nuttx-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From GitBox <>
Subject [GitHub] [incubator-nuttx] xiaoxiang781216 edited a comment on pull request #1009: Add support for preemptive scheduling in simulator build
Date Sat, 09 May 2020 13:44:34 GMT

xiaoxiang781216 edited a comment on pull request #1009:

   > > @sebastianene07 I stop to review the last several patch because it's better to
reorg your patch into a meaning set. For example, I found serveral patch fix the problem in
the previous one.
   > > And basically, the better change is enhance up_oneshot.c which is much simple
and get the tickless support free:
   > > 1.Register alarm timer in up_oneshot.c
   > > 2.Call back to NuttX in signal action
   > > 3.Modify up_irq_save/up_irq_restore to disable/enable signal
   > > I don't think it's nessary to use ucontext API.
   > I used ucontext API because it can restore the signal mask after a call to `swapcontext`.
The setjmp/longjmp mechanism doesn't seem to work with signals. Do you think it's better to
close this PR and re-open a different one with the suggested changes ?
   But if the irq save/restore work correctly, I think we don't need care about the host signal
logic. The thread in NuttX isn't 1:1 mapping to host thread. The whole NuttX simulator run
on one host thread(main thread) except you enable SMP then we have multiple threads(same number
as Simulated CPUs). The context switch inside NuttX doesn't mean we switch host thread. From
the host view, the thread is still running on the main thread, so I can't understand why we
need save/restore host signal flag here.

This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:

View raw message