Ethan Quach wrote:
> Evan,
> 
> 355 - Shouldn't this be set to 0 ?

Yes it should. In fact with it set to one if the orig BE is the first entry we 
end up with no title line...

> 
> 386 - Instead of setting num_lines = num_tmp_lines here, initialize it 
> to 0,
> and use (and increment) it as the index to entries[] at lines 392 and 399.

I added the use of num_lines and incremented it as suggested.

> 
> 431, 451 - If entering these if() clauses, you need to check if entries[]
> if populated, and if so, free it. (I would suggest setting a ret code, 
> and then
> jumping down to a cleanup section below.)

I took another look at it and I was doing that clean-up when we found a 
duplicate entry for the new BE. I've moved that down to a clean-up and return 
section similar to what you suggested.

I've retested and updated the webrev.

-evan

> 
> 
> thanks,
> -ethan
> 
> 
> Evan Layton wrote:
>> Hi Ethan,
>>
>> I've corrected these two issues, retested and updated the webrev. The 
>> code now does the collection of the menu entry lines at the point we 
>> detect that we've found the original BE's entry. Once we have that 
>> information we continue through the file making sure we don't also 
>> have a duplicate entry. Finally we collect the menu entry lline and 
>> maintain their order so that the new BE's menu entry will also match 
>> the order of the entry we're copied from.
>>
>> Thanks!
>> -evan
>>
>> Ethan Quach wrote:
>>  
>>> Evan,
>>>
>>> The existing code accounts for the case where if the new BE we're adding
>>> already has an entry in the menu.lst, it does the proper thing.  
>>> However,
>>> with your changes, it might fail to detect that.  For the case where 
>>> the new
>>> BE already has an entry but is below orig_root_ds entry, you no longer
>>> find it.
>>>
>>>
>>> Is it possible for a line to exist be between the 'title' line and 
>>> 'bootfs'?
>>> If so, that line would have been recorded in entries[] and written out
>>> after the 'bootfs' line for the new BE.  I don't know if that would 
>>> cause
>>> a problem but just something to note.
>>>
>>>
>>> thanks,
>>> -ethan
>>>
>>>
>>> Evan Layton wrote:
>>>    
>>>> I need to get a review the changes for
>>>>
>>>> 3654 libbe should keep menu.lst customizations when cloning
>>>> http://defect.opensolaris.org/bz/show_bug.cgi?id=3654
>>>>
>>>> The webrev is available from:
>>>> http://cr.opensolaris.org/~evanl/webrev.3654/
>>>>
>>>> This fix changes the way we build grub menu entries. When we're
>>>> creating a new BE we take a look at the grub entry lines from the
>>>> parent of BE we're creating and we edit the title and bootfs lines
>>>> the rest of the lines are copied directly form the parents entry.
>>>>
>>>> This has been tested using "beadm create" and "beadm activate" which
>>>> call into be_copy() and be_activate() which are the only consumers
>>>> of this functionality. I've also verified that changed entries in
>>>> menu.lst are copied from the parent BE to the entry for the
>>>> clone/copy.
>>>>
>>>> These changes also include some simple lint clean-up.
>>>>
>>>> Thanks,
>>>> -evan
>>>> _______________________________________________
>>>> caiman-discuss mailing list
>>>> caiman-discuss at opensolaris.org
>>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>>>         
>>
>> _______________________________________________
>> caiman-discuss mailing list
>> caiman-discuss at opensolaris.org
>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>   


Reply via email to