grub-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RFT,PATCH] Move embedding to appropriate partmap files


From: Vladimir 'phcoder' Serbinenko
Subject: Re: [RFT,PATCH] Move embedding to appropriate partmap files
Date: Fri, 16 Oct 2009 23:12:52 +0200
User-agent: Mozilla-Thunderbird 2.0.0.22 (X11/20090701)

Robert Millan wrote:
> On Fri, Oct 16, 2009 at 12:58:42PM +0200, Vladimir 'phcoder' Serbinenko wrote:
>   
>> Robert Millan wrote:
>>     
>>> On Thu, Aug 13, 2009 at 10:53:23PM +0200, Vladimir 'phcoder' Serbinenko 
>>> wrote:
>>>   
>>>       
>>>>>> We may want to embed more in the future. Actually I think it's not
>>>>>> ad-hoc. Basically partition map defines a function which gives back
>>>>>> the sectors available for embedding.
>>>>>>         
>>>>>>             
>>>>> Is embedding useful elsewhere?
>>>>>       
>>>>>           
>>>> Yes. Consider a world of checksummed filesystems. In such world you
>>>> can't change the contents of the file by just writing to its blocklist
>>>> since it will break the checksum. Similar problems exist with RAIDs
>>>> and LVMs. On some systems we can't put grub-env in a file. For these
>>>> cases we can embed grub-env somewhere where we can write it with ease
>>>>     
>>>>         
>>> Ok.  Feel free to use partmap/ then.  But please make sure #ifdefs only
>>> enable those functions where they are going to be used.  That'd be
>>> GRUB_MACHINE_PCBIOS for now, if later code in other ports relies on them,
>>> this can be changed.
>>>
>>>   
>>>       
>> This patch fixes an important bug - namely overwriting extended
>> partition tables
>>     
>
> Is there a simpler way to resolve this?  I don't object to the
> restructuring you propose, but it seems too intrusive to do this
> just a few hours before we release 1.97.
>
>   
Yes. It's my concern too. Solving this requires metadata info form
partition map which is normally hidden and new code may have
undiscovered bugs. On the other hand already present code has a bug but
which is unlikely to be triggered.
As a simple alternative one could keep current check + enumerte entries
in MBR - this should be enough.

-- 
Regards
Vladimir 'phcoder' Serbinenko
Personal git repository: http://repo.or.cz/w/grub2/phcoder.git 





reply via email to

[Prev in Thread] Current Thread [Next in Thread]