Please revert "use todc time functions for PPC_PREP" changeset

Please revert "use todc time functions for PPC_PREP" changeset

[Leigh Brown]

Hi Tom,

Your patch (ChangeSet@1.1557.41.25) breaks the 43p140 and Carolina, and
probably quite a few other RS/6000 based PRePs.  I was a bit surprised
to see it 2.6.3.

http://linux.bkbits.net:8080/linux-2.5/patch@1.1557.41.25?nav=index.html|ChangeSet@-3w|cset@1.1557.41.25

Can we revert this until there is a working version, please?  I don't
really know enough about this stuff to fix it myself, unfortunately,
although I'm working through the code to see how it works.

Cheers,

Leigh.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Please revert "use todc time functions for PPC_PREP" changeset

[Benjamin Herrenschmidt]

On Sat, 2004-02-28 at 19:32, Andrew Morton wrote:
> "Leigh Brown" <leigh@solinno.co.uk> wrote:
> >
> > Hi Tom,
> >
> >  Your patch (ChangeSet@1.1557.41.25) breaks the 43p140 and Carolina, and
> >  probably quite a few other RS/6000 based PRePs.  I was a bit surprised
> >  to see it 2.6.3.
> >
> >  http://linux.bkbits.net:8080/linux-2.5/patch@1.1557.41.25?nav=index.html|ChangeSet@-3w|cset@1.1557.41.25
> >
> >  Can we revert this until there is a working version, please?
>
> It wouild help if you could define "breaks".

Yup, Leigh, we have at least a week I suppose before 2.6.4 goes final, so
we can probably fix the problem rather than reverting the cleanup.

Can you tell us more about what's going on ?

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Please revert "use todc time functions for PPC_PREP" changeset

[Andrew Morton]

"Leigh Brown" <leigh@solinno.co.uk> wrote:
>
> Hi Tom,
>
>  Your patch (ChangeSet@1.1557.41.25) breaks the 43p140 and Carolina, and
>  probably quite a few other RS/6000 based PRePs.  I was a bit surprised
>  to see it 2.6.3.
>
>  http://linux.bkbits.net:8080/linux-2.5/patch@1.1557.41.25?nav=index.html|ChangeSet@-3w|cset@1.1557.41.25
>
>  Can we revert this until there is a working version, please?

It wouild help if you could define "breaks".

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Please revert "use todc time functions for PPC_PREP" changeset

[Leigh Brown]

Andrew Morton said:
>
> "Leigh Brown" <leigh@solinno.co.uk> wrote:
>>
>> Hi Tom,
>>
>>  Your patch (ChangeSet@1.1557.41.25) breaks the 43p140 and Carolina, and
>>  probably quite a few other RS/6000 based PRePs.  I was a bit surprised
>>  to see it 2.6.3.
>>
>>  http://linux.bkbits.net:8080/linux-2.5/patch@1.1557.41.25?nav=index.html|ChangeSet@-3w|cset@1.1557.41.25
>>
>>  Can we revert this until there is a working version, please?
>
> It wouild help if you could define "breaks".

Sorry!  It hangs during boot, before any output on the serial console.  I
got as far as working out that it hangs somewhere in
arch/ppc/kernel/time.c:time_init() (or one of the called functions).

I can do some more narrowing down if it helps.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Please revert "use todc time functions for PPC_PREP" changeset

[Leigh Brown]

Benjamin Herrenschmidt said:
> On Sat, 2004-02-28 at 19:32, Andrew Morton wrote:
>> "Leigh Brown" <leigh@solinno.co.uk> wrote:
>> >
>> > Hi Tom,
>> >
>> >  Your patch (ChangeSet@1.1557.41.25) breaks the 43p140 and Carolina,
>> and
>> >  probably quite a few other RS/6000 based PRePs.  I was a bit
>> surprised
>> >  to see it 2.6.3.
>> >
>> >  http://linux.bkbits.net:8080/linux-2.5/patch@1.1557.41.25?nav=index.html|ChangeSet@-3w|cset@1.1557.41.25
>> >
>> >  Can we revert this until there is a working version, please?
>>
>> It wouild help if you could define "breaks".
>
> Yup, Leigh, we have at least a week I suppose before 2.6.4 goes final, so
> we can probably fix the problem rather than reverting the cleanup.
>
> Can you tell us more about what's going on ?

Okay, great.  I have to go out now but I'll narrow down the exact function
that's hanging when I get back.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Please revert "use todc time functions for PPC_PREP" changeset

[Tom Rini]

On Sat, Feb 28, 2004 at 08:43:06AM -0000, Leigh Brown wrote:
> Andrew Morton said:
> >
> > "Leigh Brown" <leigh@solinno.co.uk> wrote:
> >>
> >> Hi Tom,
> >>
> >>  Your patch (ChangeSet@1.1557.41.25) breaks the 43p140 and Carolina, and
> >>  probably quite a few other RS/6000 based PRePs.  I was a bit surprised
> >>  to see it 2.6.3.
> >>
> >>  http://linux.bkbits.net:8080/linux-2.5/patch@1.1557.41.25?nav=index.html|ChangeSet@-3w|cset@1.1557.41.25
> >>
> >>  Can we revert this until there is a working version, please?
> >
> > It wouild help if you could define "breaks".
>
> Sorry!  It hangs during boot, before any output on the serial console.  I
> got as far as working out that it hangs somewhere in
> arch/ppc/kernel/time.c:time_init() (or one of the called functions).

Er.. Didn't we talk about this awhile ago, and get a different version
of this patch working back in 2.4?  The problem was that we couldn't use
todc_calibrate_decr we had to still grab the value from the residual
data.  And I thought I had kept that logic still..

--
Tom Rini
http://gate.crashing.org/~trini/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Please revert "use todc time functions for PPC_PREP" changeset

[Leigh Brown]

Tom Rini said:
> On Sat, Feb 28, 2004 at 08:43:06AM -0000, Leigh Brown wrote:
>> Andrew Morton said:
>> >
>> > "Leigh Brown" <leigh@solinno.co.uk> wrote:
>> >>  Your patch (ChangeSet@1.1557.41.25) breaks the 43p140 and Carolina,
>> and
>> >>  probably quite a few other RS/6000 based PRePs.  I was a bit
>> surprised
>> >>  to see it 2.6.3.
>> >>
>> >>  http://linux.bkbits.net:8080/linux-2.5/patch@1.1557.41.25?nav=index.html|ChangeSet@-3w|cset@1.1557.41.25
>> >>
>> >>  Can we revert this until there is a working version, please?
>> >
>> > It wouild help if you could define "breaks".
>>
>> Sorry!  It hangs during boot, before any output on the serial console.
>> I
>> got as far as working out that it hangs somewhere in
>> arch/ppc/kernel/time.c:time_init() (or one of the called functions).
>
> Er.. Didn't we talk about this awhile ago, and get a different version
> of this patch working back in 2.4?  The problem was that we couldn't use
> todc_calibrate_decr we had to still grab the value from the residual
> data.  And I thought I had kept that logic still..

Just a quick update on this.  Tom has a patch that I've tested that
makes it work again.  I very much hope it will be submitted shortly.

Thanks all for your help.

Leigh.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/