Re: [PATCH] lsm: cleanup repeated lsm_blob_size_update() calls in lsm_prepare()

On Thu, Jul 2, 2026 at 4:18 PM Matt Bobrowski <[email protected]> wrote:
> On Thu, Jul 02, 2026 at 03:15:02PM -0400, Paul Moore wrote:
> > On Jun 29, 2026 Matt Bobrowski <[email protected]> wrote:
> > >
> > > Centralize the definition of LSM security blob fields using an X-macro
> > > (LSM_BLOBS_LIST). This reduces repetitive boilerplate code across
> > > struct lsm_blob_sizes, blob size registration in lsm_prepare(), and
> > > debug log printing in security_init().
> > >
> > > Signed-off-by: Matt Bobrowski <[email protected]>
> > > ---
> > >  include/linux/lsm_hooks.h | 42 ++++++++++++++++------------
> > >  security/lsm_init.c       | 59 ++++++++++-----------------------------
> > >  2 files changed, 38 insertions(+), 63 deletions(-)
> > >
> > > diff --git a/include/linux/lsm_hooks.h b/include/linux/lsm_hooks.h
> > > index b4f8cad53ddb..0e73b22bdeea 100644
> > > --- a/include/linux/lsm_hooks.h
> > > +++ b/include/linux/lsm_hooks.h
> > > @@ -98,28 +98,34 @@ struct security_hook_list {
> > >     const struct lsm_id *lsmid;
> > >  } __randomize_layout;
> > >
> > > +#define LSM_BLOBS_LIST(X) \
> > > +   X(cred)          \
> > > +   X(file)          \
> > > +   X(backing_file)  \
> > > +   X(ib)            \
> > > +   X(inode)         \
> > > +   X(sock)          \
> > > +   X(superblock)    \
> > > +   X(ipc)           \
> > > +   X(key)           \
> > > +   X(msg_msg)       \
> > > +   X(perf_event)    \
> > > +   X(task)          \
> > > +   X(tun_dev)       \
> > > +   X(xattr_count)   \
> > > +   X(bdev)          \
> > > +   X(bpf_map)       \
> > > +   X(bpf_prog)      \
> > > +   X(bpf_token)
> > > +
> > >  /*
> > >   * Security blob size or offset data.
> > > + * Note: lbs_xattr_count is the number of xattr slots in new_xattrs array.
> > >   */
> > >  struct lsm_blob_sizes {
> > > -   unsigned int lbs_cred;
> > > -   unsigned int lbs_file;
> > > -   unsigned int lbs_backing_file;
> > > -   unsigned int lbs_ib;
> > > -   unsigned int lbs_inode;
> > > -   unsigned int lbs_sock;
> > > -   unsigned int lbs_superblock;
> > > -   unsigned int lbs_ipc;
> > > -   unsigned int lbs_key;
> > > -   unsigned int lbs_msg_msg;
> > > -   unsigned int lbs_perf_event;
> > > -   unsigned int lbs_task;
> > > -   unsigned int lbs_xattr_count; /* num xattr slots in new_xattrs array */
> > > -   unsigned int lbs_tun_dev;
> > > -   unsigned int lbs_bdev;
> > > -   unsigned int lbs_bpf_map;
> > > -   unsigned int lbs_bpf_prog;
> > > -   unsigned int lbs_bpf_token;
> > > +#define LSM_BLOB_SIZE(name) unsigned int lbs_##name;
> > > +   LSM_BLOBS_LIST(LSM_BLOB_SIZE);
> > > +#undef LSM_BLOB_SIZE
> > >  };
> >
> > Generally speaking I greatly prefer to see the structure fields typed out
> > as it makes it easier to find them using grep, code indexers, etc.  There
> > is a similar argument for the security_init() changes (below).
> >
> > > diff --git a/security/lsm_init.c b/security/lsm_init.c
> > > index 7c0fd17f1601..c256f1c33efa 100644
> > > --- a/security/lsm_init.c
> > > +++ b/security/lsm_init.c
> > > @@ -282,40 +282,24 @@ static void __init lsm_blob_size_update(unsigned int *sz_req,
> > >   * lsm_prepare - Prepare the LSM framework for a new LSM
> > >   * @lsm: LSM definition
> > >   */
> > > -static void __init lsm_prepare(struct lsm_info *lsm)
> > > +static void __init lsm_prepare(const struct lsm_info *lsm)
> > >  {
> > >     struct lsm_blob_sizes *blobs = lsm->blobs;
> > >
> > >     if (!blobs)
> > >             return;
> > >
> > > -   /* Register the LSM blob sizes. */
> > > -   blobs = lsm->blobs;
> > > -   lsm_blob_size_update(&blobs->lbs_cred, &blob_sizes.lbs_cred);
> > > -   lsm_blob_size_update(&blobs->lbs_file, &blob_sizes.lbs_file);
> > > -   lsm_blob_size_update(&blobs->lbs_backing_file,
> > > -                        &blob_sizes.lbs_backing_file);
> > > -   lsm_blob_size_update(&blobs->lbs_ib, &blob_sizes.lbs_ib);
> > > -   /* inode blob gets an rcu_head in addition to LSM blobs. */
> > > +   /* The inode blob (inode->i_security) gets an rcu_head in addition to
> > > +    * LSM blobs.
> > > +    */
> > >     if (blobs->lbs_inode && blob_sizes.lbs_inode == 0)
> > >             blob_sizes.lbs_inode = sizeof(struct rcu_head);
> > > -   lsm_blob_size_update(&blobs->lbs_inode, &blob_sizes.lbs_inode);
> > > -   lsm_blob_size_update(&blobs->lbs_ipc, &blob_sizes.lbs_ipc);
> > > -   lsm_blob_size_update(&blobs->lbs_key, &blob_sizes.lbs_key);
> > > -   lsm_blob_size_update(&blobs->lbs_msg_msg, &blob_sizes.lbs_msg_msg);
> > > -   lsm_blob_size_update(&blobs->lbs_perf_event,
> > > -                        &blob_sizes.lbs_perf_event);
> > > -   lsm_blob_size_update(&blobs->lbs_sock, &blob_sizes.lbs_sock);
> > > -   lsm_blob_size_update(&blobs->lbs_superblock,
> > > -                        &blob_sizes.lbs_superblock);
> > > -   lsm_blob_size_update(&blobs->lbs_task, &blob_sizes.lbs_task);
> > > -   lsm_blob_size_update(&blobs->lbs_tun_dev, &blob_sizes.lbs_tun_dev);
> > > -   lsm_blob_size_update(&blobs->lbs_xattr_count,
> > > -                        &blob_sizes.lbs_xattr_count);
> > > -   lsm_blob_size_update(&blobs->lbs_bdev, &blob_sizes.lbs_bdev);
> > > -   lsm_blob_size_update(&blobs->lbs_bpf_map, &blob_sizes.lbs_bpf_map);
> > > -   lsm_blob_size_update(&blobs->lbs_bpf_prog, &blob_sizes.lbs_bpf_prog);
> > > -   lsm_blob_size_update(&blobs->lbs_bpf_token, &blob_sizes.lbs_bpf_token);
> > > +
> > > +   /* Register the LSM blob sizes. */
> > > +#define UPDATE_LSM_BLOB_SIZE(name) \
> > > +   lsm_blob_size_update(&blobs->lbs_##name, &blob_sizes.lbs_##name);
> > > +   LSM_BLOBS_LIST(UPDATE_LSM_BLOB_SIZE);
> > > +#undef UPDATE_LSM_BLOB_SIZE
> > >  }
> >
> > This is one case where the macro approach *might* have some value, but as
> > it is the one instance, and multiple sequential blob operations like this
> > are going to be *very* limited, I'm not sure there is enough value to
> > be worth the change.
> >
> > I appreciate the initiative to search out areas that you think could use
> > improvement, but for the reasons mentioned, I'm not going to merge this
> > patch.
>
> That's not a problem. Perhaps you may want to consider only taking the
> minor cleanup targets the redundant lsm->blobs access within
> lsm_prepare().

I did think about that, but I ever *so* slightly prefer it the way it
is now, although as I said this is the one case where there may be
some value.  If there had been other areas where it was a clear cut
win I likely would have just accepted this change too, but there
wasn't (at least not now, and with this revision) so I just decided to
leave it well enough alone.

However, as I said, thank you taking the initiative on this.

-- 
paul-moore.com