* [PATCH] mmc: sunxi: fix clock and runtime PM leaks on probe failure
@ 2026-10-20 15:36 Anfernee Cao
0 siblings, 0 replies; only message in thread
From: Anfernee Cao @ 2026-10-20 15:36 UTC (permalink / raw)
To: Ulf Hansson, Chen-Yu Tsai
Cc: Jernej Skrabec, Samuel Holland, linux-mmc, linux-kernel,
linux-arm-kernel, stable
sunxi_mmc_resource_request() enables the bus, card, output and sample
clocks before returning. Several failure paths later in
sunxi_mmc_probe() leave them running:
- If devm_request_threaded_irq() fails, the clocks enabled by
sunxi_mmc_enable() are never disabled.
- If dma_alloc_coherent(), mmc_of_parse() or sunxi_mmc_init_host()
fails, probe only frees the DMA buffer and the mmc_host.
- If mmc_add_host() fails, runtime PM has already been enabled, but
pm_runtime_disable() is never called.
Besides leaking the clocks, this leaves the clock enable counts
unbalanced, so a reprobe (for example after -EPROBE_DEFER) calls
clk_prepare_enable() on the same clocks again.
Disable the IRQ and the clocks in the probe error path, and disable
runtime PM on the mmc_add_host() failure path. Also handle a failure of
devm_request_threaded_irq() in sunxi_mmc_resource_request().
Tested with QEMU's orangepi-pc machine: with forced sunxi_mmc_init_host()
and mmc_add_host() failures the MMC clocks stay disabled in clk_summary,
and a normal probe with an emulated SD card still works.
This issue was found by manual code inspection.
Fixes: 3cbcb16095f9 ("mmc: sunxi: Add driver for SD/MMC hosts found on Allwinner sunxi SoCs")
Fixes: 9a8e1e8cc2c0 ("mmc: sunxi: Add runtime_pm support")
Cc: stable@vger.kernel.org # v4.18+
Signed-off-by: Anfernee Cao <anferneemelokc5@gmail.com>
---
drivers/mmc/host/sunxi-mmc.c | 24 ++++++++++++++++++------
1 file changed, 18 insertions(+), 6 deletions(-)
diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c
index fe4c0f6d73f3..437fdadb5950 100644
--- a/drivers/mmc/host/sunxi-mmc.c
+++ b/drivers/mmc/host/sunxi-mmc.c
@@ -1354,8 +1354,13 @@ static int sunxi_mmc_resource_request(struct sunxi_mmc_host *host,
goto error_disable_mmc;
}
- return devm_request_threaded_irq(&pdev->dev, host->irq, sunxi_mmc_irq,
- sunxi_mmc_handle_manual_stop, 0, "sunxi-mmc", host);
+ ret = devm_request_threaded_irq(&pdev->dev, host->irq, sunxi_mmc_irq,
+ sunxi_mmc_handle_manual_stop, 0,
+ "sunxi-mmc", host);
+ if (ret)
+ goto error_disable_mmc;
+
+ return 0;
error_disable_mmc:
sunxi_mmc_disable(host);
@@ -1385,9 +1390,11 @@ static int sunxi_mmc_probe(struct platform_device *pdev)
host->sg_cpu = dma_alloc_coherent(&pdev->dev, PAGE_SIZE,
&host->sg_dma, GFP_KERNEL);
- if (!host->sg_cpu)
- return dev_err_probe(&pdev->dev, -ENOMEM,
- "Failed to allocate DMA descriptor mem\n");
+ if (!host->sg_cpu) {
+ ret = dev_err_probe(&pdev->dev, -ENOMEM,
+ "Failed to allocate DMA descriptor mem\n");
+ goto error_disable_mmc;
+ }
if (host->cfg->ccu_has_timings_switch) {
/*
@@ -1467,7 +1474,7 @@ static int sunxi_mmc_probe(struct platform_device *pdev)
ret = mmc_add_host(mmc);
if (ret)
- goto error_free_dma;
+ goto error_disable_rpm;
dev_info(&pdev->dev, "initialized, max. request size: %u KB%s\n",
mmc->max_req_size >> 10,
@@ -1475,8 +1482,13 @@ static int sunxi_mmc_probe(struct platform_device *pdev)
return 0;
+error_disable_rpm:
+ pm_runtime_disable(&pdev->dev);
error_free_dma:
dma_free_coherent(&pdev->dev, PAGE_SIZE, host->sg_cpu, host->sg_dma);
+error_disable_mmc:
+ disable_irq(host->irq);
+ sunxi_mmc_disable(host);
return ret;
}
--
2.54.0.windows.1
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2026-10-20 15:36 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-10-20 15:36 [PATCH] mmc: sunxi: fix clock and runtime PM leaks on probe failure Anfernee Cao
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®