2011-09-27 07:52:30

by Roland Vossen

[permalink] [raw]
Subject: Re: Fwd: [PATCH 16/20] staging: brcm80211: declared global vars in softmac phy as const

Hello Julian,

>> diff --git a/drivers/staging/brcm80211/brcmsmac/phy/phytbl_lcn.c b/drivers/staging/brcm80211/brcmsmac/phy/phytbl_lcn.c
>> index 15c5ffc..1b11b2f 100644
>> --- a/drivers/staging/brcm80211/brcmsmac/phy/phytbl_lcn.c
>> +++ b/drivers/staging/brcm80211/brcmsmac/phy/phytbl_lcn.c
>> @@ -1611,7 +1611,7 @@ const u32 dot11lcnphytbl_rx_gain_info_sz_rev0 =
>> sizeof(dot11lcnphytbl_rx_gain_info_rev0) /
>> sizeof(dot11lcnphytbl_rx_gain_info_rev0[0]);
>>
>> -static const u32 dot11lcnphytbl_rx_gain_info_sz_rev1 =
>> +const u32 dot11lcnphytbl_rx_gain_info_sz_rev1 =
>> sizeof(dot11lcnphytbl_rx_gain_info_rev1) /
>> sizeof(dot11lcnphytbl_rx_gain_info_rev1[0]);
>
> That seems like an anomaly. In a commit that is supposed
> to declare variables as const, the linkage of a variable
> changed.

The name of this commit could have been better. It should have been
'fixed sparse warnings on variable usage in softmac phy'. For what it is
worth: this variable is deleted in a later commit.

> Likewise, the removal of extern global variable
> declarations in phy_int.h also does not seem to belong
> here.

Agree.

The whole patch series will be resent anyway, let me split this commit
up in two to address your feedback.

Thanks, Roland.