On 12/06/2012 07:32 AM, Joe Perches wrote:
> On Wed, 2012-12-05 at 22:28 +0100, Arend van Spriel wrote:
>> The call to brcmf_dbg(ERROR, ...) only resulted in a log message
>> when compiled with -DDEBUG. Error messages are valuable for resolving
>> issues so this patch replaces it with brcmf_err(...) so they always
>> end up in the log.
> []
>> Reviewed-by: Joe Perches <joe@xxxxxxxxxxx>
>
> Don't add signatures unless the person actually gives one.
> I didn't.
Sorry about that.
> []
>
>> diff --git a/drivers/net/wireless/brcm80211/brcmfmac/dhd_dbg.h b/drivers/net/wireless/brcm80211/brcmfmac/dhd_dbg.h
> []
>> @@ -35,20 +34,11 @@
>>
>> #if defined(DEBUG)
>>
>> -#define brcmf_dbg(level, fmt, ...) \
>> -do { \
>> - if (BRCMF_ERROR_VAL == BRCMF_##level##_VAL) { \
>> - if (brcmf_msg_level & BRCMF_##level##_VAL) { \
>> - if (net_ratelimit()) \
>> - pr_debug("%s: " fmt, \
>> - __func__, ##__VA_ARGS__); \
>> - } \
>> - } else { \
>> - if (brcmf_msg_level & BRCMF_##level##_VAL) { \
>> - pr_debug("%s: " fmt, \
>> - __func__, ##__VA_ARGS__); \
>> - } \
>> - } \
>> +#define brcmf_err(fmt, ...) pr_err("%s: " fmt, __func__, ##__VA_ARGS__)
>
> You still lost the net_ratelimit() for the DEBUG case.
>
> This brcmf_err macro should probably be in dhd.h,
> not in a debug header and not guarded by DEBUG at all.
I thought I explained why in my email response to you, but now I don't
see that one on the mailing list archives (attached now). When we are
debugging we do not want error messages to be filtered.
Gr. AvS
Attachment:
Re: [PATCH 05_11] brcmfmac: error messages should not be suppressed.eml
Description: application/extension-eml
