Hi,
I was hoping to use HT on my new Latitude C640 (P4 @ 1.8Ghz) but at boot
both 2.4.21-pre3 and 2.5.59 (obviously with a SMP kernel) tell me
"Dell Latitude with broken BIOS detected. Refusing to enable the local APIC."
Is this anything that can be played with ? I google'd a bit but was unable to
turn up anything interesting. I flashed BIOS A05 (which is the latest on Dell
support sites), but I still get the above kernel message.
Thanks in advance for any insight,
--alessandro
On Mon, Jan 20, 2003 at 04:31:58AM -0800, Alessandro Suardi wrote:
> I was hoping to use HT on my new Latitude C640 (P4 @ 1.8Ghz) but at boot
> both 2.4.21-pre3 and 2.5.59 (obviously with a SMP kernel) tell me
I'd be surprised^Wamazed if your laptop has HT. AFAIK, no-one is
shipping such a system yet. Just because the CPU flags say 'ht' does
not mean it has >1 CPU thread per CPU package.
> "Dell Latitude with broken BIOS detected. Refusing to enable the local APIC."
Lots of Dell laptops (like other vendors) crash instantly when trying to
enable the APIC.
> Is this anything that can be played with ?
Nope.
Dave
--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs
Dave Jones wrote:
> On Mon, Jan 20, 2003 at 04:31:58AM -0800, Alessandro Suardi wrote:
> > I was hoping to use HT on my new Latitude C640 (P4 @ 1.8Ghz) but at boot
> > both 2.4.21-pre3 and 2.5.59 (obviously with a SMP kernel) tell me
>
> I'd be surprised^Wamazed if your laptop has HT. AFAIK, no-one is
> shipping such a system yet. Just because the CPU flags say 'ht' does
> not mean it has >1 CPU thread per CPU package.
I'd imagined this - but digging on the Intel website I couldn't find anything
that told me "the mobile P4 can't do HT, period". As a matter of fact not
even win2k (I'm dualbooting waiting to install Debian and RHAS...) sees
the system as a 2-CPU.
> > "Dell Latitude with broken BIOS detected. Refusing to enable the local APIC."
>
> Lots of Dell laptops (like other vendors) crash instantly when trying to
> enable the APIC.
Well my Dells power off on rebooting from 2.5... bug 119 or 134 in
http://bugme.osdl.org, no need to resort to messing with the APIC ;(
> > Is this anything that can be played with ?
>
> Nope.
Oh, okay. Giving up on this one...
Thanks for the quick reply ! Ciao,
--alessandro
On Mon, Jan 20, 2003 at 05:28:49AM -0800, Alessandro Suardi wrote:
> > > "Dell Latitude with broken BIOS detected. Refusing to enable the local APIC."
> > Lots of Dell laptops (like other vendors) crash instantly when trying to
> > enable the APIC.
> Well my Dells power off on rebooting from 2.5... bug 119 or 134 in
> http://bugme.osdl.org, no need to resort to messing with the APIC ;(
That one IMO looks like an ACPI problem (Note ACPI != APIC)
Dave
--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs
> On Mon, Jan 20, 2003 at 05:28:49AM -0800, Alessandro Suardi wrote:
>
> > > > "Dell Latitude with broken BIOS detected. Refusing to enable the local APIC."
> > >
> > > Lots of Dell laptops (like other vendors) crash instantly when trying to
> > > enable the APIC.
> >
> > Well my Dells power off on rebooting from 2.5... bug 119 or 134 in
> > http://bugme.osdl.org, no need to resort to messing with the APIC ;(
>
> That one IMO looks like an ACPI problem (Note ACPI != APIC)
More like PM in general, happens with my older Latitude with APM as well
(I was making a point of ease of crashing, not confusing APIC and ACPI ;)
--alessandro
On Mon, 20 Jan 2003 04:31:58 -0800 (GMT-08:00), Alessandro Suardi wrote:
> I was hoping to use HT on my new Latitude C640 (P4 @ 1.8Ghz) but at boot
> both 2.4.21-pre3 and 2.5.59 (obviously with a SMP kernel) tell me
>
> "Dell Latitude with broken BIOS detected. Refusing to enable the local APIC."
>
> Is this anything that can be played with ?
First of all, your 1.8GHz mobile P4 doesn't actually have HT.
The only ones to have it are the new 3.06GHz P4s, and most Xeons.
Secondly, I'm responsible for the message you quoted. Many if not all
Pentium III-based Dell laptops (including Latitude Cnnn and I8nnn)
that have local-APIC capable processors fail miserably if the OS
actually enables it. For instance, pulling or inserting the DC
power plug would hang the machine. This is a BIOS bug we can't work
around, except by refusing to enable the local APIC.
Your P4-based Latitude probably has a different BIOS than the buggy
P3-based ones, and it may work better. Try commenting out the
local_apic_kills_bios entry for "Dell Latitude" at around line 692
in arch/i386/kernel/dmi_scan.c and rebuild the kernel. If it
boots ok and enables the local APIC, try various things that
should trigger BIOS (actually SMM) events:
- pull and insert the DC power plug
- enter and exit the BIOS setup screens
- run a big heavy compile job for half an hour or so
Let us know if it survives all this, or if it hangs or crashes.
/Mikael
Mikael Pettersson wrote:
> On Mon, 20 Jan 2003 04:31:58 -0800 (GMT-08:00), Alessandro Suardi wrote:
> > I was hoping to use HT on my new Latitude C640 (P4 @ 1.8Ghz) but at boot
> > both 2.4.21-pre3 and 2.5.59 (obviously with a SMP kernel) tell me
> >
> > "Dell Latitude with broken BIOS detected. Refusing to enable the local APIC."
> >
> > Is this anything that can be played with ?
>
> First of all, your 1.8GHz mobile P4 doesn't actually have HT.
> The only ones to have it are the new 3.06GHz P4s, and most Xeons.
Ok.
> Secondly, I'm responsible for the message you quoted. Many if not all
> Pentium III-based Dell laptops (including Latitude Cnnn and I8nnn)
> that have local-APIC capable processors fail miserably if the OS
> actually enables it. For instance, pulling or inserting the DC
> power plug would hang the machine. This is a BIOS bug we can't work
> around, except by refusing to enable the local APIC.
>
> Your P4-based Latitude probably has a different BIOS than the buggy
> P3-based ones, and it may work better. Try commenting out the
> local_apic_kills_bios entry for "Dell Latitude" at around line 692
> in arch/i386/kernel/dmi_scan.c and rebuild the kernel. If it
(ahem) I had tried that in 2.5.58 already ;)
However I rebuilt 2.5.59 with that change, and I'm not going further:
No local APIC present or hardware disabled
I have both CONFIG_X86_IO_APIC and CONFIG_X86_LOCAL_APIC=y.
(while I'm using my old PIII/750 the PIV is still okay for testing ;)
Ciao,
--alessandro
Alessandro Suardi wrote:
> > Your P4-based Latitude probably has a different BIOS than the buggy
> > P3-based ones, and it may work better. Try commenting out the
> > local_apic_kills_bios entry for "Dell Latitude" at around line 692
> > in arch/i386/kernel/dmi_scan.c and rebuild the kernel. If it
>
> (ahem) I had tried that in 2.5.58 already ;)
>
> However I rebuilt 2.5.59 with that change, and I'm not going further:
>
> No local APIC present or hardware disabled
Ah, your P4 needs another patch. Edit arch/i386/kernel/apic.c, function
detect_init_APIC(), at line 631, and remove the "&& cpu_has_apic" after
the "boot_cpu_data.x86 == 15" test. (Or check if your BIOS can be set up
to boot with the local APIC enabled.)
/Mikael
Mikael Pettersson wrote:
> Alessandro Suardi wrote:
>
> > > Your P4-based Latitude probably has a different BIOS than the buggy
> > > P3-based ones, and it may work better. Try commenting out the
> > > local_apic_kills_bios entry for "Dell Latitude" at around line 692
> > > in arch/i386/kernel/dmi_scan.c and rebuild the kernel. If it
> >
> > (ahem) I had tried that in 2.5.58 already ;)
> >
> > However I rebuilt 2.5.59 with that change, and I'm not going further:
> >
> > No local APIC present or hardware disabled
>
> Ah, your P4 needs another patch. Edit arch/i386/kernel/apic.c, function
> detect_init_APIC(), at line 631, and remove the "&& cpu_has_apic" after
> the "boot_cpu_data.x86 == 15" test. (Or check if your BIOS can be set up
> to boot with the local APIC enabled.)
>
> /Mikael
The BIOS doesn't seem to have any such option... and the code change
does exactly what you imagined - pull the power cord, instant hang.
It seems even BIOS A05 of the C640 is legitimately blacklisted :(
Thanks, ciao,
--alessandro
I am not well aware of what APIC is but I was running Mandrake 8.2 on my Linux
partition of a Dell Pentium III latitude 550 MHz and I don't remember such a
dmesg message. But when I upgraded to Mandrake 9.1 here it is. The problem I
have is that I can't have any suspend mode any more neither battery
indicators and /etc/rc.d/init.d/apm start claims that apm is no more in the
kernel. Is it linked to that APIC problem and this BIOS problem, why did it
work earlier? Do you think that if I found a new bios from Dell it will help?
Any information will be greatly appreciated (I am sorry no to be able to
change my laptop so frequently...)
--
Nicolas Brouard
Brouard Nicolas writes:
> I am not well aware of what APIC is but I was running Mandrake 8.2 on my Linux
> partition of a Dell Pentium III latitude 550 MHz and I don't remember such a
> dmesg message. But when I upgraded to Mandrake 9.1 here it is. The problem I
> have is that I can't have any suspend mode any more neither battery
> indicators and /etc/rc.d/init.d/apm start claims that apm is no more in the
> kernel. Is it linked to that APIC problem and this BIOS problem, why did it
> work earlier? Do you think that if I found a new bios from Dell it will help?
The "$machine with broken BIOS detected, refusing to enable the local APIC"
message only affects the local APIC and the few services using it like the
NMI watchdog and some performance measurement/profiling tools.
It has no impact on whether APM works or not.
Possibly Mandrake 9.1 detects ACPI (not APIC) which would disable APM.
Try booting with "noacpi" or "acpi=off" or whatever the option is called.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Friday 23 May 2003 9:55 am, Brouard Nicolas wrote:
> I am not well aware of what APIC is but I was running Mandrake 8.2 on my
> Linux partition of a Dell Pentium III latitude 550 MHz and I don't remember
> such a dmesg message. But when I upgraded to Mandrake 9.1 here it is. The
> problem I have is that I can't have any suspend mode any more neither
> battery indicators and /etc/rc.d/init.d/apm start claims that apm is no
> more in the kernel. Is it linked to that APIC problem and this BIOS
> problem, why did it work earlier? Do you think that if I found a new bios
> from Dell it will help?
>
> Any information will be greatly appreciated (I am sorry no to be able to
> change my laptop so frequently...)
On my Latitute C610, I have to boot with ACPI=off under Mandrake 9.1 otherwise
nothing power related works. This allows the kernel to revery back to APM and
you get battery status things et al.
Mark.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)
iD8DBQE+zl1Va0XO123WNagRAuCFAJ49frC2WTjqEPyGUlD0RkUe3gULPQCfd0bY
caCqMvILiCxP8I3eEcs3pOU=
=0nu8
-----END PGP SIGNATURE-----
Yes acpi=off solved the APM problem.
I am able to suspend my laptop again!
In Mandrake Control Center, in the Lilo parameter options the acpi=off or
acpi=on is prepared! There is a second option which can be set: "noapic".
There is no other option.
But by default, it was acpi=on and apic.
Thanks a lot!
Nicolas Brouard
Le Vendredi 23 Mai 2003 11:02, [email protected] a ?crit :
> Brouard Nicolas writes:
> > I am not well aware of what APIC is but I was running Mandrake 8.2 on my
> > Linux partition of a Dell Pentium III latitude 550 MHz and I don't
> > remember such a dmesg message. But when I upgraded to Mandrake 9.1 here
> > it is. The problem I have is that I can't have any suspend mode any more
> > neither battery indicators and /etc/rc.d/init.d/apm start claims that
> > apm is no more in the kernel. Is it linked to that APIC problem and this
> > BIOS problem, why did it work earlier? Do you think that if I found a
> > new bios from Dell it will help?
>
> The "$machine with broken BIOS detected, refusing to enable the local APIC"
> message only affects the local APIC and the few services using it like the
> NMI watchdog and some performance measurement/profiling tools.
> It has no impact on whether APM works or not.
>
> Possibly Mandrake 9.1 detects ACPI (not APIC) which would disable APM.
> Try booting with "noacpi" or "acpi=off" or whatever the option is called.
--
Nicolas Brouard