| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID:
<TYCPR01MB11332395B91D672D6BD46CECD8687A@TYCPR01MB11332.jpnprd01.prod.outlook.com>
Date: Tue, 6 Jan 2026 15:41:10 +0000
From: Biju Das <biju.das.jz@...renesas.com>
To: Frank Li <Frank.li@....com>
CC: Tommaso Merciai <tommaso.merciai.xr@...renesas.com>, Tommaso Merciai
<tomm.merciai@...il.com>, "linux-renesas-soc@...r.kernel.org"
<linux-renesas-soc@...r.kernel.org>, wsa+renesas
<wsa+renesas@...g-engineering.com>, Alexandre Belloni
<alexandre.belloni@...tlin.com>, Philipp Zabel <p.zabel@...gutronix.de>,
Geert Uytterhoeven <geert+renesas@...der.be>, magnus.damm
<magnus.damm@...il.com>, "linux-i3c@...ts.infradead.org"
<linux-i3c@...ts.infradead.org>, "linux-kernel@...r.kernel.org"
<linux-kernel@...r.kernel.org>
Subject: RE: [PATCH v4 1/4] i3c: renesas: Switch to clk_bulk API and store
clocks in private data
Hi Frank Li,
> -----Original Message-----
> From: Frank Li <Frank.li@....com>
> Sent: 06 January 2026 15:17
> Subject: Re: [PATCH v4 1/4] i3c: renesas: Switch to clk_bulk API and store clocks in private data
>
> On Tue, Jan 06, 2026 at 03:01:49PM +0000, Biju Das wrote:
> > Hi Frank Li,
> >
> > Thanks for the feedback.
> >
> > > -----Original Message-----
> > > From: Frank Li <Frank.li@....com>
> > > Sent: 06 January 2026 14:49
> > > Subject: Re: [PATCH v4 1/4] i3c: renesas: Switch to clk_bulk API and
> > > store clocks in private data
> > >
> > > On Mon, Jan 05, 2026 at 06:06:19PM +0000, Biju Das wrote:
> > > > Hi Frank Li,
> > > >
> > > > > -----Original Message-----
> > > > > From: Frank Li <Frank.li@....com>
> > > > > Sent: 05 January 2026 17:12
> > > > > Subject: Re: [PATCH v4 1/4] i3c: renesas: Switch to clk_bulk API
> > > > > and store clocks in private data
> > > > >
> > > > > On Mon, Jan 05, 2026 at 11:49:59AM +0100, Tommaso Merciai wrote:
> > > > > > Replace individual devm_clk_get_enabled() calls with the
> > > > > > clk_bulk API and store the clock handles in the driver's private data structure.
> > > > > >
> > > > > > All clocks required by the controller are now acquired and
> > > > > > enabled using devm_clk_bulk_get_all_enabled(), removing the
> > > > > > need for per-SoC clock handling and the renesas_i3c_config data.
> > > > > > The TCLK is accessed via a fixed index in the bulk clock array.
> > > > > >
> > > > > > Simplify the code and prepare the driver for upcoming
> > > > > > suspend/resume support.
> > > > > >
> > > > > > No functional change intended.
> > > > > >
> > > > > > Reviewed-by: Biju Das <biju.das.jz@...renesas.com>
> > > > > > Signed-off-by: Tommaso Merciai
> > > > > > <tommaso.merciai.xr@...renesas.com>
> > > > > > ---
> > > > > > v3->v4:
> > > > > > - Collected Biju Das tag.
> > > > > >
> > > > > > v2->v3:
> > > > > > - Added define for TCLK index.
> > > > > > - Use devm_clk_bulk_get_all_enabled() into renesas_i3c_probe().
> > > > > > - Improved commit body.
> > > > > > - Dropped unnecessary static const char * const renesas_i3c_clks[].
> > > > > > - Removed the need for per-SoC clock handling and the renesas_i3c_config data.
> > > > > >
> > > > > > v1->v2:
> > > > > > - New patch.
> > > > > >
> > > > > > drivers/i3c/master/renesas-i3c.c | 43
> > > > > > ++++++++------------------------
> > > > > > 1 file changed, 11 insertions(+), 32 deletions(-)
> > > > > >
> > > > > > diff --git a/drivers/i3c/master/renesas-i3c.c
> > > > > > b/drivers/i3c/master/renesas-i3c.c
> > > > > > index 426a418f29b6..1b8f4be9ad67 100644
> > > > > > --- a/drivers/i3c/master/renesas-i3c.c
> > > > > > +++ b/drivers/i3c/master/renesas-i3c.c
> > > > > > @@ -198,6 +198,8 @@
> > > > > > #define RENESAS_I3C_MAX_DEVS 8
> > > > > > #define I2C_INIT_MSG -1
> > > > > >
> > > > > > +#define RENESAS_I3C_TCLK_IDX 1
> > > > > > +
> > > > > > enum i3c_internal_state {
> > > > > > I3C_INTERNAL_STATE_DISABLED,
> > > > > > I3C_INTERNAL_STATE_CONTROLLER_IDLE,
> > > > > > @@ -259,7 +261,8 @@ struct renesas_i3c {
> > > > > > u8 addrs[RENESAS_I3C_MAX_DEVS];
> > > > > > struct renesas_i3c_xferqueue xferqueue;
> > > > > > void __iomem *regs;
> > > > > > - struct clk *tclk;
> > > > > > + struct clk_bulk_data *clks;
> > > > > > + u8 num_clks;
> > > > > > };
> > > > > >
> > > > > > struct renesas_i3c_i2c_dev_data { @@ -272,10 +275,6 @@ struct
> > > > > > renesas_i3c_irq_desc {
> > > > > > const char *desc;
> > > > > > };
> > > > > >
> > > > > > -struct renesas_i3c_config {
> > > > > > - unsigned int has_pclkrw:1;
> > > > > > -};
> > > > > > -
> > > > > > static inline void renesas_i3c_reg_update(void __iomem *reg,
> > > > > > u32 mask, u32 val) {
> > > > > > u32 data = readl(reg);
> > > > > > @@ -489,7 +488,7 @@ static int renesas_i3c_bus_init(struct i3c_master_controller *m)
> > > > > > int od_high_ticks, od_low_ticks, i2c_total_ticks;
> > > > > > int ret;
> > > > > >
> > > > > > - rate = clk_get_rate(i3c->tclk);
> > > > > > + rate = clk_get_rate(i3c->clks[RENESAS_I3C_TCLK_IDX].clk);
> > > > > > if (!rate)
> > > > > > return -EINVAL;
> > > > > >
> > > > > > @@ -1302,13 +1301,8 @@ static int renesas_i3c_probe(struct
> > > > > > platform_device *pdev) {
> > > > > > struct renesas_i3c *i3c;
> > > > > > struct reset_control *reset;
> > > > > > - struct clk *clk;
> > > > > > - const struct renesas_i3c_config *config = of_device_get_match_data(&pdev->dev);
> > > > > > int ret, i;
> > > > > >
> > > > > > - if (!config)
> > > > > > - return -ENODATA;
> > > > > > -
> > > > > > i3c = devm_kzalloc(&pdev->dev, sizeof(*i3c), GFP_KERNEL);
> > > > > > if (!i3c)
> > > > > > return -ENOMEM;
> > > > > > @@ -1317,19 +1311,11 @@ static int renesas_i3c_probe(struct platform_device *pdev)
> > > > > > if (IS_ERR(i3c->regs))
> > > > > > return PTR_ERR(i3c->regs);
> > > > > >
> > > > > > - clk = devm_clk_get_enabled(&pdev->dev, "pclk");
> > > > > > - if (IS_ERR(clk))
> > > > > > - return PTR_ERR(clk);
> > > > > > -
> > > > > > - if (config->has_pclkrw) {
> > > > > > - clk = devm_clk_get_enabled(&pdev->dev, "pclkrw");
> > > > > > - if (IS_ERR(clk))
> > > > > > - return PTR_ERR(clk);
> > > > > > - }
> > > > > > + ret = devm_clk_bulk_get_all_enabled(&pdev->dev, &i3c->clks);
> > > > > > + if (ret < 0)
> > > > > > + return ret;
> > > > > >
> > > > > > - i3c->tclk = devm_clk_get_enabled(&pdev->dev, "tclk");
> > > > > > - if (IS_ERR(i3c->tclk))
> > > > > > - return PTR_ERR(i3c->tclk);
> > > > > > + i3c->num_clks = ret;
> > > > >
> > > > > Need check num_clks > RENESAS_I3C_TCLK_IDX to avoid outbound
> > > > > access at
> > > > > i3c->clks[RENESAS_I3C_TCLK_IDX].clk
> > > >
> > > > I guess dt binding check validate this as well. Eg: a single clk
> > > > defined in the DT instead of minimum 2, will give DT warnings.
> > > >
> > > > Do you expect additional check in C code as well?
> > >
> > > Yes, worry about a wrong dtb cause kernel crash. Direct access
> > > i3c->clks[RENESAS_I3C_TCLK_IDX] without check is risk. if clock
> > > i3c->wrong in
> > > dtb,generally, it just impact function. but this may cause crash. So
> > > I suggest add addtional check here.
> >
> > OK, to avoid a crash num_clks > RENESAS_I3C_TCLK_IDX check is
> > sufficient
>
> Yes,
>
> > But it does not check whether tclk in 0th entry or first entry.
> >
> > >
> > > Or search 'tclk' in array i3c->clks.
> > > for (i = 0; i < i3c->num_clks; i++)
> > > if (!strcmp("tclk", i3c->clks[i]->id))
> > > break;
> >
> > But this returns correct index of the "tclk"
>
> This is just more flexiable for clk schema. If your schema is simple enough check num_clks >
> RENESAS_I3C_TCLK_IDX should be enough.
For avoiding crash that check is sufficient, but as you said wrong dtb like crash
can also lead to non-functional i3c device
Eg:
>From bindings, we expect the below entries in DTS for RZ/G3S
"pclk", "tclk" for RZ/G3S
But user mistakenly just added "pclk", that will lead to kernel crash
Or
Swap the clocks
"tclk", "pclk" this will lead to non-functional i3c device
On both cases, user ignored DT binding check warnings.
As you said crash is fatal, maybe stick with just num_clks > RENESAS_I3C_TCLK_IDX
Cheers,
Biju
Powered by blists - more mailing lists