Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
References: <20221107202114.823975-1-y.oudjana@protonmail.com>
From:   Aidan MacDonald <aidanmacdonald.0x0@gmail.com>
To:     Yassine Oudjana <yassine.oudjana@gmail.com>
Cc:     Mark Brown <broonie@kernel.org>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        "Rafael J. Wysocki" <rafael@kernel.org>,
        linux-kernel@vger.kernel.org,
        Yassine Oudjana <y.oudjana@protonmail.com>
Subject: Re: [PATCH] regmap-irq: Use the new num_config_regs property in
 regmap_add_irq_chip_fwnode
In-reply-to: <20221107202114.823975-1-y.oudjana@protonmail.com>
Date:   Mon, 07 Nov 2022 23:58:01 +0000
Message-ID: <YRVTxVzmyHg9azsCXIwib22zfh2O3Xr6@localhost>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk


Hi Yassine,

> Commit faa87ce9196d ("regmap-irq: Introduce config registers for irq
> types") added the num_config_regs, then commit 9edd4f5aee84 ("regmap-irq:
> Deprecate type registers and virtual registers") suggested to replace
> num_type_reg with it. However, regmap_add_irq_chip_fwnode wasn't modified
> to use the new property. Later on, commit 255a03bb1bb3 ("ASoC: wcd9335:
> Convert irq chip to config regs") removed the old num_type_reg property
> from the WCD9335 driver's struct regmap_irq_chip, causing a null pointer
> dereference in regmap_irq_set_type when it tried to index d->type_buf as
> it was never allocated in regmap_add_irq_chip_fwnode:

Whoops... My bad. Actually, d->type_buf shouldn't be accessed in this
configuration, the bug is in regmap_irq_set_type(). The access should
be guarded by "if (d->chip->type_in_mask || d->chip->num_type_reg)" to
prevent the NULL deref.

>
> Use num_config_regs in regmap_add_irq_chip_fwnode instead of num_type_reg,
> and fall back to it if num_config_regs isn't defined to maintain backward
> compatibility.
>
> Fixes: faa87ce9196d ("regmap-irq: Introduce config registers for irq types")
> Signed-off-by: Yassine Oudjana <y.oudjana@protonmail.com>
> ---
>  drivers/base/regmap/regmap-irq.c | 15 +++++++++++----
>  1 file changed, 11 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/base/regmap/regmap-irq.c b/drivers/base/regmap/regmap-irq.c
> index 4ef9488d05cd..3de89795f584 100644
> --- a/drivers/base/regmap/regmap-irq.c
> +++ b/drivers/base/regmap/regmap-irq.c
> @@ -722,6 +722,7 @@ int regmap_add_irq_chip_fwnode(struct fwnode_handle *fwnode,
>  	int i;
>  	int ret = -ENOMEM;
>  	int num_type_reg;
> +	int num_regs;
>  	u32 reg;
>
>  	if (chip->num_regs <= 0)
> @@ -796,14 +797,20 @@ int regmap_add_irq_chip_fwnode(struct fwnode_handle *fwnode,
>  			goto err_alloc;
>  	}
>
> -	num_type_reg = chip->type_in_mask ? chip->num_regs : chip->num_type_reg;
> -	if (num_type_reg) {
> -		d->type_buf_def = kcalloc(num_type_reg,
> +	/*
> +	 * Use num_config_regs if defined, otherwise fall back to num_type_reg
> +	 * to maintain backward compatibility.
> +	 */
> +	num_type_reg = chip->num_config_regs ? chip->num_config_regs
> +			: chip->num_type_reg;

A chip that uses config regs doesn't need type_buf. If you check
regmap_irq_sync_unlock(), you'll see the contents of type_buf aren't
written to the registers unless num_type_reg > 0.

If type_in_mask is set, type_buf has a completely different meaning.
There are no type registers -- instead each trigger has its own IRQ
enable bit. The type_buf records which bit (= trigger type) is used
for the logical IRQ.

Regards,
Aidan

> +	num_regs = chip->type_in_mask ? chip->num_regs : num_type_reg;
> +	if (num_regs) {
> +		d->type_buf_def = kcalloc(num_regs,
>  					  sizeof(*d->type_buf_def), GFP_KERNEL);
>  		if (!d->type_buf_def)
>  			goto err_alloc;
>
> -		d->type_buf = kcalloc(num_type_reg, sizeof(*d->type_buf),
> +		d->type_buf = kcalloc(num_regs, sizeof(*d->type_buf),
>  				      GFP_KERNEL);
>  		if (!d->type_buf)
>  			goto err_alloc;