* Some cache questions
@ 2004-12-27 16:35 Thomas Petazzoni
0 siblings, 0 replies; 4+ messages in thread
From: Thomas Petazzoni @ 2004-12-27 16:35 UTC (permalink / raw)
To: linux-mips
[-- Attachment #1: Type: text/plain, Size: 1165 bytes --]
Hello,
I'm using an RM9000 dual-core processor, buggy revisions (the one that
doesn't support the "Shared" cache state if I understood correctly).
When going through the CVS logs, I saw that Ralf quite recently changed
the cache mode from 4 to 5 in pgtable-bits.h. Is that change involved in
the use of the "Shared" cache state with newer RM9000 revisions that
don't have the bug ?
Currently, the KSEG0 cache coherency mode (2 lower bits of the CONFIG
register) is set to 3 during PMON (start.S file). When I write something
to the memory through KSEG0 with the first core, it doesn't appear to be
read by the second core. This indicates, in my opinion, that the cache
line of the first core hasn't been written to memory so that the second
core could use it. Am I right ?
If I want to correctly use both cores using KSEG0, should I set the mode
in the CONFIG register to 4 (so that I can work with buggy processors) ?
Thanks,
Thomas
--
PETAZZONI Thomas - thomas.petazzoni@enix.org
http://thomas.enix.org - Jabber: thomas.petazzoni@jabber.dk
http://kos.enix.org, http://sos.enix.org
Fingerprint : 0BE1 4CF3 CEA4 AC9D CC6E 1624 F653 CB30 98D3 F7A7
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]
^ permalink raw reply [flat|nested] 4+ messages in thread
* RE: Some cache questions
@ 2004-12-27 22:11 Brad Larson
2004-12-28 4:02 ` Manish Lachwani
0 siblings, 1 reply; 4+ messages in thread
From: Brad Larson @ 2004-12-27 22:11 UTC (permalink / raw)
To: 'Thomas Petazzoni', linux-mips
You haven't mentioned which board. If its Yosemite then you may have one of the few not upgraded to 1.2 silicon. If so it won't work with the changes committed by Ralf which requires the shared state for SMP boot. For further discussion contact the apps@pmc-sierra.com
--Brad
-----Original Message-----
From: linux-mips-bounce@linux-mips.org
[mailto:linux-mips-bounce@linux-mips.org]On Behalf Of Thomas Petazzoni
Sent: Monday, December 27, 2004 8:35 AM
To: linux-mips@linux-mips.org
Subject: Some cache questions
Hello,
I'm using an RM9000 dual-core processor, buggy revisions (the one that
doesn't support the "Shared" cache state if I understood correctly).
When going through the CVS logs, I saw that Ralf quite recently changed
the cache mode from 4 to 5 in pgtable-bits.h. Is that change involved in
the use of the "Shared" cache state with newer RM9000 revisions that
don't have the bug ?
Currently, the KSEG0 cache coherency mode (2 lower bits of the CONFIG
register) is set to 3 during PMON (start.S file). When I write something
to the memory through KSEG0 with the first core, it doesn't appear to be
read by the second core. This indicates, in my opinion, that the cache
line of the first core hasn't been written to memory so that the second
core could use it. Am I right ?
If I want to correctly use both cores using KSEG0, should I set the mode
in the CONFIG register to 4 (so that I can work with buggy processors) ?
Thanks,
Thomas
--
PETAZZONI Thomas - thomas.petazzoni@enix.org
http://thomas.enix.org - Jabber: thomas.petazzoni@jabber.dk
http://kos.enix.org, http://sos.enix.org
Fingerprint : 0BE1 4CF3 CEA4 AC9D CC6E 1624 F653 CB30 98D3 F7A7
^ permalink raw reply [flat|nested] 4+ messages in thread
* RE: Some cache questions
2004-12-27 22:11 Some cache questions Brad Larson
@ 2004-12-28 4:02 ` Manish Lachwani
2004-12-28 8:21 ` Thomas Petazzoni
0 siblings, 1 reply; 4+ messages in thread
From: Manish Lachwani @ 2004-12-28 4:02 UTC (permalink / raw)
To: Brad Larson, 'Thomas Petazzoni', linux-mips
Hello !
For chip revisions 1.0 and 1.1, there are some changes
to the memory management subsystem for the kernel to
work on the board (dual core). As already known, these
versions dont support Shared state.
I had made those changes to the 2.4.21 kernel. Maybe
you can take a look at those changes and port them to
2.6 appropriately. However, there is more sanity in
1.2 version
Thanks
Manish Lachwani
--- Brad Larson <Brad_Larson@pmc-sierra.com> wrote:
> You haven't mentioned which board. If its Yosemite
> then you may have one of the few not upgraded to 1.2
> silicon. If so it won't work with the changes
> committed by Ralf which requires the shared state
> for SMP boot. For further discussion contact the
> apps@pmc-sierra.com
>
> --Brad
>
> -----Original Message-----
> From: linux-mips-bounce@linux-mips.org
> [mailto:linux-mips-bounce@linux-mips.org]On Behalf
> Of Thomas Petazzoni
> Sent: Monday, December 27, 2004 8:35 AM
> To: linux-mips@linux-mips.org
> Subject: Some cache questions
>
>
> Hello,
>
> I'm using an RM9000 dual-core processor, buggy
> revisions (the one that
> doesn't support the "Shared" cache state if I
> understood correctly).
>
> When going through the CVS logs, I saw that Ralf
> quite recently changed
> the cache mode from 4 to 5 in pgtable-bits.h. Is
> that change involved in
> the use of the "Shared" cache state with newer
> RM9000 revisions that
> don't have the bug ?
>
> Currently, the KSEG0 cache coherency mode (2 lower
> bits of the CONFIG
> register) is set to 3 during PMON (start.S file).
> When I write something
> to the memory through KSEG0 with the first core, it
> doesn't appear to be
> read by the second core. This indicates, in my
> opinion, that the cache
> line of the first core hasn't been written to memory
> so that the second
> core could use it. Am I right ?
>
> If I want to correctly use both cores using KSEG0,
> should I set the mode
> in the CONFIG register to 4 (so that I can work with
> buggy processors) ?
>
> Thanks,
>
> Thomas
> --
> PETAZZONI Thomas - thomas.petazzoni@enix.org
> http://thomas.enix.org - Jabber:
> thomas.petazzoni@jabber.dk
> http://kos.enix.org, http://sos.enix.org
> Fingerprint : 0BE1 4CF3 CEA4 AC9D CC6E 1624 F653
> CB30 98D3 F7A7
>
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Some cache questions
2004-12-28 4:02 ` Manish Lachwani
@ 2004-12-28 8:21 ` Thomas Petazzoni
0 siblings, 0 replies; 4+ messages in thread
From: Thomas Petazzoni @ 2004-12-28 8:21 UTC (permalink / raw)
To: linux-mips
[-- Attachment #1: Type: text/plain, Size: 1315 bytes --]
Hello,
Manish Lachwani a écrit :
> For chip revisions 1.0 and 1.1, there are some changes
> to the memory management subsystem for the kernel to
> work on the board (dual core). As already known, these
> versions dont support Shared state.
>
> I had made those changes to the 2.4.21 kernel. Maybe
> you can take a look at those changes and port them to
> 2.6 appropriately. However, there is more sanity in
> 1.2 version
Actually, my question was not really Linux-specific. On the second core,
I will not use the MMU, because this core will not run Linux, but a
custom code. Both cores will share informations through KSEG0, so I need
to maintain coherency between caches. What should I do in order to do
that ? Is it enough to set cache mode for KSEG0 to 4 (in the CONFIG
register) ?
I have only 1.0 and 1.1 cores, on home-made boards, so there's no way to
switch to 1.2.
BTW, do you have pointers, papers, information about a system running
Linux on a core, and some custom code on a second core, in order to have
real-time on the second core with very low latency ?
Thanks,
Thomas
--
PETAZZONI Thomas - thomas.petazzoni@enix.org
http://thomas.enix.org - Jabber: thomas.petazzoni@jabber.dk
http://kos.enix.org, http://sos.enix.org
Fingerprint : 0BE1 4CF3 CEA4 AC9D CC6E 1624 F653 CB30 98D3 F7A7
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2004-12-28 8:19 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-12-27 22:11 Some cache questions Brad Larson
2004-12-28 4:02 ` Manish Lachwani
2004-12-28 8:21 ` Thomas Petazzoni
-- strict thread matches above, loose matches on Subject: below --
2004-12-27 16:35 Thomas Petazzoni
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox