Re: [Qemu-devel] [PATCH 16/19] qemu-char: move msmouse registeration to msmouse.c

Re: [Qemu-devel] [PATCH 16/19] qemu-char: move msmouse registeration to msmouse.c

[Amit Shah]

On (Mon) 18 Feb 2013 [15:48:13], Anthony Liguori wrote:

> +static CharDriverState *qemu_chr_open_pp(QemuOpts *opts)
> +{
> +    return NULL;
> +}

I guess this was to fix up the compile broken by a previous commit.
I've handled this in my rebase.

		Amit

Re: [Qemu-devel] [PATCH 00/19] chardev flow control

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 793 bytes --]

On 2013-02-18 22:47, Anthony Liguori wrote:
> This series implements an idea from Paolo to introduce flow control
> in the char layer by converting all char backends to use GIOChannels
> internally.  Then we can just use the existing IO watch support in glib
> to implement flow control.
> 
> This is based on a rebased version of an old series of mine plus a
> number of improvements from Amit Shah.

There is a problem with these changes, maybe just a small bug but I do
not see a simple solution yet:

If an input channel signals readiness but the frontend has no space
(can_read returns 0), the main loop will start busy-waiting, causing
100% CPU load.

It's trivial to reproduce: qemu-system-x86_64 -serial stdio -S, and then
hit a key twice on that console.

Jan



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 263 bytes --]