RAS/AMD/ATL: Add debug prints for DF register reads

[PATCH] RAS/AMD/ATL: Add debug prints for DF register reads

Posted by Yazen Ghannam 1 year, 3 months ago

The ATL will fail early if the DF register access fails due to missing
PCI IDs in the amd_nb code. There aren't any clear indicators on why the
ATL will fail to load in this case.

Add a couple of debug print statements to highlight reasons for failure.

A common scenario is missing support for new hardware. If the ATL fails
to load on a system, and there is interest to support it, then dynamic
debugging can be enabled to help find the cause for failure. If there is
no interest in supporting ATL on a new system, then these failures will
be silent.

Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
---
 drivers/ras/amd/atl/access.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/ras/amd/atl/access.c b/drivers/ras/amd/atl/access.c
index ee4661ed28ba..c2334f8f9add 100644
--- a/drivers/ras/amd/atl/access.c
+++ b/drivers/ras/amd/atl/access.c
@@ -70,12 +70,16 @@ static int __df_indirect_read(u16 node, u8 func, u16 reg, u8 instance_id, u32 *l
 	u32 ficaa = 0;
 
 	node = get_accessible_node(node);
-	if (node >= amd_nb_num())
+	if (node >= amd_nb_num()) {
+		pr_debug("Node %u is out of bounds\n", node);
 		goto out;
+	}
 
 	F4 = node_to_amd_nb(node)->link;
-	if (!F4)
+	if (!F4) {
+		pr_debug("DF function 4 not found\n");
 		goto out;
+	}
 
 	/* Enable instance-specific access. */
 	if (instance_id != DF_BROADCAST) {
-- 
2.43.0

Re: [PATCH] RAS/AMD/ATL: Add debug prints for DF register reads

Posted by Borislav Petkov 1 year, 3 months ago

On Mon, Oct 21, 2024 at 03:21:58PM +0000, Yazen Ghannam wrote:
> The ATL will fail early if the DF register access fails due to missing
> PCI IDs in the amd_nb code. There aren't any clear indicators on why the
> ATL will fail to load in this case.
> 
> Add a couple of debug print statements to highlight reasons for failure.
> 
> A common scenario is missing support for new hardware. If the ATL fails
> to load on a system, and there is interest to support it, then dynamic
> debugging can be enabled to help find the cause for failure. If there is
> no interest in supporting ATL on a new system, then these failures will
> be silent.
> 
> Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
> ---
>  drivers/ras/amd/atl/access.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)

Applied, thanks.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette