Suspend/resume - Request for Testing - Slackware

This is a discussion on Suspend/resume - Request for Testing - Slackware ; I've just finished packaging pm-utils 1.2.2 for Slackware (unofficially, of course), and I'd like to get some more widespread testing with it. This new release automatically queries the HAL "quirks" database, so in theory, suspend/resume should work on almost all ...

+ Reply to Thread
Results 1 to 20 of 20

Thread: Suspend/resume - Request for Testing

  1. Suspend/resume - Request for Testing

    I've just finished packaging pm-utils 1.2.2 for Slackware (unofficially,
    of course), and I'd like to get some more widespread testing with it.
    This new release automatically queries the HAL "quirks" database, so in
    theory, suspend/resume should work on almost all laptops with it.

    Testing is relatively straightforward, but due to its experimental
    nature, backups are expected. If you don't have backups and you lose
    data, don't blame me or pm-utils.

    Package here:
    http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    Sources here:
    http://rlworkman.net/pkgs/sources/12.1/pm-utils/

    Once it's installed, running "pm-suspend" as root should do a suspend
    to ram, and "pm-hibernate" should suspend to disk. Note that you
    must have a swap partition (or file, but partition is easier) which is
    large enough (2x ram is good here) and the relevant initrd lines and
    such for this to work. If you're not familiar with all of that, don't
    test pm-hibernate. [1]

    If it works for you, and/or you want to automate it a bit, you'll find
    some sample acpi event declarations and scripts here:
    http://rlworkman.net/conf/acpi/
    The "s2ram.sh" and "s2disk.sh" scripts are currently written to use the
    default kernel methods, but they're commented with how to change them
    to use pm-utils methods.


    [1] Here's the short version of using suspend to disk:

    Stanza in /etc/lilo.conf should look something like this:
    image = /boot/vmlinuz-generic-smp-2.6.26.5-smp
    initrd = /boot/initrd.gz
    append = "resume=/dev/hda2"
    root = /dev/hda1
    label = Slackware
    read-only

    /etc/mkinitrd.conf should look something like this:
    MODULE_LIST="jfs"
    ROOTDEV="/dev/hda1"
    ROOTFS="jfs"
    RESUMEDEV="/dev/hda2"

    mkinitrd command invocation would be this:
    mkinitrd -c -k 2.6.26.5-smp -F

    All of the above assumes at least a cursory understanding of what those
    commands and parameters do. If it's not clear, don't attempt it on a
    system that you can't easily reinstall without concern.

    -RW



  2. Re: Suspend/resume - Request for Testing

    On 2008-10-06, Robby Workman wrote:
    > I've just finished packaging pm-utils 1.2.2 for Slackware (unofficially,
    > of course), and I'd like to get some more widespread testing with it.
    > This new release automatically queries the HAL "quirks" database, so in
    > theory, suspend/resume should work on almost all laptops with it.



    For the record, my testing ability on this is quite limited due to the
    hardware I have available. All of my laptops are Thinkpads, and the
    kernel suspend/resume is pretty much flawless on them; that's why I'm
    requesting more widespread testing...

    -RW

  3. Re: Suspend/resume - Request for Testing

    On Mon, 06 Oct 2008 03:53:31 +0000, Robby Workman wrote:

    >I've just finished packaging pm-utils 1.2.2 for Slackware (unofficially,
    >of course), and I'd like to get some more widespread testing with it.
    >This new release automatically queries the HAL "quirks" database, so in
    >theory, suspend/resume should work on almost all laptops with it.
    >
    >Testing is relatively straightforward, but due to its experimental
    >nature, backups are expected. If you don't have backups and you lose
    >data, don't blame me or pm-utils.
    >
    >Package here:
    > http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >Sources here:
    > http://rlworkman.net/pkgs/sources/12.1/pm-utils/
    >
    >Once it's installed, running "pm-suspend" as root should do a suspend
    >to ram, and "pm-hibernate" should suspend to disk. Note that you
    >must have a swap partition (or file, but partition is easier) which is
    >large enough (2x ram is good here) and the relevant initrd lines and
    >such for this to work. If you're not familiar with all of that, don't
    >test pm-hibernate. [1]


    Well, I just did a fresh install of 12.1 and compiled a new kernel for
    the laptop and nothing happened:

    root@toshi:~# installpkg pm-utils-1.2.2-i486-1_rlw.tgz
    Installing package pm-utils-1.2.2-i486-1_rlw...
    PACKAGE DESCRIPTION:
    pm-utils: pm-utils (Power Management Utilities)
    pm-utils:
    pm-utils: The pm-utils package provides simple shell command line tools
    pm-utils: to suspend and hibernate computers that can be used to run
    pm-utils: vendor or distro supplied scripts on suspend and resume.
    pm-utils:
    pm-utils: Homepage: http://webcvs.freedesktop.org/pm-utils/pm-utils/
    pm-utils:
    Executing install script for pm-utils-1.2.2-i486-1_rlw...

    root@toshi:~# pm-suspend
    root@toshi:~# pm-hibernate
    root@toshi:~# uname -r
    2.6.26.5a

    Not even a crash... May try another day, when I have the lappy more
    setup.

    Was this thing meant to work for slack-12.1 or -current?
    >
    >If it works for you, and/or you want to automate it a bit, you'll find
    >some sample acpi event declarations and scripts here:
    > http://rlworkman.net/conf/acpi/
    >The "s2ram.sh" and "s2disk.sh" scripts are currently written to use the
    >default kernel methods, but they're commented with how to change them
    >to use pm-utils methods.
    >
    >
    >[1] Here's the short version of using suspend to disk:
    >
    > Stanza in /etc/lilo.conf should look something like this:
    > image = /boot/vmlinuz-generic-smp-2.6.26.5-smp
    > initrd = /boot/initrd.gz
    > append = "resume=/dev/hda2"
    > root = /dev/hda1
    > label = Slackware
    > read-only
    >
    > /etc/mkinitrd.conf should look something like this:
    > MODULE_LIST="jfs"
    > ROOTDEV="/dev/hda1"
    > ROOTFS="jfs"
    > RESUMEDEV="/dev/hda2"
    >
    > mkinitrd command invocation would be this:
    > mkinitrd -c -k 2.6.26.5-smp -F
    >
    > All of the above assumes at least a cursory understanding of what those
    > commands and parameters do. If it's not clear, don't attempt it on a
    > system that you can't easily reinstall without concern.


    I assume it should work wihout an initrd? Lappy wont boot the generic
    slack kernel. The huge is okay, except fot loading a botload of irrelevant
    drivers -- but you know I favour custom kernels And yes, I enabled
    s2r + s2d in the kernel as well as nomiate the swap partitioon, but
    I've never seen linus go sleepy-byes -- just read the horror stories
    about how it doesn't work.

    Grant.
    --
    http://bugsplatter.id.au/

  4. Re: Suspend/resume - Request for Testing

    On 2008-10-07, Grant wrote:
    >
    > Well, I just did a fresh install of 12.1 and compiled a new kernel for
    > the laptop and nothing happened:
    >
    > root@toshi:~# pm-suspend
    > root@toshi:~# pm-hibernate
    > root@toshi:~# uname -r
    > 2.6.26.5a
    >
    > Not even a crash... May try another day, when I have the lappy more
    > setup.
    >
    > Was this thing meant to work for slack-12.1 or -current?



    Should work on either one.

    One of these routines should tell you whether any of the operations
    are supported by your current kernel/userspace at all:

    for i in hibernate suspend suspend-hybrid ; do
    pm-is-supported --$i && echo "$i is supported" || echo "$i is not supported" ;
    done

    cat /sys/power/state

    This is another of those things that, because I'm on a Thinkpad, I may
    not be aware of all the system-specific knobs to tweak. Perhaps there's
    some special module(s) that need(s) to be loaded.

    -RW

  5. Re: Suspend/resume - Request for Testing

    Robby Workman wrote:

    > I've just finished packaging pm-utils 1.2.2 for Slackware (unofficially,
    > of course), and I'd like to get some more widespread testing with it.
    > [...]


    Installs and runs perfectly on my T60 with ATIx1400 with slack 12.1 .

    It finally solved the problem with the blank console after suspend to
    RAM. Nor the kernel parameter acpi_sleep=s3_bios nor the vbetool utility
    helped, but this "pm-suspend --quirks-s3-bios" works great.

    Thank you so much.

    --
    enos76 from Italy

  6. Re: Suspend/resume - Request for Testing

    On 2008-10-06, Robby Workman wrote:

    > Package here:
    > http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    > Sources here:
    > http://rlworkman.net/pkgs/sources/12.1/pm-utils/



    Update:

    http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    Sources in same place.

    -RW

  7. Re: Suspend/resume - Request for Testing

    On Tue, 07 Oct 2008 04:43:02 +0000, Robby Workman wrote:
    ....
    >> Was this thing meant to work for slack-12.1 or -current?

    >
    >
    >Should work on either one.
    >
    >One of these routines should tell you whether any of the operations
    >are supported by your current kernel/userspace at all:
    >
    > for i in hibernate suspend suspend-hybrid ; do
    > pm-is-supported --$i && echo "$i is supported" || echo "$i is not supported" ;
    > done
    >
    > cat /sys/power/state


    root@toshi:~# for i in hibernate suspend suspend-hybrid; \
    do pm-is-supported --$i && echo "$i is supported" || \
    echo "$i is not supported" ; done
    hibernate is supported
    suspend is not supported
    suspend-hybrid is not supported
    root@toshi:~# cat /sys/power/state
    standby disk

    Okay, so I need to check kernel options and other stuff, config and dmesg
    and lappy info are here:
    http://bugsplatter.id.au/kernel/boxen/toshi/
    >
    >This is another of those things that, because I'm on a Thinkpad, I may
    >not be aware of all the system-specific knobs to tweak. Perhaps there's
    >some special module(s) that need(s) to be loaded.


    This is a nine-year-old Toshiba, 500MHz coppermine celeron, 192MB memory.

    Grant.
    --
    http://bugsplatter.id.au/

  8. Re: Suspend/resume - Request for Testing

    On Tue, 07 Oct 2008 20:38:16 +0000, Robby Workman wrote:

    >On 2008-10-06, Robby Workman wrote:
    >
    >> Package here:
    >> http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >> Sources here:
    >> http://rlworkman.net/pkgs/sources/12.1/pm-utils/

    >
    >
    >Update:
    >
    >http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >Sources in same place.


    Okay,I'm reading and learning more about suspend2disk (hibernate),
    and would like to know if you've considered this little gem:

    "In principle it does not require ACPI or APM, although for example
    ACPI will be used for the final steps when it is available. One
    of the reasons to use software suspend is that the firmware hooks
    for suspend states like suspend-to-RAM (STR) often don't work very
    well with Linux.

    It creates an image which is saved in your active swap. Upon the next
    boot, pass the 'resume=/dev/swappartition' argument to the kernel to
    have it detect the saved image, restore memory state from it, and
    continue to run as before. If you do not want the previous state to
    be reloaded, then use the 'noresume' kernel command line argument.
    Note, however, that fsck will be run on your filesystems and you will
    need to run mkswap against the swap partition used for the suspend.

    It also works with swap files to a limited extent (for details see
    ocumentation/power/swsusp-and-swap-files.txt>).

    Right now you may boot without resuming and resume later but in the
    meantime you cannot use the swap partition(s)/file(s) involved in
    suspending. Also in this case you must not use the filesystems
    that were mounted before the suspend. In particular, you MUST NOT
    MOUNT any journaled filesystems mounted before the suspend or they
    will get corrupted in a nasty way."

    -- kernel .config Hibernation (aka 'suspend to disk') help text


    From this text I would like to see a 'mkswap -a' done in the
    non-resume boot path to make sure that any prior hibernate is
    invalidated so filesystem corruption cannot happen due to user
    does a:

    s2d -> normal boot -> write filesystem -> shutdown -> resume

    sequence. Dunno if this has been addressed anywhere in the s2d /
    resume scripts or documentation but it looks too scary to let users
    even open the possibility. A 'mkswap -a' is fairly quick and
    harmless in normal boot path to be unnoticeable? Thing is how to
    flag not to do that on the resume boot path?

    At least this issue is not slack-specific. It's a kernel function.

    Quick thought is to mark s2d kernel with say 'linux-2.6.26.5-s2d'
    and test for that trailing '-s2d' and do the mkswap -a if it's missing.

    Safe option -- disabling resume prevents possible filesystem
    corruption at cost of user perhaps losing work-in-progress? Think
    I'd prefer to lose some work rather than the filesystem.

    Or, have I misunderstood the risk to the filesystem here? Got another
    box I might be able to test on too, as it has a little more swap than
    RAM 4GB -- what could go wrong? Though it's split over two spindles.

    Grant.
    --
    http://bugsplatter.id.au/

  9. Re: Suspend/resume - Request for Testing

    On Tue, 07 Oct 2008 20:38:16 +0000, Robby Workman wrote:

    >On 2008-10-06, Robby Workman wrote:
    >
    >> Package here:
    >> http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >> Sources here:
    >> http://rlworkman.net/pkgs/sources/12.1/pm-utils/

    >
    >
    >Update:
    >
    >http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >Sources in same place.


    Sorry can't help with Toshiba lappy -- doesn't resume properly from
    the:
    echo disk > /sys/power/state

    kernel command, so it seems pointless to go further with whatever pm-utils
    does, unless I've missed something? The lappy seems to resume okay except
    the NIC doesn't come back up, yet the dmesg says all was well on resume,
    had debug turned on so it was fairly clear, I put the file here:

    http://bugsplatter.id.au/kernel/boxe....6.26.5-s2d.gz

    in case you're curious.

    Grant.
    --
    http://bugsplatter.id.au/

  10. Re: Suspend/resume - Request for Testing

    Robby Workman wrote:
    > I've just finished packaging pm-utils 1.2.2 for Slackware (unofficially,
    > of course), and I'd like to get some more widespread testing with it.
    > This new release automatically queries the HAL "quirks" database, so in
    > theory, suspend/resume should work on almost all laptops with it.


    [snip]

    > Once it's installed, running "pm-suspend" as root should do a suspend
    > to ram, and "pm-hibernate" should suspend to disk. Note that you
    > must have a swap partition (or file, but partition is easier) which is
    > large enough (2x ram is good here) and the relevant initrd lines and
    > such for this to work. If you're not familiar with all of that, don't
    > test pm-hibernate. [1]

    [snip]

    Hi, great to see these efforts for getting some suspend automation on
    Slackware!

    I tested both pm-suspend and pm-hibernate, and both worked properly (better
    than with the hibernate-script I had installed). This on a Thinkpad t43p.

    Incidentally, I also tested pm-utils with uswsusp (I installed uswsusp, and
    added a text-file with SLEEP_MODULE=uswsusp in /etc/pm/config.d/ ),
    pm-suspend and pm-suspend-hybrid, and these also worked as advertised.

    Regards,

    PJ

  11. Re: Suspend/resume - Request for Testing

    On 2008-10-07, enos76 wrote:
    > Robby Workman wrote:
    >
    >> I've just finished packaging pm-utils 1.2.2 for Slackware (unofficially,
    >> of course), and I'd like to get some more widespread testing with it.
    >> [...]

    >
    > Installs and runs perfectly on my T60 with ATIx1400 with slack 12.1 .
    >
    > It finally solved the problem with the blank console after suspend to
    > RAM. Nor the kernel parameter acpi_sleep=s3_bios nor the vbetool utility
    > helped, but this "pm-suspend --quirks-s3-bios" works great.



    Hrm, I'm surprised that was needed. Can you email the complete "lshal"
    output to me, please? A general description of the hardware would be
    good too (i.e. what you said above, just so I don't forget if the mail
    gets unlinked with this thread) :-)

    Mail address is in headers.

    -RW

  12. Re: Suspend/resume - Request for Testing

    On 2008-10-08, Grant wrote:
    >
    > Okay,I'm reading and learning more about suspend2disk (hibernate),
    > and would like to know if you've considered this little gem:
    >
    > "In principle it does not require ACPI or APM, although for example
    > ACPI will be used for the final steps when it is available. One
    > of the reasons to use software suspend is that the firmware hooks
    > for suspend states like suspend-to-RAM (STR) often don't work very
    > well with Linux.
    >
    > It creates an image which is saved in your active swap. Upon the next
    > boot, pass the 'resume=/dev/swappartition' argument to the kernel to
    > have it detect the saved image, restore memory state from it, and
    > continue to run as before. If you do not want the previous state to
    > be reloaded, then use the 'noresume' kernel command line argument.
    > Note, however, that fsck will be run on your filesystems and you will
    > need to run mkswap against the swap partition used for the suspend.
    >
    > It also works with swap files to a limited extent (for details see
    > ocumentation/power/swsusp-and-swap-files.txt>).
    >
    > Right now you may boot without resuming and resume later but in the
    > meantime you cannot use the swap partition(s)/file(s) involved in
    > suspending. Also in this case you must not use the filesystems
    > that were mounted before the suspend. In particular, you MUST NOT
    > MOUNT any journaled filesystems mounted before the suspend or they
    > will get corrupted in a nasty way."
    >
    > -- kernel .config Hibernation (aka 'suspend to disk') help text
    >
    >
    > From this text I would like to see a 'mkswap -a' done in the
    > non-resume boot path to make sure that any prior hibernate is
    > invalidated so filesystem corruption cannot happen due to user
    > does a:
    >
    > s2d -> normal boot -> write filesystem -> shutdown -> resume
    >
    > sequence. Dunno if this has been addressed anywhere in the s2d /
    > resume scripts or documentation but it looks too scary to let users
    > even open the possibility. A 'mkswap -a' is fairly quick and
    > harmless in normal boot path to be unnoticeable? Thing is how to
    > flag not to do that on the resume boot path?
    >
    > At least this issue is not slack-specific. It's a kernel function.
    >
    > Quick thought is to mark s2d kernel with say 'linux-2.6.26.5-s2d'
    > and test for that trailing '-s2d' and do the mkswap -a if it's missing.
    >
    > Safe option -- disabling resume prevents possible filesystem
    > corruption at cost of user perhaps losing work-in-progress? Think
    > I'd prefer to lose some work rather than the filesystem.
    >
    > Or, have I misunderstood the risk to the filesystem here? Got another
    > box I might be able to test on too, as it has a little more swap than
    > RAM 4GB -- what could go wrong? Though it's split over two spindles.



    IMHO, if a user chooses to boot in a "non-resume" mode, then that user
    damn well better understand what he/she is doing (and risking). As I
    understand it, that situation will never "just happen" (i.e. it will
    not happen accidentally).

    -RW

  13. Re: Suspend/resume - Request for Testing

    On 2008-10-08, Grant wrote:
    > On Tue, 07 Oct 2008 20:38:16 +0000, Robby Workman wrote:
    >
    >>On 2008-10-06, Robby Workman wrote:
    >>
    >>> Package here:
    >>> http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >>> Sources here:
    >>> http://rlworkman.net/pkgs/sources/12.1/pm-utils/

    >>
    >>
    >>Update:
    >>
    >>http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >>Sources in same place.

    >
    > Sorry can't help with Toshiba lappy -- doesn't resume properly from
    > the:
    > echo disk > /sys/power/state
    >
    > kernel command, so it seems pointless to go further with whatever pm-utils
    > does, unless I've missed something? The lappy seems to resume okay except
    > the NIC doesn't come back up, yet the dmesg says all was well on resume,
    > had debug turned on so it was fairly clear, I put the file here:
    >
    > http://bugsplatter.id.au/kernel/boxe....6.26.5-s2d.gz


    This might give you some ideas if you're so inclined:
    http://people.freedesktop.org/~hughs...end-index.html

    -RW

  14. Re: Suspend/resume - Request for Testing

    On 2008-10-08, PJ Beers wrote:
    >
    > Hi, great to see these efforts for getting some suspend automation on
    > Slackware!



    Well, I can't promise that it will ever reach a state where I can feel
    confident recommending it for inclusion, but here's hoping. Even if it
    doesn't, I plan to maintain unofficial packages until/unless something
    better comes along.


    > I tested both pm-suspend and pm-hibernate, and both worked properly (better
    > than with the hibernate-script I had installed). This on a Thinkpad t43p.



    For clarification, you did not have to pass any extra quirks on the
    command line (IOW, you used only "pm-suspend" or "pm-hibernate") ?

    The reason I ask is: I've already gotten one report where running with
    no arguments didn't quite work out, but specifying additional quirks
    on the command line worked fine. I've got a patch to hal-info queued
    for that one already.

    -RW

  15. Re: Suspend/resume - Request for Testing

    On Wed, 08 Oct 2008 20:44:51 +0000, Robby Workman wrote:

    >On 2008-10-08, Grant wrote:

    ....
    >> Right now you may boot without resuming and resume later but in the
    >> meantime you cannot use the swap partition(s)/file(s) involved in
    >> suspending. Also in this case you must not use the filesystems
    >> that were mounted before the suspend. In particular, you MUST NOT
    >> MOUNT any journaled filesystems mounted before the suspend or they
    >> will get corrupted in a nasty way."
    >>
    >> -- kernel .config Hibernation (aka 'suspend to disk') help text

    ....
    >> Or, have I misunderstood the risk to the filesystem here? Got another
    >> box I might be able to test on too, as it has a little more swap than
    >> RAM 4GB -- what could go wrong? Though it's split over two spindles.

    >
    >
    >IMHO, if a user chooses to boot in a "non-resume" mode, then that user
    >damn well better understand what he/she is doing (and risking). As I
    >understand it, that situation will never "just happen" (i.e. it will
    >not happen accidentally).


    Okay -- it gave me quite a scare on first reading To be honest I'm not
    a fan of suspend or standby, it is not reliable on windoze (due to improper
    network handling) and I've seen too many issues over the years thrashed out
    on LKML.

    As far as how easy it is to hose a filesystem, I agree if you bootup into
    a default lilo stanza with the resume clause there should be no issue at
    all. But I've seen these multi-boot options in other distros and it leaves
    me wondering how safe they are in their boot menu options.

    I suppose slackers are safer because it requires some thought to add this
    option to lilo.

    Grant.
    --
    http://bugsplatter.id.au/

  16. Re: Suspend/resume - Request for Testing

    On 2008-10-07 at 22:15 ADT, Grant wrote:
    > On Tue, 07 Oct 2008 20:38:16 +0000, Robby Workman wrote:
    >
    >>On 2008-10-06, Robby Workman wrote:
    >>
    >>> Package here:
    >>> http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >>> Sources here:
    >>> http://rlworkman.net/pkgs/sources/12.1/pm-utils/

    >>
    >>
    >>Update:
    >>
    >>http://rlworkman.net/pkgs/12.1/TESTI...i486-1_rlw.tgz
    >>Sources in same place.

    >
    > Okay,I'm reading and learning more about suspend2disk (hibernate),
    > and would like to know if you've considered this little gem:
    >
    > "In principle it does not require ACPI or APM, although for example
    > ACPI will be used for the final steps when it is available. One
    > of the reasons to use software suspend is that the firmware hooks
    > for suspend states like suspend-to-RAM (STR) often don't work very
    > well with Linux.


    I wonder how recently that comment has been updated. I've been using
    S2R flawlessly on a variety of laptops (8 or 9 different models) over
    the last few years. I can't imagine that I've been unusually lucky.

    Cheers.
    Jim

  17. Re: Suspend/resume - Request for Testing

    On 2008-10-08, Grant wrote:
    > On Wed, 08 Oct 2008 20:44:51 +0000, Robby Workman wrote:
    >
    >>IMHO, if a user chooses to boot in a "non-resume" mode, then that user
    >>damn well better understand what he/she is doing (and risking). As I
    >>understand it, that situation will never "just happen" (i.e. it will
    >>not happen accidentally).

    >
    > Okay -- it gave me quite a scare on first reading To be honest I'm not
    > a fan of suspend or standby, it is not reliable on windoze (due to improper
    > network handling) and I've seen too many issues over the years thrashed out
    > on LKML.



    Yeah, and while it's come a *long* way over the years, it's obviously
    still not (and likely never will be) perfect. I've spent a bit of time
    lately wading through hal-info source (creating patches in response to
    feedback on all of this - adding new machine signatures and default
    quirks and such), and it's a *lot* of information. There's not any
    "standard" way to implement suspend/resume at a low level, and while
    part of me wishes there *were* standards in place, my experience with
    "standards" says that the people who write them lose sight of the fact
    that they have to *implemented* in the real world, so maybe that's not
    such a good idea either.

    All that being said, I think the best we can hope for at this point
    is to continue merging known-good quirks for machine signatures into
    hal-info and hope that eventually the hardware manufacturers settle
    into designs that either eliminate the need for quirks or make it so
    that entire product lines (e.g. all Thinkpad T6x series) use the same
    set of quirks. In some cases, this seems to be improving, but in
    others, it seems to be getting worse...


    > As far as how easy it is to hose a filesystem, I agree if you bootup into
    > a default lilo stanza with the resume clause there should be no issue at
    > all. But I've seen these multi-boot options in other distros and it leaves
    > me wondering how safe they are in their boot menu options.



    That seems to be the intent of the 00grub hook in pm-utils.


    > I suppose slackers are safer because it requires some thought to add this
    > option to lilo.



    Indeed. There's only so far you can go in "braindead-proofing" before
    things get unnecessarily complex.

    -RW

  18. Re: Suspend/resume - Request for Testing

    I've been using the TuxonIce stuff (formerly suspend2) with
    varying degrees of success on my Dell D820. Real catastrophies
    on a regular basis with Slack 12.1 and the 2.6.24.5 kernel;
    things *much* better with the 2.6.26.1 kernel.

    So - couple of questions: what is the relationship between
    your pm-utils and the TuxonIce stuff (in terms of safety,
    functionality, etc)? How do I "bind" the pm-utilities
    to the Klaptop buttons in KDE (if possible)?

    I've gone through a fair amount of trauma messing around
    with suspend/hibernate with this machine; my ldconfig
    mystery (eventually "solved" by blasting everything to
    bedrock, repartitioning and reinstalling Slack 12.1)
    occurred after a post-hibernation meltdown with the
    2.6.24.5 kernel.


    Steve

  19. Re: Suspend/resume - Request for Testing

    On Thu, 09 Oct 2008 08:03:54 -0500, Stephen Bloom wrote:


    > I've gone through a fair amount of trauma messing around with
    > suspend/hibernate with this machine; my ldconfig mystery (eventually
    > "solved" by blasting everything to bedrock, repartitioning and
    > reinstalling Slack 12.1) occurred after a post-hibernation meltdown with
    > the 2.6.24.5 kernel.


    been there with Mandriva on an R51...a guru in that ng suggested
    disabling write caching IIRC...might be worth a google

  20. Re: Suspend/resume - Request for Testing

    Robby Workman wrote:

    > On 2008-10-08, PJ Beers wrote:
    >>
    >> Hi, great to see these efforts for getting some suspend automation on
    >> Slackware!

    >
    >
    > Well, I can't promise that it will ever reach a state where I can feel
    > confident recommending it for inclusion, but here's hoping. Even if it
    > doesn't, I plan to maintain unofficial packages until/unless something
    > better comes along.


    At least the pm-utils framework seems more widespread and simple to maintain
    than the hibernate-script or the laptop-mode-tools. And it would by nice
    not to have to maintain my own packages... ...mean, thanks for the
    good job :-)

    >> I tested both pm-suspend and pm-hibernate, and both worked properly
    >> (better than with the hibernate-script I had installed). This on a
    >> Thinkpad t43p.

    >
    >
    > For clarification, you did not have to pass any extra quirks on the
    > command line (IOW, you used only "pm-suspend" or "pm-hibernate") ?


    No quirks needed in any of the tests. To be precise: I tested

    # pm-suspend

    and

    # pm-hibernate

    and both worked out of the box with your package, without quirks.

    Then, for good measure, I also tested pm-utils with SLEEP_MODULE=uswsusp:

    # pm-suspend-hybrid

    and

    # pm-suspend

    and both worked flawlessly (I did not bother to test pm-hibernate with
    uswsusp).

    PJ

+ Reply to Thread