Charles Manning wrote:
> On Monday 01 August 2005 11:04, Luc Van Oostenryck wrote:
> 
>>Here are some other problems that I detect when running some tests:
>>
>>- Trying to change the permission of the root -> crash.
>>   -> problem with SetAttributes() forcing the update of the header
>>      but the header does not exist, the root is a fake dir.
>>   Possible solution:
>>   - do not force the update of the header if it doesn't exists
>>   - don't use a fake header for the root directory so that any change
>>     of permissions made to it are permanent.
>>     Of course, I much prefer this one.
> 
> 
> Hmm this did work at one stage, and there is even special handling code in the 
> scanning for this.
> 
> I'll investigate it.
>

Yes, this is/was working on YAFFS1.

> 
>>- There is a problem related to truncation and rewrite of files but I
>>   have not yet investigated this one.
>>   May be related to not using USE_HEADER_SIZE ???
> 
> 
> If you can make a test case I will investigate further.
> 

I need to narrow it; I run a test (stolen from the FS part of LTP) and it just report a failure
after a:
	truncate
	write
	read
	read
and the content of the test file was different that the reference file.
I stop the test there.
I will know more in a few hours.

> 
>>- The problem related to NULL name in yaffs_FindObjectByName() reported
>>   in a previous mail.
> 
> 
> I have provisionally fixed this in CVS.
> 
> I will investigate and check further as part of some tests I am doing tonight.
> 

OK, thanks.


Luc