Patchwork hw/i386/pc: fix possible segment fault for port92_write

login
register
mail settings
Submitter liguang
Date March 22, 2013, 9:12 a.m.
Message ID <1363943551-31150-1-git-send-email-lig.fnst@cn.fujitsu.com>
Download mbox | patch
Permalink /patch/229932/
State New
Headers show

Comments

liguang - March 22, 2013, 9:12 a.m.
e.g.
$qemu-system-x86_64 -device port92
will report segment fault,
for port92_write try a un-allocated
assignment for a20_out.
so check before this assignment.

Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
---
 hw/i386/pc.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
Paolo Bonzini - March 22, 2013, 11:07 a.m.
Il 22/03/2013 10:12, liguang ha scritto:
> e.g.
> $qemu-system-x86_64 -device port92
> will report segment fault,
> for port92_write try a un-allocated
> assignment for a20_out.
> so check before this assignment.
> 
> Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
> ---
>  hw/i386/pc.c |    3 ++-
>  1 files changed, 2 insertions(+), 1 deletions(-)
> 
> diff --git a/hw/i386/pc.c b/hw/i386/pc.c
> index ed7d9ba..a0e8ee0 100644
> --- a/hw/i386/pc.c
> +++ b/hw/i386/pc.c
> @@ -440,7 +440,8 @@ static void port92_write(void *opaque, hwaddr addr, uint64_t val,
>  
>      DPRINTF("port92: write 0x%02x\n", val);
>      s->outport = val;
> -    qemu_set_irq(*s->a20_out, (val >> 1) & 1);
> +    if (s->a20_out)
> +        qemu_set_irq(*s->a20_out, (val >> 1) & 1);
>      if (val & 1) {
>          qemu_system_reset_request();
>      }
> 

Unfortunately, this is a very common problem.  The correct fix is to
change port92 to use the GPIO mechanism instead.

Paolo
Andreas Färber - March 22, 2013, 11:20 a.m.
Am 22.03.2013 10:12, schrieb liguang:
> e.g.
> $qemu-system-x86_64 -device port92
> will report segment fault,
> for port92_write try a un-allocated
> assignment for a20_out.
> so check before this assignment.
> 
> Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
> ---
>  hw/i386/pc.c |    3 ++-
>  1 files changed, 2 insertions(+), 1 deletions(-)
> 
> diff --git a/hw/i386/pc.c b/hw/i386/pc.c
> index ed7d9ba..a0e8ee0 100644
> --- a/hw/i386/pc.c
> +++ b/hw/i386/pc.c
> @@ -440,7 +440,8 @@ static void port92_write(void *opaque, hwaddr addr, uint64_t val,
>  
>      DPRINTF("port92: write 0x%02x\n", val);
>      s->outport = val;
> -    qemu_set_irq(*s->a20_out, (val >> 1) & 1);
> +    if (s->a20_out)
> +        qemu_set_irq(*s->a20_out, (val >> 1) & 1);

Missing braces.

But I think it would be better to proceed with my QOM'ification [1] and
return an Error on realize here since these IRQs don't change while
realized and qdev init doesn't allow to return a textual error.

Andreas

[1] https://github.com/afaerber/qemu-cpu/commits/realize-isa

>      if (val & 1) {
>          qemu_system_reset_request();
>      }
>
liguang - March 26, 2013, 8:47 a.m.
在 2013-03-22五的 12:20 +0100,Andreas Färber写道:
> Am 22.03.2013 10:12, schrieb liguang:
> > e.g.
> > $qemu-system-x86_64 -device port92
> > will report segment fault,
> > for port92_write try a un-allocated
> > assignment for a20_out.
> > so check before this assignment.
> > 
> > Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
> > ---
> >  hw/i386/pc.c |    3 ++-
> >  1 files changed, 2 insertions(+), 1 deletions(-)
> > 
> > diff --git a/hw/i386/pc.c b/hw/i386/pc.c
> > index ed7d9ba..a0e8ee0 100644
> > --- a/hw/i386/pc.c
> > +++ b/hw/i386/pc.c
> > @@ -440,7 +440,8 @@ static void port92_write(void *opaque, hwaddr addr, uint64_t val,
> >  
> >      DPRINTF("port92: write 0x%02x\n", val);
> >      s->outport = val;
> > -    qemu_set_irq(*s->a20_out, (val >> 1) & 1);
> > +    if (s->a20_out)
> > +        qemu_set_irq(*s->a20_out, (val >> 1) & 1);
> 
> Missing braces.

Yes, Thanks!

> 
> But I think it would be better to proceed with my QOM'ification [1] and
> return an Error on realize here since these IRQs don't change while
> realized and qdev init doesn't allow to return a textual error.
> 
> Andreas
> 
> [1] https://github.com/afaerber/qemu-cpu/commits/realize-isa
> 
> >      if (val & 1) {
> >          qemu_system_reset_request();
> >      }
> > 
> 
>
liguang - March 26, 2013, 8:49 a.m.
在 2013-03-22五的 12:07 +0100,Paolo Bonzini写道:
> Il 22/03/2013 10:12, liguang ha scritto:
> > e.g.
> > $qemu-system-x86_64 -device port92
> > will report segment fault,
> > for port92_write try a un-allocated
> > assignment for a20_out.
> > so check before this assignment.
> > 
> > Signed-off-by: liguang <lig.fnst@cn.fujitsu.com>
> > ---
> >  hw/i386/pc.c |    3 ++-
> >  1 files changed, 2 insertions(+), 1 deletions(-)
> > 
> > diff --git a/hw/i386/pc.c b/hw/i386/pc.c
> > index ed7d9ba..a0e8ee0 100644
> > --- a/hw/i386/pc.c
> > +++ b/hw/i386/pc.c
> > @@ -440,7 +440,8 @@ static void port92_write(void *opaque, hwaddr addr, uint64_t val,
> >  
> >      DPRINTF("port92: write 0x%02x\n", val);
> >      s->outport = val;
> > -    qemu_set_irq(*s->a20_out, (val >> 1) & 1);
> > +    if (s->a20_out)
> > +        qemu_set_irq(*s->a20_out, (val >> 1) & 1);
> >      if (val & 1) {
> >          qemu_system_reset_request();
> >      }
> > 
> 
> Unfortunately, this is a very common problem.  The correct fix is to
> change port92 to use the GPIO mechanism instead.
> 

Sorry, Paolo,
I'm not so familiar with 'GPIO mechanism', 
can you specify more?

Thanks!

Patch

diff --git a/hw/i386/pc.c b/hw/i386/pc.c
index ed7d9ba..a0e8ee0 100644
--- a/hw/i386/pc.c
+++ b/hw/i386/pc.c
@@ -440,7 +440,8 @@  static void port92_write(void *opaque, hwaddr addr, uint64_t val,
 
     DPRINTF("port92: write 0x%02x\n", val);
     s->outport = val;
-    qemu_set_irq(*s->a20_out, (val >> 1) & 1);
+    if (s->a20_out)
+        qemu_set_irq(*s->a20_out, (val >> 1) & 1);
     if (val & 1) {
         qemu_system_reset_request();
     }