Hello,

Eric W. Biederman wrote:
quoted text
>> More specifically, d_off field.  It's a bit twisted.  For the last
>> entry, filp->f_pos gets written into the field and gets wrapped while
>> being copied out to userland or in glibc.
> 
> That could do it, and glibc is crunching it.  Oh well, it is
> easy enough to avoid as long as our inode numbers are small which
> the idr allocator seems to ensure.

Yeah, now I think about it.  glibc throws out entries which don't fit in
the data structure specified by the called API, so it probably threw out
the last entry which has UINT_MAX in d_off which doesn't fit in the
readdir() return structure.  Using INT_MAX should be just fine as IDA
always allocates the first empty slot.  We can add paranoia check in ino
allocation path.

-- 
tejun
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/