linux-kernel - Re: [PATCH v4 02/35] staging: rtl8192e: Declare ethernet addresses as _

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: <5564D6C6.9040400@gmail.com>
Date:	Tue, 26 May 2015 22:25:42 +0200
From:	Mateusz Kulikowski <mateusz.kulikowski@...il.com>
To:	Dan Carpenter <dan.carpenter@...cle.com>
CC:	gregkh@...uxfoundation.org, joe@...ches.com,
	devel@...verdev.osuosl.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4 02/35] staging: rtl8192e: Declare ethernet addresses
 as __aligned(2)

On 25.05.2015 15:53, Dan Carpenter wrote:
> On Thu, May 21, 2015 at 05:53:15PM +0200, Mateusz Kulikowski wrote:
>> Add __aligned(2) into ethernet addresses allocated on stack or in non-packed
>> structures. Use ETH_ALEN as array length in places where it was hardcoded to 6.
> 
> To be honest, this patch makes me hate the stupid ether_addr_copy()
> checkpatch warnings even more than I hated it before.  Do we really need
> to add __aligned(2) to everything?  Is there no other option which isn't
> as horrible?
> 
> We could introduce a new typedef:
> 
> typedef struct {
>         char addr[6] __aligned(2);
> } eth_addr_t;
> 
> Introducing a new kind of type is maybe a bit much just to make
> checkpatch.pl happy but so is putting __aligned(2) everywhere.

True. I was a bit overzealous here with __aligned(2) and assumed someone will
break it by accident.

Proposal for v5: I'll verify alignment of each array, 
and reorder variables where possible/needed to keep the alignment.
Then I'll hopefully drop this patch.

> Or maybe we could make a new ether_addr_copy() implementation that uses
> preprocessor macros and chooses the right kind of implementation?

I'm not sure it can be handled compile time for generic case.

> 
>> diff --git a/drivers/staging/rtl8192e/rtl8192e/r8192E_dev.c b/drivers/staging/rtl8192e/rtl8192e/r8192E_dev.c
>> index aad5cc9..9c6dc82 100644
>> --- a/drivers/staging/rtl8192e/rtl8192e/r8192E_dev.c
>> +++ b/drivers/staging/rtl8192e/rtl8192e/r8192E_dev.c
>> @@ -321,7 +321,9 @@ static void rtl8192_read_eeprom_info(struct net_device *dev)
>>  	u8 ICVer8192, ICVer8256;
>>  	u16 i, usValue, IC_Version;
>>  	u16 EEPROMId;
>> -	u8 bMac_Tmp_Addr[6] = {0x00, 0xe0, 0x4c, 0x00, 0x00, 0x01};
>> +	static const u8 bMac_Tmp_Addr[ETH_ALEN] __aligned(2) = {
>> +					0x00, 0xe0, 0x4c, 0x00, 0x00, 0x01
>> +	};
> 
> What's up with the whacky indenting?

It was proposal from Joe; perhaps breaking 80-column line would be even better in that case 
(for sure will be once I remove __aligned)

Regards,
Mateusz

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/