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
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <b883059e51c97d34196a1ad15bbec66a89283c8e.camel@pengutronix.de>
Date:   Mon, 21 Sep 2020 16:15:01 +0200
From:   Philipp Zabel <p.zabel@...gutronix.de>
To:     Bernard <bernard@...o.com>, dri-devel@...ts.freedesktop.org,
        linux-kernel@...r.kernel.org
Cc:     opensource.kernel@...o.com
Subject: Re: [Re-send][PATCH] gpu/ipu-v3:reduce protected code area in ipu
 idmac get/put

Hi Bernard,

On Mon, 2020-09-21 at 19:11 +0800, Bernard wrote:
> This change will speed-up a bit these ipu_idmac_get &
> ipu_idmac_put processing and there is no need to protect
> kzalloc & kfree.

I don't think that will be measurable, the channel lock is very unlikely
to be contended. It might make more sense to replace the list walk with
a bitfield.

> Signed-off-by: Bernard Zhao <bernard@...o.com>
> ---
>  drivers/gpu/ipu-v3/ipu-common.c | 24 +++++++++++++-----------
>  1 file changed, 13 insertions(+), 11 deletions(-)
> 
> diff --git a/drivers/gpu/ipu-v3/ipu-common.c b/drivers/gpu/ipu-v3/ipu-common.c
> index b3dae9ec1a38..8b3a57864c2e 100644
> --- a/drivers/gpu/ipu-v3/ipu-common.c
> +++ b/drivers/gpu/ipu-v3/ipu-common.c
> @@ -267,29 +267,30 @@ EXPORT_SYMBOL_GPL(ipu_rot_mode_to_degrees);
>  struct ipuv3_channel *ipu_idmac_get(struct ipu_soc *ipu, unsigned num)
>  {
>  	struct ipuv3_channel *channel;
> +	struct ipuv3_channel *entry;
>  
>  	dev_dbg(ipu->dev, "%s %d\n", __func__, num);
>  
>  	if (num > 63)
>  		return ERR_PTR(-ENODEV);
>  
> +	channel = kzalloc(sizeof(*channel), GFP_KERNEL);
> +	if (!channel)
> +		return ERR_PTR(-ENOMEM);
> +
> +	channel->num = num;
> +	channel->ipu = ipu;
> +
>  	mutex_lock(&ipu->channel_lock);
>  
> -	list_for_each_entry(channel, &ipu->channels, list) {
> -		if (channel->num == num) {
> +	list_for_each_entry(entry, &ipu->channels, list) {
> +		if (entry->num == num) {
> +			kfree(channel);
>  			channel = ERR_PTR(-EBUSY);
>  			goto out;

This leaks the channel memory allocated above.

>  		}
>  	}
> 
> -	channel = kzalloc(sizeof(*channel), GFP_KERNEL);
> -	if (!channel) {
> -		channel = ERR_PTR(-ENOMEM);
> -		goto out;
> -	}
> -
> -	channel->num = num;
> -	channel->ipu = ipu;
>  	list_add(&channel->list, &ipu->channels);
>  
>  out:
> @@ -308,9 +309,10 @@ void ipu_idmac_put(struct ipuv3_channel *channel)
>  	mutex_lock(&ipu->channel_lock);
>  
>  	list_del(&channel->list);
> -	kfree(channel);
>  
>  	mutex_unlock(&ipu->channel_lock);
> +
> +	kfree(channel);
>  }
>  EXPORT_SYMBOL_GPL(ipu_idmac_put);

regards
Philipp

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ