On Thu, Feb 07, 2002 at 04:20:36PM +0100, Dave Kleikamp wrote:
> Andi Kleen wrote:
> >JFS sets a s_maxbytes value that is bigger than the page cache limit
> >on 32bit machines. This could lead to silent wrapping when a big file
> >write reaches the page cache limit, corrupting data at the beginning. 
> 
> Yes, this would be a problem.  But shouldn't the upper limit be
> (u64)PAGE_CACHE_SIZE << 32 ?

I did use 31 to be safe against sign extension (better safe than sorry) 

In 2.5.4pre Linus did also merge a patch (other fs had this problem too)
that adds a MAX_LFS_FILESIZE constant for it if you prefer that.

-Andi
_______________________________________________
Jfs-discussion mailing list
[EMAIL PROTECTED]
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion

Reply via email to