[bug #44555] "make" fails to use parallelism

[Jeff Epler]

Follow-up Comment #9, bug #44555 (project make):

While it doesn't account for the full 2.5+ minutes difference, "strace -c"'s
summary of syscalls shows that vfork() in a good version of make is less than
.02s, while fork() (show as clone()) in a bad version is about 75s
(547us/call), a difference which becomes worse as the memory size of make
increases (this is why having a lot of dependency information, as in my
generated makefile, is relevant).  Furthermore, the total number of clone()
syscalls is higher, I believe because more of them are interrupted by
SIGCHLD.

Slow trace:
% time     seconds  usecs/call     calls    errors syscall
------ ----------- ----------- --------- --------- ----------------
 99.97   78.092824         547    142754     47753 clone
  0.02    0.018415           0    105026           wait4
  0.01    0.004928           0    200005           rt_sigprocmask

Fast trace:
 74.59    0.013843           0    100334      5333 vfork
  7.51    0.001394           0    200005           rt_sigprocmask
  7.19    0.001335           0    110296           wait4


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?44555>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/