[PATCH v2 1/3] integrity: Replace all uses of integrity_audit_msg() with integrity_audit_message()

Mimi Zohar zohar at linux.ibm.com
Tue Sep 15 21:32:12 UTC 2026


On Tue, 2026-09-15 at 16:08 -0500, Frederick Lawler wrote:
> Hi Mimi,
> 
> On Tue, Sep 15, 2026 at 04:00:34PM -0400, Mimi Zohar wrote:
> > On Mon, 2026-07-27 at 17:21 -0500, Frederick Lawler wrote:
> > > integrity_audit_msg() wraps integrity_audit_message() such that
> > > error codes are hidden by passing an additional parameter to
> > > integrity_audit_message(). Since this is a wrapper, we can replace
> > > this function with integrity_audit_message() introduced in commit
> > > 2f845882ecd2 ("integrity: Add errno field in audit message")
> > > 
> > > No functional change intended.
> > 
> > There's an unintentional change in ima_write_policy(), which replaces the
> > existing '1' with result.
> 
> Oof. Good catch. That was intended for the later patch.
> 
> > 
> > > 
> > > Signed-off-by: Frederick Lawler <fred at cloudflare.com>
> > 
> > Is renaming integrity_audit_msg() to integrity_audit_message() actually the
> > right approach here?  integrity_audit_message() currently has only 2 callers vs.
> > ~20+ for integrity_audit_msg(), and the rename forces re-wrapping of every
> > multi-line call site since integrity_audit_message is 4 chars longer, bloating
> > the first patch.
> > 
> > Why not simply add errno to the integrity_audit_msg() definition and callers.  
> > 
> 
> I can switch that around. I assumed we'd want to keep _message() since
> that was wrapped by _msg(). And not dealing with the line-wrapping is
> better.

No, making integrity_audit_msg() a wrapper for integrity_audit_message was done
for expediency, instead of changing ALL the integrity_audit_msg() call sites.

Thank you for taking the time to review all the integrity_audit_msg() call sites
and adding errno..

Mimi










More information about the Linux-security-module-archive mailing list