Re: [PATCH 2/2] RAS: Introduce the FRU Memory Poison Manager

[Borislav Petkov]

On Wed, Feb 14, 2024 at 09:28:54AM -0500, Yazen Ghannam wrote:
> > That's a good thing to have here.

Up to here. __packed still needs clarification.

diff --git a/drivers/ras/Kconfig b/drivers/ras/Kconfig
index 782951aa302f..f5dde88a3188 100644
--- a/drivers/ras/Kconfig
+++ b/drivers/ras/Kconfig
@@ -37,14 +37,13 @@ source "drivers/ras/amd/atl/Kconfig"
 config RAS_FMPM
 	tristate "FRU Memory Poison Manager"
 	default m
-	depends on X86_MCE
-	imply AMD_ATL
+	depends on AMD_ATL
 	help
 	  Support saving and restoring memory error information across reboot
-	  cycles using ACPI ERST as persistent storage. Error information is
-	  saved with the UEFI CPER "FRU Memory Poison" section format.
+	  using ACPI ERST as persistent storage. Error information is saved with
+	  the UEFI CPER "FRU Memory Poison" section format.
 
-	  Memory may be retired during boot time and run time depending on
+	  Memory will be retired during boot time and run time depending on
 	  platform-specific policies.
 
 endif
diff --git a/drivers/ras/amd/fmpm.c b/drivers/ras/amd/fmpm.c
index d6a963aca093..901a1f0018fc 100644
--- a/drivers/ras/amd/fmpm.c
+++ b/drivers/ras/amd/fmpm.c
@@ -12,7 +12,7 @@
  *
  * Implementation notes, assumptions, and limitations:
  *
- * - FRU Memory Poison Section and Memory Poison Descriptor definitions are not yet
+ * - FRU memory poison section and memory poison descriptor definitions are not yet
  *   included in the UEFI specification. So they are defined here. Afterwards, they
  *   may be moved to linux/cper.h, if appropriate.
  *
@@ -23,16 +23,13 @@
  *   AMD MI300-based platform(s) assumptions:
  *   - Memory errors are reported through x86 MCA.
  *   - The entire DRAM row containing a memory error should be retired.
- *   - There will be (1) FRU Memory Poison Section per CPER.
- *   - The FRU will be the CPU Package (Processor Socket).
- *   - The default number of Memory Poison Descriptor entries should be (8).
- *   - The Platform will use ACPI ERST for persistent storage.
+ *   - There will be (1) FRU memory poison section per CPER.
+ *   - The FRU will be the CPU package (processor socket).
+ *   - The default number of memory poison descriptor entries should be (8).
+ *   - The platform will use ACPI ERST for persistent storage.
  *   - All FRU records should be saved to persistent storage. Module init will
  *     fail if any FRU record is not successfully written.
  *
- * - Source code will be under 'drivers/ras/amd/' unless and until there is interest
- *   to use this module for other vendors.
- *
  * - Boot time memory retirement may occur later than ideal due to dependencies
  *   on other libraries and drivers. This leaves a gap where bad memory may be
  *   accessed during early boot stages.

-- 
Regards/Gruss,
    Boris.

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