On Wednesday 22 September 2004 09:29 am, Oliver Neukum wrote:
> > On 2.6.0 /proc/sys/vm/min_free_kbytes = 1024
> > On 2.6.8 /proc/sys/vm/min_free_kbytes = 128
> >
> > Changing 2.6.8 to 1024 appears to resolve this issue.
>
> What else is running? What network connection?
> That would mean that more
Am Mittwoch, 22. September 2004 15:20 schrieb Kyle Harris:
> On Wednesday 22 September 2004 02:59 am, Oliver Neukum wrote:
> > Does it work if you replace the umount with a delay of equal
> > duration? Will sync have the same effect has umount?
>
> On 2.6.0 /proc/sys/vm/min_free_kbytes = 1024
> On
On Wednesday 22 September 2004 02:59 am, Oliver Neukum wrote:
> Does it work if you replace the umount with a delay of equal
> duration? Will sync have the same effect has umount?
On 2.6.0 /proc/sys/vm/min_free_kbytes = 1024
On 2.6.8 /proc/sys/vm/min_free_kbytes = 128
Changing 2.6.8 to 1024 appea
Am Dienstag, 21. September 2004 22:44 schrieb Kyle Harris:
> > I don't think memory usage patterns in the USB subsystem have changed
> > significantly since 2.6.0. The new HCD is definitely the most likely
> > suspect. But not being familiar with either the old or the new one, I
> > can't offer a
On Tuesday 21 September 2004 07:41 pm, Pete Zaitcev wrote:
> Yes. If you don't see it in /proc/meminfo, maybe your kswapd is not
> working. It's not going to free much though, just the size of the atomic
> pool. This reminds me, if you do an embedded system work, you have to
> double check that you
On Tue, 21 Sep 2004 19:13:57 -0400
Kyle Harris <[EMAIL PROTECTED]> wrote:
> If this were the case, should memory recover once the file transfer is
> complete and all USB activity has ceased?
Yes. If you don't see it in /proc/meminfo, maybe your kswapd is not working.
It's not going to free much
On Tuesday 21 September 2004 06:21 pm, Pete Zaitcev wrote:
>
> Don't forget that the atomic pool has to be maintained by kswapd.
> One easy way to deplete it is to starve kswapd out of CPU.
>
> Just to make for excitement, writing to a drive dirties pages,
> and then you need an extra thread to run
On Tue, 21 Sep 2004 17:15:26 -0400
Kyle Harris <[EMAIL PROTECTED]> wrote:
> I have not ported old HCD to 2.6.8. I must run it on 2.6.0. Using old
> HCD/2.6.0, free memory never drops below 2M when looping and writing 1M files
> to usbdrive. If I umount between each write, free memory hovers arou
On Tuesday 21 September 2004 05:05 pm, Alan Stern wrote:
> On Tue, 21 Sep 2004, Kyle Harris wrote:
> > I agree. But this is what makes no sense. I can loop writing a 1 M file
> > to the usbdrive. The system will run out of memory after 6-7 loops. But
> > if I umount the usbdrive between each write
On Tue, 21 Sep 2004, Kyle Harris wrote:
> I agree. But this is what makes no sense. I can loop writing a 1 M file to
> the usbdrive. The system will run out of memory after 6-7 loops. But if I
> umount the usbdrive between each write the system works fine. I don't see how
> performing an umount
On Tuesday 21 September 2004 03:13 pm, Alan Stern wrote:
> On Tue, 21 Sep 2004, Kyle Harris wrote:
> > I tested this on 2.6.0 with old HCD and free memory never drops below
> > about 2M. On 2.6.8.1 with new HCD it drops very low and it appears
> > urb_dequeue is called when this occurs. I wonder wh
On Tue, 21 Sep 2004, Kyle Harris wrote:
> I tested this on 2.6.0 with old HCD and free memory never drops below about
> 2M. On 2.6.8.1 with new HCD it drops very low and it appears urb_dequeue is
> called when this occurs. I wonder what has changed (if it is related to new
> HCD) and what can b
Thank you Alan:)
I'll keep you posted:)
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Alan
Stern
Sent: Tuesday, September 21, 2004 10:47 AM
To: Kyle Harris
Cc: [EMAIL PROTECTED]
Subject: Re: [linux-usb-devel] usb-storage: page allocation failur
On Tuesday 21 September 2004 10:47 am, Alan Stern wrote:
>
> The memory usage that you see while the device is mounted doesn't come
> from usb-storage; it must come from the filesystem code.
>
I tested this on 2.6.0 with old HCD and free memory never drops below about
2M. On 2.6.8.1 with new HCD
On Tue, 21 Sep 2004, Kyle Harris wrote:
> While debugging my new hcd 2.6.8.1 HCD on PXA platform I get the following
> error:
>
> usb-storage: page allocation failure. order:0, mode:0x20
>
> I've checked for memory leaks in the hcd but pretty sure there are none
> (count mallocs and frees). I
While debugging my new hcd 2.6.8.1 HCD on PXA platform I get the following
error:
usb-storage: page allocation failure. order:0, mode:0x20
I've checked for memory leaks in the hcd but pretty sure there are none
(count mallocs and frees). I can copy small files to the usbdrive and see
free memo
16 matches
Mail list logo
