quoted hunk ↗ jump to hunk
Subject: Re: [PATCH v2 6/8] EDAC/aspeed: Replace regmap with direct register
access
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to
consider:
- [Low] Mixing scope-based cleanup helpers with traditional goto-based error
handling within the same function.
--
--- Patch [6]: [PATCH v2 6/8] EDAC/aspeed: Replace regmap with direct register
access --- commit 707abf914ceaccc2e73503abf17a6380b0fc949b
Author: Ryan Chen [off-list ref]
EDAC/aspeed: Replace regmap with direct register access
The driver instantiates its own regmap purely as an MMIO wrapper: it has no
register cache, uses custom .reg_read()/.reg_write() callbacks, and is not
shared as a syscon with other drivers. So it brings nothing here beyond the
spinlock that regmap takes around each access when fast_io is set.
quoted
diff --git a/drivers/edac/aspeed_edac.c b/drivers/edac/aspeed_edac.c
index 352910e1defc5..26d2c456cc0d6 100644
--- a/drivers/edac/aspeed_edac.c
+++ b/drivers/edac/aspeed_edac.c
[ ... ]
quoted
@@ -274,13 +246,11 @@ static int aspeed_probe(struct platform_device
*pdev)
quoted
if (IS_ERR(regs))
return PTR_ERR(regs);
- aspeed_regmap = devm_regmap_init(dev, NULL, (__force void *)regs,
- &aspeed_regmap_config);
- if (IS_ERR(aspeed_regmap))
- return PTR_ERR(aspeed_regmap);
+ scoped_guard(raw_spinlock_irqsave, &aspeed_lock)
+ aspeed_regs = regs;
[Severity: Low]
Is it okay to mix scoped_guard() with goto-based error handling in this function?
According to include/linux/cleanup.h, the expectation is that usage of goto and
cleanup helpers should not be mixed in the same function.
In aspeed_probe(), we introduce scoped_guard() here, but the function still has
goto probe_exit01 and goto probe_exit02 later on:
aspeed_probe() {
...
scoped_guard(raw_spinlock_irqsave, &aspeed_lock)
aspeed_regs = regs;
...
rc = init_csrows(mci);
if (rc) {
dev_err(&pdev->dev, "failed to init csrows\n");
goto probe_exit02;
}
...
probe_exit01:
edac_mc_del_mc(&pdev->dev);
probe_exit02:
edac_mc_free(mci);
...
}
Should the remaining error paths be converted to scope-based cleanups, or
would it be better to avoid using scope-based cleanup helpers here?
No goto crosses the guard: the scoped_guard() covers a single
assignment and its scope ends about thirty lines before the first goto
becomes reachable, so a goto cannot skip a cleanup here. The labels
unwind mci, which no cleanup helper holds.
Converting them isn't attractive either - they implement a two-stage
teardown, edac_mc_del_mc() then edac_mc_free(), entered at two
different stages, which doesn't map onto a single __free(). And
dropping the guard would mean open-coding lock/unlock around one store
just to satisfy the __guarded_by(&aspeed_lock) check, in a path where
nothing can race yet.
quoted
/* bail out if ECC mode is not configured */
- regmap_read(aspeed_regmap, ASPEED_MCR_CONF, ®04);
+ reg04 = readl(regs + ASPEED_MCR_CONF);
if (!(reg04 & ASPEED_MCR_CONF_ECC)) {
--
Sashiko AI review ·
https://sashiko.dev/#/patchset/20260824-edac-v2-0-c8d8bb693586@aspeedte
ch.com?part=6