Hi Jean,

Thanks for the review.   Comments inline ...


thanks,
-ethan


Jean McCormack wrote:
> be_mount.c:
>     line 348: I'm confused as to why you do the be_get_uuid call. uu  is 
> never used after it and
>        it's not clear there's any side effect. So why do so? If there's 
> a reason a comment would help.
>   

Its required that the BE has a uuid for zones to be supported.  The process
of getting the uuid validates it has one and is in the right form.
I'll add a comment to note this.

>     line 351: Do you need the if(gen_tmp_altroot) free(tmp_altroot); 
> code here?
>   

yes, thanks.

>     line 477: ditto
>   

There's no gen_tmp_altroot in _be_unmount().

>    line 545: Aren't we supposed to use strncmp for security reasons?
>   
>    line 598: ditto
>   

It may be a security issue if one of the strings is a not a null 
terminated string,
but we could still run into that with strncmp can't we?

>    line 645: Shouldn't these really be be_errno_t?
>   
>    line 858  & 861: shouldn't this really be be_errno_t?  (Actually this 
> is going to be a generic comment
>        since I see it many places)
>   

I would rather leave them all as ints.  The definition of the be_errno_t 
is the
enumeration of possible error codes, and the functions return one of them.

>    line 679: Nit (there always has to be at least one nit). Remove the 
> get, it's already on the line above.
>   

okay.

>    line 828- 841: What happens if the strlcpy actually truncates dir?
>   

I suppose that if a truncation happened at exactly a point that left a valid
directory name, the code could lookup up the wrong dataset.

We'll validate the length of dir upon entry.  Thanks.

>    line 851: Is this really none? Maybe something here or in the returns 
> about tmp_mp.
>   

Yes.  I will update this.

>    line 867: should this be strncpy?
>   

We explicitly allocated *tmp_mp at line 863 to be the exact length
required for BE_TMP_MNTPNT, so I think a strcpy here should be
sufficient.

>    line 940 & 942: should this be strncmp?
>   

Same comment as for 545,598

>    line 1447: Huh?
>   

Should be " at the altroot"

>
>    line 1895: do we need to free altroot?
>   

If the call to make_tmp_mountpoint() failed, we don't have
an altroot.

>    line 2175: Maybe file a bug on this?
>   

We will once we implement shared file systems for zones.



thanks,
-ethan

Reply via email to