[net-next] airo: work around stack usage warning

From: Arnd Bergmann <arnd@arndb.de>


gcc-11 with KASAN on 32-bit arm produces a warning about a function
that needs a lot of stack space:

drivers/net/wireless/cisco/airo.c: In function 'setup_card.constprop':
drivers/net/wireless/cisco/airo.c:3960:1: error: the frame size of 1512 bytes is larger than 1400 bytes [-Werror=frame-larger-than=]

Most of this is from a single large structure that could be dynamically
allocated or moved into the per-device structure.  However, as the callers
all seem to have a fairly well bounded call chain, the easiest change
is to pull out the part of the function that needs the large variables
into a separate function and mark that as noinline_for_stack. This does
not reduce the total stack usage, but it gets rid of the warning and
requires minimal changes otherwise.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>

---
 drivers/net/wireless/cisco/airo.c | 117 +++++++++++++++++-------------
 1 file changed, 65 insertions(+), 52 deletions(-)

-- 
2.29.2

On 3/23/21 6:16 AM, Arnd Bergmann wrote:
> From: Arnd Bergmann <arnd@arndb.de>
>
> gcc-11 with KASAN on 32-bit arm produces a warning about a function
> that needs a lot of stack space:
>
> drivers/net/wireless/cisco/airo.c: In function 'setup_card.constprop':
> drivers/net/wireless/cisco/airo.c:3960:1: error: the frame size of 1512 bytes is larger than 1400 bytes [-Werror=frame-larger-than=]
>
> Most of this is from a single large structure that could be dynamically
> allocated or moved into the per-device structure.  However, as the callers
> all seem to have a fairly well bounded call chain, the easiest change
> is to pull out the part of the function that needs the large variables
> into a separate function and mark that as noinline_for_stack. This does
> not reduce the total stack usage, but it gets rid of the warning and
> requires minimal changes otherwise.
>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/net/wireless/cisco/airo.c | 117 +++++++++++++++++-------------
>  1 file changed, 65 insertions(+), 52 deletions(-)
>
> diff --git a/drivers/net/wireless/cisco/airo.c b/drivers/net/wireless/cisco/airo.c
> index e35e1380ae43..540ba694899c 100644
> --- a/drivers/net/wireless/cisco/airo.c
> +++ b/drivers/net/wireless/cisco/airo.c
> @@ -3818,6 +3818,68 @@ static inline void set_auth_type(struct airo_info *local, int auth_type)
>  		local->last_auth = auth_type;
>  }
>  
> +static int noinline_for_stack airo_readconfig(struct airo_info *ai, u8 *mac, int lock)
> +{
> +	int i, status;
> +	/* large variables, so don't inline this function,
> +	 * maybe change to kmalloc
> +	 */
> +	tdsRssiRid rssi_rid;
> +	CapabilityRid cap_rid;
> +
> +	kfree(ai->SSID);
> +	ai->SSID = NULL;
> +	// general configuration (read/modify/write)
> +	status = readConfigRid(ai, lock);
> +	if (status != SUCCESS) return ERROR;
> +
> +	status = readCapabilityRid(ai, &cap_rid, lock);
> +	if (status != SUCCESS) return ERROR;
> +
> +	status = PC4500_readrid(ai, RID_RSSI, &rssi_rid, sizeof(rssi_rid), lock);
> +	if (status == SUCCESS) {
> +		if (ai->rssi || (ai->rssi = kmalloc(512, GFP_KERNEL)) != NULL)
> +			memcpy(ai->rssi, (u8*)&rssi_rid + 2, 512); /* Skip RID length member */
> +	}
> +	else {
> +		kfree(ai->rssi);
> +		ai->rssi = NULL;
> +		if (cap_rid.softCap & cpu_to_le16(8))
> +			ai->config.rmode |= RXMODE_NORMALIZED_RSSI;
> +		else
> +			airo_print_warn(ai->dev->name, "unknown received signal "
> +					"level scale");
> +	}
> +	ai->config.opmode = adhoc ? MODE_STA_IBSS : MODE_STA_ESS;
> +	set_auth_type(ai, AUTH_OPEN);
> +	ai->config.modulation = MOD_CCK;
> +
> +	if (le16_to_cpu(cap_rid.len) >= sizeof(cap_rid) &&
> +	    (cap_rid.extSoftCap & cpu_to_le16(1)) &&
> +	    micsetup(ai) == SUCCESS) {
> +		ai->config.opmode |= MODE_MIC;
> +		set_bit(FLAG_MIC_CAPABLE, &ai->flags);
> +	}
> +
> +	/* Save off the MAC */
> +	for (i = 0; i < ETH_ALEN; i++) {
> +		mac[i] = ai->config.macAddr[i];
> +	}
> +
> +	/* Check to see if there are any insmod configured
> +	   rates to add */
> +	if (rates[0]) {
> +		memset(ai->config.rates, 0, sizeof(ai->config.rates));
> +		for (i = 0; i < 8 && rates[i]; i++) {
> +			ai->config.rates[i] = rates[i];
> +		}
> +	}
> +	set_bit (FLAG_COMMIT, &ai->flags);
> +
> +	return SUCCESS;
> +}
> +
> +
>  static u16 setup_card(struct airo_info *ai, u8 *mac, int lock)
>  {
>  	Cmd cmd;
> @@ -3864,58 +3926,9 @@ static u16 setup_card(struct airo_info *ai, u8 *mac, int lock)
>  	if (lock)
>  		up(&ai->sem);
>  	if (ai->config.len == 0) {
> -		int i;
> -		tdsRssiRid rssi_rid;
This is mostly an array, and what I guess is the problem.
> -		CapabilityRid cap_rid;
> -
> -		kfree(ai->SSID);
> -		ai->SSID = NULL;
> -		// general configuration (read/modify/write)
> -		status = readConfigRid(ai, lock);
> -		if (status != SUCCESS) return ERROR;
> -
> -		status = readCapabilityRid(ai, &cap_rid, lock);
> -		if (status != SUCCESS) return ERROR;
> -
> -		status = PC4500_readrid(ai, RID_RSSI,&rssi_rid, sizeof(rssi_rid), lock);
This is reading into a stack temp
> -		if (status == SUCCESS) {
> -			if (ai->rssi || (ai->rssi = kmalloc(512, GFP_KERNEL)) != NULL)
> -				memcpy(ai->rssi, (u8*)&rssi_rid + 2, 512); /* Skip RID length member */

Here is the temp being copied to heap memory.

This is the only use of rssi_rid

Why not do the alloc in PC4500_read ?

The call would be something like

status = PC4500_readrid(ai, RID_RSSI, &ai->rssi, sizeof(), lock)

or since ai, is common reduce the signature of the function and make the call

PC4500_readid(ai, RID_RSSI, lock)

Tom

> -		}
> -		else {
> -			kfree(ai->rssi);
> -			ai->rssi = NULL;
> -			if (cap_rid.softCap & cpu_to_le16(8))
> -				ai->config.rmode |= RXMODE_NORMALIZED_RSSI;
> -			else
> -				airo_print_warn(ai->dev->name, "unknown received signal "
> -						"level scale");
> -		}
> -		ai->config.opmode = adhoc ? MODE_STA_IBSS : MODE_STA_ESS;
> -		set_auth_type(ai, AUTH_OPEN);
> -		ai->config.modulation = MOD_CCK;
> -
> -		if (le16_to_cpu(cap_rid.len) >= sizeof(cap_rid) &&
> -		    (cap_rid.extSoftCap & cpu_to_le16(1)) &&
> -		    micsetup(ai) == SUCCESS) {
> -			ai->config.opmode |= MODE_MIC;
> -			set_bit(FLAG_MIC_CAPABLE, &ai->flags);
> -		}
> -
> -		/* Save off the MAC */
> -		for (i = 0; i < ETH_ALEN; i++) {
> -			mac[i] = ai->config.macAddr[i];
> -		}
> -
> -		/* Check to see if there are any insmod configured
> -		   rates to add */
> -		if (rates[0]) {
> -			memset(ai->config.rates, 0, sizeof(ai->config.rates));
> -			for (i = 0; i < 8 && rates[i]; i++) {
> -				ai->config.rates[i] = rates[i];
> -			}
> -		}
> -		set_bit (FLAG_COMMIT, &ai->flags);
> +		status = airo_readconfig(ai, mac, lock);
> +		if (status != SUCCESS)
> +			return ERROR;
>  	}
>  
>  	/* Setup the SSIDs if present */

Arnd Bergmann <arnd@kernel.org> wrote:

> From: Arnd Bergmann <arnd@arndb.de>
> 
> gcc-11 with KASAN on 32-bit arm produces a warning about a function
> that needs a lot of stack space:
> 
> drivers/net/wireless/cisco/airo.c: In function 'setup_card.constprop':
> drivers/net/wireless/cisco/airo.c:3960:1: error: the frame size of 1512 bytes is larger than 1400 bytes [-Werror=frame-larger-than=]
> 
> Most of this is from a single large structure that could be dynamically
> allocated or moved into the per-device structure.  However, as the callers
> all seem to have a fairly well bounded call chain, the easiest change
> is to pull out the part of the function that needs the large variables
> into a separate function and mark that as noinline_for_stack. This does
> not reduce the total stack usage, but it gets rid of the warning and
> requires minimal changes otherwise.
> 
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>

Patch applied to wireless-drivers-next.git, thanks.

7909a590eba6 airo: work around stack usage warning