| Message ID | 1596034621-4714-1-git-send-email-penguin-kernel@I-love.SAKURA.ne.jp (mailing list archive) |
|---|---|
| State | Superseded, archived |
| Headers | show |
| Series | vt: defer kfree() of vc_screenbuf in vc_do_resize() | expand |
Do you think this approach is acceptable? Or, do we need to modify set_origin() ? On 2020/07/29 23:57, Tetsuo Handa wrote: > syzbot is reporting UAF bug in set_origin() from vc_do_resize() [1], for > vc_do_resize() calls kfree(vc->vc_screenbuf) before calling set_origin(). > > Unfortunately, in set_origin(), vc->vc_sw->con_set_origin() might access > vc->vc_pos when scroll is involved in order to manipulate cursor, but > vc->vc_pos refers already released vc->vc_screenbuf until vc->vc_pos gets > updated based on the result of vc->vc_sw->con_set_origin(). > > Preserving old buffer and tolerating outdated vc members until set_origin() > completes would be easier than preventing vc->vc_sw->con_set_origin() from > accessing outdated vc members. > > [1] https://syzkaller.appspot.com/bug?id=6649da2081e2ebdc65c0642c214b27fe91099db3 > > Reported-by: syzbot <syzbot+9116ecc1978ca3a12f43@syzkaller.appspotmail.com> > Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> > --- > drivers/tty/vt/vt.c | 5 +++-- > 1 file changed, 3 insertions(+), 2 deletions(-) > > diff --git a/drivers/tty/vt/vt.c b/drivers/tty/vt/vt.c > index 42d8c67..c9ee8e9 100644 > --- a/drivers/tty/vt/vt.c > +++ b/drivers/tty/vt/vt.c > @@ -1196,7 +1196,7 @@ static int vc_do_resize(struct tty_struct *tty, struct vc_data *vc, > unsigned int old_rows, old_row_size, first_copied_row; > unsigned int new_cols, new_rows, new_row_size, new_screen_size; > unsigned int user; > - unsigned short *newscreen; > + unsigned short *oldscreen, *newscreen; > struct uni_screen *new_uniscr = NULL; > > WARN_CONSOLE_UNLOCKED(); > @@ -1294,10 +1294,11 @@ static int vc_do_resize(struct tty_struct *tty, struct vc_data *vc, > if (new_scr_end > new_origin) > scr_memsetw((void *)new_origin, vc->vc_video_erase_char, > new_scr_end - new_origin); > - kfree(vc->vc_screenbuf); > + oldscreen = vc->vc_screenbuf; > vc->vc_screenbuf = newscreen; > vc->vc_screenbuf_size = new_screen_size; > set_origin(vc); > + kfree(oldscreen); > > /* do part of a reset_terminal() */ > vc->vc_top = 0; >
On Tue, Aug 04, 2020 at 08:15:43PM +0900, Tetsuo Handa wrote: > Do you think this approach is acceptable? Or, do we need to modify set_origin() ? > > On 2020/07/29 23:57, Tetsuo Handa wrote: > > syzbot is reporting UAF bug in set_origin() from vc_do_resize() [1], for > > vc_do_resize() calls kfree(vc->vc_screenbuf) before calling set_origin(). > > > > Unfortunately, in set_origin(), vc->vc_sw->con_set_origin() might access > > vc->vc_pos when scroll is involved in order to manipulate cursor, but > > vc->vc_pos refers already released vc->vc_screenbuf until vc->vc_pos gets > > updated based on the result of vc->vc_sw->con_set_origin(). > > > > Preserving old buffer and tolerating outdated vc members until set_origin() > > completes would be easier than preventing vc->vc_sw->con_set_origin() from > > accessing outdated vc members. > > > > [1] https://syzkaller.appspot.com/bug?id=6649da2081e2ebdc65c0642c214b27fe91099db3 > > > > Reported-by: syzbot <syzbot+9116ecc1978ca3a12f43@syzkaller.appspotmail.com> > > Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> > > --- > > drivers/tty/vt/vt.c | 5 +++-- > > 1 file changed, 3 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/tty/vt/vt.c b/drivers/tty/vt/vt.c > > index 42d8c67..c9ee8e9 100644 > > --- a/drivers/tty/vt/vt.c > > +++ b/drivers/tty/vt/vt.c > > @@ -1196,7 +1196,7 @@ static int vc_do_resize(struct tty_struct *tty, struct vc_data *vc, > > unsigned int old_rows, old_row_size, first_copied_row; > > unsigned int new_cols, new_rows, new_row_size, new_screen_size; > > unsigned int user; > > - unsigned short *newscreen; > > + unsigned short *oldscreen, *newscreen; > > struct uni_screen *new_uniscr = NULL; > > > > WARN_CONSOLE_UNLOCKED(); > > @@ -1294,10 +1294,11 @@ static int vc_do_resize(struct tty_struct *tty, struct vc_data *vc, > > if (new_scr_end > new_origin) > > scr_memsetw((void *)new_origin, vc->vc_video_erase_char, > > new_scr_end - new_origin); > > - kfree(vc->vc_screenbuf); > > + oldscreen = vc->vc_screenbuf; > > vc->vc_screenbuf = newscreen; > > vc->vc_screenbuf_size = new_screen_size; > > set_origin(vc); > > + kfree(oldscreen); > > > > /* do part of a reset_terminal() */ > > vc->vc_top = 0; I think what you have here is fine, as cleaning up set_orgin() might be hard to do at this point in time. thanks, greg k-h
On 2020/08/04 21:58, Greg Kroah-Hartman wrote: > On Tue, Aug 04, 2020 at 08:15:43PM +0900, Tetsuo Handa wrote: >> Do you think this approach is acceptable? Or, do we need to modify set_origin() ? >> > I think what you have here is fine, as cleaning up set_orgin() might be > hard to do at this point in time. > Seems that there is no comment. Greg, will you pick up this patch?
On Wed, Aug 12, 2020 at 12:02:03AM +0900, Tetsuo Handa wrote: > On 2020/08/04 21:58, Greg Kroah-Hartman wrote: > > On Tue, Aug 04, 2020 at 08:15:43PM +0900, Tetsuo Handa wrote: > >> Do you think this approach is acceptable? Or, do we need to modify set_origin() ? > >> > > I think what you have here is fine, as cleaning up set_orgin() might be > > hard to do at this point in time. > > > > Seems that there is no comment. Greg, will you pick up this patch? Yes, will do after -rc1 is out. thanks, greg k-h
diff --git a/drivers/tty/vt/vt.c b/drivers/tty/vt/vt.c index 42d8c67..c9ee8e9 100644 --- a/drivers/tty/vt/vt.c +++ b/drivers/tty/vt/vt.c @@ -1196,7 +1196,7 @@ static int vc_do_resize(struct tty_struct *tty, struct vc_data *vc, unsigned int old_rows, old_row_size, first_copied_row; unsigned int new_cols, new_rows, new_row_size, new_screen_size; unsigned int user; - unsigned short *newscreen; + unsigned short *oldscreen, *newscreen; struct uni_screen *new_uniscr = NULL; WARN_CONSOLE_UNLOCKED(); @@ -1294,10 +1294,11 @@ static int vc_do_resize(struct tty_struct *tty, struct vc_data *vc, if (new_scr_end > new_origin) scr_memsetw((void *)new_origin, vc->vc_video_erase_char, new_scr_end - new_origin); - kfree(vc->vc_screenbuf); + oldscreen = vc->vc_screenbuf; vc->vc_screenbuf = newscreen; vc->vc_screenbuf_size = new_screen_size; set_origin(vc); + kfree(oldscreen); /* do part of a reset_terminal() */ vc->vc_top = 0;
syzbot is reporting UAF bug in set_origin() from vc_do_resize() [1], for vc_do_resize() calls kfree(vc->vc_screenbuf) before calling set_origin(). Unfortunately, in set_origin(), vc->vc_sw->con_set_origin() might access vc->vc_pos when scroll is involved in order to manipulate cursor, but vc->vc_pos refers already released vc->vc_screenbuf until vc->vc_pos gets updated based on the result of vc->vc_sw->con_set_origin(). Preserving old buffer and tolerating outdated vc members until set_origin() completes would be easier than preventing vc->vc_sw->con_set_origin() from accessing outdated vc members. [1] https://syzkaller.appspot.com/bug?id=6649da2081e2ebdc65c0642c214b27fe91099db3 Reported-by: syzbot <syzbot+9116ecc1978ca3a12f43@syzkaller.appspotmail.com> Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> --- drivers/tty/vt/vt.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)