2003-08-18 13:09:47

by Stian Jordet

[permalink] [raw]
Subject: 2.6.0-test3 latest bk hangs when enabling IO-APIC

Hello,

latest bk of 2.6.0-test3 hangs with these three lines:

ENABLING IO-APIC IRQs
Setting 2 in the phys_id_present_map
...changing IO-APIC physical APIC ID to 2 ... ok.

And there it stays forever. 2.6.0-test3 worked like a charm. This is a
Asus CUV265-DLS motherboard. Dual P3.

Should I file a bugreport at bugzilla?

Best regards,
Stian


2003-08-18 17:53:53

by Brown, Len

[permalink] [raw]
Subject: RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

Try booting with pci=noacpi, and if that doesn't work acpi=off
If either of those work, then file in bugzilla with component=ACPI and
assign it to [email protected]

Thanks,
-Len

> -----Original Message-----
> From: Stian Jordet [mailto:[email protected]]
> Sent: Monday, August 18, 2003 9:10 AM
> To: [email protected]
> Subject: 2.6.0-test3 latest bk hangs when enabling IO-APIC
>
>
> Hello,
>
> latest bk of 2.6.0-test3 hangs with these three lines:
>
> ENABLING IO-APIC IRQs
> Setting 2 in the phys_id_present_map
> ...changing IO-APIC physical APIC ID to 2 ... ok.
>
> And there it stays forever. 2.6.0-test3 worked like a charm. This is a
> Asus CUV265-DLS motherboard. Dual P3.
>
> Should I file a bugreport at bugzilla?
>
> Best regards,
> Stian
>
> -
> To unsubscribe from this list: send the line "unsubscribe
> linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>

2003-08-18 19:24:49

by Stian Jordet

[permalink] [raw]
Subject: RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

man, 18.08.2003 kl. 19.53 skrev Brown, Len:
> Try booting with pci=noacpi, and if that doesn't work acpi=off
> If either of those work, then file in bugzilla with component=ACPI and
> assign it to [email protected]

It works when booting with noapic, but not with acpi=off nor pci=noapic.
Does that mean I can't blame you for it?

Best regards,
Stian

> > --Original Message--
> > From: Stian Jordet [mailto:[email protected]]
> > Sent: Monday, August 18, 2003 9:10 AM
> > To: [email protected]
> > Subject: 2.6.0-test3 latest bk hangs when enabling IO-APIC
> >
> >
> > Hello,
> >
> > latest bk of 2.6.0-test3 hangs with these three lines:
> >
> > ENABLING IO-APIC IRQs
> > Setting 2 in the phys_id_present_map
> > ...changing IO-APIC physical APIC ID to 2 ... ok.
> >
> > And there it stays forever. 2.6.0-test3 worked like a charm. This is a
> > Asus CUV265-DLS motherboard. Dual P3.
> >
> > Should I file a bugreport at bugzilla?
> >
> > Best regards,
> > Stian

2003-08-18 20:05:58

by Stian Jordet

[permalink] [raw]
Subject: RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

man, 18.08.2003 kl. 21.23 skrev Stian Jordet:
> man, 18.08.2003 kl. 19.53 skrev Brown, Len:
> > Try booting with pci=noacpi, and if that doesn't work acpi=off
> > If either of those work, then file in bugzilla with component=ACPI and
> > assign it to [email protected]
>
> It works when booting with noapic, but not with acpi=off nor pci=noapic.
> Does that mean I can't blame you for it?

Just to confuse myself (and whoever reading my mails). When I disabled
the second onboard ide-port, the kernel booted, but usb didn't work.
Absolutely no problem what so ever with -test3.

Best regards,
Stian

2003-08-18 23:06:32

by Stian Jordet

[permalink] [raw]
Subject: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

man, 18.08.2003 kl. 22.06 skrev Stian Jordet:
> man, 18.08.2003 kl. 21.23 skrev Stian Jordet:
> > man, 18.08.2003 kl. 19.53 skrev Brown, Len:
> > > Try booting with pci=noacpi, and if that doesn't work acpi=off
> > > If either of those work, then file in bugzilla with component=ACPI and
> > > assign it to [email protected]
> >
> > It works when booting with noapic, but not with acpi=off nor pci=noapic.
> > Does that mean I can't blame you for it?
>
> Just to confuse myself (and whoever reading my mails). When I disabled
> the second onboard ide-port, the kernel booted, but usb didn't work.
> Absolutely no problem what so ever with -test3.

I had to enable CONFIG_ACPI_HT=y. But you must have screwed something
up, since my P3's really, really don't have any Hyper Threading :) ACPI
was already enabled, and from the help texts, there shouldn't be any
difference with ACPI_HT then. Nevermind, I'm happy :)

Best regards,
Stian

2003-08-18 23:12:06

by Stian Jordet

[permalink] [raw]
Subject: Re: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

tir, 19.08.2003 kl. 01.06 skrev Stian Jordet:
> man, 18.08.2003 kl. 22.06 skrev Stian Jordet:
> > man, 18.08.2003 kl. 21.23 skrev Stian Jordet:
> > > man, 18.08.2003 kl. 19.53 skrev Brown, Len:
> > > > Try booting with pci=noacpi, and if that doesn't work acpi=off
> > > > If either of those work, then file in bugzilla with component=ACPI and
> > > > assign it to [email protected]
> > >
> > > It works when booting with noapic, but not with acpi=off nor pci=noapic.
> > > Does that mean I can't blame you for it?
> >
> > Just to confuse myself (and whoever reading my mails). When I disabled
> > the second onboard ide-port, the kernel booted, but usb didn't work.
> > Absolutely no problem what so ever with -test3.
>
> I had to enable CONFIG_ACPI_HT=y. But you must have screwed something
> up, since my P3's really, really don't have any Hyper Threading :) ACPI
> was already enabled, and from the help texts, there shouldn't be any
> difference with ACPI_HT then. Nevermind, I'm happy :)

Sorry for spamming, but when I think of it, I don't think acpi is
enabled at all without the CONFIG_ACPI_HT. Because my computer actually
uses to hang when enabling io-apic without acpi support enabled (in
other words, my pc is useless without acpi enabled).

Once again, sorry for flooding.

Best regards,
Stian

2003-08-19 01:15:54

by Brown, Len

[permalink] [raw]
Subject: RE: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

CONFIG_ACPI_HT is mostly just an alias for CONFIG_ACPI_BOOT -- the early
boot part of ACPI without the run-time events included in the full ACPI
implementation. Unless I screwed up the config dependencies, it should
be impossible to enable the full CONFIG_ACPI without including
CONFIG_ACPI_HT.

When full CONFIG_ACPI is used, the cmdline "acpi=ht" can be used to
revert to just the CONFIG_ACPI_HT behaviour.

The help built into config should explain this clearly, let me know if
it doesn't.

Thanks,
-Len

Ps. The "noapic" flag is actually broken in the baseline kernel in some
cases (ACPI enabled), a fix will come with the next batch...

> -----Original Message-----
> From: Stian Jordet [mailto:[email protected]]
> Sent: Monday, August 18, 2003 7:12 PM
> To: Brown, Len
> Cc: [email protected]
> Subject: Re: [SOLVED] RE: 2.6.0-test3 latest bk hangs when
> enabling IO-APIC
>
>
> tir, 19.08.2003 kl. 01.06 skrev Stian Jordet:
> > man, 18.08.2003 kl. 22.06 skrev Stian Jordet:
> > > man, 18.08.2003 kl. 21.23 skrev Stian Jordet:
> > > > man, 18.08.2003 kl. 19.53 skrev Brown, Len:
> > > > > Try booting with pci=noacpi, and if that doesn't work acpi=off
> > > > > If either of those work, then file in bugzilla with
> component=ACPI and
> > > > > assign it to [email protected]
> > > >
> > > > It works when booting with noapic, but not with
> acpi=off nor pci=noapic.
> > > > Does that mean I can't blame you for it?
> > >
> > > Just to confuse myself (and whoever reading my mails).
> When I disabled
> > > the second onboard ide-port, the kernel booted, but usb
> didn't work.
> > > Absolutely no problem what so ever with -test3.
> >
> > I had to enable CONFIG_ACPI_HT=y. But you must have screwed
> something
> > up, since my P3's really, really don't have any Hyper
> Threading :) ACPI
> > was already enabled, and from the help texts, there shouldn't be any
> > difference with ACPI_HT then. Nevermind, I'm happy :)
>
> Sorry for spamming, but when I think of it, I don't think acpi is
> enabled at all without the CONFIG_ACPI_HT. Because my
> computer actually
> uses to hang when enabling io-apic without acpi support enabled (in
> other words, my pc is useless without acpi enabled).
>
> Once again, sorry for flooding.
>
> Best regards,
> Stian
>
>

2003-08-19 01:57:40

by Jeff Garzik

[permalink] [raw]
Subject: Re: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

Brown, Len wrote:
> implementation. Unless I screwed up the config dependencies, it should
> be impossible to enable the full CONFIG_ACPI without including
> CONFIG_ACPI_HT.


You got it right...

...unfortunately the current options are very non-obvious. Linux kernel
config has always presented a "subsystem switch", and under that, you
have a bunch of options and modules and such. It is not obvious to me
at all that _disabling_ CONFIG_ACPI_HT means that CONFIG_ACPI will
simply never appear. And users who want ACPI, but don't have
HyperThreading, would probably think similarly. I think the current
thread shows it generates at least a little bit of confusion, too.

To me, it makes sense for CONFIG_ACPI_HT to require CONFIG_ACPI -- from
a Kconfig standpoint -- but not the other way around. My first time
configuring with the new options, I wanted to turn off CONFIG_ACPI_HT
and turn on CONFIG_ACPI, but was frustrated when the config system
wouldn't allow it.

I actually think the old method was a bit less confusing --
CONFIG_ACPI=N would never prompt me for any other ACPI config option.
But if I wanted only the HT stuff, you had CONFIG_ACPI_HT_ONLY :)

So... concrete suggestions? Overall, IMO, move everything under
CONFIG_ACPI, or, make CONFIG_ACPI_BOOT a _peer_ option, whose selection
or lackthereof doesn't affect CONFIG_ACPI visibility at all.

Jeff



2003-08-19 03:17:49

by Brown, Len

[permalink] [raw]
Subject: RE: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

> So... concrete suggestions? Overall, IMO, move everything under
> CONFIG_ACPI, or, make CONFIG_ACPI_BOOT a _peer_ option, whose
> selection
> or lackthereof doesn't affect CONFIG_ACPI visibility at all.

Simply re-naming CONFIG_ACPI_HT to be CONFIG_ACPI_BOOT might help, as it
would be more clear that it is necessary for the rest of ACPI. However,
it may not be obvious that it provides the minimal config to enable HT.

Re: peers
Unfortunately ACPI doesn't work so well if CONFIG_ACPI_BOOT is left out.
Yes, it's conceivable, but I spent several hours tinkering with it in
search of a "noht" build option, but ditched it b/c it seemed like a
build option very few would use.

Re: CONFIG_ACPI is the the master switch, and all other ACPI options
subservient...
If implemented literally, this is sort of a pain, as CONFIG_ACPI appears
all over the code. However, a dummy config master ACPI config option
could be used to enable the menu that contains all the rest of ACPI...

-Len

2003-08-19 07:22:52

by Stian Jordet

[permalink] [raw]
Subject: RE: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

tir, 19.08.2003 kl. 03.15 skrev Brown, Len:
> CONFIG_ACPI_HT is mostly just an alias for CONFIG_ACPI_BOOT -- the early
> boot part of ACPI without the run-time events included in the full ACPI
> implementation. Unless I screwed up the config dependencies, it should
> be impossible to enable the full CONFIG_ACPI without including
> CONFIG_ACPI_HT.

Hmm. I just ran "make oldconfig" and when I got the question about
"CONFIG_ACPI_HT" I just chose no, since I don't have HT. When I _now_
look at it with "make menuconfig" I ofcourse see that without the first,
I never get the latter one, but that was really non-obvious. But thanks
:)

Best regards,
Stian

2003-08-19 12:54:54

by Stian Jordet

[permalink] [raw]
Subject: RE: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

tir, 19.08.2003 kl. 05.17 skrev Brown, Len:
> > So... concrete suggestions? Overall, IMO, move everything under
> > CONFIG_ACPI, or, make CONFIG_ACPI_BOOT a _peer_ option, whose
> > selection
> > or lackthereof doesn't affect CONFIG_ACPI visibility at all.
>
> Simply re-naming CONFIG_ACPI_HT to be CONFIG_ACPI_BOOT might help, as it
> would be more clear that it is necessary for the rest of ACPI. However,
> it may not be obvious that it provides the minimal config to enable HT.
>
> Re: peers
> Unfortunately ACPI doesn't work so well if CONFIG_ACPI_BOOT is left out.
> Yes, it's conceivable, but I spent several hours tinkering with it in
> search of a "noht" build option, but ditched it b/c it seemed like a
> build option very few would use.
>
> Re: CONFIG_ACPI is the the master switch, and all other ACPI options
> subservient...
> If implemented literally, this is sort of a pain, as CONFIG_ACPI appears
> all over the code. However, a dummy config master ACPI config option
> could be used to enable the menu that contains all the rest of ACPI...

Btw, (a little off-topic) should I file a bug-report that my motherboard
doesn't boot without acpi (never has, not even with 2.4), or should I
just smile and be happy because acpi works like a charm? (I already do
that :)

Thanks.

Regards,
Stian

2003-08-19 15:27:03

by Jeff Garzik

[permalink] [raw]
Subject: Re: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

On Tue, Aug 19, 2003 at 02:54:01PM +0200, Stian Jordet wrote:
> tir, 19.08.2003 kl. 05.17 skrev Brown, Len:
> > > So... concrete suggestions? Overall, IMO, move everything under
> > > CONFIG_ACPI, or, make CONFIG_ACPI_BOOT a _peer_ option, whose
> > > selection
> > > or lackthereof doesn't affect CONFIG_ACPI visibility at all.
> >
> > Simply re-naming CONFIG_ACPI_HT to be CONFIG_ACPI_BOOT might help, as it
> > would be more clear that it is necessary for the rest of ACPI. However,
> > it may not be obvious that it provides the minimal config to enable HT.
> >
> > Re: peers
> > Unfortunately ACPI doesn't work so well if CONFIG_ACPI_BOOT is left out.
> > Yes, it's conceivable, but I spent several hours tinkering with it in
> > search of a "noht" build option, but ditched it b/c it seemed like a
> > build option very few would use.
> >
> > Re: CONFIG_ACPI is the the master switch, and all other ACPI options
> > subservient...
> > If implemented literally, this is sort of a pain, as CONFIG_ACPI appears
> > all over the code. However, a dummy config master ACPI config option
> > could be used to enable the menu that contains all the rest of ACPI...
>
> Btw, (a little off-topic) should I file a bug-report that my motherboard
> doesn't boot without acpi (never has, not even with 2.4), or should I
> just smile and be happy because acpi works like a charm? (I already do
> that :)

Nah... you are actually a member of the growing number of people with
computers that require ACPI to boot. It's something that will become
more and more common and times goes on. Nothing really to "fix" in the
kernel besides turning on ACPI...

Jeff



2003-08-19 15:44:31

by Stian Jordet

[permalink] [raw]
Subject: Re: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

tir, 19.08.2003 kl. 17.26 skrev Jeff Garzik:
> On Tue, Aug 19, 2003 at 02:54:01PM +0200, Stian Jordet wrote:
> > tir, 19.08.2003 kl. 05.17 skrev Brown, Len:
> > > > So... concrete suggestions? Overall, IMO, move everything under
> > > > CONFIG_ACPI, or, make CONFIG_ACPI_BOOT a _peer_ option, whose
> > > > selection
> > > > or lackthereof doesn't affect CONFIG_ACPI visibility at all.
> > >
> > > Simply re-naming CONFIG_ACPI_HT to be CONFIG_ACPI_BOOT might help, as it
> > > would be more clear that it is necessary for the rest of ACPI. However,
> > > it may not be obvious that it provides the minimal config to enable HT.
> > >
> > > Re: peers
> > > Unfortunately ACPI doesn't work so well if CONFIG_ACPI_BOOT is left out.
> > > Yes, it's conceivable, but I spent several hours tinkering with it in
> > > search of a "noht" build option, but ditched it b/c it seemed like a
> > > build option very few would use.
> > >
> > > Re: CONFIG_ACPI is the the master switch, and all other ACPI options
> > > subservient...
> > > If implemented literally, this is sort of a pain, as CONFIG_ACPI appears
> > > all over the code. However, a dummy config master ACPI config option
> > > could be used to enable the menu that contains all the rest of ACPI...
> >
> > Btw, (a little off-topic) should I file a bug-report that my motherboard
> > doesn't boot without acpi (never has, not even with 2.4), or should I
> > just smile and be happy because acpi works like a charm? (I already do
> > that :)
>
> Nah... you are actually a member of the growing number of people with
> computers that require ACPI to boot. It's something that will become
> more and more common and times goes on. Nothing really to "fix" in the
> kernel besides turning on ACPI...

Ok, thanks. Just a note, this isn't a new board, it's almost precisely
two years old.

Thanks :)

Regards,
Stian

2003-08-22 02:44:10

by Mark W. Alexander

[permalink] [raw]
Subject: RE: [SOLVED] RE: 2.6.0-test3 latest bk hangs when enabling IO-APIC

On Mon, 18 Aug 2003, Brown, Len wrote:

> CONFIG_ACPI_HT is mostly just an alias for CONFIG_ACPI_BOOT -- the early
> boot part of ACPI without the run-time events included in the full ACPI
> implementation. Unless I screwed up the config dependencies, it should
> be impossible to enable the full CONFIG_ACPI without including
> CONFIG_ACPI_HT.

Then one of us screwed up. I've configured everything ACPI I could
find via gconf and there's no sign on CONFIG_ACPI_HT on my .config. I
_do_ have CONFIG_ACPI_BOOT, though, so what to you mean by "mostly"
just an alias.

(And please be gentle, I'm just figuring out ACPI out of necessity
since my new HP ze4400 barely has a BIOS, let alone APM ;)

mwa
--
Mark W. Alexander
[email protected]