[coreboot] lar copy patch

Carl-Daniel Hailfinger c-d.hailfinger.devel.2006 at gmx.net
Fri Mar 14 21:15:01 CET 2008


On 13.03.2008 04:35, Myles Watson wrote:
> Carl-Daniel Hailfinger wrote:
>   
>> Sorry, had I known the greedy match didn't match at path boundaries, I
>> would have NACKed it. Having "normal/in" match "normal/initram/segment0"
>> as well as "normal/in" as well as "normal/interesting" is horrible.
>>     
>
> This is easy to fix.  It was a toss-up to me because there are so few
> entries in a lar that there will be few cases where one would choose to name
> something normal/in and normal/interesting and not want them both out.
>
> I also wondered about having it only match if the nesting wasn't too deep.
> For example normal wouldn't match normal/payload/segment0, but would match
> normal/initram.  I could argue that one either way.
>   

I'd mirror tar behaviour. Could you post a patch which accomplishes
that? (Match at path boundaries regardless of nesting depth.) This is
independent of the LAR copy work, though.


>>>> How does LAR know that you want to copy the bootblock instead of an
>>>> empty operation?
>>>>
>>>>         
>>> Since it is a create operation with a source lar, it copies the
>>>       
>>> bootblock. It made sense to me that if you want a new lar starting
>>>       
>>> with an old one, you were asking for the bootblock.
>>>
>>>       
>> That's an implicit undocumented action. It will trigger a lot of head
>> scratching in the future.
>>     
>
> I could document it with a warning when no bootblock is specified, so that
> it's clear what's happening.
>   

What about using the "-b" flag for the bootblock?


>>>>> I think it would be nice if lar would warn you (or stop you) when you
>>>>> try to add a duplicate entry.  Maybe that will be the next patch.
>>>>>
>>>>> Myles
>>>>>
>>>>> Signed-off-by: Myles Watson <mylesgw at gmail.com>
>>>>>
>>>>>           
>> One thing I didn't check yet: If you specify that compression of a file
>> should be changed, how does this affect the "zeroes" compression?
>>     
>
> It doesn't.  Zeroes is a special case, and doesn't get changed.
>   

I like that.

Regards,
Carl-Daniel

-- 
http://www.hailfinger.org/





More information about the coreboot mailing list