Re: [PATCH v7 3/5] Add debugfs based silicon debug support in DWC

From: Krzysztof Wilczyński
Date: Wed Mar 05 2025 - 14:10:47 EST


Hello,

[...]
> > > Even though debugfs_init() failure is not supposed to fail the probe(),
> > > dwc_pcie_rasdes_debugfs_init() has a devm_kzalloc() and propagating that
> > > failure would be canolically correct IMO.
> >
> > I'm not sure about this. What's the requirement to propagate
> > devm_kzalloc() failures? I think devres will free any allocs that
> > were successful regardless.
> >
> > IIUC, we resolved the Gray Hawk Single issue by changing
> > dwc_pcie_rasdes_debugfs_init() to return success without doing
> > anything when there's no RAS DES Capability.
> >
> > But dwc_pcie_debugfs_init() can still return failure, and that still
> > causes dw_pcie_ep_init_registers() to fail, which breaks the "don't
> > propagate debugfs issues upstream" rule:
> >
> > int dw_pcie_ep_init_registers(struct dw_pcie_ep *ep)
> > {
> > ...
> > ret = dwc_pcie_debugfs_init(pci);
> > if (ret)
> > goto err_remove_edma;
> >
> > return 0;
> >
> > err_remove_edma:
> > dw_pcie_edma_remove(pci);
> >
> > return ret;
> > }
> >
> > We can say that kzalloc() failure should "never" happen, and therefore
> > it's OK to fail the driver probe if it happens, but that doesn't seem
> > like a strong argument for breaking the "don't propagate debugfs
> > issues" rule. And someday there may be other kinds of failures from
> > dwc_pcie_debugfs_init().
> >
>
> Fine with me. I was not too sure about propagating failure either.

What if we do this?

diff --git i/drivers/pci/controller/dwc/pcie-designware-debugfs.c w/drivers/pci/controller/dwc/pcie-designware-debugfs.c
index 586a9d107434..fddf71f014c4 100644
--- i/drivers/pci/controller/dwc/pcie-designware-debugfs.c
+++ w/drivers/pci/controller/dwc/pcie-designware-debugfs.c
@@ -162,7 +162,7 @@ void dwc_pcie_debugfs_deinit(struct dw_pcie *pci)
debugfs_remove_recursive(pci->debugfs->debug_dir);
}

-int dwc_pcie_debugfs_init(struct dw_pcie *pci)
+void dwc_pcie_debugfs_init(struct dw_pcie *pci)
{
char dirname[DWC_DEBUGFS_BUF_MAX];
struct device *dev = pci->dev;
@@ -174,17 +174,15 @@ int dwc_pcie_debugfs_init(struct dw_pcie *pci)
snprintf(dirname, DWC_DEBUGFS_BUF_MAX, "dwc_pcie_%s", dev_name(dev));
dir = debugfs_create_dir(dirname, NULL);
debugfs = devm_kzalloc(dev, sizeof(*debugfs), GFP_KERNEL);
- if (!debugfs)
- return -ENOMEM;
+ if (!debugfs) {
+ dev_err(dev, "failed to allocate memory for debugfs\n");
+ return;
+ }

debugfs->debug_dir = dir;
pci->debugfs = debugfs;
err = dwc_pcie_rasdes_debugfs_init(pci, dir);
- if (err) {
- dev_err(dev, "failed to initialize RAS DES debugfs, err=%d\n",
- err);
- return err;
- }
-
- return 0;
+ if (err)
+ dev_warn(dev, "failed to initialize RAS DES debugfs, err=%d",
+ err);
}
diff --git i/drivers/pci/controller/dwc/pcie-designware-ep.c w/drivers/pci/controller/dwc/pcie-designware-ep.c
index c6e76a07c2c9..11ff292ca87d 100644
--- i/drivers/pci/controller/dwc/pcie-designware-ep.c
+++ w/drivers/pci/controller/dwc/pcie-designware-ep.c
@@ -838,9 +838,7 @@ int dw_pcie_ep_init_registers(struct dw_pcie_ep *ep)

dw_pcie_ep_init_non_sticky_registers(pci);

- ret = dwc_pcie_debugfs_init(pci);
- if (ret)
- goto err_remove_edma;
+ dwc_pcie_debugfs_init(pci);

return 0;

diff --git i/drivers/pci/controller/dwc/pcie-designware-host.c w/drivers/pci/controller/dwc/pcie-designware-host.c
index 2081e8c72d12..6501fb062c70 100644
--- i/drivers/pci/controller/dwc/pcie-designware-host.c
+++ w/drivers/pci/controller/dwc/pcie-designware-host.c
@@ -548,9 +548,7 @@ int dw_pcie_host_init(struct dw_pcie_rp *pp)
if (pp->ops->post_init)
pp->ops->post_init(pp);

- ret = dwc_pcie_debugfs_init(pci);
- if (ret)
- goto err_stop_link;
+ dwc_pcie_debugfs_init(pci);

return 0;

diff --git i/drivers/pci/controller/dwc/pcie-designware.h w/drivers/pci/controller/dwc/pcie-designware.h
index 7f9807d4e5de..dd129718fb41 100644
--- i/drivers/pci/controller/dwc/pcie-designware.h
+++ w/drivers/pci/controller/dwc/pcie-designware.h
@@ -815,12 +815,11 @@ dw_pcie_ep_get_func_from_ep(struct dw_pcie_ep *ep, u8 func_no)
#endif

#ifdef CONFIG_PCIE_DW_DEBUGFS
-int dwc_pcie_debugfs_init(struct dw_pcie *pci);
+void dwc_pcie_debugfs_init(struct dw_pcie *pci);
void dwc_pcie_debugfs_deinit(struct dw_pcie *pci);
#else
-static inline int dwc_pcie_debugfs_init(struct dw_pcie *pci)
+static inline void dwc_pcie_debugfs_init(struct dw_pcie *pci)
{
- return 0;
}
static inline void dwc_pcie_debugfs_deinit(struct dw_pcie *pci)
{

I think this would be fine, especially given the rules around debugfs and
a quick look around Git history to see what the prefernce would be typically.

Thank you!

Krzysztof