Re: [PATCH 1/2] stmmac: intel: Add a missing clk_disable_unprepare() call in intel_eth_pci_remove()

[Christophe JAILLET]

Le 30/07/2022 à 22:17, Andy Shevchenko a écrit :
> On Sat, Jul 30, 2022 at 08:19:47PM +0200, Christophe JAILLET wrote:
> > Commit 09f012e64e4b ("stmmac: intel: Fix clock handling on error and remove
> > paths") removed this clk_disable_unprepare()
> >
> > This was partly revert by commit ac322f86b56c ("net: stmmac: Fix clock
> > handling on remove path") which removed this clk_disable_unprepare()
> > because:
> > "
> >     While unloading the dwmac-intel driver, clk_disable_unprepare() is
> >     being called twice in stmmac_dvr_remove() and
> >     intel_eth_pci_remove(). This causes kernel panic on the second call.
> > "
> >
> > However later on, commit 5ec55823438e8 ("net: stmmac: add clocks management
> > for gmac driver") has updated stmmac_dvr_remove() which do not call
> > clk_disable_unprepare() anymore.
> >
> > So this call should now be called from intel_eth_pci_remove().
>
> The correct way of fixing it (which might be very well end up functionally
> the same as this patch), is to introduce ->quit() in struct stmmac_pci_info
> and assign it correctly, because not all platforms enable clocks.

I won't be able to propose anything like that.

By the way, in the first sentence of my log, s/removed/added/.
(I hope that it can be fixed when/if the patch is applied)


Thanks for the review.

CJ

> Perhaps, we may leave this patch as is (for the sake of easy backporting) and
> apply another one as I explained above to avoid similar mistakes in the future.
>
> Reviewed-by: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
>
> > Fixes: 5ec55823438e8 ("net: stmmac: add clocks management for gmac driver")
> > Signed-off-by: Christophe JAILLET <christophe.jaillet@xxxxxxxxxx>
> > ---
> > /!\     This patch is HIGHLY speculative.     /!\
> >
> > The corresponding clk_disable_unprepare() is still called within the pm
> > related stmmac_bus_clks_config() function.
> >
> > However, with my limited understanding of the pm API, I think it that the
> > patch is valid.
> > (in other word, does the pm_runtime_put() and/or pm_runtime_disable()
> > and/or stmmac_dvr_remove() can end up calling .runtime_suspend())
> >
> > So please review with care, as I'm not able to test the change by myself.
> >
> >
> > If I'm wrong, maybe a comment explaining why it is safe to have this
> > call in the error handling path of the probe and not in the remove function
> > would avoid erroneous patches generated from static code analyzer to be
> > sent.
> > ---
> >   drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c | 1 +
> >   1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c
> > index 52f9ed8db9c9..9f38642f86ce 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-intel.c
> > @@ -1134,6 +1134,7 @@ static void intel_eth_pci_remove(struct pci_dev *pdev)
> >   
> >   	stmmac_dvr_remove(&pdev->dev);
> >   
> > +	clk_disable_unprepare(plat->stmmac_clk);
> >   	clk_unregister_fixed_rate(priv->plat->stmmac_clk);
> >   
> >   	pcim_iounmap_regions(pdev, BIT(0));
> > --
> > 2.34.1