linux-kernel - Re: [PATCH net-next 3/6] net: lan966x: add support for interrupts from analyzer

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <20211206182456.4494c5f6@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date:   Mon, 6 Dec 2021 18:24:56 -0800
From:   Jakub Kicinski <kuba@...nel.org>
To:     Horatiu Vultur <horatiu.vultur@...rochip.com>
Cc:     <davem@...emloft.net>, <robh+dt@...nel.org>,
        <UNGLinuxDriver@...rochip.com>, <linux@...linux.org.uk>,
        <netdev@...r.kernel.org>, <devicetree@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH net-next 3/6] net: lan966x: add support for interrupts
 from analyzer

On Fri, 3 Dec 2021 11:46:42 +0100 Horatiu Vultur wrote:
> This patch adds support for handling the interrupts generated by the
> analyzer. Currently, only the MAC table generates these interrupts.
> The MAC table will generate an interrupt whenever it learns or forgets
> an entry in the table. It is the SW responsibility figure out which
> entries were added/removed.
> 
> Signed-off-by: Horatiu Vultur <horatiu.vultur@...rochip.com>

> +static struct lan966x_mac_entry *lan966x_mac_alloc_entry(struct lan966x *lan966x,
> +							 const unsigned char *mac,
> +							 u16 vid, u16 port_index)
> +{
> +	struct lan966x_mac_entry *mac_entry;
> +
> +	mac_entry = devm_kzalloc(lan966x->dev,
> +				 sizeof(*mac_entry), GFP_ATOMIC);

Is it really necessary to use devm_ allocation for the mac entries?
It's 2x memory overhead.

Also why GFP_ATOMIC? Memory allocations are _a lot_ less likely with
GFP_KERNEL.

> +	if (!mac_entry)
> +		return NULL;
> +
> +	memcpy(mac_entry->mac, mac, ETH_ALEN);
> +	mac_entry->vid = vid;
> +	mac_entry->port_index = port_index;
> +	mac_entry->row = LAN966X_MAC_INVALID_ROW;
> +	return mac_entry;
> +}

> +static void lan966x_mac_process_raw_entry(struct lan966x_mac_raw_entry *raw_entry,
> +					  u8 *mac, u16 *vid, u32 *dest_idx)
> +{
> +	mac[0] = (raw_entry->mach >> 8)  & 0xff;
> +	mac[1] = (raw_entry->mach >> 0)  & 0xff;
> +	mac[2] = (raw_entry->macl >> 24) & 0xff;
> +	mac[3] = (raw_entry->macl >> 16) & 0xff;
> +	mac[4] = (raw_entry->macl >> 8)  & 0xff;
> +	mac[5] = (raw_entry->macl >> 0)  & 0xff;
> +
> +	*vid = (raw_entry->mach >> 16) & 0xfff;
> +	*dest_idx  = ANA_MACACCESS_DEST_IDX_GET(raw_entry->maca);

Double space before =

> +}
> +
> +static void lan966x_mac_irq_process(struct lan966x *lan966x, u32 row,
> +				    struct lan966x_mac_raw_entry *raw_entries)
> +{
> +	struct lan966x_mac_entry *mac_entry, *tmp;
> +	unsigned long flags;
> +	char mac[ETH_ALEN];
> +	u32 dest_idx;
> +	u32 column;
> +	u16 vid;
> +
> +	spin_lock_irqsave(&lan966x->mac_lock, flags);
> +	list_for_each_entry_safe(mac_entry, tmp, &lan966x->mac_entries, list) {
> +		bool founded = false;

s/founded/found/

> +		if (mac_entry->row != row)
> +			continue;
> +
> +		for (column = 0; column < LAN966X_MAC_COLUMNS; ++column) {
> +			/* All the valid entries are at the start of the row,
> +			 * so when get one invalid entry it can just skip the
> +			 * rest of the columns
> +			 */
> +			if (!ANA_MACACCESS_VALID_GET(raw_entries[column].maca))
> +				break;
> +
> +			lan966x_mac_process_raw_entry(&raw_entries[column],
> +						      mac, &vid, &dest_idx);
> +			WARN_ON(dest_idx > lan966x->num_phys_ports);
> +
> +			/* If the entry in SW is found, then there is nothing
> +			 * to do
> +			 */
> +			if (mac_entry->vid == vid &&
> +			    ether_addr_equal(mac_entry->mac, mac) &&

You need to add __aligned(2) to mac, ether_addr_equal() needs aligned
arguments.

> +			    mac_entry->port_index == dest_idx) {
> +				raw_entries[column].process = true;
> +				founded = true;
> +				break;
> +			}
> +		}