[Cluster-devel] [PATCH v2][next] dlm: Replace one-element array with flexible-array member

[Paulo Miguel Almeida <paulo.miguel.almeida.rodenas@gmail.com>]

On Sat, Oct 08, 2022 at 05:18:35PM -0700, Kees Cook wrote:
> >$ diff <(objdump -M intel -j .text -D dlm.old) <(objdump -M intel -j
> >.text -D dlm.new)
> 
> I'd suggest different options here, this is harder to map back to the source line.
> See https://outflux.net/blog/archives/2022/06/24/finding-binary-differences/
> for lots of details. :)
> 

Just read the blog entry, it's really interesting. I will be using it
from now onwards :)

> >
> >13778c13778
> ><     c693:	49 8d bc 24 c0 08 00 	lea    rdi,[r12+0x8c0]
> >---
> >>     c693:	49 8d bc 24 c1 08 00 	lea    rdi,[r12+0x8c1]
> 
> This implies something unexpected changed.
> 

I will add more details about this line at the other point you made
below to avoid repeating myself. But to cut a long story, short.. this
[reg + displacement + 1] difference is caused because I deliberately add
the NUL-terminator space to the kzalloc() call.

> This has trailing padding, so the struct size didn't actually change.
> 
> >-	ls = kzalloc(sizeof(struct dlm_ls) + namelen, GFP_NOFS);
> >+	ls = kzalloc(sizeof(struct dlm_ls) + namelen + 1, GFP_NOFS);
> 
> This is allocating 1 more byte than before, since the struct size didn't change. But this has always allocated too much space, due to the struct padding. For a "no binary changes" patch, the above "+ 1" needs to be left off.

That's true. I agree that leaving "+ 1" would work and produce a
no-binary-changes patch due to the existing padding that the structure
has. OTOH, I thought that relying on that space could bite us in the
future if anyone tweaks the struct again...so my reaction was to ensure 
that the NUL-terminator space was always guaranteed to be there.
Hence, the change on c693 (objdump above).

What do you think? Should we keep or leave the above
"+ 1" after the rationale above?

> 
> I would expect the correct allocation size to be:
> offsetof(typeof(*ls), ls_name) + namelen

Fair point, I will make this change.

> 
> Question, though: is ls_name _expected_ to be %NUL terminated

Yes, it is. I tracked down ls_name's utilisations and it is passed down to 
a bunch of routines that expects it to be NUL-terminated such as
snprintf and vsnprintf.

>, and was the prior 3 bytes of extra allocation accidentally required?
> 

I am assuming that you are refering to ls_namelen in the struct dlm_ls
(please correct me if this isn't what you meant).

ls_namelen member is only used within the routine which kzalloc
the space for the struct (fs/dlm/lockspace.c:new_lockspace). 

There are no external references to this member outside of that method in the
kernel. One could say that ls_namelen can be removed without side effects but 
I wouldn't suggest doing it in this patch, that's why I didn't touch it :)

Paulo A.