[Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Hervé Poussineau]

With those two patches, I'm able to run a real BIOS in QEMU (Iwill P55XB2)
However, QEMU crashes when rebooting the computer after leaving BIOS settings.

Hervé Poussineau (2):
  pam: partly fix write-only mode
  pckbd: return 'keyboard enabled' on read input port command

 hw/input/pckbd.c  |    2 +-
 hw/pci-host/pam.c |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

-- 
1.7.10.4

[Qemu-devel] [PATCH 1/2] pam: partly fix write-only mode

[Hervé Poussineau]

In write-only mode, writes are forwarded to RAM, while reads should not be
handled (ie should return 0xff).
Assume that in this mode, no read access is ever done, as they shouldn't
give any sensible result.

So, in write-only mode, alias PAM region to RAM, instead of PCI memory
(which can even be mapped to some device!)

This fixes Award BIOS, which use this mode to shadow system BIOS and video BIOS.

Signed-off-by: Hervé Poussineau <hpoussin@reactos.org>
---
 hw/pci-host/pam.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/hw/pci-host/pam.c b/hw/pci-host/pam.c
index ec6be46..e1e95aa 100644
--- a/hw/pci-host/pam.c
+++ b/hw/pci-host/pam.c
@@ -68,7 +68,7 @@ void init_pam(DeviceState *dev, MemoryRegion *ram_memory,
     /* XXX: should distinguish read/write cases */
     memory_region_init_alias(&mem->alias[0], OBJECT(dev), "pam-pci", pci_address_space,
                              start, size);
-    memory_region_init_alias(&mem->alias[2], OBJECT(dev), "pam-pci", pci_address_space,
+    memory_region_init_alias(&mem->alias[2], OBJECT(dev), "pam-pci", ram_memory,
                              start, size);
 
     for (i = 0; i < 4; ++i) {
-- 
1.7.10.4

[Qemu-devel] [PATCH 2/2] pckbd: return 'keyboard enabled' on read input port command

[Hervé Poussineau]

Bit 7 of Input Port is the keyboard inhibit switch.
0 means keyboard inhibited, while 1 means keyboard enabled.

Incidentaly, this also fixes an error encountered while booting
an Award BIOS: "Keyboard is locked out - Unlock the key".

Signed-off-by: Hervé Poussineau <hpoussin@reactos.org>
---
 hw/input/pckbd.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/hw/input/pckbd.c b/hw/input/pckbd.c
index 655b8c5..29af3d7 100644
--- a/hw/input/pckbd.c
+++ b/hw/input/pckbd.c
@@ -281,7 +281,7 @@ static void kbd_write_command(void *opaque, hwaddr addr,
         kbd_update_irq(s);
         break;
     case KBD_CCMD_READ_INPORT:
-        kbd_queue(s, 0x00, 0);
+        kbd_queue(s, 0x80, 0);
         break;
     case KBD_CCMD_READ_OUTPORT:
         kbd_queue(s, s->outport, 0);
-- 
1.7.10.4

Re: [Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Hervé Poussineau]

Ping.

QEMU crash at reboot is due to not resetting PAM registers at i440fx 
reset, but a patch has already been rejected on ML: 
http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02928.html

Hervé

Hervé Poussineau a écrit :
> With those two patches, I'm able to run a real BIOS in QEMU (Iwill P55XB2)
> However, QEMU crashes when rebooting the computer after leaving BIOS settings.
> 
> Hervé Poussineau (2):
>   pam: partly fix write-only mode
>   pckbd: return 'keyboard enabled' on read input port command
> 
>  hw/input/pckbd.c  |    2 +-
>  hw/pci-host/pam.c |    2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
> 

Re: [Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Michael S. Tsirkin]

On Mon, Mar 03, 2014 at 11:05:17PM +0100, Hervé Poussineau wrote:
> Ping.
> 
> QEMU crash at reboot is due to not resetting PAM registers at i440fx
> reset, but a patch has already been rejected on ML:
> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02928.html
> 
> Hervé

I thought an alternative fix is being worked on?
http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02967.html


> Hervé Poussineau a écrit :
> >With those two patches, I'm able to run a real BIOS in QEMU (Iwill P55XB2)
> >However, QEMU crashes when rebooting the computer after leaving BIOS settings.
> >
> >Hervé Poussineau (2):
> >  pam: partly fix write-only mode
> >  pckbd: return 'keyboard enabled' on read input port command
> >
> > hw/input/pckbd.c  |    2 +-
> > hw/pci-host/pam.c |    2 +-
> > 2 files changed, 2 insertions(+), 2 deletions(-)
> >

Re: [Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Paolo Bonzini]

Il 05/03/2014 00:49, Michael S. Tsirkin ha scritto:
> On Mon, Mar 03, 2014 at 11:05:17PM +0100, Hervé Poussineau wrote:
>> Ping.
>>
>> QEMU crash at reboot is due to not resetting PAM registers at i440fx
>> reset, but a patch has already been rejected on ML:
>> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02928.html
>>
>> Hervé
>
> I thought an alternative fix is being worked on?
> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02967.html

Hervé, how does the BIOS do the reset?

Paolo

Re: [Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Michael S. Tsirkin]

On Tue, Feb 11, 2014 at 11:46:01PM +0100, Hervé Poussineau wrote:
> With those two patches, I'm able to run a real BIOS in QEMU (Iwill P55XB2)
> However, QEMU crashes when rebooting the computer after leaving BIOS settings.
> 
> Hervé Poussineau (2):
>   pam: partly fix write-only mode
>   pckbd: return 'keyboard enabled' on read input port command

Thanks, I've applied this.

>  hw/input/pckbd.c  |    2 +-
>  hw/pci-host/pam.c |    2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
> 
> -- 
> 1.7.10.4

Re: [Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Andreas Färber]

Am 05.03.2014 00:49, schrieb Michael S. Tsirkin:
> On Mon, Mar 03, 2014 at 11:05:17PM +0100, Hervé Poussineau wrote:
>> Ping.
>>
>> QEMU crash at reboot is due to not resetting PAM registers at i440fx
>> reset, but a patch has already been rejected on ML:
>> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02928.html
>>
>> Hervé
> 
> I thought an alternative fix is being worked on?
> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02967.html

And I was told an alternative fix had been applied - was I misinformed?

Andreas

-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg

Re: [Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Hervé Poussineau]

Le mer. 05 mars 2014 19:50:04 CET, Andreas Färber a écrit :
> Am 05.03.2014 00:49, schrieb Michael S. Tsirkin:
>> On Mon, Mar 03, 2014 at 11:05:17PM +0100, Hervé Poussineau wrote:
>>> Ping.
>>>
>>> QEMU crash at reboot is due to not resetting PAM registers at i440fx
>>> reset, but a patch has already been rejected on ML:
>>> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02928.html
>>>
>>> Hervé
>>
>> I thought an alternative fix is being worked on?
>> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02967.html
>
> And I was told an alternative fix had been applied - was I misinformed?
>
> Andreas
>

In current master, I don't see any reset of PAM registers in 
hw/pci-host/piix.c .
However, piix3 has a reset function.

Hervé

Re: [Qemu-devel] [PATCH 0/2] PC: improve emulation correctness

[Hervé Poussineau]

Le 05/03/2014 09:15, Paolo Bonzini a écrit :
> Il 05/03/2014 00:49, Michael S. Tsirkin ha scritto:
>> On Mon, Mar 03, 2014 at 11:05:17PM +0100, Hervé Poussineau wrote:
>>> Ping.
>>>
>>> QEMU crash at reboot is due to not resetting PAM registers at i440fx
>>> reset, but a patch has already been rejected on ML:
>>> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02928.html
>>>
>>> Hervé
>>
>> I thought an alternative fix is being worked on?
>> http://lists.gnu.org/archive/html/qemu-devel/2013-02/msg02967.html
>
> Hervé, how does the BIOS do the reset?
>
> Paolo

At cold boot, BIOS is shadowed to RAM, and then hot-patched, so a warm boot won't do the same thing.
At BIOS exit, BIOS jumps to the new warm boot path.
This warm path doesn't work and crashes QEMU with "qemu: fatal: Trying to execute code outside RAM or ROM at 0x000a1249" ; I've no idea why exactly.
Resetting PAM registers prevents the shadowed copy to be visible, and BIOS does again a cold boot, which works.

Hervé