Prev: (none)
Next: Removing dead CONFIG_I2C_PNX0105
From: Yedire, Sandeep on 9 Jun 2010 07:40
Hi ALL,
I am currently using o_nonblock flag for writing a file of 64MB over a
flash device. Flash is formatted with ext3 filesystem with default
options.
When instrumented the driver code for collecting the erase/write count
at sector level. I noticed that o_nonblock has no effect on this
count. There is no difference in erase/write count with o_nonblock
flag or without this flag.
When selected Ext2 as filesystem I noticed a significant difference in
the erase/write count.
There were more updates with o_nonblock and test application was
faster. I could confirm this by checking the Dirty page limit in the
background. It was updating faster in case of Ext2 not incase of Ext3.

Can any one explain on o_nonblock flag in case of Ext3 filesystem?

Regards,
Sandeep.Yedire

----------------------------------------------------------
T.Jefferson, 'Victory and defeat are each of the same price"
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Jeff Moyer on 11 Jun 2010 15:00
"Yedire, Sandeep" <syedire(a)gmail.com> writes:

> Hi ALL,
> I am currently using o_nonblock flag for writing a file of 64MB over a
> flash device. Flash is formatted with ext3 filesystem with default
> options.
> When instrumented the driver code for collecting the erase/write count
> at sector level. I noticed that o_nonblock has no effect on this
> count. There is no difference in erase/write count with o_nonblock
> flag or without this flag.
> When selected Ext2 as filesystem I noticed a significant difference in
> the erase/write count.
> There were more updates with o_nonblock and test application was
> faster. I could confirm this by checking the Dirty page limit in the
> background. It was updating faster in case of Ext2 not incase of Ext3.
>
> Can any one explain on o_nonblock flag in case of Ext3 filesystem?

The O_NONBLOCK flag should be ignored when opening files on a file
system. I'm not sure how to explain your observations.

What exactly were you trying to accomplish?

Cheers,
Jeff
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Yedire, Sandeep on 12 Jun 2010 03:10
On 12 June 2010 00:26, Jeff Moyer <jmoyer(a)redhat.com> wrote:
> "Yedire, Sandeep" <syedire(a)gmail.com> writes:
>
>> Hi ALL,
>> I am currently using o_nonblock flag for writing a file of 64MB over a
>> flash device. Flash is formatted with ext3 filesystem with default
>> options.
>> When instrumented the driver code for collecting the erase/write count
>> at sector level. I noticed that o_nonblock has no effect on this
>> count. There is no difference in erase/write count with o_nonblock
>> flag or without this flag.
>> When selected Ext2 as filesystem I noticed a significant difference in
>> the erase/write count.
>> There were more updates with o_nonblock and test application was
>> faster. I could confirm this by checking the Dirty page limit in the
>> background. It was updating faster in case of Ext2 not incase of Ext3.
>>
>> Can any one explain on o_nonblock flag in case of �Ext3 filesystem?
>
> The O_NONBLOCK flag should be ignored when opening files on a file
> system. �I'm not sure how to explain your observations.
>
> What exactly were you trying to accomplish?
>
> Cheers,
> Jeff
>
--
[Sandeep]
> What exactly were you trying to accomplish?
>
I am only trying to observe the difference with O_NONBLOCK flag in the
updates to filesystem blocks and data blocks in Ext2 and Ext3
filesystem.

It was observed that with O_NONBLOCK flag in Ext2 there were less
updates to Filesystem block at sector 0, and frequent updates to data
blocks. This is not the case with Ext3.

Sandeep
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: (none)
Next: Removing dead CONFIG_I2C_PNX0105