From: Ivan Kuten <ivan.kuten@promwad.com>
To: linux-wireless@vger.kernel.org
Cc: Johannes Berg <johannes@sipsolutions.net>,
Yauhen Kharuzhy <yauhen.kharuzhy@promwad.com>
Subject: Re: [PATCH] wireless: Compare ethernet addresses by unaligned safe way
Date: Thu, 23 Jul 2009 20:56:53 +0300 [thread overview]
Message-ID: <4A68A465.6020501@promwad.com> (raw)
In-Reply-To: <1246262320.5947.3.camel@johannes.local>
> On Sun, 2009-06-28 at 15:18 +0300, Ivan Kuten wrote:
>> Hello,
>>
>> In net/wireless/scan.c : cfg80211_wext_siwscan there seems also unaligned allocations
>> for creq->ssids and creq->channels. Should it be something like that?
>
> Seems alright, but there is more than one instance of this, maybe you
> can make a function to allocate a scan request properly and have it be
> called from all the places it's needed.
>
> johannes
>
Hello Johannes,
Can you point to that multiple scan allocations? I see only one kzalloc
with followed possible alignment violation - it's in cfg80211_wext_siwscan in scan.c
Regards,
Ivan
>> Modified: trunk/uClinux-dist-2008R1-RC8/compat-wireless-2009-06-11/net/wireless/scan.c
>> ==============================================================================
>> --- trunk/uClinux-dist-2008R1-RC8/compat-wireless-2009-06-11/net/wireless/scan.c (original)
>> +++ trunk/uClinux-dist-2008R1-RC8/compat-wireless-2009-06-11/net/wireless/scan.c Fri Jun 26 14:00:52 2009
>> @@ -619,7 +619,7 @@
>> if (wiphy->bands[band])
>> n_channels += wiphy->bands[band]->n_channels;
>>
>> - creq = kzalloc(sizeof(*creq) + sizeof(struct cfg80211_ssid) +
>> + creq = kzalloc(roundup(sizeof(*creq), 4) + roundup(sizeof(struct cfg80211_ssid), 4) +
>> n_channels * sizeof(void *),
>> GFP_ATOMIC);
>> if (!creq) {
>> @@ -629,8 +629,8 @@
>>
>> creq->wiphy = wiphy;
>> creq->ifidx = dev->ifindex;
>> - creq->ssids = (void *)(creq + 1);
>> - creq->channels = (void *)(creq->ssids + 1);
>> + creq->ssids = (void *)creq + roundup(sizeof(*creq), 4);
>> + creq->channels = (void *)creq->ssids + roundup(sizeof(*creq->ssids), 4);
>> creq->n_channels = n_channels;
>> creq->n_ssids = 1;
>>
>> Regards,
>> Ivan
>>
>>> On Tue, 2009-06-16 at 13:54 +0300, Yauhen Kharuzhy wrote:
>>>> When we try to run RTL8187 driver on AD BlackFin platform, we got
>>>> messages from kernel about unaligned memory access at
>>>> compare_ether_addr() calls.
>>>>
>>>> Replacing of compare_ether_addr() by memcmp() fixes this problem.
>>> This shouldn't be necessary. Which operand is unaligned?
>>>
>>>> --- a/net/mac80211/ibss.c
>>>> +++ b/net/mac80211/ibss.c
>>>> @@ -395,7 +395,7 @@ struct sta_info *ieee80211_ibss_add_sta(struct ieee80211_sub_if_data *sdata,
>>>> return NULL;
>>>> }
>>>>
>>>> - if (compare_ether_addr(bssid, sdata->u.ibss.bssid))
>>>> + if (memcmp(bssid, sdata->u.ibss.bssid, ETH_ALEN))
>>>> return NULL;
>>> So in this case it seems that it is possible that u.ibss.bssid is not
>>> aligned, consider fixing by doing
>>>
>>> --- ieee80211_i.h
>>> +++ ieee80211_i.h
>>> - u8 bssid[ETH_ALEN];
>>> + u8 bssid[ETH_ALEN] __align(2);
>>>
>>> or so instead.
>>>
>>>> --- a/net/wireless/scan.c
>>>> +++ b/net/wireless/scan.c
>>>> @@ -134,7 +134,7 @@ static bool is_bss(struct cfg80211_bss *a,
>>>> {
>>>> const u8 *ssidie;
>>>>
>>>> - if (bssid && compare_ether_addr(a->bssid, bssid))
>>>> + if (bssid && memcmp(a->bssid, bssid, ETH_ALEN))
>>> Since a->bssid is after a pointer I can't see how it would be unaligned,
>>> and bssid should be unaligned only if the call trace shows it's coming
>>> from the above u.ibss.bssid.
>>>
>>> johannes
>>
next prev parent reply other threads:[~2009-07-23 17:57 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-06-16 10:54 [PATCH] wireless: Compare ethernet addresses by unaligned safe way Yauhen Kharuzhy
2009-06-16 11:14 ` Johannes Berg
2009-06-28 12:18 ` Ivan Kuten
2009-06-29 7:58 ` Johannes Berg
2009-07-23 17:56 ` Ivan Kuten [this message]
2009-07-23 17:59 ` Johannes Berg
2009-08-07 15:34 ` John W. Linville
2009-06-29 8:23 ` Johannes Berg
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4A68A465.6020501@promwad.com \
--to=ivan.kuten@promwad.com \
--cc=johannes@sipsolutions.net \
--cc=linux-wireless@vger.kernel.org \
--cc=yauhen.kharuzhy@promwad.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).