Greg,

> So the timing as to when the parameters is not what I was addressing.

>From your previous reply, it looked like you are not agreeing with my first
statement

"As I can remember, Linux uses BIOS only at the bootup time, after that on a
running kernel, there is no role of BIOS routines."

so this bootup time code involves all these sort of initialization, perhaps
I used 'bootup time' as very abstract word. I think we are on same page.

BTW with your reply, looks you are also in storage domain.

Rajat


2010/11/6 Greg Freemyer <[email protected]>

> 2010/11/6 Rajat Sharma <[email protected]>:
> > Greg,
> > you are right about Intel AHCI, I also had to change it once in BIOS for
> > detecting some Intel's SSD drives. But I think that was again for boot
> time
> > hardware initialization. Are you sure that these parameters are accessed
> > from BIOS routines on a running Linux kernel?
> > Rajat
> >
>
> Rajat,
>
> The OP said "I am just exploring how much the linux is dependent on BIOS."
>
> So the timing as to when the parameters is not what I was addressing.
>
> I am saying that there are BIOS controlled parameters such as PCI ID
> that impact linux in a fundamental way because they drive which ATA
> driver is used.
>
> Then further there are BIOS settings that are read by the kernel that
> impact the operation of the drivers.  That the parameters are only
> read from the BIOS area on startup doesn't change the fact they impact
> the operation during the entire time linux is running.
>
> Note, while many BIOS parameters are used exclusively only for startup
> hardware initialization, the ones I am referring to are used to
> control the logic flow of the driver code in many routine operational
> situations.
>
> One simple example is for a fake-raid sata controller that implements
> a meta-data area at the start of the disk in certain modes, but not
> all.  The linux driver for this must interrogate the system to
> determine which mode the controller card is in and then set an
> internal offset field to ensure that "sector 0" is what it is expected
> to be.
>
> If this is not done, then if the system is booted into Windows with a
> dual-boot, the offsets will be wrong and everything falls apart in a
> major way.
>
> You might argue this is not a "BIOS" setting, but with many
> motherboards today, the SATA controller is built in and the BIOS
> directly controls which mode the controller operates in.  Thus I would
> call this a situation where "linux is dependent on BIOS".
>
> Greg
>

Reply via email to