[Qemu-devel] AHCI in SeaBIOS

[Qemu-devel] AHCI in SeaBIOS

[Alexander Graf]

Hi Kevin,

Do you remember why you put AHCI in with default=n? I'd like to see it enabled in Qemu 0.14 and IIUC we use the default configuration for that.

Speaking of which - it would be awesome of we had a companion SeaBIOS release with Qemu 0.14, so we don't have to fetch a random git snapshot but potentially even could maintain a stable SeaBIOS for during the lifetime of 0.14 :).


Alex

Re: [Qemu-devel] AHCI in SeaBIOS

[Anthony Liguori]

On 02/07/2011 05:57 PM, Alexander Graf wrote:
> Hi Kevin,
>
> Do you remember why you put AHCI in with default=n? I'd like to see it enabled in Qemu 0.14 and IIUC we use the default configuration for that.
>
> Speaking of which - it would be awesome of we had a companion SeaBIOS release with Qemu 0.14, so we don't have to fetch a random git snapshot but potentially even could maintain a stable SeaBIOS for during the lifetime of 0.14 :).
>    

We don't use a random git snapshot.  We're shipping with 0.6.1.3 which 
is the latest stable release of SeaBIOS.

Regards,

Anthony Liguori

>
> Alex
>
>
>    

Re: [Qemu-devel] AHCI in SeaBIOS

[Jonathan A. Kollasch]

On Tue, Feb 08, 2011 at 12:57:41AM +0100, Alexander Graf wrote:
> Hi Kevin,
> 
> Do you remember why you put AHCI in with default=n? I'd like to see it enabled in Qemu 0.14 and IIUC we use the default configuration for that.

IIRC we haven't had conclusively successful tests of it on real
hardware.  Still, might as well enable it anyway.

	Jonathan Kollasch

Re: [Qemu-devel] AHCI in SeaBIOS

[Gleb Natapov]

On Mon, Feb 07, 2011 at 08:27:43PM -0600, Anthony Liguori wrote:
> On 02/07/2011 05:57 PM, Alexander Graf wrote:
> >Hi Kevin,
> >
> >Do you remember why you put AHCI in with default=n? I'd like to see it enabled in Qemu 0.14 and IIUC we use the default configuration for that.
> >
> >Speaking of which - it would be awesome of we had a companion SeaBIOS release with Qemu 0.14, so we don't have to fetch a random git snapshot but potentially even could maintain a stable SeaBIOS for during the lifetime of 0.14 :).
> 
> We don't use a random git snapshot.  We're shipping with 0.6.1.3
> which is the latest stable release of SeaBIOS.
> 
Unfortunately this version is too outdated for use with 0.14. It lacks
support fo AHCI and boot order and may be other features that are
present in qemu 0.14. We need new stable Seabios branch for 0.14 use.

--
			Gleb.

[Qemu-devel] Re: AHCI in SeaBIOS

[Kevin O'Connor]

On Tue, Feb 08, 2011 at 12:57:41AM +0100, Alexander Graf wrote:
> Hi Kevin,
> 
> Do you remember why you put AHCI in with default=n? I'd like to see
> it enabled in Qemu 0.14 and IIUC we use the default configuration
> for that.

Hi Alex,

Sorry - I've gotten behind on emails.

The reason for not enabling it by default was for two reasons - it had
not been tested on real hardware, and it was unclear if doing the jump
into 32bit mode would have adverse impact.

> Speaking of which - it would be awesome of we had a companion
> SeaBIOS release with Qemu 0.14, so we don't have to fetch a random
> git snapshot but potentially even could maintain a stable SeaBIOS
> for during the lifetime of 0.14 :).

It's about time to make a new release of seabios, so I'll see if that
can be done in a week or so.

-Kevin

[Qemu-devel] Re: AHCI in SeaBIOS

[Alexander Graf]

Kevin O'Connor wrote:
> On Tue, Feb 08, 2011 at 12:57:41AM +0100, Alexander Graf wrote:
>   
>> Hi Kevin,
>>
>> Do you remember why you put AHCI in with default=n? I'd like to see
>> it enabled in Qemu 0.14 and IIUC we use the default configuration
>> for that.
>>     
>
> Hi Alex,
>
> Sorry - I've gotten behind on emails.
>
> The reason for not enabling it by default was for two reasons - it had
> not been tested on real hardware, and it was unclear if doing the jump
> into 32bit mode would have adverse impact.
>   

Do you have to do the jump even when AHCI is unused? If no, it shouldn't
hurt, right?

Anthony, can we enable it only for the Qemu build?

>> Speaking of which - it would be awesome of we had a companion
>> SeaBIOS release with Qemu 0.14, so we don't have to fetch a random
>> git snapshot but potentially even could maintain a stable SeaBIOS
>> for during the lifetime of 0.14 :).
>>     
>
> It's about time to make a new release of seabios, so I'll see if that
> can be done in a week or so.
>   

Very nice, thank you! It would be great if we could sync that up with
the 0.14 release. Anthony?


Alex

[Qemu-devel] Re: AHCI in SeaBIOS

[Kevin O'Connor]

On Thu, Feb 10, 2011 at 04:25:11PM +0100, Alexander Graf wrote:
> Kevin O'Connor wrote:
> > On Tue, Feb 08, 2011 at 12:57:41AM +0100, Alexander Graf wrote:
> >> Do you remember why you put AHCI in with default=n? I'd like to see
> >> it enabled in Qemu 0.14 and IIUC we use the default configuration
> >> for that.
> > The reason for not enabling it by default was for two reasons - it had
> > not been tested on real hardware, and it was unclear if doing the jump
> > into 32bit mode would have adverse impact.
> 
> Do you have to do the jump even when AHCI is unused? If no, it shouldn't
> hurt, right?

The 32bit jump is only done if an AHCI drive is found and one tries to
read/write to it.

-Kevin

[Qemu-devel] Re: AHCI in SeaBIOS

[Alexander Graf]

Kevin O'Connor wrote:
> On Thu, Feb 10, 2011 at 04:25:11PM +0100, Alexander Graf wrote:
>   
>> Kevin O'Connor wrote:
>>     
>>> On Tue, Feb 08, 2011 at 12:57:41AM +0100, Alexander Graf wrote:
>>>       
>>>> Do you remember why you put AHCI in with default=n? I'd like to see
>>>> it enabled in Qemu 0.14 and IIUC we use the default configuration
>>>> for that.
>>>>         
>>> The reason for not enabling it by default was for two reasons - it had
>>> not been tested on real hardware, and it was unclear if doing the jump
>>> into 32bit mode would have adverse impact.
>>>       
>> Do you have to do the jump even when AHCI is unused? If no, it shouldn't
>> hurt, right?
>>     
>
> The 32bit jump is only done if an AHCI drive is found and one tries to
> read/write to it.
>   

Very good, so it really shouldn't hurt. We could add another option that
only kicks off the detection when Qemu is found, no? That way we're sure
to not break real hardware, but have the functionality in Qemu :)


Alex

Re: [Qemu-devel] Re: AHCI in SeaBIOS

[Anthony Liguori]

On 02/10/2011 04:25 PM, Alexander Graf wrote:
> Kevin O'Connor wrote:
>    
>> On Tue, Feb 08, 2011 at 12:57:41AM +0100, Alexander Graf wrote:
>>
>>      
>>> Hi Kevin,
>>>
>>> Do you remember why you put AHCI in with default=n? I'd like to see
>>> it enabled in Qemu 0.14 and IIUC we use the default configuration
>>> for that.
>>>
>>>        
>> Hi Alex,
>>
>> Sorry - I've gotten behind on emails.
>>
>> The reason for not enabling it by default was for two reasons - it had
>> not been tested on real hardware, and it was unclear if doing the jump
>> into 32bit mode would have adverse impact.
>>
>>      
> Do you have to do the jump even when AHCI is unused? If no, it shouldn't
> hurt, right?
>
> Anthony, can we enable it only for the Qemu build?
>
>    
>>> Speaking of which - it would be awesome of we had a companion
>>> SeaBIOS release with Qemu 0.14, so we don't have to fetch a random
>>> git snapshot but potentially even could maintain a stable SeaBIOS
>>> for during the lifetime of 0.14 :).
>>>
>>>        
>> It's about time to make a new release of seabios, so I'll see if that
>> can be done in a week or so.
>>
>>      
> Very nice, thank you! It would be great if we could sync that up with
> the 0.14 release. Anthony?
>    

We're right around the corner from -rc2 so let's wait until after 0.14 
is released so we have a full release cycle to test it.

Regards,

Anthony Liguori

> Alex
>
>
>    

[Qemu-devel] Re: AHCI in SeaBIOS

[Paolo Bonzini]

On 02/10/2011 04:37 PM, Anthony Liguori wrote:
>
> We're right around the corner from -rc2 so let's wait until after 0.14
> is released so we have a full release cycle to test it.

And have unusable or only partly usable features (already mentioned 
multiple times: boot order, AHCI) in 0.14?

Paolo