| 1 |
On Sun, Dec 27, 2009 at 1:46 PM, Paul Hartman |
| 2 |
<paul.hartman+gentoo@×××××.com> wrote: |
| 3 |
> On Sun, Dec 27, 2009 at 12:10 PM, Paul Hartman |
| 4 |
> <paul.hartman+gentoo@×××××.com> wrote: |
| 5 |
>> On Sun, Dec 27, 2009 at 12:01 AM, Michael Holmes |
| 6 |
>> <holmesmich@××××××××××.com> wrote: |
| 7 |
>>> 2009/12/27 Paul Hartman <paul.hartman+gentoo@×××××.com>: |
| 8 |
>>>> On Sat, Dec 26, 2009 at 7:00 PM, Hung Dang <hungptit@×××××.com> wrote: |
| 9 |
>>>>> On 12/26/09 15:54, Paul Hartman wrote: |
| 10 |
>>>>>> Hi, |
| 11 |
>>>>>> |
| 12 |
>>>>>> I got a Nokia N900 linux internet tablet/phone a few days ago, and |
| 13 |
>>>>>> when I connect it in USB Mass Storage mode to a Windows Vista computer |
| 14 |
>>>>>> I can write at 17MB/sec, but when I connect it to my Gentoo box my |
| 15 |
>>>>>> writes are really slow, between 500-900kb/sec depending on if I mount |
| 16 |
>>>>>> in "sync" mode or not. As far as I know it should be just a totally |
| 17 |
>>>>>> standard/generic mass storage device. (there were no drivers or |
| 18 |
>>>>>> software install needed in windows, it just worked) |
| 19 |
>>>>>> |
| 20 |
>>>>>> Other USB devices plugged into the same port go full speed, and AFAIK |
| 21 |
>>>>>> everything appears as if it should be high speed USB 2.0. Has anyone |
| 22 |
>>>>>> seen something like this before? I'm not sure what the deal is. It |
| 23 |
>>>>>> takes 20 minutes to copy 1 gigabyte from Linux and takes just under 1 |
| 24 |
>>>>>> minute to do the same in Windows. |
| 25 |
>>>>>> |
| 26 |
>>>>>> I'm not sure about debugging USB or what the options are. Everything |
| 27 |
>>>>>> I've used previously has worked without any hassle. |
| 28 |
>>>>> Have you received a lot of debugging messages at the output of dmesg |
| 29 |
>>>>> when copying files? |
| 30 |
>>>>> |
| 31 |
>>>>> Hung |
| 32 |
>>>> |
| 33 |
>>>> No errors, no strange messages at all, it seems normal (only slow). I |
| 34 |
>>>> have other USB devices like SD card reader, external HDD, and they |
| 35 |
>>>> perform at full speed when plugged into the same port, so it's weird |
| 36 |
>>>> to me. |
| 37 |
>>>> |
| 38 |
>>>> Thanks |
| 39 |
>>>> Paul |
| 40 |
>>>> |
| 41 |
>>>> |
| 42 |
>>> |
| 43 |
>>> This isn't that helpful, but in Windows I get good thoroughputs in |
| 44 |
>>> mass storage mode, but compartively weak ones in sync mode. But this |
| 45 |
>>> shouldn't be the problem because as far as I know, there are no |
| 46 |
>>> ync-mode drivers for good old Linux (which is ironic if you consider |
| 47 |
>>> it). |
| 48 |
>> |
| 49 |
>> When I said "sync" mode I mean mounting the device in mass storage |
| 50 |
>> mode with with "-o sync" (as opposed to the default cached/async |
| 51 |
>> mode), which I believe is the default in Windows Vista & Win7 (because |
| 52 |
>> most windows users cannot be bothered to unmount before pulling the |
| 53 |
>> plug). |
| 54 |
>> |
| 55 |
>> Thanks |
| 56 |
> |
| 57 |
> Well, after a bit more plugging/unplugging of all my USB devices into |
| 58 |
> various ports in different orders, it seems to be going fast now. I |
| 59 |
> hate USB :) |
| 60 |
> |
| 61 |
> Thanks for the help |
| 62 |
|
| 63 |
Maybe I spoke too soon. It seems what's happening is when I write a |
| 64 |
large amount of data, there are several "pdflush" threads at near 100% |
| 65 |
i/o wait. I'm thinking it's writing multiple streams over USB which is |
| 66 |
causing the massive slow-down. |
| 67 |
|
| 68 |
I'm using 2.6.31, and I see in 2.6.32 there is something called |
| 69 |
"Per-backing-device based writeback" which may help me here... I'll |
| 70 |
try the new kernel and report back :) |
Archives