2011-05-02 13:57:06

by werner

[permalink] [raw]
Subject: 2.6.39-rc5-git2 boot crashs

A middle-big config (starting from Slackware huge-smp, or
from a L.T. werner-config) works. A more simple one, or a
more bigger one, both don't work, but probably by
different reasons.

A very basical config (starting from Slackware normal-smp)
stops booting after 2 seconds; it don't find the root
device (even if I told to it at the command line by the
bios or dev name), and although the bootwait option, it
crashs rather than simply wait.

An everythingyes config get several problems, reported
here (and in my 2.6.39-rc3 reclamations). Some of them
are: crashs on zipping big files; secondary
reset-resistent crashs wrongly understanding anything as
ata device.

wl
---
Professional hosting for everyone - http://www.host.ru


2011-05-02 15:16:52

by Steven Rostedt

[permalink] [raw]
Subject: Re: 2.6.39-rc5-git2 boot crashs

On Mon, 2011-05-02 at 09:56 -0400, werner wrote:
> A middle-big config (starting from Slackware huge-smp, or
> from a L.T. werner-config) works. A more simple one, or a
> more bigger one, both don't work, but probably by
> different reasons.

Could you email me privately the three configs.

Thanks!

-- Steve

2011-05-02 15:18:56

by Linus Torvalds

[permalink] [raw]
Subject: Re: 2.6.39-rc5-git2 boot crashs

On Sun, May 1, 2011 at 11:54 PM, werner <[email protected]> wrote:
>
> Then I gone back almost to my normal / everythingyes config, but I disabled
> CONFIG_MSC_FILESYSTEMS and CONFIG-MTD , also ACPI-APEI because it needs
> PSTORE of the 1st one.

Ok, good.

> With them disabled: ?The 1st booting in grafic mode was normally. I also
> could unzip a 120 MB file (what before with the everythingyes config
> crashed). ? However, if I tried to unzip an appr. 500 MB file, it crashed.

So that didn't help. Let's disable some other things too.

Start off by disabling CONFIG_IDE, because you don't need it and maybe
it does stomp on the SATA stuff somehow (on your computer, you
_should_ be using just SATA_NV, I think).

But I also notice that you have enabled the X86 "extended platform"
support, and in particular CONFIG_X86_ELAN, and that will disable all
the normal CPU logic. So please turn off CONFIG_X86_EXTENDED_PLATFORM
too, and use the "normal" x86 support.

Turn off CONFIG_EISA and CONFIG_MCA too, in order to limit things a bit.

Let's see if that makes any difference.

Linus

2011-05-02 22:12:00

by werner

[permalink] [raw]
Subject: Re: 2.6.39-rc5-git2 boot crashs

I disabled now CONFIG_IDE , CONFIG-X86_ELAN ,
CONFIG_X86_EXTENDED-PATFORM . Alias, moorestown I enabled
since 2.6.37 because it gave big problems on my computer
too.

On the other hand, I re-enabled CONFIG_MTD and
CONFIG-MISC_FILESYSTEMS , what I disabled yesterday.


On booting, it came back the non-crash error reported
earlier, see the enclosed syslog. Thus, this may be
caused by MTS oMISC_FILESYSTEMS.

The unzip problem now didn't occure, I could unzip a 140
MB and a 460 MB file without it crashed.

Actually it works stable? I have to continue to observe
what happens.

wl

1 enclosed file, syslog
=======================================================================
On Mon, 2 May 2011 08:17:58 -0700
Linus Torvalds <[email protected]> wrote:
> On Sun, May 1, 2011 at 11:54 PM, werner
><[email protected]> wrote:
>>
>> Then I gone back almost to my normal / everythingyes
>>config, but I disabled
>> CONFIG_MSC_FILESYSTEMS and CONFIG-MTD , also ACPI-APEI
>>because it needs
>> PSTORE of the 1st one.
>
> Ok, good.
>
>> With them disabled: ?The 1st booting in grafic mode was
>>normally. I also
>> could unzip a 120 MB file (what before with the
>>everythingyes config
>> crashed). ? However, if I tried to unzip an appr. 500 MB
>>file, it crashed.
>
> So that didn't help. Let's disable some other things
>too.
>
> Start off by disabling CONFIG_IDE, because you don't
>need it and maybe
> it does stomp on the SATA stuff somehow (on your
>computer, you
> _should_ be using just SATA_NV, I think).
>
> But I also notice that you have enabled the X86
>"extended platform"
> support, and in particular CONFIG_X86_ELAN, and that
>will disable all
> the normal CPU logic. So please turn off
>CONFIG_X86_EXTENDED_PLATFORM
> too, and use the "normal" x86 support.
>
> Turn off CONFIG_EISA and CONFIG_MCA too, in order to
>limit things a bit.
>
> Let's see if that makes any difference.
>
> Linus
>
>

"werner" <[email protected]>
---
Professional hosting for everyone - http://www.host.ru


Attachments:
error.txt (18.41 kB)

2011-05-02 23:23:27

by Linus Torvalds

[permalink] [raw]
Subject: Re: 2.6.39-rc5-git2 boot crashs

2011/5/2 werner <[email protected]>:
>
> The unzip problem now didn't occure, I could unzip a 140 MB and a 460 MB
> file without it crashed.
>
> Actually it works stable? I have to continue to observe what happens.

Ok, please do continue to test to make sure.

Onc eyou are sure - and _if_ that kernel is stable - please then
enable CONFIG_IDE again, just to see whether it's CONFIG_IDE or the
CONFIG_X86_ELAN that causes the problems.

I suspect IDE (that would make more sense from the standpoint of where
the problem occurs), but we'd better make sure.

Linus

2011-05-03 01:22:53

by werner

[permalink] [raw]
Subject: Re: 2.6.39-rc5-git2 boot crashs

Now I tested the config with:
Of your suggestions of today: IDE off , X86ELAN = OFF ,
X86_EXTended_PLATFORM = OFF
Of your suggestions of yesterday: MTD off ,
MISC-FILESYSTEMS = on

in order to find out what of the last two may cause the
error message during boot.



The (not killing) boot error message still occurs. See
the enclosed syslog and dmesg. Thus, probably its caused
by MISC-FILESYSTEMS.

sync hangs. See also the enclosed syslog. This also seems
being caused by MISC-FILESYSTEMS (but I'm not sure if it
didnt occur when I switched off this, soon I'll check
this, run again the previous compilation both MTD and
MISC-FILESYSTEMS switched off).

unzip 140 MB and 460 MB works, thus this error seems to
be caused by IDE, X86_ELAN or X86_EXTENDED_PLATFORM.



At this oportunity, I still add a boot error message by a
Toshiba Osmio Laptop, booting 2.6.38.3, some ten days ago,
for inspection. On my own computer, nor on my neighbours
Packard-Bell laptop, the same kernel package gave an error
message.


Now I continue to test some time the previous compilation,
and then I switch on IDE and compile the kernel again.


wl



3 enclosed files
---
Professional hosting for everyone - http://www.host.ru


Attachments:
dmesg.txt (60.34 kB)
syslog.txt (21.20 kB)
dmesg-osmio-26383 (60.50 kB)
Download all attachments