Re: [PATCH] x86/amd_node: Fix PCI device reference counting in amd_smn_init()

[Yazen Ghannam]

On Tue, Aug 25, 2026 at 09:52:06AM -0400, Yazen Ghannam wrote:
> On Mon, Aug 24, 2026 at 10:59:28PM -0500, Serge Hallyn (AMD) wrote:
> > On Mon, Aug 24, 2026 at 12:50:03PM -0500, Yazen Ghannam wrote:
> > > The local "root" pointer is a temporary variable used during the device
> > > search. Therefore, refcount related to the search iterators should be
> > > cleaned up after the search is complete.
> > > 
> > > Use the __free() cleanup macro to ensure the refcount is decremented
> > > when the temporary pointer goes out of scope.
> > > 
> > > Additionally, increment the refcount when caching a root pointer. This
> > > ensures the in-use refcount is separate from the temporary search
> > > refcounting.
> > > 
> > > Fixes: 0a4b61d9c2e4 ("x86/amd_node: Fix AMD root device caching")
> > > Reported-by: Sashiko <sashiko-bot@xxxxxxxxxx>
> > > Closes: https://sashiko.dev/#/patchset/20260806160159.230453-1-jason.andryuk%40amd.com
> > > Assisted-by: Claude-Code:claude-opus-5
> > > Signed-off-by: Yazen Ghannam <yazen.ghannam@xxxxxxx>
> > > ---
> > >  arch/x86/kernel/amd_node.c | 5 ++---
> > >  1 file changed, 2 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/arch/x86/kernel/amd_node.c b/arch/x86/kernel/amd_node.c
> > > index 0be01725a2a4..312adf73313b 100644
> > > --- a/arch/x86/kernel/amd_node.c
> > > +++ b/arch/x86/kernel/amd_node.c
> > > @@ -247,7 +247,7 @@ __setup("amd_smn_debugfs_enable", amd_smn_enable_dfs);
> > >  static int __init amd_smn_init(void)
> > >  {
> > >  	u16 count, num_roots, roots_per_node, node, num_nodes;
> > > -	struct pci_dev *root;
> > > +	struct pci_dev *root __free(pci_dev_put) = NULL;
> > >  
> > >  	if (!cpu_feature_enabled(X86_FEATURE_ZEN))
> > >  		return 0;
> > > @@ -258,7 +258,6 @@ static int __init amd_smn_init(void)
> > >  		return 0;
> > >  
> > >  	num_roots = 0;
> > > -	root = NULL;
> > >  	while ((root = get_next_root(root))) {
> > >  		pci_dbg(root, "Reserving PCI config space\n");
> > >  
> > 
> > Will this leak the ref taken on the last get_next_root(root) in the
> > first loop?  You might need a pci_dev_put(root) before the root = NULL
> > above the second loop.  Or I could be wrong.
> > 
> 
> Yes, I think you're right. Good catch.
> 
> I'll send another revision with your suggestion.
> 

Actually, this isn't an issue. An exhaustive search ends with a NULL
pointer. The PCI search helpers would have done the 'put' on the last
device.

This also means that the second 'root = NULL' is redundant. I'll remove
it in the next revision.

Please let me know what you think.

Thanks,
Yazen