qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH] monitor/hmp: schedule qemu_chr_fe_accept_input() after read


From: Volker Rümelin
Subject: Re: [PATCH] monitor/hmp: schedule qemu_chr_fe_accept_input() after read
Date: Tue, 10 Aug 2021 07:18:11 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.12.0


    Since commit 584af1f1d9 (ui/gtk: add a keyboard fifo to the VTE
    consoles) a GTK VTE console chardev backend relies on the
    connected chardev frontend to call qemu_chr_fe_accept_input()
    whenever it can receive new characters. The HMP monitor doesn't
    do this. It only schedules a call to qemu_chr_fe_accept_input()
    after it handled a HMP command in monitor_command_cb().

    This is a problem if you paste a few characters into the GTK
    monitor console. Even entering a UTF-8 multibyte character leads
    to the same problem.

    Schedule a call to qemu_chr_fe_accept_input() after handling the
    received bytes to fix the HMP monitor.

    Signed-off-by: Volker Rümelin <vr_qemu@t-online.de
    <mailto:vr_qemu@t-online.de>>


Wouldn't it work to make gd_vc_send_chars() write in a loop (until it fails)?

Hi Marc-André,

yes that works. I found similar code in the udp_chr_flush_buffer() function in chardev/char-udp.c. I will send a new patch within the next few hours.

With best regards,
Volker


If not, I think monitor/qmp.c should be adjusted too.

    ---
     monitor/hmp.c              |  1 +
     monitor/monitor-internal.h |  1 +
     monitor/monitor.c          | 19 +++++++++++++++++--
     3 files changed, 19 insertions(+), 2 deletions(-)

    diff --git a/monitor/hmp.c b/monitor/hmp.c
    index d50c3124e1..470f56a71d 100644
    --- a/monitor/hmp.c
    +++ b/monitor/hmp.c
    @@ -1349,6 +1349,7 @@ static void monitor_read(void *opaque, const
    uint8_t *buf, int size)
             for (i = 0; i < size; i++) {
                 readline_handle_byte(mon->rs, buf[i]);
             }
    +        monitor_accept_input(&mon->common);
         } else {
             if (size == 0 || buf[size - 1] != 0) {
                 monitor_printf(&mon->common, "corrupted command\n");
    diff --git a/monitor/monitor-internal.h b/monitor/monitor-internal.h
    index 9c3a09cb01..af33c3c617 100644
    --- a/monitor/monitor-internal.h
    +++ b/monitor/monitor-internal.h
    @@ -170,6 +170,7 @@ int monitor_puts(Monitor *mon, const char *str);
     void monitor_data_init(Monitor *mon, bool is_qmp, bool skip_flush,
                            bool use_io_thread);
     void monitor_data_destroy(Monitor *mon);
    +void monitor_accept_input(Monitor *mon);
     int monitor_can_read(void *opaque);
     void monitor_list_append(Monitor *mon);
     void monitor_fdsets_cleanup(void);
    diff --git a/monitor/monitor.c b/monitor/monitor.c
    index 46a171bca6..8e3cf4ad98 100644
    --- a/monitor/monitor.c
    +++ b/monitor/monitor.c
    @@ -519,13 +519,28 @@ int monitor_suspend(Monitor *mon)
         return 0;
     }

    -static void monitor_accept_input(void *opaque)
    +static void monitor_accept_input_bh(void *opaque)
     {
         Monitor *mon = opaque;

         qemu_chr_fe_accept_input(&mon->chr);
     }

    +void monitor_accept_input(Monitor *mon)
    +{
    +    if (!qatomic_mb_read(&mon->suspend_cnt)) {
    +        AioContext *ctx;
    +
    +        if (mon->use_io_thread) {
    +            ctx = iothread_get_aio_context(mon_iothread);
    +        } else {
    +            ctx = qemu_get_aio_context();
    +        }
    +
    +        aio_bh_schedule_oneshot(ctx, monitor_accept_input_bh, mon);
    +    }
    +}
    +
     void monitor_resume(Monitor *mon)
     {
         if (monitor_is_hmp_non_interactive(mon)) {
    @@ -547,7 +562,7 @@ void monitor_resume(Monitor *mon)
                 readline_show_prompt(hmp_mon->rs);
             }

    -        aio_bh_schedule_oneshot(ctx, monitor_accept_input, mon);
    +        aio_bh_schedule_oneshot(ctx, monitor_accept_input_bh, mon);
         }

         trace_monitor_suspend(mon, -1);
-- 2.26.2




--
Marc-André Lureau




reply via email to

[Prev in Thread] Current Thread [Next in Thread]