Re: [PATCH] introduce range check for extent pblock references

[Thiemo Nagel <thiemo.nagel@ph.tum.de>]

Aneesh Kumar K.V wrote:
> On Sat, Feb 07, 2009 at 04:36:58PM +0100, Thiemo Nagel wrote:
>> This time I have aimed to catch all cases in which an invalid physical  
>> block might be used and implemented checks directly in ext_pblock() and  
>> idx_pblock() following the assumption that most of the times one of  
>> these functions is called a device access to that address will follow.  
>> If you think this is too heavy, I could also split the check from the  
>> pblock calculation, but in that case I could only guess at which of the  
>> several accesses to *_pblock() in extents.c a check would be necessary  
>> and where it wouldn't and there would be the possibility of missing  
>> something.
> 
> 
> Do we want to check for validity every time we look at the physical
> block of the extent. I guess that would be bad performance wise. I guess
> we should check only once when we read the extent from the disk. ??

Then we need to be careful not to miss a case.  We would need to check 
every time we either a) read from the physical block ourselves or b) 
return the physical block number to a caller outside of ext4.  On the 
other hand I wonder if there are cases where one looks at the physical 
block number which are not a) or b) and which would not benefit from the 
added sanity check?

For repeated accesses to the same physical block number I cannot think 
of a way to bookkeep whether the check already has been done, except if 
that is implicit in the code flow and I don't know how frequent that 
case is.  If you think the performance gain is worth the risk of missing 
a case (either now or during some future change), I can try to rewrite 
the patch to implement the check on a case-by-case basis.

Kind regards,

Thiemo