Re: [PATCH] clk: Guard clk_round_rate() against error pointers

[Brian Masney <bmasney@redhat.com>]

On Fri, Jul 24, 2026 at 11:40:19PM +0200, Sebastian Reichel wrote:
> On Fri, Jul 24, 2026 at 11:44:55AM -0400, Brian Masney wrote:
> > On Thu, Jul 23, 2026 at 10:10:06PM +0530, Praveen Talari wrote:
> > > On 23-07-2026 19:59, Brian Masney wrote:
> > > > On Thu, Jul 23, 2026 at 11:40:47AM +0530, Praveen Talari wrote:
> > > > > clk_round_rate() only checks for a NULL clk pointer before
> > > > > dereferencing it, but callers such as dev_pm_opp_set_rate() can pass
> > > > > it an error pointer (e.g. ERR_PTR(-ENOENT) left behind by
> > > > > clk_get() when a device has no Linux clock and is instead managed by
> > > > > firmware via a genpd/OPP performance domain).
> > > > > 
> > > > > Dereferencing that error pointer to read clk->exclusive_count
> > > > > crashes with an unhandled kernel NULL pointer dereference, since
> > > > > ERR_PTR(-ENOENT) plus the field's offset lands on a small, unmapped
> > > > > address:
> > > > > 
> > > > >    Unable to handle kernel NULL pointer dereference at virtual
> > > > >    address 000000000000002e
> > > > >    ...
> > > > >    pc : clk_round_rate+0x3c/0x188
> > > > >    ...
> > > > >    Call trace:
> > > > >     clk_round_rate+0x3c/0x188 (P)
> > > > >     dev_pm_opp_set_rate+0x114/0x33c
> > > > > 
> > > > > Change the guard from "if (!clk)" to "if (IS_ERR_OR_NULL(clk))",
> > > > > matching the pattern already used by other clk consumer API
> > > > > functions such as clk_unprepare(), so an error pointer is rejected
> > > > > the same way a NULL pointer is.
> > > > > 
> > > > > Signed-off-by: Praveen Talari <praveen.talari@oss.qualcomm.com>
> > > > > ---
> > > > >   drivers/clk/clk.c | 2 +-
> > > > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > 
> > > > > diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> > > > > index 048adfa86a5d..8c1ad3d10284 100644
> > > > > --- a/drivers/clk/clk.c
> > > > > +++ b/drivers/clk/clk.c
> > > > > @@ -1780,7 +1780,7 @@ long clk_round_rate(struct clk *clk, unsigned long rate)
> > > > >   	struct clk_rate_request req;
> > > > >   	int ret;
> > > > > -	if (!clk)
> > > > > +	if (IS_ERR_OR_NULL(clk))
> > > > Can you provide more details about the clk_get() call point that starts this
> > > > error? Specifically which driver this occurs in and the exact scenario that
> > > > triggers this.
> > > 
> > > On SA8255P platform there is no Linux
> > > clock for the SE, and the perf domain device's OPPs are populated entirely
> > > from firmware via devm_pm_opp_of_add_table() (through
> > > of_genpd_add_provider_simple()/onecell()), so the perf domain's OPP table
> > > has entries even though no clk_get() ever succeeds for it.
> > > 
> > > The clk_get(-ENOENT) case comes from _update_opp_table_clk() in
> > > drivers/opp/core.c:
> > > 
> > >     opp_table->clk = clk_get(dev, NULL);
> > >     ret = PTR_ERR_OR_ZERO(opp_table->clk);
> > >     ...
> > >     if (ret == -ENOENT) {
> > >         /* ... no clk provided ... */
> > >         opp_table->clk_count = 1;
> > >         return opp_table;   /* opp_table->clk left as ERR_PTR(-ENOENT) */
> > >     }
> > > 
> > > Because the perf domain device has no "clocks" property (its OPPs are
> > > supplied purely as performance states by firmware/genpd), clk_get()
> > > returns -ENOENT, and opp_table->clk is left holding that error pointer
> > > rather than being reset to NULL.
> > 
> > This looks to be a reasonable change. Thanks.
> > 
> > Reviewed-by: Brian Masney <bmasney@redhat.com>
> 
> I suggest to instead change the OPP code, so that it calls
> clk_get_optional() instead of clk_get() and thus properly
> "document" that the clock is optional and use the NULL dummy
> clock (it's also shorter):
> 
> /*
>  * There are few platforms which don't want the OPP core to
>  * manage device's clock settings. In such cases neither the
>  * platform provides the clks explicitly to us, nor the DT
>  * contains a valid clk entry. The OPP nodes in DT may still
>  * contain "opp-hz" property though, which we need to parse and
>  * allow the platform to find an OPP based on freq later on.
>  *
>  * This is a simple solution to take care of such corner cases,
>  * i.e. make the clk_count 1, which lets us allocate space for
>  * frequency in opp->rates and also parse the entries in DT.
>  */
> opp_table->clk = clk_get_optional(dev, NULL);
> 
> ret = PTR_ERR_OR_ZERO(opp_table->clk);
> if (ret) {
>     dev_pm_opp_put_opp_table(opp_table);
>     dev_err_probe(dev, ret, "Couldn't find clock\n");
>     return ERR_PTR(ret);
> }
> 
> if (opp_table->clk)
>     opp_table->config_clks = _opp_config_clk_single;
> opp_table->clk_count = 1;
> return opp_table;

Yes, that makes sense. Thanks!

Brian