Received-SPF: pass (google.com: domain of linux-kernel+bounces-70363-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Message-ID: <7260ca36-fd92-4bf5-86b1-fb91393ad064@redhat.com>
Date: Sun, 18 Feb 2024 12:01:44 +0100
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH 2/2] platform/x86/amd/pmf: Fixup error handling for
 amd_pmf_init_smart_pc()
Content-Language: en-US, nl
To: Mario Limonciello <mario.limonciello@amd.com>, Shyam-sundar.S-k@amd.com
Cc: platform-driver-x86@vger.kernel.org, linux-kernel@vger.kernel.org,
 Linux regressions mailing list <regressions@lists.linux.dev>
References: <20240217012205.113614-1-mario.limonciello@amd.com>
 <20240217012205.113614-3-mario.limonciello@amd.com>
From: Hans de Goede <hdegoede@redhat.com>
In-Reply-To: <20240217012205.113614-3-mario.limonciello@amd.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit

Hi Mario,

Thank you for your patches.

On 2/17/24 02:22, Mario Limonciello wrote:
> amd_pmf_init_smart_pc() calls out to amd_pmf_get_bios_buffer() but
> the error handling flow doesn't clean everything up all allocated
> memory.
> 
> As amd_pmf_get_bios_buffer() is only called by amd_pmf_init_smart_pc(),
> fold it into the function and add labels to clean up any step that
> can fail along the way.
> 
> Fixes: 7c45534afa44 ("platform/x86/amd/pmf: Add support for PMF Policy Binary")
> Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
> ---
>  drivers/platform/x86/amd/pmf/tee-if.c | 67 ++++++++++++++++++---------
>  1 file changed, 44 insertions(+), 23 deletions(-)
> 
> diff --git a/drivers/platform/x86/amd/pmf/tee-if.c b/drivers/platform/x86/amd/pmf/tee-if.c
> index 1359ab340f7c..feb9dfafea30 100644
> --- a/drivers/platform/x86/amd/pmf/tee-if.c
> +++ b/drivers/platform/x86/amd/pmf/tee-if.c
> @@ -338,25 +338,6 @@ static void amd_pmf_remove_pb(struct amd_pmf_dev *dev) {}
>  static void amd_pmf_hex_dump_pb(struct amd_pmf_dev *dev) {}
>  #endif
>  
> -static int amd_pmf_get_bios_buffer(struct amd_pmf_dev *dev)
> -{
> -	dev->policy_buf = kzalloc(dev->policy_sz, GFP_KERNEL);
> -	if (!dev->policy_buf)
> -		return -ENOMEM;
> -
> -	dev->policy_base = devm_ioremap(dev->dev, dev->policy_addr, dev->policy_sz);
> -	if (!dev->policy_base)
> -		return -ENOMEM;
> -
> -	memcpy(dev->policy_buf, dev->policy_base, dev->policy_sz);
> -
> -	amd_pmf_hex_dump_pb(dev);
> -	if (pb_side_load)
> -		amd_pmf_open_pb(dev, dev->dbgfs_dir);
> -
> -	return amd_pmf_start_policy_engine(dev);
> -}
> -
>  static int amd_pmf_amdtee_ta_match(struct tee_ioctl_version_data *ver, const void *data)
>  {
>  	return ver->impl_id == TEE_IMPL_ID_AMDTEE;
> @@ -454,14 +435,54 @@ int amd_pmf_init_smart_pc(struct amd_pmf_dev *dev)
>  	if (ret)
>  		return ret;
>  
> +	ret = amd_pmf_set_dram_addr(dev, true);
> +	if (ret)
> +		goto out_dram;
> +
> +	dev->policy_base = devm_ioremap(dev->dev, dev->policy_addr, dev->policy_sz);
> +	if (!dev->policy_base) {
> +		ret = -ENOMEM;
> +		goto out_policy_base;
> +	}
> +
> +	dev->policy_buf = kzalloc(dev->policy_sz, GFP_KERNEL);
> +	if (!dev->policy_buf) {
> +		ret = -ENOMEM;
> +		goto out_policy_buf;
> +	}
> +
> +	memcpy(dev->policy_buf, dev->policy_base, dev->policy_sz);
> +
> +	amd_pmf_hex_dump_pb(dev);
> +	if (pb_side_load)
> +		amd_pmf_open_pb(dev, dev->dbgfs_dir);
> +
>  	INIT_DELAYED_WORK(&dev->pb_work, amd_pmf_invoke_cmd);
> -	amd_pmf_set_dram_addr(dev, true);
> -	amd_pmf_get_bios_buffer(dev);
> +
> +	ret = amd_pmf_start_policy_engine(dev);
> +	if (ret)
> +		goto out_start_engine;
> +
>  	dev->prev_data = kzalloc(sizeof(*dev->prev_data), GFP_KERNEL);
>  	if (!dev->prev_data)

I just checked and dev->prev_data gets used by dev->pb_work
which gets queued by amd_pmf_start_policy_engine() so there
is a (pre-existing) race here and dev->prev_data should be allocated
before amd_pmf_start_policy_engine().

Note kfree(NULL) is a no-op as is cancel_delayed_work_sync()
on a non-queued work.

So I think you can just use a single error_exit: label 
(or any other label-name you like) and do:

error_exit:
	cancel_delayed_work_sync(&dev->pb_work);
	kfree(dev->prev_data);
	dev->prev_data = NULL;
	kfree(dev->policy_buf);
	dev->policy_buf = NULL;
	kfree(dev->buf);
	dev->buf = NULL;
	amd_pmf_tee_deinit(dev);
	return ret;

There as long as you do the INIT_DELAYED_WORK() before
any of the code which may fail with a goto error_exit.

Note I also added clearing of the pointers after freeing
them, at least for dev->buf this is important because
that also gets used in non smart-pc paths and those
count on it either being NULL or a valid pointer.

Regards,

Hans


> -		return -ENOMEM;
> +		goto out_prev_data;
> +
> +	return 0;
>  
> -	return dev->smart_pc_enabled;
> +out_prev_data:
> +	cancel_delayed_work_sync(&dev->pb_work);
> +
> +out_start_engine:
> +	kfree(dev->policy_buf);
> +
> +out_policy_buf:
> +out_policy_base:
> +	kfree(dev->buf);
> +
> +out_dram:
> +	amd_pmf_tee_deinit(dev);
> +
> +	return ret;
>  }
>  
>  void amd_pmf_deinit_smart_pc(struct amd_pmf_dev *dev)