Nicholas Mc Guire <der.herr@xxxxxxx> writes:
> On Sat, 13 Jun 2015, Jes Sorensen wrote:
>
>> Nicholas Mc Guire <hofrat@xxxxxxxxx> writes:
>> > scanning for trivial bug-patters with coccinelle spatches returned:
>> > ./drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c:1395
>> > WARNING: condition with no effect (if branch == else)
>> >
>> > drivers/staging/rtl8723au/hal/rtl8723a_hal_init.c - line numbers
>> > from 4.1-rc7
>> > 1395 if (bWithoutHWSM) {
>> > 1396 /* value16 |= (APDM_HOST | FSM_HSUS |/PFM_ALDN); */
>> > 1397 /* 2010/08/31 According to Filen description, we need to
>> > 1398 use HW to shut down 8051 automatically. */
>> > 1399 /* Because suspend operation need the asistance of 8051
>> > 1400 to wait for 3ms. */
>> > 1401 value16 = APDM_HOST | AFSM_HSUS | PFM_ALDN;
>> > 1402 } else {
>> > 1403 value16 = APDM_HOST | AFSM_HSUS | PFM_ALDN;
>> > 1404 }
>> > 1405
>> > 1406 rtl8723au_write16(padapter, REG_APS_FSMCO, value16); /* 0x4802 */
>> >
>> > Not clear what the intent is but this looks like a typo/bug in the assigment
>> > of value16 as the condition here has no effect.
>>
>> Doesn't look like a typo, looks like someone at some point had commented
>> out PFM_ALDN in the bWithoutHWSM case. Why they did that and then later
>> overrode it, I have no idea.
>>
> as far as its traceable in git log if == else was in there from the
> very beginning (including that comment) - both the if and else were
> updated in commit cffca68d7b2f ("staging: rtl8723au: _DisableAnalog(): Avoid
> zero-init variables unnecessaril") but without changing PFM_ALDN - as there
> are no comments to the meaning of these bits in the header file there is
> no way to really come up with a resonable patch. anyway its either wrong
> settings or the condition should be removed.
There's a much longer history of this code going back to the vendor
origin - it shouldn't be removed unless we can get clarification from
them why that was.
Jes
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
