[Bug 2783] Patch for disabling progress meter after file has finished uploading

bugzilla-daemon at bugzilla.mindrot.org bugzilla-daemon at bugzilla.mindrot.org
Mon Sep 25 07:55:57 AEST 2017


https://bugzilla.mindrot.org/show_bug.cgi?id=2783

--- Comment #2 from Damien Miller <djm at mindrot.org> ---
Comment on attachment 3059
  --> https://bugzilla.mindrot.org/attachment.cgi?id=3059
patch that fixes issue

This bug report doesn't make any sense: version 7.6p1 isn't released
yet and progressmeter.c doesn't contain the code you're modifying:

>-	if (cur_pos < end_pos) {
>-		signal(SIGALRM, update_progress_meter);
>-		alarm(UPDATE_INTERVAL);
>-		errno = save_errno;
>-	}
>+	signal(SIGALRM, update_progress_meter);
>+	alarm(UPDATE_INTERVAL);
>+	errno = save_errno;

The existing code already reads as after your patch is applied:

   241          if (can_output())
   242                  refresh_progress_meter();
   243  
   244          signal(SIGALRM, update_progress_meter);
   245          alarm(UPDATE_INTERVAL);
   246          errno = save_errno;

Generally though, apparent "hangs" at 100% of the progress meter
display are caused by buffering happening in the sending system ssh
process, kernel socket buffers, the network or at the receiving side.
Stopping the clock until the remote end has acknowledged receipt of the
entire file is *correcting* the time and bandwidth calculations that
are being distorted by the aforementioned buffering.

-- 
You are receiving this mail because:
You are watching the assignee of the bug.


More information about the openssh-bugs mailing list