[Skiboot] [PATCH] sensors: dts: Fix uninitialised data in dts result when reading asynchronously

[Cédric Le Goater]

On 03/26/2018 07:39 AM, Nicholas Piggin wrote:
> On Sun, 25 Mar 2018 17:53:17 +1000
> Nicholas Piggin <npiggin at gmail.com> wrote:
> 
>> dts_async_read_temp() must zero its on-stack dts structure, similarly
>> to dts_sensor_read().
>>
>> Cc: Shilpasri G Bhat <shilpa.bhat at linux.vnet.ibm.com>
>> Cc: Cédric Le Goater <clg at kaod.org>
>> Signed-off-by: Nicholas Piggin <npiggin at gmail.com>
>> ---
>>
>> The used uninitialized warning showed up when I tried to compile
>> skiboot with -O2. Too bad -Os and other static checkers didn't
>> catch it.
>>
>> I excitedly thought this might also be the cause of the FIXME:,
>> but it looks like that predates this bug, and came over from the
>> P8 implementation. Has it been tested that the trip bits are still
>> wrong? Maybe we can remove it for p9 now?
>>
>> Thanks,
>> Nick
>> ---
>>
>>  hw/dts.c | 6 +-----
>>  1 file changed, 1 insertion(+), 5 deletions(-)
>>
>> diff --git a/hw/dts.c b/hw/dts.c
>> index 949791b0..38811fa9 100644
>> --- a/hw/dts.c
>> +++ b/hw/dts.c
>> @@ -228,11 +228,6 @@ static int dts_read_core_temp_p9(uint32_t pir, struct dts *dts)
>>  	prlog(PR_TRACE, "DTS: Chip %x Core %x temp:%dC trip:%x\n",
>>  	      chip_id, core, dts->temp, dts->trip);
>>  
>> -	/*
>> -	 * FIXME: The trip bits are always set ?! Just discard
>> -	 * them for the moment until we understand why.
>> -	 */
>> -	dts->trip = 0;
>>  	return 0;
>>  }
>>  
> 
> Oops, I intended to remove this hunk from the patch before submitting
> it. If it is good, it should be a separate patch.

The trip bits are still set to 0b11 == fatal from what I can see on a 
witherspoon.   

C.