ui/gtk: Negative Page number is not valid
PATCH 1/1 ui/gtk: Negative Page number is not valid
From: Dongwon Kim <[email protected]>
Negative page number means the page with that number does not belong to the notebook so it shouldn't be used as a valid page number in gd_vc_find_by_page. This function should just return null in such case.
This change, however, will cause a segfault during detaching /untabifying process in gtk_release_modifiers because the current VC's page number suddenly becomes '-1' as soon as the VC is detached, which makes gd_vc_find_by_page return null. So gtk_release_modifiers should do the null check on VC returned from gd_vc_find_by_page.
--- ui/gtk.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
ui/gtk.c +2 -2
@@ -164,7 +164,7 @@ static VirtualConsole *gd_vc_find_by_page(GtkDisplayState *s, gint page)for (i =0 ; i < s->nb_vcs; i++) { vc = &s->vc[i]; p =gtk_notebook_page_num (GTK_NOTEBOOK (s->notebook), vc->tab_item); -if (p == page) { +if (p >-1 && p == page) {return vc; } }
@@ -357,7 +357,7 @@ static void gtk_release_modifiers(GtkDisplayState *s) {VirtualConsole *vc =gd_vc_find_current (s); -if (vc->type !=GD_VC_GFX || +if (!vc || vc->type !=GD_VC_GFX || !qemu_console_is_graphic (vc->gfx.dcl.con)) {return ; }
Signature · 1 line
2.34.1
Hi Dongwon,
ui/gtk.c static VirtualConsole *gd_vc_find_by_page(GtkDisplayState *s, gint page)
Quoted 27 lines · On 26/6/24 02:08, [email protected] wrote:
From: Dongwon Kim <[email protected]>
Negative page number means the page with that number does not belong to the notebook so it shouldn't be used as a valid page number in
gd_vc_find_by_page. This function should just return null in such case.This change, however, will cause a segfault during detaching /untabifying process in
gtk_release_modifiersbecause the current VC's page number suddenly becomes '-1' as soon as the VC is detached, which makesgd_vc_find_by_pagereturn null. Sogtk_release_modifiersshould do the null check on VC returned fromgd_vc_find_by_page.Cc Gerd Hoffmann <[email protected]>Cc Marc-André Lureau <[email protected]>Signed-off-by Vivek Kasireddy <[email protected]>Signed-off-by Dongwon Kim <[email protected]>--- ui/gtk.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)ui/gtk.c +0
-0@@ -164,7 +164,7 @@ static VirtualConsole *gd_vc_find_by_page(GtkDisplayState *s, gint page)
The caller should check gtk_notebook_get_current_page() != -1.
We might assert(page >= 0) here.
ui/gtk.c static VirtualConsole *gd_vc_find_by_page(GtkDisplayState *s, gint page) · line 167
Quoted 5 lines
ui/gtk.c +1
-1for (i =0 ; i < s->nb_vcs; i++) { vc = &s->vc[i]; p =gtk_notebook_page_num (GTK_NOTEBOOK (s->notebook), vc->tab_item); -if (p == page) { +if (p >-1 && p == page) {
Then this is not necessary.
ui/gtk.c static VirtualConsole *gd_vc_find_by_page(GtkDisplayState *s, gint page) · line 170
Quoted 3 lines
ui/gtk.c +0
-0return vc; } }
return NULL;
I wonder about returning NULL, maybe just
g_assert_not_reached();
ui/gtk.c static void gtk_release_modifiers(GtkDisplayState *s) · line 360
Quoted 6 lines
ui/gtk.c +1
-1@@ -357,7 +357,7 @@ static void gtk_release_modifiers(GtkDisplayState *s) {VirtualConsole *vc =gd_vc_find_current (s); - if (vc->type != GD_VC_GFX || + if (!vc || vc->type != GD_VC_GFX ||
Then this is not necessary.
Quoted 3 lines
ui/gtk.c +0
-0!qemu_console_is_graphic(vc->gfx.dcl.con)) {return ; }
Quoted 34 lines · Philippe Mathieu-Daudé
Hi Dongwon,
Quoted 27 lines · On 26/6/24 02:08, [email protected] wrote:
From: Dongwon Kim <[email protected]>
Negative page number means the page with that number does not belong to the notebook so it shouldn't be used as a valid page number in
gd_vc_find_by_page. This function should just return null in such case.This change, however, will cause a segfault during detaching /untabifying process in
gtk_release_modifiersbecause the current VC's page number suddenly becomes '-1' as soon as the VC is detached, which makesgd_vc_find_by_pagereturn null. Sogtk_release_modifiersshould do the null check on VC returned fromgd_vc_find_by_page.Cc Gerd Hoffmann <[email protected]>Cc Marc-André Lureau <[email protected]>Signed-off-by Vivek Kasireddy <[email protected]>Signed-off-by Dongwon Kim <[email protected]>---
ui/gtk.c| 4 ++--1 file changed, 2 insertions(+), 2 deletions(-)ui/gtk.c +0
-0ui/gtk.c static VirtualConsole · line 164
Quoted 1 line
ui/gtk.c +0
-0@@ -164,7 +164,7 @@ static VirtualConsole *gd_vc_find_by_page(GtkDisplayState *s, gint page)The caller should check
gtk_notebook_get_current_page()!= -1.
Quoted 1 line
We might assert(page >= 0) here.
We could do that but it means there should be more checks in other functions where gd_vc_find_by_page is called, like gd_vc_find_current. And we just can't do assert in\gd_vc_find_current because detached VC has the page number of -1.
Quoted 30 lines
ui/gtk.c static VirtualConsole *gd_vc_find_by_page(GtkDisplayState *s, gint page) · line 167
Quoted 6 lines
ui/gtk.c +1
-1for (i = 0; i < s->nb_vcs; i++) { vc = &s->vc[i]; p = gtk_notebook_page_num(GTK_NOTEBOOK(s->notebook), vc->tab_item); - if (p == page) { + if (p > -1 && p == page) {Then this is not necessary.
Quoted 3 lines
ui/gtk.c +0
-0return vc; }}
return NULL;
I wonder about returning NULL, maybe just
g_assert_not_reached();ui/gtk.c static void gtk_release_modifiers(GtkDisplayState *s) · line 359
Quoted 5 lines
ui/gtk.c +1
-1@@ -357,7 +357,7 @@ static void gtk_release_modifiers(GtkDisplayState *s) { VirtualConsole *vc = gd_vc_find_current(s); - if (vc->type != GD_VC_GFX || + if (!vc || vc->type != GD_VC_GFX ||Then this is not necessary.
Quoted 3 lines
ui/gtk.c +0
-0!qemu_console_is_graphic(vc->gfx.dcl.con)) { return; }