* [PATCH] EDAC/aspeed: do not fail probe when ECC is disabled
@ 2026-11-18 15:19 Eric Yun Lin Chiang
0 siblings, 0 replies; only message in thread
From: Eric Yun Lin Chiang @ 2026-11-18 15:19 UTC (permalink / raw)
To: Stefan Schaeckeler, Borislav Petkov, Tony Luck
Cc: Joel Stanley, Andrew Jeffery, Ryan Chen, Billy Tsai, linux-edac,
linux-arm-kernel, linux-aspeed, linux-kernel,
Eric Yun Lin Chiang
The AST2600 memory-controller EDAC node is enabled by default in
aspeed-g6.dtsi (it has no status property), so aspeed_edac binds on
every AST2600 board. aspeed_probe() reads the memory-controller
configuration and, when DRAM ECC was not turned on by firmware,
prints an error and returns -EPERM:
aspeed-edac 1e6e0000.sdram: ECC mode is not configured in u-boot
aspeed-edac 1e6e0000.sdram: probe with driver aspeed-edac failed with error -1
Running without DRAM ECC is a normal platform choice, not a failure.
-EPERM is both the wrong errno (this is not a permission problem) and
one the driver core reports as a probe error, so every non-ECC AST2600
board logs a spurious "failed with error -1" at boot.
Return -ENODEV instead, which the driver core treats as "no device to
bind" and does not log as an error, and lower the message to dev_info so
the reason EDAC did not register is still visible once without looking
like a fault. The AST2500 node is disabled by default, so only boards
that explicitly enable it are affected there. No other error path is
changed.
This change was prepared with the help of an AI coding assistant, which
analysed the probe path and drafted this changelog; the author has
reviewed it and stands behind the change.
Fixes: 9b7e6242ee4e ("EDAC, aspeed: Add an Aspeed AST2500 EDAC driver")
Signed-off-by: Eric Yun Lin Chiang <ericyunlinchiang@gmail.com>
---
Tested: the pre-fix "failed with error -1" was reproduced on an AST2600
(Facebook Bletchley) BMC by triggering the aspeed-edac probe over sysfs;
the fixed driver builds (W=1, no new warnings) and passes checkpatch,
and the resulting behaviour -- aspeed_probe() now returns -ENODEV, which
the driver core does not log as an error -- was confirmed by disassembly
of the built object. NOT boot-tested on hardware (the kernel was not
booted).
drivers/edac/aspeed_edac.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/edac/aspeed_edac.c b/drivers/edac/aspeed_edac.c
index 6e069b255..e6f8cf78a 100644
--- a/drivers/edac/aspeed_edac.c
+++ b/drivers/edac/aspeed_edac.c
@@ -295,8 +295,8 @@ static int aspeed_probe(struct platform_device *pdev)
/* bail out if ECC mode is not configured */
regmap_read(aspeed_regmap, ASPEED_MCR_CONF, ®04);
if (!(reg04 & ASPEED_MCR_CONF_ECC)) {
- dev_err(&pdev->dev, "ECC mode is not configured in u-boot\n");
- return -EPERM;
+ dev_info(&pdev->dev, "ECC mode is not configured in u-boot\n");
+ return -ENODEV;
}
edac_op_state = EDAC_OPSTATE_INT;
--
2.53.0
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2026-11-18 15:19 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-18 15:19 [PATCH] EDAC/aspeed: do not fail probe when ECC is disabled Eric Yun Lin Chiang
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®