            > > > The "hardware clock" is perfect,
            > >
            > > Assuming you've set it perfectly.
            >
            > And nothing's perfect - the hardware clock may keep time acceptably
            > but I seriously doubt it is perfect.

            OK, _perfect_ is not literal in this sense, how about _accurate_? You
            know the hardware clock is accurate when you are setting it, but all
            hardware clocks will drift a little over time. That's why I
            periodically go into the BIOS to verify my hardware clock time.

            > > > and win2k just gets its time
            > > > from this hardware clock all the time, instead of trying to
            > > > do it itself.
            > >
            > > Nope, Windows uses a software clock as well, but this is, erh,
            > > a linux list.
            >
            > You keep saying that like some kind of mantra. It is a Linux list,
            > but comparing Linux with Windows is not out of the scope of the list
            > charter, if I remember correctly.

            OK, so please tell me what purpose will the comparisons serve for this
            problem and others?

            > > > Even more unfortunately, (probably because of linux's background
            > > > of being developed to run on crappy old hardware, and broken
            > > > PC-clones with bug-ridden BIOS's and hardware that the
            > > > manufacturers only support under windows)
            > >
            > > Damn I guess we should all be using Windows then.

            Perhaps this was unnecessarily smartass.

            > No, maybe we should try to figure out how to make Linux keep time at
            > least as well as Windows does on the same hardware.

            Fine, we share the same goal: fixing the Linux system time drift on
            Dell laptops, but Windows is still irrelevant in this regard.

            > I think the point that was being made may be relevant.
            > I think your attitude is counter productive.

            Smartass remarks? Yes, counter-productive. But I think that citing
            an exagerated and partially false version of Linux's development
            history as an excuse for a minor problem is even more
            counter-productive. Again, the workaround is already known. A more
            permanent fix requires identifying the cause. We don't have evidence
            that the problem is even in the kernel. For instance, what happens if
            you disable apm and related applets that pole it?

            > The original poster was not slamming Linux as far as I could tell,
            > he was just stating his case and doing a reasonable job of it
            > (ie time is kept well when running Windows, time is not kept well
            > running Linux on the same hardware, perhaps Linux could be improved
            > somehow to keep time as well as Windows). There may be a few flaws
            > in the argument (ie Windows uses the hardware clock all the time -
            > I am pretty certain that is not true) but the flaws I see don't
            > discredit the basic argument. And the original poster NEVER seemed
            > to suggest that we should all be using Windows.

            OK, I think this is beating a dead horse now.

            > > > all linux distributions and most software assume that the
            > > > "hardware clock" is not to be trusted, and that the linux
            > > > "system clock" is always perfect.
            > >
            > > Uh, the system clock is set from the hardware clock at boot time,
            > > and the kernel doesn't know or care if it is inaccurate.
            >
            > And then some Linux distributions set the hardware clock from the
            > system clock when shutting down, which means that any inaccuracy in
            > the Linux time will now be passed on to the hardware.

            Yes, so comment out the 'hwclock --systohc' call in /etc/init.d/halt,
            or any other shutdown scripts (see original thread).

            Enough said.

            -DDH
          • Pierre Lombard
            ... IIRC, there have some report that enabling in your kernel config: general setup / Allow interrupts during APM BIOS calls may fix this issue too. I cannot
            Message 5 of 22 , Dec 5, 2002
              * James D Strandboge <jstrand1@...> [2002-12-06 06:11]:
              > On Thu, 2002-12-05 at 08:52, ya_hoo_com wrote:
              > > Like apparently many others in this group, I had massive problems with
              > > linux + dell laptop being totally rubbish at keeping time. At one
              > > point, my laptop was losing 2 minutes every hour. Two days was enough
              > > to be almost an hour wrong!

              > Partially right. Are you using a battery monitor? If so, this is the
              > source of your time problems. Win2k uses acpi to monitor the battery,
              > which does not require interrupts to query the battery status. Linux,
              > by default, uses apm, which DOES require interrupts to poll the battery
              > status. Poll every couple of seconds, and you get a lot of drift.
              > After removing my battery monitor, my i8200 keeps good time (2 sec drift
              > every 8 hours).
              >
              > What to do?
              >
              > 1) Poll battery status far less frequently and/or use ntp to reset it.
              > Not great, but as you will see, the only viable 'solution' at present.
              >
              > 2) use ACPI. Unfortunately, though ACPI does not cause clock drift
              > when polling the battery, polling the battery repeatedly causes a system
              > hang. Not much better, but there is talk of this issue on the ACPI
              > development lists. It appears to be Dell specific-- I will post to the
              > list my findings.
              >
              > So the only real option is to use ntp to set the clock or not use a
              > battery monitor. You should know that using apm also causes other
              > problems like usb hiccups. The acpi implementation on linux is coming
              > along, and is quite solid in my experience, other than this battery
              > issue-- hopefully it will be resolved soon.

              IIRC, there have some report that enabling in your kernel config:
              general setup / Allow interrupts during APM BIOS calls
              may fix this issue too.

              I cannot confirm though (I use ntp and did not bother fixing this yet).

              --
              Pierre Lombard <p-l-42@...>
            • Ian Bell
              ... That much is obvious. He was clearly using it to illustrate it is a Linux problem or did you miss that? ... He was clearly using it to illustrate it is a
              Message 6 of 22 , Dec 6, 2002
                On Friday 06 Dec 2002 5:52 am, ddhummel wrote:
                > --- In linux-dell-laptops@y..., jsc3@g... wrote:
                > > ddhummel wrote:
                > > > --- "ya_hoo_com" <ya_hoo_com@y...> wrote:
                > > > > All this was whilst Win2k kept perfect time for months on end.
                > > >
                > > > This is a linux list.
                > >
                > > Really? I guess that must be why they call it "linux-dell-laptops".
                >
                > ???
                >
                > > What's your point? He's talking about a time issue with Linux.
                >
                > Right, but Windows has nothing to do with the solution.

                That much is obvious. He was clearly using it to illustrate it is a Linux
                problem or did you miss that?

                >
                > > but a comparison with Windows performance on the same hardware
                > > certainly seems reasonable for this forum.
                >
                > What purpose does it serve for a Linux list? I've been on lots of
                > Linux lists for a long time. These kinds of comparisons are useless
                > for arriving at solutions to specific linux problems.

                He was clearly using it to illustrate it is a Linus problem rather tan one
                with his PC hardware or did you miss that?

                Talk about mountains out of molehills.

                Ian
              • Nils-Anders Duesund N√łttseter
                ... ... I just wanted to mention what we discussed here a few months back, at least it fits my case perfectly. When I have an apm-applet of any sort in
                Message 7 of 22 , Dec 6, 2002
                  > -----Original Message-----
                  > From: ya_hoo_com [mailto:ya_hoo_com@...]
                  > Sent: Thursday, December 05, 2002 2:52 PM
                  > To: linux-dell-laptops@yahoogroups.com
                  > Subject: [linux-dell-laptops] re: Solution to time/clock
                  > problems with Inspiron laptops in linux
                  >
                  >
                  > Like apparently many others in this group, I had massive
                  > problems with linux + dell laptop being totally rubbish at
                  > keeping time. At one point, my laptop was losing 2 minutes
                  > every hour. Two days was enough to be almost an hour wrong!
                  >

                  <snip>

                  >
                  > I have since discovered (after lots of experimentation) that
                  > the problem (for me at least) is due to bugs in the Linux
                  > kernel. 2.4.18 (and possibly many other versions) have big
                  > problems with variable-speed laptop CPUs (allegedly).
                  >
                  > Linux has two "clocks". The "hardware clock"[1*] which is
                  > CPU/BIOS/chipset controlled, and continues running even
                  > whilst the laptop is powered off, and the "System clock"
                  > which is entirely in software, and is purely maintained by
                  > the linux kernel without using any hardware clock at all.
                  >
                  > The problem is that bugs in the linux kernel mean that
                  > "system time" is often completely wrong, losing and gaining
                  > large numbers of minutes each hour. The "hardware clock" is
                  > perfect, and win2k just gets its time from this hardware
                  > clock all the time, instead of trying to do it itself.
                  >

                  I just wanted to mention what we discussed here a few months back, at
                  least it fits my case perfectly. When I have an apm-applet of any sort
                  in X, the time slows about 2 minutes every hour. When I disable all
                  apm-applets (everything that polls apm for status) the time is kept
                  perfectly. I have no idea why this happens.

                  --
                  Nils-Anders Duesund N√łttseter
                • jsc3@gtf.org
                  ... Windows comparison shows the solution should be possible, by providing a point of reference. Basic troubleshooting technique, in my opinion. ... They are
                  Message 8 of 22 , Dec 6, 2002
                    ddhummel wrote:
                    >
                    > --- In linux-dell-laptops@y..., jsc3@g... wrote:
                    > > ddhummel wrote:
                    > > > --- "ya_hoo_com" <ya_hoo_com@y...> wrote:
                    > > > > All this was whilst Win2k kept perfect time for months on end.
                    > > >
                    > > > This is a linux list.
                    > >
                    > > Really? I guess that must be why they call it "linux-dell-laptops".
                    >
                    > ???
                    >
                    > > What's your point? He's talking about a time issue with Linux.
                    >
                    > Right, but Windows has nothing to do with the solution.

                    Windows comparison shows the solution should be possible, by
                    providing a point of reference. Basic troubleshooting technique,
                    in my opinion.

                    > > but a comparison with Windows performance on the same hardware
                    > > certainly seems reasonable for this forum.
                    >
                    > What purpose does it serve for a Linux list? I've been on lots of
                    > Linux lists for a long time. These kinds of comparisons are useless
                    > for arriving at solutions to specific linux problems.

                    They are useful for showing that SOME software solution for the
                    particulary hardware does exist. If NO OS can keep time reasonably
                    on a given hardware platform, then it is easy to blame the hardware
                    platform. If some OS does keep time reasonably on that same platform,
                    then a hopefully reasonable software solution does exist. If you have
                    some insight into how one OS managed to solve the problem, perhaps a
                    similar technique can be used on the other platform.

                    I guess this is a general approach which can be used to approach a
                    specific solution, which is what the original poster was trying to do.

                    > > Are you such a Linux zealot that any mention
                    > > of a Linux flaw or any comparison with how some other OS
                    > > (such as Windows) performs a similar task is not acceptable to you?
                    >
                    > Zealot? Sometimes ;->.

                    I believe in Linux - a lot. I believe that Linux is going to replace
                    most proprietary Unix variants, and give Microsoft real trouble in the
                    server market. I think eventually Linux will challenge Microsoft on the
                    desktop, particularly in places that are ill-equipped or just unwilling
                    to put up with Microsoft's licensing requirements (ie all of the
                    license tracking, the threats, the leasing rather than purchase model,
                    and the requirements to phone in whenever you want to change your
                    hardware).

                    That said, I don't think being unwilling to even mention the words
                    Microsoft Windows, or to accept that Linux does have flaws and may
                    in fact be inferior to Windows in some respects is productive.

                    > I was just asking him to clarify or provide
                    > proof of the kernel flaws or bugs he claimed to have discovered, for
                    > the benefit of everyone that might be experiencing this problem, and
                    > so that more permanent fixes could be found.

                    It was not what you were asking that I found mildly annoying, it was
                    the manner in which you seemed to ask. Of course, it is hard to
                    interpret tone accurately with no cues other than some letters on a
                    screen.

                    > > His experimentation and level of experience may not enable him
                    > > to point to specific kernel source files.
                    >
                    > Then he shouldn't have claimed to have found bugs in the kernel.

                    Perhaps he doesn't have enough experience to know that. Your type of
                    critical response does not seem likely to push him towards the path of
                    enlightenment - it is more like giving a newbie sh*t. Or maybe he has
                    found a kernel bug but he doesn't know how to hunt it down.

                    In most people's experience, they don't distinguish between the kernel
                    and the rest of the OS. They start out by saying things like Linux 8.0,
                    etc. The original poster seems to have progressed past that point.

                    Rather than slam him for not being specific enough, and perhaps laying
                    blame where it should not be laid, you could offer some suggestions to
                    encourage him toward the path of righteousness.

                    > > It is perfectly possible to find flaws without examining the source
                    > > - people do it with Windows all the time.
                    >
                    > Yep, and these flaws are usually due to misconfigurations...not kernel
                    > bugs.

                    True enough, but there are kernel bugs too.

                    > One of the strong points of Linux is that is almost infinitely
                    > configurable. The trick is finding the exact configuration that works
                    > best with your system.

                    True enough. On the other hand, to get mass acceptance, Linux is going
                    to have to be as self configuring as possible.

                    > > A URL or some other solid reference for the laptop CPU part of this
                    > > issues would be quite useful, and it would be nice if the original
                    > > poster could direct us to more details on this.
                    >
                    > Agreed.
                    >
                    > > > > The "hardware clock" is perfect,
                    > > >
                    > > > Assuming you've set it perfectly.
                    > >
                    > > And nothing's perfect - the hardware clock may keep time acceptably
                    > > but I seriously doubt it is perfect.
                    >
                    > OK, _perfect_ is not literal in this sense, how about _accurate_? You
                    > know the hardware clock is accurate when you are setting it, but all
                    > hardware clocks will drift a little over time. That's why I
                    > periodically go into the BIOS to verify my hardware clock time.

                    I was probably being a bit too pedantic - my turn to be a smart ass I
                    guess - nobody's perfect!

                    [snip]

                    > > > Nope, Windows uses a software clock as well, but this is, erh,
                    > > > a linux list.
                    > >
                    > > You keep saying that like some kind of mantra. It is a Linux list,
                    > > but comparing Linux with Windows is not out of the scope of the list
                    > > charter, if I remember correctly.
                    >
                    > OK, so please tell me what purpose will the comparisons serve for this
                    > problem and others?

                    It shows that without a bunch of twiddling that Windows is able to keep
                    reasonable time on the platform, so it should be possible to keep decent
                    time running Linux without having to run NTP or running hwclock out of
                    cron every hour (the former is useful only if you are connected to a
                    network with an NTP server on it - it doesn't address the root cause,
                    and the latter is basically a hack to work around the basic problem.

                    I am not one of these folks who thinks that Linux should only be for
                    people who are smart enough to figure out how to make it run right.
                    The vast majority of humanity wants the computer to do the work - that's
                    what automation is all about. It's all about making things easier,
                    which often means making the software smarter.

                    In the long run, a REAL solution to the problem would be nice.

                    > > > > Even more unfortunately, (probably because of linux's background
                    > > > > of being developed to run on crappy old hardware, and broken
                    > > > > PC-clones with bug-ridden BIOS's and hardware that the
                    > > > > manufacturers only support under windows)
                    > > >
                    > > > Damn I guess we should all be using Windows then.
                    >
                    > Perhaps this was unnecessarily smartass.

                    Agreed. Seriously, I guess all I was looking for was a more civil tone.
                    I am all for slamming trolls and idiots, but the original poster did not
                    seem to be either.

                    > > No, maybe we should try to figure out how to make Linux keep time at
                    > > least as well as Windows does on the same hardware.
                    >
                    > Fine, we share the same goal: fixing the Linux system time drift on
                    > Dell laptops, but Windows is still irrelevant in this regard.

                    It is not irrelevant as proof of concept that the hardware is not
                    hopelessly broken. With PCs, you can't assume that the hardware is
                    not broken.

                    > > I think the point that was being made may be relevant.
                    > > I think your attitude is counter productive.
                    >
                    > Smartass remarks? Yes, counter-productive. But I think that citing
                    > an exagerated and partially false version of Linux's development
                    > history as an excuse for a minor problem is even more
                    > counter-productive.

                    Actually, I thought his history in a nutshell was largely accurate.
                    I also thought the same could be said for Windows, and even DOS. The
                    first IBM PC was broken and buggy out of the gate. Many of the clones
                    that followed were not as good as that. The fact that the OS's can cope
                    with this is a testament to the software, not a slam on it. That Linux
                    has gotten as far as it has without the same manufacturer support
                    advantages that Windows has had is also a testament to the developers.
                    Further, many of these developers DID get their start on crappy OLD
                    PC-clones. I knew a lot of them in college, working on old 386 and
                    486 castoffs. Some of these folks are Linux developers and kernel
                    hackers today.

                    I am not sure what you found so offensive, actually.

                    > Again, the workaround is already known. A more
                    > permanent fix requires identifying the cause. We don't have evidence
                    > that the problem is even in the kernel. For instance, what happens if
                    > you disable apm and related applets that pole it?

                    On this we agree 100%. We are almost certainly past the point of useful
                    Windows comparisons now.

                    > > The original poster was not slamming Linux as far as I could tell,
                    > > he was just stating his case and doing a reasonable job of it
                    > > (ie time is kept well when running Windows, time is not kept well
                    > > running Linux on the same hardware, perhaps Linux could be improved
                    > > somehow to keep time as well as Windows). There may be a few flaws
                    > > in the argument (ie Windows uses the hardware clock all the time -
                    > > I am pretty certain that is not true) but the flaws I see don't
                    > > discredit the basic argument. And the original poster NEVER seemed
                    > > to suggest that we should all be using Windows.
                    >
                    > OK, I think this is beating a dead horse now.

                    Agreed.

                    > > > > all linux distributions and most software assume that the
                    > > > > "hardware clock" is not to be trusted, and that the linux
                    > > > > "system clock" is always perfect.
                    > > >
                    > > > Uh, the system clock is set from the hardware clock at boot time,
                    > > > and the kernel doesn't know or care if it is inaccurate.
                    > >
                    > > And then some Linux distributions set the hardware clock from the
                    > > system clock when shutting down, which means that any inaccuracy in
                    > > the Linux time will now be passed on to the hardware.
                    >
                    > Yes, so comment out the 'hwclock --systohc' call in /etc/init.d/halt,
                    > or any other shutdown scripts (see original thread).
                    >
                    > Enough said.

                    Well, not quite, but serious work would need to be done to take it
                    further. Right now, all we have is a workaround, not a solution for
                    the root cause of the problem.

                    I don't have the time or inclination to take it further, and frankly
                    the work-around is enough for me. At some point it would be nice if
                    a real fix was found so that everybody who is using one of these laptops
                    and doesn't use NTP all the time doesn't have to find the workaround and
                    implement it themselves. Thankfully lots of folks have the source code
                    and at some point somebody will probably get annoyed enough to pursue
                    this further and fix the root cause.

                    I am glad this didn't turn into a flame war, by the way. I just believe
                    in encouraging folks who are trying, and moving in productive
                    directions. I think you and I honestly disagree on what is productive
                    in some cases, which is the way life works sometimes.

                    --
                    John Cronin
                    mailto: `echo NjsOc3@... | sed 's/[NOSPAM]//g'`
                  • Gregory Gulik
                    I know this is slightly off topic... I ve been hearing this as long as I ve been using Linux and it s starting to bug me. We Linux users are treated as second
                    Message 9 of 22 , Dec 6, 2002
                      I know this is slightly off topic...

                      I've been hearing this as long as I've been using Linux and it's
                      starting to bug me.

                      We Linux users are treated as second class citizens and from what I can
                      tell is mostly because these vendors don't think there are many of us.

                      I'm starting to think that part of that perception is that we're too
                      quiet, we discuss issues only amongst ourselves and rely on each other
                      to solve our problems.

                      I for one have recently started to E-mail feedback departments of
                      various hardware/software vendors, asking about Linux.

                      I have a feeling that if we stop hiding in the shadows and start making
                      our presence known we'll get a little more respect.



                      ddhummel wrote:
                      >>The last time I was receiving emails from this group, it seemed Dell
                      >>were offering no help, and no-one quite knew the problem.
                      >
                      > You can pretty much count on this from Dell these days for anything
                      > having to do with Linux. Don't bother.
                    • James D Strandboge
                      ... Intel is heading up the ACPI development on linux. They work hard. I suggest you join the acpi development mailing list and provide your feedback.
                      Message 10 of 22 , Dec 6, 2002
                        On Fri, 2002-12-06 at 09:32, Gregory Gulik wrote:
                        > I know this is slightly off topic...
                        >
                        > I've been hearing this as long as I've been using Linux and it's
                        > starting to bug me.
                        >
                        > We Linux users are treated as second class citizens and from what I can
                        > tell is mostly because these vendors don't think there are many of us.
                        >
                        > I'm starting to think that part of that perception is that we're too
                        > quiet, we discuss issues only amongst ourselves and rely on each other
                        > to solve our problems.
                        >
                        > I for one have recently started to E-mail feedback departments of
                        > various hardware/software vendors, asking about Linux.
                        >
                        > I have a feeling that if we stop hiding in the shadows and start making
                        > our presence known we'll get a little more respect.

                        Intel is heading up the ACPI development on linux. They work hard. I
                        suggest you join the acpi development mailing list and provide your
                        feedback.

                        https://lists.sourceforge.net/lists/listinfo/acpi-devel

                        Jamie Strandboge

                        --
                        Email: jstrand1@...
                        GPG/PGP ID: 26384A3A
                        Fingerprint: D9FF DF4A 2D46 A353 A289 E8F5 AA75 DCBE 2638 4A3A
                      • James D Strandboge
                        ... I wish this were true, but it made no effect on my laptop (i8200). Some people report strange sounds coming from their machine if they enable interrupts
                        Message 11 of 22 , Dec 6, 2002
                          On Fri, 2002-12-06 at 02:15, Pierre Lombard wrote:
                          > * James D Strandboge <jstrand1@...> [2002-12-06 06:11]:
                          > > On Thu, 2002-12-05 at 08:52, ya_hoo_com wrote:
                          > > > Like apparently many others in this group, I had massive problems with
                          > > > linux + dell laptop being totally rubbish at keeping time. At one
                          > > > point, my laptop was losing 2 minutes every hour. Two days was enough
                          > > > to be almost an hour wrong!
                          >
                          > > Partially right. Are you using a battery monitor? If so, this is the
                          > > source of your time problems. Win2k uses acpi to monitor the battery,
                          > > which does not require interrupts to query the battery status. Linux,
                          > > by default, uses apm, which DOES require interrupts to poll the battery
                          > > status. Poll every couple of seconds, and you get a lot of drift.
                          > > After removing my battery monitor, my i8200 keeps good time (2 sec drift
                          > > every 8 hours).
                          > >
                          > > What to do?
                          > >
                          > > 1) Poll battery status far less frequently and/or use ntp to reset it.
                          > > Not great, but as you will see, the only viable 'solution' at present.
                          > >
                          > > 2) use ACPI. Unfortunately, though ACPI does not cause clock drift
                          > > when polling the battery, polling the battery repeatedly causes a system
                          > > hang. Not much better, but there is talk of this issue on the ACPI
                          > > development lists. It appears to be Dell specific-- I will post to the
                          > > list my findings.
                          > >
                          > > So the only real option is to use ntp to set the clock or not use a
                          > > battery monitor. You should know that using apm also causes other
                          > > problems like usb hiccups. The acpi implementation on linux is coming
                          > > along, and is quite solid in my experience, other than this battery
                          > > issue-- hopefully it will be resolved soon.
                          >
                          > IIRC, there have some report that enabling in your kernel config:
                          > general setup / Allow interrupts during APM BIOS calls
                          > may fix this issue too.

                          I wish this were true, but it made no effect on my laptop (i8200). Some
                          people report strange sounds coming from their machine if they enable
                          interrupts during apm calls (though I don't).

                          Jamie Strandboge

                          --
                          Email: jstrand1@...
                          GPG/PGP ID: 26384A3A
                          Fingerprint: D9FF DF4A 2D46 A353 A289 E8F5 AA75 DCBE 2638 4A3A
                        • jsc3@gtf.org
                          ... This is true not only from a Windows-centric perspective, but from an enterprise level Unix perspective too. I am currently working at a large Fortune 100
                          Message 12 of 22 , Dec 6, 2002
                            Gregory Gulik wrote:
                            >
                            > I know this is slightly off topic...
                            >
                            > I've been hearing this as long as I've been using Linux and it's
                            > starting to bug me.
                            >
                            > We Linux users are treated as second class citizens and from what I can
                            > tell is mostly because these vendors don't think there are many of us.

                            This is true not only from a Windows-centric perspective, but from
                            an enterprise level Unix perspective too. I am currently working
                            at a large Fortune 100 telecom and their security folks came out
                            with a ridiculous policy against Open Source; it was absurd. Oracle
                            has gone over to Linux though, and many others are beginning to see
                            the light (ie Veritas).

                            > I'm starting to think that part of that perception is that we're too
                            > quiet, we discuss issues only amongst ourselves and rely on each other
                            > to solve our problems.

                            No, the problem has been that there really weren't that many of us.
                            However, the numbers grow exponentially. There may be a huge wave of
                            newbies riding the backs of the $200 Walmart PCs.

                            That said, continuing to ask for support from vendors is something we
                            should be doing. I do it regularly.

                            > I for one have recently started to E-mail feedback departments of
                            > various hardware/software vendors, asking about Linux.

                            When I go to Veritas software training, and fill out the course
                            evaluation, and it asks what other training I might be interested in, I
                            note that Linux is not one of the choices and add some comments that I
                            would be interested in it. Veritas now offers Veritas Cluster Server
                            and the Foundation Suite (Volume Manager and Filesystem) on Linux, and
                            I ask them when they will port the rest of it. When I am asked to go to
                            a user focus group with the developers, I ask the developers themselves
                            about Linux.

                            > I have a feeling that if we stop hiding in the shadows and start making
                            > our presence known we'll get a little more respect.

                            I have made my views known to Dell as well. I work as a consultant, and
                            I advocate Linux to my clients (though most of my bread and butter comes
                            from Solaris, I don't expect it to remain that way).

                            > ddhummel wrote:
                            > >>The last time I was receiving emails from this group, it seemed Dell
                            > >>were offering no help, and no-one quite knew the problem.
                            > >
                            > > You can pretty much count on this from Dell these days for anything
                            > > having to do with Linux. Don't bother.

                            I guess I see the point - if we don't bother to ask, they'll never
                            bother to support it.

                            --
                            John Cronin
                            mailto: `echo NjsOc3@... | sed 's/[NOSPAM]//g'`
                          • ddhummel
                            ... This is taken out of context. I agree that we should make as much noise as possible about Linux. My comment above had to do with getting Linux support
                            Message 13 of 22 , Dec 6, 2002
                              --- In linux-dell-laptops@y..., jsc3@g... wrote:
                              > > ddhummel wrote:
                              > > >>The last time I was receiving emails from this group, it seemed
                              > > >>Dell were offering no help, and no-one quite knew the problem.
                              > > >
                              > > > You can pretty much count on this from Dell these days for
                              > > > anything having to do with Linux. Don't bother.
                              >
                              > I guess I see the point - if we don't bother to ask, they'll never
                              > bother to support it.

                              This is taken out of context. I agree that we should make as much
                              noise as possible about Linux. My comment above had to do with
                              getting Linux support from Dell about specific problems, right now.
                              Unfortunately, I don't think this will be possible until Linux enjoys
                              a greater percentage of the market in the Desktop space, at which
                              point, vendors big and small will begin supporting Linux on their
                              hardware. Keep making noise I guess.

                              -DDH
                            • AthlonRob
                              ... I think it is a bit of a chicken-and-egg problem there - we *must* rely on each other, because hardware (and many software) manufacturers won t offer us
                              Message 14 of 22 , Dec 6, 2002
                                On Fri, 2002-12-06 at 06:32, Gregory Gulik wrote:

                                > We Linux users are treated as second class citizens and from what I
                                > can tell is mostly because these vendors don't think there are many of
                                > us.
                                >
                                > I'm starting to think that part of that perception is that we're too
                                > quiet, we discuss issues only amongst ourselves and rely on each other
                                > to solve our problems.

                                I think it is a bit of a chicken-and-egg problem there - we *must* rely
                                on each other, because hardware (and many software) manufacturers won't
                                offer us support.

                                I can't think of any user-level hardware makers (ie; not server makers)
                                who will offer Linux users any kind of support, at all. They may
                                release drivers, but most don't even do that.

                                > I for one have recently started to E-mail feedback departments of
                                > various hardware/software vendors, asking about Linux.
                                >
                                > I have a feeling that if we stop hiding in the shadows and start
                                > making our presence known we'll get a little more respect.

                                I've been fighting with one software maker for years over this - and,
                                being just one voice, have gotten nowhere. The funny thing is, this
                                company has a Linux version of their software already developed, they
                                just don't want to release it - because they would then have to support
                                it.

                                I definitely agree with you - we need to be a bit louder and contact
                                these folks. Getting more users is one thing.... but we need the
                                support before they'd want to use Linux... a chicken-and-egg problem
                                again.

                                Whenever I run in to hardware that just doesn't work right in Linux due
                                to the manufacturer's lack of interest, you can bet I do email them.
                                :-)

                                --
                                Rob | If not safe,
                                Email and Jabber: | one can never be free.
                                athlonrob at data dot 4t3 dot com |
                              • Jim Diamond
                                ... Well, every now and then you can pleasantly surprised. Way Back When I went to Adobe s web site, to a page where you could report bugs. I put in a bug
                                Message 15 of 22 , Dec 6, 2002
                                  On Fri, 6 Dec 2002 at 09:54 (-0800), AthlonRob wrote:

                                  > On Fri, 2002-12-06 at 06:32, Gregory Gulik wrote:

                                  >> We Linux users are treated as second class citizens and from what I
                                  >> can tell is mostly because these vendors don't think there are many of
                                  >> us.

                                  >> I'm starting to think that part of that perception is that we're too
                                  >> quiet, we discuss issues only amongst ourselves and rely on each other
                                  >> to solve our problems.

                                  > I think it is a bit of a chicken-and-egg problem there - we *must* rely
                                  > on each other, because hardware (and many software) manufacturers won't
                                  > offer us support.

                                  Well, every now and then you can pleasantly surprised. Way Back When
                                  I went to Adobe's web site, to a page where you could report bugs. I
                                  put in a bug report for Acrobat Reader, my bug being that they did not
                                  have a Linux version available. Not long after (a couple of weeks?) a
                                  Linux version appeared on their web site. I have no idea whether it
                                  was already in the works or not when I complained, but if it was,
                                  perhaps that is because someone complained the day/week/month before I
                                  did.

                                  I read somewhere that there are more Linux users than Mac users
                                  (although I'm sure this is not easy to verify), and yet many, many h/w
                                  and third-party s/w companies support Macs. So why not Linux? Why
                                  should I accept a colour printer (for example) that has minimal or no
                                  Linux support? Unlike all the free software we all use, printer
                                  manufactures know that we have to pay money for h/w, and presumably
                                  our money is as good as any Mac user's. Maybe if enough of us make
                                  enough noise, skeptical sales and marketing droids will begin to
                                  realize that there is another group out there.

                                  Jim

                                  P.S. Are you the same AthlonRob as in a.o.l.s.? Trying out a
                                  (generally) civil mailing list for a change? :-)
                                • AthlonRob
                                  ... Now I wish I could have luck like that... what s your secret? :-) ... I read several years ago there were more Linux users than Mac users - before OS X - I
                                  Message 16 of 22 , Dec 6, 2002
                                    On Fri, 2002-12-06 at 16:15, Jim Diamond wrote:

                                    > Well, every now and then you can pleasantly surprised. Way Back When
                                    > I went to Adobe's web site, to a page where you could report bugs. I
                                    > put in a bug report for Acrobat Reader, my bug being that they did not
                                    > have a Linux version available. Not long after (a couple of weeks?) a
                                    > Linux version appeared on their web site. I have no idea whether it
                                    > was already in the works or not when I complained, but if it was,
                                    > perhaps that is because someone complained the day/week/month before I
                                    > did.

                                    Now I wish I could have luck like that... what's your secret? :-)

                                    > I read somewhere that there are more Linux users than Mac users
                                    > (although I'm sure this is not easy to verify), and yet many, many h/w
                                    > and third-party s/w companies support Macs. So why not Linux? Why
                                    > should I accept a colour printer (for example) that has minimal or no
                                    > Linux support? Unlike all the free software we all use, printer
                                    > manufactures know that we have to pay money for h/w, and presumably
                                    > our money is as good as any Mac user's. Maybe if enough of us make
                                    > enough noise, skeptical sales and marketing droids will begin to
                                    > realize that there is another group out there.

                                    I read several years ago there were more Linux users than Mac users -
                                    before OS X - I don't know how things have changed since then, though.
                                    But for those software makers that support OS X, porting over to Linux
                                    shouldn't be too much of a hassle, really. Much less than porting from
                                    Windows to OS X!

                                    As somebody else said, maybe those $200 walmart PCs with Lycoris,
                                    Mandrake, or (*ghasp*) Lindows will help spur more support.

                                    > P.S. Are you the same AthlonRob as in a.o.l.s.? Trying out a
                                    > (generally) civil mailing list for a change? :-)

                                    Small world... that is, indeed, me. I do many mailing lists and very
                                    few newsgroups... basically aols and alt.sci.seti. I still really like
                                    AOLS, though. :-)

                                    I just ordered a Dell laptop Tuesday night. It's in the 'Testing' phase
                                    (has been since Wednesday! they must test the hell out of these
                                    suckers) ... I wanted to get prepared for any nuances I might run in to,
                                    so joined this list, which I do enjoy. :-)

                                    --
                                    Rob | If not safe,
                                    Email and Jabber: | one can never be free.
                                    athlonrob at data dot 4t3 dot com |
                                  • Dossy
                                    ... Funny, I just ordered mine on Wednesday 12/04 ... and it s sitting in the Testing phase of the build process, too. I ended up ordering a Latitude C840
                                    Message 17 of 22 , Dec 7, 2002
                                      On 2002.12.06, AthlonRob <athlonrob@...> wrote:
                                      > I just ordered a Dell laptop Tuesday night. It's in the 'Testing' phase
                                      > (has been since Wednesday! they must test the hell out of these
                                      > suckers) ... I wanted to get prepared for any nuances I might run in to,
                                      > so joined this list, which I do enjoy. :-)

                                      Funny, I just ordered mine on Wednesday 12/04 ... and it's sitting in
                                      the "Testing" phase of the build process, too.

                                      I ended up ordering a Latitude C840 instead of the Inspiron 8200 -- was
                                      this a bad move? I was eagerly looking at the Precision M50, but for
                                      the extra $1,000 for just dual-head support, I'd rather buy a PCMCIA
                                      VGA card and save myself a lot of money ...

                                      <sarcasm> I can't /wait/ to find out all the gotchas of running Linux
                                      on this thing. </sarcasm> :-)

                                      -- Dossy

                                      --
                                      Dossy Shiobara mail: dossy@...
                                      Panoptic Computer Network web: http://www.panoptic.com/
                                      "He realized the fastest way to change is to laugh at your own
                                      folly -- then you can let go and quickly move on." (p. 70)
                                    • Jim Diamond
                                      ... I wish I knew... it was a once-in-a-lifetime coincidence, probably. ... But I think it s easier than that. A lot of people (i.e., companies) support
                                      Message 18 of 22 , Dec 9, 2002
                                        On Sat, 6 Dec 2002 at 21:36 (-0800), AthlonRob wrote:

                                        > On Fri, 2002-12-06 at 16:15, Jim Diamond wrote:

                                        >> Well, every now and then you can pleasantly surprised. Way Back When
                                        >> I went to Adobe's web site, to a page where you could report bugs. I
                                        >> put in a bug report for Acrobat Reader, my bug being that they did not
                                        >> have a Linux version available. Not long after (a couple of weeks?) a
                                        >> Linux version appeared on their web site. I have no idea whether it
                                        >> was already in the works or not when I complained, but if it was,
                                        >> perhaps that is because someone complained the day/week/month before I
                                        >> did.

                                        > Now I wish I could have luck like that... what's your secret? :-)

                                        I wish I knew... it was a once-in-a-lifetime coincidence, probably.

                                        >> I read somewhere that there are more Linux users than Mac users
                                        > I read several years ago there were more Linux users than Mac users -
                                        > before OS X - I don't know how things have changed since then, though.
                                        > But for those software makers that support OS X, porting over to Linux
                                        > shouldn't be too much of a hassle, really. Much less than porting from
                                        > Windows to OS X!
                                        But I think it's easier than that. A lot of people (i.e., companies)
                                        support Solaris (or other commercial Unixes), and unless they've gone
                                        out of their way to write non-portable code, it shouldn't (always) be
                                        a major development effort to port their stuff to Linux. In the case
                                        of Acrobat reader, they already had versions out for multiple flavours
                                        of Unix, so the port to Linux was probably a breeze. But someone had
                                        to ask them!

                                        >> P.S. Are you the same AthlonRob as in a.o.l.s.? Trying out a
                                        >> (generally) civil mailing list for a change? :-)
                                        > Small world... that is, indeed, me. I do many mailing lists and
                                        > very few newsgroups... basically aols and alt.sci.seti. I still
                                        > really like AOLS, though. :-)
                                        Yeah, me too, but it could stand more civility. But that's not a
                                        l-d-l issue.

                                        > I just ordered a Dell laptop Tuesday night. It's in the 'Testing' phase
                                        > (has been since Wednesday! they must test the hell out of these
                                        > suckers) ...
                                        Not mine... it was through the system in no time flat.

                                        > I wanted to get prepared for any nuances I might run in to, so
                                        > joined this list, which I do enjoy. :-)
                                        You want nuances? Then I hope you ordered the Nvidea card, cuz boy,
                                        if you did, you'll have nuances. :-)

                                        Jim
                                      Your message has been successfully submitted and would be delivered to recipients shortly.