On Mon, Aug 3, 2020 at 12:09 PM enh <e...@google.com> wrote: > On Mon, Aug 3, 2020 at 3:52 AM Rob Landley <r...@landley.net> wrote: > > > > On 7/31/20 4:19 PM, Mark Salyzyn via Toybox wrote: > > > On 7/31/20 12:30 PM, Mark Salyzyn wrote: > > >> -P flag was fake just to provide compatibility. Add support for -P > > >> maximum process count to actually create parallelism. Does not > > >> support SIGUSR1 or SIGUSR2 increment/decrement signals as publicly > > >> documented for other variants. If max-proc is 0, run as many > > >> processes as possible simultaneously. > > >> > > > The Bug: referenced were scrubbed, and regardless hides some history. > When using > > > toybox xargs for a kernel build, it added 5 minutes to the total build > time when > > > it ignored the -P8 flag and exec'd only one at a time. > > > > > > Sincerely -- Mark Salyzyn > > > > I did a slightly smaller implementation (I'd looked at implementing -P > before, > > but waited for somebody to complain about its absence), which supports > SIGUSR1 > > and SIGUSR2. > > > > Does what I checked in work for you? > > it passes the tests on Android (i synced AOSP this morning), but i'll > let salyzyn report specifically on the kernel build times. >
(the SoC vendor that originally brought this up say they're happy now...) > (i've attached a patch to fix the -P help text alignment.) > > > Thanks, > > > > Rob > > _______________________________________________ > > Toybox mailing list > > Toybox@lists.landley.net > > http://lists.landley.net/listinfo.cgi/toybox-landley.net >
_______________________________________________ Toybox mailing list Toybox@lists.landley.net http://lists.landley.net/listinfo.cgi/toybox-landley.net
