Hi Jeffrey,
thanks a lot for this analysis. Any chance that somebody could
turn this into a patch I could try?
Kind regards
Andreas.
On Thu, Apr 30, 2020 at 03:40:12PM -0400, Jeffrey Walton wrote:
> On Fri, Apr 17, 2020 at 7:21 AM Andreas Tille <andr...@an3as.eu> wrote:
> > ...
> > So it seems the bus error occures somehow here:
> >
> >
> > https://salsa.debian.org/med-team/clustalo/-/blob/master/src/clustal/pair_dist.c#L346
>
> NewProgress is at
> https://salsa.debian.org/med-team/clustalo/-/blob/master/src/clustal/progress.c#L53.
> It uses a progress_t at
> https://salsa.debian.org/med-team/clustalo/-/blob/master/src/clustal/progress.h#L25.
>
> typedef struct {
> /* where to write to */
> FILE *prFile;
> /* prefix printed before each step */
> char *pcPrefix;
> bool bPrintCR;
> char pcLastLogMsg[1024];
> Stopwatch_t *prStopwatch;
> } progress_t;
>
> And stopwatch_t at
> https://salsa.debian.org/med-team/clustalo/-/blob/master/src/squid/stopwatch.h:
>
> struct stopwatch_s {
> time_t t0; /* Wall clock time, ANSI time() */
> #ifdef SRE_STRICT_ANSI
> clock_t cpu0; /* CPU time, ANSI clock() */
> #else
> struct tms cpu0; /* CPU/system time, POSIX times()*/
> #endif
>
> double elapsed; /* elapsed time, seconds */
> double user; /* CPU time, seconds */
> double sys; /* system time, seconds */
> };
>
> There are your doubles that are likely the problem.
>
> And what do we have here at
> https://salsa.debian.org/med-team/clustalo/-/blob/master/src/squid/stopwatch.c#L276
> ...
>
> void
> StopwatchPVMPack(Stopwatch_t *w)
> {
> pvm_pkdouble(&(w->elapsed), 1, 1);
> pvm_pkdouble(&(w->user), 1, 1);
> pvm_pkdouble(&(w->sys), 1, 1);
> }
> void
> StopwatchPVMUnpack(Stopwatch_t *w)
> {
> pvm_upkdouble(&(w->elapsed), 1, 1);
> pvm_upkdouble(&(w->user), 1, 1);
> pvm_upkdouble(&(w->sys), 1, 1);
> }
>
> I can't track the trail any further. The source code is missing for
> pvm_pkdouble and pvm_upkdouble. I would be very suspect of
> pvm_pkdouble and pvm_upkdouble.
>
> Jeff
>
--
http://fam-tille.de
