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

USB storage slowed down on purpose?

Expand Messages
  • dtaub001
    I was moving some big files to an external 250GB drive yesterday, and noticed that klogd and syslogd were taking up a fair amount of cpu (40-ish percent). It
    Message 1 of 12 , Oct 11, 2004
    View Source
    • 0 Attachment
      I was moving some big files to an external 250GB drive yesterday, and
      noticed that klogd and syslogd were taking up a fair amount of cpu
      (40-ish percent).

      It looks like some debug messages were left in or put into the
      usb-storage driver. A "cat /proc/kmsg" will produce a flood of output
      while the external device is being accessed (something about "switchin").

      This driver is compiled into the kernel (not loadable), so I'll see if
      I can get a new kernel compiled with the kurobox tools...

      Derek
    • dtaub001
      Found the source that does this in drivers/usb/storage/transport.c: /* did we send all the data? */ if (partial == length) {
      Message 2 of 12 , Oct 11, 2004
      View Source
      • 0 Attachment
        Found the source that does this in drivers/usb/storage/transport.c:

        /* did we send all the data? */
        if (partial == length) {
        US_DEBUGP("usb_stor_transfer_partial(): transfer
        complete\n");
        printk(KERN_DEBUG USB_STORAGE "task-switchin\n");
        return US_BULK_TRANSFER_GOOD;
        }

        That printk sucks up a lot of cpu...

        Derek


        --- In LinkStation_General@yahoogroups.com, "dtaub001" <taubert@g...>
        wrote:
        >
        >
        > I was moving some big files to an external 250GB drive yesterday, and
        > noticed that klogd and syslogd were taking up a fair amount of cpu
        > (40-ish percent).
        >
        > It looks like some debug messages were left in or put into the
        > usb-storage driver. A "cat /proc/kmsg" will produce a flood of output
        > while the external device is being accessed (something about
        "switchin").
        >
        > This driver is compiled into the kernel (not loadable), so I'll see if
        > I can get a new kernel compiled with the kurobox tools...
        >
        > Derek
      • James Ronald
        Derek, Cool, good work! Keep us posted or even post a rough how-to in the files section until we get something better setup. I just ordered a 250GB
        Message 3 of 12 , Oct 11, 2004
        View Source
        • 0 Attachment
          Derek,

          Cool, good work! Keep us posted or even post a rough how-to in the files
          section until we get something better setup. I just ordered a 250GB
          DriveStation to use with my Linkstation today so I'm interested. Although,
          I'm probably going swap the 120GB in my LinkStation for the 250GB drive in
          the DriveStation first.

          JR

          ----- Original Message -----
          From: "dtaub001" <taubert@...>
          To: <LinkStation_General@yahoogroups.com>
          Sent: Monday, October 11, 2004 6:43 PM
          Subject: [LinkStation_General] Re: USB storage slowed down on purpose?


          >
          >
          > Found the source that does this in drivers/usb/storage/transport.c:
          >
          > /* did we send all the data? */
          > if (partial == length) {
          > US_DEBUGP("usb_stor_transfer_partial(): transfer
          > complete\n");
          > printk(KERN_DEBUG USB_STORAGE "task-switchin\n");
          > return US_BULK_TRANSFER_GOOD;
          > }
          >
          > That printk sucks up a lot of cpu...
          >
          > Derek
          >
          >
          > --- In LinkStation_General@yahoogroups.com, "dtaub001" <taubert@g...>
          > wrote:
          > >
          > >
          > > I was moving some big files to an external 250GB drive yesterday, and
          > > noticed that klogd and syslogd were taking up a fair amount of cpu
          > > (40-ish percent).
          > >
          > > It looks like some debug messages were left in or put into the
          > > usb-storage driver. A "cat /proc/kmsg" will produce a flood of output
          > > while the external device is being accessed (something about
          > "switchin").
          > >
          > > This driver is compiled into the kernel (not loadable), so I'll see if
          > > I can get a new kernel compiled with the kurobox tools...
          > >
          > > Derek
          >
          >
          >
          >
          >
          >
          >
          > Yahoo! Groups Links
          >
          >
          >
          >
          >
          >
          >
        • mphoust2
          Is there any other way to disable this CPU-hog without a new kernal? I just don t think I want to replace the kernal right now :). ... files ... Although, ...
          Message 4 of 12 , Oct 14, 2004
          View Source
          • 0 Attachment
            Is there any other way to disable this CPU-hog without a new kernal?
            I just don't think I want to replace the kernal right now :).




            --- In LinkStation_General@yahoogroups.com, "James Ronald"
            <jronald@c...> wrote:
            > Derek,
            >
            > Cool, good work! Keep us posted or even post a rough how-to in the
            files
            > section until we get something better setup. I just ordered a 250GB
            > DriveStation to use with my Linkstation today so I'm interested.
            Although,
            > I'm probably going swap the 120GB in my LinkStation for the 250GB
            drive in
            > the DriveStation first.
            >
            > JR
            >
            > ----- Original Message -----
            > From: "dtaub001" <taubert@g...>
            > To: <LinkStation_General@yahoogroups.com>
            > Sent: Monday, October 11, 2004 6:43 PM
            > Subject: [LinkStation_General] Re: USB storage slowed down on purpose?
            >
            >
            > >
            > >
            > > Found the source that does this in drivers/usb/storage/transport.c:
            > >
            > > /* did we send all the data? */
            > > if (partial == length) {
            > > US_DEBUGP("usb_stor_transfer_partial(): transfer
            > > complete\n");
            > > printk(KERN_DEBUG USB_STORAGE "task-switchin\n");
            > > return US_BULK_TRANSFER_GOOD;
            > > }
            > >
            > > That printk sucks up a lot of cpu...
            > >
            > > Derek
            > >
            > >
            > > --- In LinkStation_General@yahoogroups.com, "dtaub001" <taubert@g...>
            > > wrote:
            > > >
            > > >
            > > > I was moving some big files to an external 250GB drive
            yesterday, and
            > > > noticed that klogd and syslogd were taking up a fair amount of cpu
            > > > (40-ish percent).
            > > >
            > > > It looks like some debug messages were left in or put into the
            > > > usb-storage driver. A "cat /proc/kmsg" will produce a flood of
            output
            > > > while the external device is being accessed (something about
            > > "switchin").
            > > >
            > > > This driver is compiled into the kernel (not loadable), so I'll
            see if
            > > > I can get a new kernel compiled with the kurobox tools...
            > > >
            > > > Derek
            > >
            > >
            > >
            > >
            > >
            > >
            > >
            > > Yahoo! Groups Links
            > >
            > >
            > >
            > >
            > >
            > >
            > >
          • Rick Flower
            Nope.. The kernel is where that offending line of code is.. I just disabled it last night (but haven t had a chance to build a new kernel yet) -- Rick
            Message 5 of 12 , Oct 14, 2004
            View Source
            • 0 Attachment
              Nope.. The kernel is where that offending line of code is.. I just
              disabled it last night (but haven't had a chance to
              build a new kernel yet)

              -- Rick

              On Oct 14, 2004, at 4:15 AM, mphoust2 wrote:

              >
              >
              > Is there any other way to disable this CPU-hog without a new kernal?
              > I just don't think I want to replace the kernal right now :).
              >
              >
              >
              >
              > --- In LinkStation_General@yahoogroups.com, "James Ronald"
              > <jronald@c...> wrote:
              >> Derek,
              >>
              >> Cool, good work! Keep us posted or even post a rough how-to in the
              > files
              >> section until we get something better setup. I just ordered a 250GB
              >> DriveStation to use with my Linkstation today so I'm interested.
              > Although,
              >> I'm probably going swap the 120GB in my LinkStation for the 250GB
              > drive in
              >> the DriveStation first.
              >>
              >> JR
              >>
              >> ----- Original Message -----
              >> From: "dtaub001" <taubert@g...>
              >> To: <LinkStation_General@yahoogroups.com>
              >> Sent: Monday, October 11, 2004 6:43 PM
              >> Subject: [LinkStation_General] Re: USB storage slowed down on purpose?
              >>
              >>
              >>>
              >>>
              >>> Found the source that does this in drivers/usb/storage/transport.c:
              >>>
              >>> /* did we send all the data? */
              >>> if (partial == length) {
              >>> US_DEBUGP("usb_stor_transfer_partial(): transfer
              >>> complete\n");
              >>> printk(KERN_DEBUG USB_STORAGE "task-switchin\n");
              >>> return US_BULK_TRANSFER_GOOD;
              >>> }
              >>>
              >>> That printk sucks up a lot of cpu...
              >>>
              >>> Derek
              >>>
              >>>
              >>> --- In LinkStation_General@yahoogroups.com, "dtaub001" <taubert@g...>
              >>> wrote:
              >>>>
              >>>>
              >>>> I was moving some big files to an external 250GB drive
              > yesterday, and
              >>>> noticed that klogd and syslogd were taking up a fair amount of cpu
              >>>> (40-ish percent).
              >>>>
              >>>> It looks like some debug messages were left in or put into the
              >>>> usb-storage driver. A "cat /proc/kmsg" will produce a flood of
              > output
              >>>> while the external device is being accessed (something about
              >>> "switchin").
              >>>>
              >>>> This driver is compiled into the kernel (not loadable), so I'll
              > see if
              >>>> I can get a new kernel compiled with the kurobox tools...
              >>>>
              >>>> Derek
              >>>
              >>>
              >>>
              >>>
              >>>
              >>>
              >>>
              >>> Yahoo! Groups Links
              >>>
              >>>
              >>>
              >>>
              >>>
              >>>
              >>>
              >
              >
              >
              >
              >
              >
              >
              > Yahoo! Groups Links
              >
              >
              >
              >
              >
              >
            • tad94564
              ... and ... output ... switchin ). ... if ... After suffering from this problem last night, I found one possible cause of the poor performance. There may be
              Message 6 of 12 , Oct 29, 2004
              View Source
              • 0 Attachment
                --- In LinkStation_General@yahoogroups.com, "dtaub001" <taubert@g...>
                wrote:
                >
                >
                > I was moving some big files to an external 250GB drive yesterday,
                and
                > noticed that klogd and syslogd were taking up a fair amount of cpu
                > (40-ish percent).
                >
                > It looks like some debug messages were left in or put into the
                > usb-storage driver. A "cat /proc/kmsg" will produce a flood of
                output
                > while the external device is being accessed (something about
                "switchin").
                >
                > This driver is compiled into the kernel (not loadable), so I'll see
                if
                > I can get a new kernel compiled with the kurobox tools...
                >

                After suffering from this problem last night, I found one possible
                cause of the poor performance. There may be more, but this one is
                HUGE. Too bad just compiling a new kernel will not fix this problem.
                Someone added a hardcoded printk to the usb-storage subsystem that is
                not in the kernel-2.4 sources I have.

                Setting /proc/sys/kernel/printk to a higher value doesn't stop it
                either.

                To fix this, you have to get the HD-LAN_GPL source, along with
                installation of the kuro box stuff, break apart the zip file, untar
                the kernel source, goto drivers/usb/storage, edit the file
                transport.c (do a search for 'task-switch', around line 531 in my
                sources), and DELETE that line, build a new kernel/ramdisk, and then
                install.. and pray that it comes back.

                thomas
              • Rick Flower
                I ve already done this on my updated kernel with no ill effects yet.. Of course, I ve not actually tried using the USB yet, so I can t really vouch for whether
                Message 7 of 12 , Oct 29, 2004
                View Source
                • 0 Attachment
                  I've already done this on my updated kernel with no ill effects yet..
                  Of course, I've not actually tried using the USB yet, so I can't really
                  vouch for whether or not my changes solve the problem or not.

                  -- Rick


                  On Fri, 29 Oct 2004, tad94564 wrote:

                  >
                  >
                  > --- In LinkStation_General@yahoogroups.com, "dtaub001" <taubert@g...>
                  > wrote:
                  > >
                  > >
                  > > I was moving some big files to an external 250GB drive yesterday,
                  > and
                  > > noticed that klogd and syslogd were taking up a fair amount of cpu
                  > > (40-ish percent).
                  > >
                  > > It looks like some debug messages were left in or put into the
                  > > usb-storage driver. A "cat /proc/kmsg" will produce a flood of
                  > output
                  > > while the external device is being accessed (something about
                  > "switchin").
                  > >
                  > > This driver is compiled into the kernel (not loadable), so I'll see
                  > if
                  > > I can get a new kernel compiled with the kurobox tools...
                  > >
                  >
                  > After suffering from this problem last night, I found one possible
                  > cause of the poor performance. There may be more, but this one is
                  > HUGE. Too bad just compiling a new kernel will not fix this problem.
                  > Someone added a hardcoded printk to the usb-storage subsystem that is
                  > not in the kernel-2.4 sources I have.
                  >
                  > Setting /proc/sys/kernel/printk to a higher value doesn't stop it
                  > either.
                  >
                  > To fix this, you have to get the HD-LAN_GPL source, along with
                  > installation of the kuro box stuff, break apart the zip file, untar
                  > the kernel source, goto drivers/usb/storage, edit the file
                  > transport.c (do a search for 'task-switch', around line 531 in my
                  > sources), and DELETE that line, build a new kernel/ramdisk, and then
                  > install.. and pray that it comes back.
                  >
                  > thomas
                  >
                  >
                  >
                  >
                  >
                  >
                  >
                  >
                  > Yahoo! Groups Links
                  >
                  >
                  >
                  >
                  >
                  >
                  >
                • tad94564
                  It s probably still in there if your kernel is based on the HD-LAN_GPL source. I ve recieved word this fix will be in the next firmware release, which is
                  Message 8 of 12 , Oct 29, 2004
                  View Source
                  • 0 Attachment
                    It's probably still in there if your kernel is based on the
                    HD-LAN_GPL source.

                    I've recieved word this fix will be in the next firmware release,
                    which is supposed to be RSN.

                    thomas

                    --- In LinkStation_General@yahoogroups.com, Rick Flower <rickf@c...>
                    wrote:
                    >
                    > I've already done this on my updated kernel with no ill effects
                    yet..
                    > Of course, I've not actually tried using the USB yet, so I can't
                    really
                    > vouch for whether or not my changes solve the problem or not.
                    >
                    > -- Rick
                    >
                    >
                    > On Fri, 29 Oct 2004, tad94564 wrote:
                    >
                    > >
                    > >
                    > > --- In LinkStation_General@yahoogroups.com, "dtaub001"
                    <taubert@g...>
                    > > wrote:
                    > > >
                    > > >
                    > > > I was moving some big files to an external 250GB drive
                    yesterday,
                    > > and
                    > > > noticed that klogd and syslogd were taking up a fair amount of
                    cpu
                    > > > (40-ish percent).
                    > > >
                    > > > It looks like some debug messages were left in or put into the
                    > > > usb-storage driver. A "cat /proc/kmsg" will produce a flood
                    of
                    > > output
                    > > > while the external device is being accessed (something about
                    > > "switchin").
                    > > >
                    > > > This driver is compiled into the kernel (not loadable), so
                    I'll see
                    > > if
                    > > > I can get a new kernel compiled with the kurobox tools...
                    > > >
                    > >
                    > > After suffering from this problem last night, I found one
                    possible
                    > > cause of the poor performance. There may be more, but this one
                    is
                    > > HUGE. Too bad just compiling a new kernel will not fix this
                    problem.
                    > > Someone added a hardcoded printk to the usb-storage subsystem
                    that is
                    > > not in the kernel-2.4 sources I have.
                    > >
                    > > Setting /proc/sys/kernel/printk to a higher value doesn't stop
                    it
                    > > either.
                    > >
                    > > To fix this, you have to get the HD-LAN_GPL source, along with
                    > > installation of the kuro box stuff, break apart the zip file,
                    untar
                    > > the kernel source, goto drivers/usb/storage, edit the file
                    > > transport.c (do a search for 'task-switch', around line 531 in
                    my
                    > > sources), and DELETE that line, build a new kernel/ramdisk, and
                    then
                    > > install.. and pray that it comes back.
                    > >
                    > > thomas
                    > >
                    > >
                    > >
                    > >
                    > >
                    > >
                    > >
                    > >
                    > > Yahoo! Groups Links
                    > >
                    > >
                    > >
                    > >
                    > >
                    > >
                    > >
                  • James Ronald
                    What does RSN stand for?
                    Message 9 of 12 , Oct 29, 2004
                    View Source
                    • 0 Attachment
                      What does RSN stand for?

                      >
                      >
                      > It's probably still in there if your kernel is
                      > based on the
                      > HD-LAN_GPL source.
                      >
                      > I've recieved word this fix will be in the next
                      > firmware release,
                      > which is supposed to be RSN.
                      >
                      > thomas
                      >
                      > --- In LinkStation_General@yahoogroups.com, Rick
                      > Flower <rickf@c...>
                      > wrote:
                      >>
                      >> I've already done this on my updated kernel
                      >> with no ill effects
                      > yet..
                      >> Of course, I've not actually tried using the
                      >> USB yet, so I can't
                      > really
                      >> vouch for whether or not my changes solve the
                      >> problem or not.
                      >>
                      >> -- Rick
                      >>
                      >>
                      >> On Fri, 29 Oct 2004, tad94564 wrote:
                      >>
                      >> >
                      >> >
                      >> > --- In LinkStation_General@yahoogroups.com,
                      >> > "dtaub001"
                      > <taubert@g...>
                      >> > wrote:
                      >> > >
                      >> > >
                      >> > > I was moving some big files to an external
                      >> > > 250GB drive
                      > yesterday,
                      >> > and
                      >> > > noticed that klogd and syslogd were taking
                      >> > > up a fair amount of
                      > cpu
                      >> > > (40-ish percent).
                      >> > >
                      >> > > It looks like some debug messages were left
                      >> > > in or put into the
                      >> > > usb-storage driver. A "cat /proc/kmsg"
                      >> > > will produce a flood
                      > of
                      >> > output
                      >> > > while the external device is being accessed
                      >> > > (something about
                      >> > "switchin").
                      >> > >
                      >> > > This driver is compiled into the kernel
                      >> > > (not loadable), so
                      > I'll see
                      >> > if
                      >> > > I can get a new kernel compiled with the
                      >> > > kurobox tools...
                      >> > >
                      >> >
                      >> > After suffering from this problem last night,
                      >> > I found one
                      > possible
                      >> > cause of the poor performance. There may be
                      >> > more, but this one
                      > is
                      >> > HUGE. Too bad just compiling a new kernel
                      >> > will not fix this
                      > problem.
                      >> > Someone added a hardcoded printk to the
                      >> > usb-storage subsystem
                      > that is
                      >> > not in the kernel-2.4 sources I have.
                      >> >
                      >> > Setting /proc/sys/kernel/printk to a higher
                      >> > value doesn't stop
                      > it
                      >> > either.
                      >> >
                      >> > To fix this, you have to get the HD-LAN_GPL
                      >> > source, along with
                      >> > installation of the kuro box stuff, break
                      >> > apart the zip file,
                      > untar
                      >> > the kernel source, goto drivers/usb/storage,
                      >> > edit the file
                      >> > transport.c (do a search for 'task-switch',
                      >> > around line 531 in
                      > my
                      >> > sources), and DELETE that line, build a new
                      >> > kernel/ramdisk, and
                      > then
                      >> > install.. and pray that it comes back.
                      >> >
                      >> > thomas
                    • Rick Flower
                      Unless there was another issue of the same type, that problem should be solved in my kernel.. I went in and removed the kprintf call or whatever it was. As
                      Message 10 of 12 , Oct 29, 2004
                      View Source
                      • 0 Attachment
                        Unless there was another issue of the same type, that problem should
                        be solved in my kernel.. I went in and removed the kprintf call or
                        whatever it was. As for the kernel update, I'll probably not be doing
                        any firmware updates anymore since my system isn't exactly stock anymore.

                        -- Rick

                        On Fri, 29 Oct 2004, tad94564 wrote:

                        >
                        >
                        > It's probably still in there if your kernel is based on the
                        > HD-LAN_GPL source.
                        >
                        > I've recieved word this fix will be in the next firmware release,
                        > which is supposed to be RSN.
                        >
                        > thomas
                        >
                        > --- In LinkStation_General@yahoogroups.com, Rick Flower <rickf@c...>
                        > wrote:
                        >>
                        >> I've already done this on my updated kernel with no ill effects
                        > yet..
                        >> Of course, I've not actually tried using the USB yet, so I can't
                        > really
                        >> vouch for whether or not my changes solve the problem or not.
                        >>
                        >> -- Rick
                        >>
                        >>
                        >> On Fri, 29 Oct 2004, tad94564 wrote:
                        >>
                        >>>
                        >>>
                        >>> --- In LinkStation_General@yahoogroups.com, "dtaub001"
                        > <taubert@g...>
                        >>> wrote:
                        >>>>
                        >>>>
                        >>>> I was moving some big files to an external 250GB drive
                        > yesterday,
                        >>> and
                        >>>> noticed that klogd and syslogd were taking up a fair amount of
                        > cpu
                        >>>> (40-ish percent).
                        >>>>
                        >>>> It looks like some debug messages were left in or put into the
                        >>>> usb-storage driver. A "cat /proc/kmsg" will produce a flood
                        > of
                        >>> output
                        >>>> while the external device is being accessed (something about
                        >>> "switchin").
                        >>>>
                        >>>> This driver is compiled into the kernel (not loadable), so
                        > I'll see
                        >>> if
                        >>>> I can get a new kernel compiled with the kurobox tools...
                        >>>>
                        >>>
                        >>> After suffering from this problem last night, I found one
                        > possible
                        >>> cause of the poor performance. There may be more, but this one
                        > is
                        >>> HUGE. Too bad just compiling a new kernel will not fix this
                        > problem.
                        >>> Someone added a hardcoded printk to the usb-storage subsystem
                        > that is
                        >>> not in the kernel-2.4 sources I have.
                        >>>
                        >>> Setting /proc/sys/kernel/printk to a higher value doesn't stop
                        > it
                        >>> either.
                        >>>
                        >>> To fix this, you have to get the HD-LAN_GPL source, along with
                        >>> installation of the kuro box stuff, break apart the zip file,
                        > untar
                        >>> the kernel source, goto drivers/usb/storage, edit the file
                        >>> transport.c (do a search for 'task-switch', around line 531 in
                        > my
                        >>> sources), and DELETE that line, build a new kernel/ramdisk, and
                        > then
                        >>> install.. and pray that it comes back.
                        >>>
                        >>> thomas
                        >>>
                        >>>
                        >>>
                        >>>
                        >>>
                        >>>
                        >>>
                        >>>
                        >>> Yahoo! Groups Links
                        >>>
                        >>>
                        >>>
                        >>>
                        >>>
                        >>>
                        >>>
                        >
                        >
                        >
                        >
                        >
                        >
                        >
                        > Yahoo! Groups Links
                        >
                        >
                        >
                        >
                        >
                        >
                        >
                      • Roy M. Silvernail
                        ... Real Soon Now. Not sure of the origin. I think I first saw it in Pournelle s Byte column. -- Roy M. Silvernail is roy@rant-central.com, and you re not
                        Message 11 of 12 , Oct 29, 2004
                        View Source
                        • 0 Attachment
                          On Fri, 2004-10-29 at 15:25 -0400, James Ronald wrote:
                          > What does RSN stand for?

                          Real Soon Now. Not sure of the origin. I think I first saw it in
                          Pournelle's Byte column.
                          --
                          Roy M. Silvernail is roy@..., and you're not
                          "It's just this little chromium switch, here." - TFS
                          SpamAssassin->procmail->/dev/null->bliss
                          http://www.rant-central.com
                        • sneed_siding
                          This issue does appear to be fixed with firmware 1.45. Yay!
                          Message 12 of 12 , Jan 2, 2005
                          View Source
                          • 0 Attachment
                            This issue does appear to be fixed with firmware 1.45. Yay!


                            --- In LinkStation_General@yahoogroups.com, "tad94564" <tdavis@b...>
                            wrote:
                            >
                            > It's probably still in there if your kernel is based on the
                            > HD-LAN_GPL source.
                            >
                            > I've recieved word this fix will be in the next firmware release,
                            > which is supposed to be RSN.
                            >
                            > thomas
                          Your message has been successfully submitted and would be delivered to recipients shortly.