Re: [PATCH] libparted: remove now-worse-than-useless _kernel_reread_part_table

[Curtis Gedak]

Jim Meyering wrote:
> Does anyone know of a reason to retain this function?
> Currently it is producing unwarranted failures,
> and removing it causes no problem I can see.
>
> If you know of a case where it's required, please
> outline it in enough detail that I can reproduce it --
> then I'll add a test case for it.  

In my opinion the key objective is to ensure that the kernel is kept 
informed of partition changes.

If this is not possible, then it is important to notify the user when 
the actual disk partition table and the kernel's view of the partition 
table differ.

As long as these objectives are accomplished then I think all is good.

From tracking problems in GParted with resizing file systems when the 
kernel view differs from the actual partition table, I can say that in 
the past the error message was correct when a failure to inform the 
kernel did occur.

I will try running my resizing test using the latest parted code in the 
git repository (May 1st, 2010) and fedora 12 to see if the problem 
reoccurs.  If successful (no problems encountered), this test will take 
several days to complete.