From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:45743) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Wh1Ct-0003Vv-18 for qemu-devel@nongnu.org; Sun, 04 May 2014 14:32:18 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Wh1Cl-0006z2-GZ for qemu-devel@nongnu.org; Sun, 04 May 2014 14:32:10 -0400 Received: from mout.web.de ([212.227.17.11]:52562) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Wh1Cl-0006yw-7L for qemu-devel@nongnu.org; Sun, 04 May 2014 14:32:03 -0400 Message-ID: <536687A0.1070705@web.de> Date: Sun, 04 May 2014 20:32:00 +0200 From: Jan Kiszka MIME-Version: 1.0 References: <1398764774-22595-1-git-send-email-kraxel@redhat.com> <1398764774-22595-8-git-send-email-kraxel@redhat.com> <5365FBD4.90108@web.de> <536673EA.7090402@redhat.com> <5366860E.5090301@web.de> In-Reply-To: <5366860E.5090301@web.de> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="arGgROOHmKbeX4VmKwcpRUKG8SrOhxCdf" Subject: Re: [Qemu-devel] [PULL 7/8] gtk: Fix -serial vc List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Cole Robinson , Gerd Hoffmann , qemu-devel@nongnu.org Cc: Anthony Liguori This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --arGgROOHmKbeX4VmKwcpRUKG8SrOhxCdf Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable On 2014-05-04 20:25, Jan Kiszka wrote: > On 2014-05-04 19:07, Cole Robinson wrote: >> On 05/04/2014 04:35 AM, Jan Kiszka wrote: >>> On 2014-04-29 11:46, Gerd Hoffmann wrote: >>>> From: Cole Robinson >>>> >>>> Try kicking off a rhel5 text install over serial, the text menu navi= gation >>>> is all messed up, and some of the kernel boot messages are randomly >>>> corrupted. >>>> >>>> Drop use of a pty and just use vte infrastructure for reading and wr= iting. >>>> This fixes the above corruption, and is simpler to boot. >>>> >>>> (I don't know what was wrong with the original code though. FWIW thi= s is >>>> what virt-manager has done for years). >>>> >>>> Signed-off-by: Cole Robinson >>>> Signed-off-by: Gerd Hoffmann >>>> --- >>>> ui/gtk.c | 41 +++++++++-------------------------------- >>>> 1 file changed, 9 insertions(+), 32 deletions(-) >>>> >>>> diff --git a/ui/gtk.c b/ui/gtk.c >>>> index c85aea3..1465a38 100644 >>>> --- a/ui/gtk.c >>>> +++ b/ui/gtk.c >>>> @@ -115,7 +115,6 @@ typedef struct VirtualConsole >>>> GtkWidget *scrolled_window; >>>> CharDriverState *chr; >>>> #endif >>>> - int fd; >>>> } VirtualConsole; >>>> =20 >>>> typedef struct GtkDisplayState >>>> @@ -1162,9 +1161,12 @@ static gboolean gd_focus_out_event(GtkWidget = *widget, >>>> =20 >>>> static int gd_vc_chr_write(CharDriverState *chr, const uint8_t *buf= , int len) >>>> { >>>> +#if defined(CONFIG_VTE) >>>> VirtualConsole *vc =3D chr->opaque; >>>> =20 >>>> - return vc ? write(vc->fd, buf, len) : len; >>>> + vte_terminal_feed(VTE_TERMINAL(vc->terminal), (const char *)buf= , len); >>>> +#endif >>>> + return len; >>>> } >>>> =20 >>>> static int nb_vcs; >>>> @@ -1190,19 +1192,12 @@ void early_gtk_display_init(void) >>>> } >>>> =20 >>>> #if defined(CONFIG_VTE) >>>> -static gboolean gd_vc_in(GIOChannel *chan, GIOCondition cond, void = *opaque) >>>> +static gboolean gd_vc_in(VteTerminal *terminal, gchar *text, guint = size, >>>> + gpointer user_data) >>>> { >>>> - VirtualConsole *vc =3D opaque; >>>> - uint8_t buffer[1024]; >>>> - ssize_t len; >>>> - >>>> - len =3D read(vc->fd, buffer, sizeof(buffer)); >>>> - if (len <=3D 0) { >>>> - return FALSE; >>>> - } >>>> - >>>> - qemu_chr_be_write(vc->chr, buffer, len); >>>> + VirtualConsole *vc =3D user_data; >>>> =20 >>>> + qemu_chr_be_write(vc->chr, (uint8_t *)text, (unsigned int)size= ); >>>> return TRUE; >>>> } >>>> #endif >>>> @@ -1214,13 +1209,8 @@ static GSList *gd_vc_init(GtkDisplayState *s,= VirtualConsole *vc, int index, GSL >>>> const char *label; >>>> char buffer[32]; >>>> char path[32]; >>>> -#if VTE_CHECK_VERSION(0, 26, 0) >>>> - VtePty *pty; >>>> -#endif >>>> - GIOChannel *chan; >>>> GtkWidget *scrolled_window; >>>> GtkAdjustment *vadjustment; >>>> - int master_fd, slave_fd; >>>> =20 >>>> snprintf(buffer, sizeof(buffer), "vc%d", index); >>>> snprintf(path, sizeof(path), "/View/VC%d", index); >>>> @@ -1239,16 +1229,7 @@ static GSList *gd_vc_init(GtkDisplayState *s,= VirtualConsole *vc, int index, GSL >>>> gtk_accel_map_add_entry(path, GDK_KEY_2 + index, HOTKEY_MODIFIE= RS); >>>> =20 >>>> vc->terminal =3D vte_terminal_new(); >>>> - >>>> - master_fd =3D qemu_openpty_raw(&slave_fd, NULL); >>>> - g_assert(master_fd !=3D -1); >>>> - >>>> -#if VTE_CHECK_VERSION(0, 26, 0) >>>> - pty =3D vte_pty_new_foreign(master_fd, NULL); >>>> - vte_terminal_set_pty_object(VTE_TERMINAL(vc->terminal), pty); >>>> -#else >>>> - vte_terminal_set_pty(VTE_TERMINAL(vc->terminal), master_fd); >>>> -#endif >>>> + g_signal_connect(vc->terminal, "commit", G_CALLBACK(gd_vc_in), = vc); >>>> =20 >>>> vte_terminal_set_scrollback_lines(VTE_TERMINAL(vc->terminal), -= 1); >>>> =20 >>>> @@ -1263,7 +1244,6 @@ static GSList *gd_vc_init(GtkDisplayState *s, = VirtualConsole *vc, int index, GSL >>>> =20 >>>> vte_terminal_set_size(VTE_TERMINAL(vc->terminal), 80, 25); >>>> =20 >>>> - vc->fd =3D slave_fd; >>>> vc->chr->opaque =3D vc; >>>> vc->scrolled_window =3D scrolled_window; >>>> =20 >>>> @@ -1281,9 +1261,6 @@ static GSList *gd_vc_init(GtkDisplayState *s, = VirtualConsole *vc, int index, GSL >>>> vc->chr->init(vc->chr); >>>> } >>>> =20 >>>> - chan =3D g_io_channel_unix_new(vc->fd); >>>> - g_io_add_watch(chan, G_IO_IN, gd_vc_in, vc); >>>> - >>>> #endif /* CONFIG_VTE */ >>>> return group; >>>> } >>>> >>> >>> This commit somehow messes up the monitor vc: Fire up qemu-system-x86= _64 >>> and switch to console 2 (monitor). You'll find it formatted as if the= >>> console was only ~10 chars wide during printout of the monitor >>> greetings. When typing, everything is fine again. Maybe an ordering >>> issue that was only revealed by this commit, dunno yet. >>> >> >> Check out gerd's ui-gtk-next branch, there's a few extra patches relat= ed to >> vte sizing that might fix it. >=20 > Looks better thanks to "gtk: zap scrolled_window" (monitor is properly > formatted again). =2E..err - no. The price of this patch is that the window is no longer properly resized on all guest mode changes. There is still more broken. Jan --arGgROOHmKbeX4VmKwcpRUKG8SrOhxCdf Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEUEARECAAYFAlNmh6AACgkQitSsb3rl5xQ0/gCYmIfNaQ5tOJ1lCMkTw3bsylXs 5wCeOTOvJQEV/xVKh1WPT+HavO6wFMQ= =gKH9 -----END PGP SIGNATURE----- --arGgROOHmKbeX4VmKwcpRUKG8SrOhxCdf--