Re: Parted 1.6.5-pre1

[Veerapuram Varadhan]

> On Mon, Jan 06, 2003 at 03:10:34PM +0530, Veerapuram Varadhan wrote:
>> Hi Andrew,
>>
>> libparted 1.6.5.pre1 is so far so good for those RAID related changes.
>>  However,  i have doubt related to the following code fragment of
>> libparted/linux.c.
>>
>> Suppose if my /proc/partition has entry like rd/c0d0, according to the
>> above code, the pattern string pointed by the variable "pos" will be
>> "/c0d0",
>
> This is the intended behaviour.
>
>> as strrchr returns pointer to the character that it matches in
>> the string from reverse, which, i feel, may be logically wrong, though
>> the code works/will work fine as it uses "isdigit" for the pattern
>> matching.
>
> Why is it logically wrong?
>

Well, according to the "literal logic", we are matching the "device names"
(not including the "/", i hope so) against a specific pattern, that's why.

>> How about adding that line marked by /*========> pos++; <==========*/?
>
> Why?
>

though not necessarily needed as the existing code works fine, just to
reflect the "literal logic".

>> I need one more clarification.  In my /dev/rd directory, i found only
>> 7 device nodes for c0d0 starting from c0d0p1 to c0d0p7.  Why is it so?
>>  Is it a kernel limitation?
>
> Perhaps... I don't know.  What are the major/minor numbers?  (You
> can get this with "ls -l").  You (or I) can check in
> /usr/src/linux/Documentation/devices.txt how many partition numbers are
> allocated.
>

I did that and found only 7 minors for each c0dx.. :(

>> I created 8 partitions using parted.  Parted is able to show all the 8
>> partitions including c0d0p8, but the entry corresponding to this
>> partition was not found in the /proc/partitions.  Can anybody throw
>> some light on this, as to how to solve this?
>
> That sounds bad.  Linux has crazy minor number limitations.  It
> has been the subject of many flamewars, and I think the plan is
> to have it fixed...
>

Will libparted reflect that Linux's limitations?


> Cheers,
> Andrew

Cheers,
V. Varadhan.