Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Jason Baron <jbaron@akamai.com>
To:     bp@alien8.de
Cc:     linux-edac@vger.kernel.org, linux-kernel@vger.kernel.org,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        Tony Luck <tony.luck@intel.com>,
        James Morse <james.morse@arm.com>,
        Robert Richter <rric@kernel.org>,
        "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>,
        Shuai Xue <xueshuai@linux.alibaba.com>, stable@vger.kernel.org
Subject: [PATCH] EDAC/edac_module: order edac_init() before ghes_edac_register()
Date:   Tue, 15 Nov 2022 19:37:29 -0500
Message-Id: <20221116003729.194802-1-jbaron@akamai.com>
Precedence: bulk

Currently, ghes_edac_register() is called via ghes_init() from acpi_init()
at the subsys_initcall() level. However, edac_init() is also called from
the subsys_initcall(), leaving the ordering ambiguous.

If ghes_edac_register() is called first, then 'mc0' ends up at:
/sys/devices/mc0/, instead of the expected:
/sys/devices/system/edac/mc/mc0.

So while everything seems ok, other than the unexpected sysfs location, it
seems like 'edac_init()' should be called before any drivers start
registering. So have 'edac_init()' called earlier via arch_initcall().

However, this moves edac_pci_clear_parity_errors() up as well. Seems like
this wants to be called after pci bus scan, so keep
edac_pci_clear_parity_errors() at subsys_init(). That said, it seems like
pci bus scan happens at subsys_init() level, so really the parity clearing
should be moved later. But that can be left as a separate patch.

Fixes: dc4e8c07e9e2 ("ACPI: APEI: explicit init of HEST and GHES in apci_init()")
Signed-off-by: Jason Baron <jbaron@akamai.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Mauro Carvalho Chehab <mchehab@kernel.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: James Morse <james.morse@arm.com>
Cc: Robert Richter <rric@kernel.org>
Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>
Cc: Shuai Xue <xueshuai@linux.alibaba.com>
Cc: stable@vger.kernel.org
---
 drivers/edac/edac_module.c | 33 +++++++++++++++++++++++----------
 1 file changed, 23 insertions(+), 10 deletions(-)

diff --git a/drivers/edac/edac_module.c b/drivers/edac/edac_module.c
index 32a931d0cb71..407d4a5fce7a 100644
--- a/drivers/edac/edac_module.c
+++ b/drivers/edac/edac_module.c
@@ -109,15 +109,6 @@ static int __init edac_init(void)
 	if (err)
 		return err;
 
-	/*
-	 * Harvest and clear any boot/initialization PCI parity errors
-	 *
-	 * FIXME: This only clears errors logged by devices present at time of
-	 *      module initialization.  We should also do an initial clear
-	 *      of each newly hotplugged device.
-	 */
-	edac_pci_clear_parity_errors();
-
 	err = edac_mc_sysfs_init();
 	if (err)
 		goto err_sysfs;
@@ -157,12 +148,34 @@ static void __exit edac_exit(void)
 	edac_subsys_exit();
 }
 
+static void __init edac_init_clear_parity_errors(void)
+{
+	/*
+	 * Harvest and clear any boot/initialization PCI parity errors
+	 *
+	 * FIXME: This only clears errors logged by devices present at time of
+	 *      module initialization.  We should also do an initial clear
+	 *      of each newly hotplugged device.
+	 */
+	edac_pci_clear_parity_errors();
+
+	return 0;
+}
+
 /*
  * Inform the kernel of our entry and exit points
+ *
+ * ghes_edac_register() is call via acpi_init() -> ghes_init()
+ * at the subsys_initcall level so edac_init() must come first
  */
-subsys_initcall(edac_init);
+arch_initcall(edac_init);
 module_exit(edac_exit);
 
+/*
+ * Clear parity errors after PCI subsys is initialized
+ */
+subsys_initcall(edac_init_clear_parity_errors);
+
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Doug Thompson www.softwarebitmaker.com, et al");
 MODULE_DESCRIPTION("Core library routines for EDAC reporting");
-- 
2.17.1