Bug#956324: Clustalo bus error on mipsel (Was: Bug#956324: python-biopython: FTBFS on mipsel)

> On May 1, 2020, at 00:27, Jeffrey Walton <[email protected]> wrote:
> 
> On Fri, May 1, 2020 at 3:05 AM Jeffrey Walton <[email protected] <mailto:[email protected]>> wrote:
>> 
>> On Fri, May 1, 2020 at 2:14 AM Andreas Tille <[email protected]> wrote:
>>> 
>>> ...
>>> ==13209== Process terminating with default action of signal 10 (SIGBUS)
>>> ==13209==    at 0x12D5CC: PairDistances (pair_dist.c:346)
>>> ==13209==    by 0x119410: AlignmentOrder (clustal-omega.c:835)
>>> ==13209==    by 0x11A6C4: Align (clustal-omega.c:1221)
>>> ==13209==    by 0x1171C8: MyMain (mymain.c:1192)
>>> ==13209==    by 0x113CCC: main (main.cpp:469)
>> 
>> Here is line 346 in
>> https://salsa.debian.org/med-team/clustalo/-/blob/master/src/clustal/pair_dist.c#L346:
>> 
>>            NewProgress(&prProgress, LogGetFP(&rLog, LOG_INFO),
>>                        "Ktuple-distance calculation progress", bPrintCR);
>> 
>> For testing, change LogGetFP(&rLog, LOG_INFO) for stdout for testing. I.e.,
>> 
>>            NewProgress(&prProgress, stdout,,
>>                        "Ktuple-distance calculation progress", bPrintCR);
>> 
>> It looks like LogGetFP retrieves an entry in an array of FILE*. From
>> https://salsa.debian.org/med-team/clustalo/-/blob/master/src/clustal/log.h:
>> 
>> typedef struct {
>>    /* the higher the level, the more priority it has. numbers must be
>>     *  sequential
>>     */
>> 
>>    /* array of function pointers */
>>    void (*prFunc[LOG_NUM_LEVELS]) (FILE *prFP, char *pcFormat,
>> va_list rVArgList);
>>    FILE *prFP[LOG_NUM_LEVELS];
>>    char *prPrefix[LOG_NUM_LEVELS];
>> 
>>    /* everything above this level will be printed */
>>    int iLogLevelEnabled;
>> } log_t;
>> 
>> And https://salsa.debian.org/med-team/clustalo/-/blob/master/src/clustal/log.c:
>> 
>> FILE *
>> LogGetFP(log_t *prLog, int iLevel)
>> {
>>    assert(iLevel>=0 && iLevel<=LOG_NUM_LEVELS);
>>    return prLog->prFP[iLevel];
>> }
>> 
>> That should help determine if something is sideways in the log_t structure.
> 
> Also, I think this should be:
> 
>> FILE *
>> LogGetFP(log_t *prLog, int iLevel)
>> {
>>    assert(iLevel>=0 && iLevel<LOG_NUM_LEVELS);
>>    return prLog->prFP[iLevel];
>> }
> 
> That is, 'iLevel<LOG_NUM_LEVELS' (not 'iLevel<=LOG_NUM_LEVELS’).

Some further patches attached:
  1. the log level correction Jeff noted.
  2. remove a prototype for a non-existent function
  3. delete a lot of dead code, aiming to minimise the number of lines we have to suspect in ongoing debugging. This is for debugging only; not intended to be applied during packaging.