Bug 1379

Summary: Failure to boot without acpi=off.
Product: Platform Specific/Hardware Reporter: Robert Scott (bugs)
Component: i386Assignee: Zwane Mwaikambo (zwane)
Status: REJECTED INSUFFICIENT_DATA    
Severity: normal CC: mbligh
Priority: P2    
Hardware: i386   
OS: Linux   
Kernel Version: 2.6.0-test8 Subsystem:
Regression: --- Bisected commit-id:
Attachments: Non-booting .config
another config

Description Robert Scott 2003-10-19 10:57:39 UTC 
Distribution: Gentoo 1.2
Hardware Environment: Dell Precision 210 (Dual Pentium II 400Mhz)
Software Environment: gcc 2.95.3, binutils 2.13.90.0.16, glibc 2.2.5

Problem Description:

Kernel fails to boot.

Gets to:

---
...
[Linux-bzImage, setup=0xc00, size=0x17e76d]

Uncompressing Linux... Ok, booting the kernel.
---

and then... nothing.

I had this problem before with kernel 2.6.0-test1. I solved it then by changing
many, many things in the config at random until it finally worked. 

However, with test8, I cannot get anything to work. I the past 24 hours I must
have compiled the kernel 25 times with different configurations, but to no
avail. I've tried systematically taking things out, putting things in, but I
haven't had a single boot. I'll attach my .config after this.

So, I set about trying to find where it stops the flow of execution by putting
my own printk()s into the source, and it seems like it is not able to "jump to
the new kernel..." after decompress_kernel() returns in
arch/i386/boot/compressed/head.S

If you want I could probably give you _many_ .configs which dont work, as I've
tried a lot of them.
Comment 1 Robert Scott 2003-10-19 10:59:16 UTC 
Created attachment 1099 [details]
Non-booting .config
Comment 2 Robert Scott 2003-10-19 11:16:35 UTC 
Created attachment 1100 [details]
another config

this one didn't work either
Comment 3 Martin J. Bligh 2003-10-20 16:13:09 UTC 
ftp://ftp.kernel.org/pub/linux/kernel/people/mbligh/patches/2.6.0-test6/2.6.0-test6-mjb1/100-early_printk

Slap this patch on, and enable it (I forget how, but it's pretty obvious). 
At least you should see how far it gets that way ....
Comment 4 Robert Scott 2003-10-20 17:00:42 UTC 
Narrowed it down.

Boots fine with acpi=off.



So, must be related to 1315. Or not. Mine doesn't even get that far.

Yes: my amateur diagnosis is probably rubbish then. Ignore that.
Comment 5 Robert Scott 2003-10-21 10:42:02 UTC 
Right, I ran it with the patch and output was as follows:


...
...
Debug: sleeping function called from invalid context at
incluse/asm/semaphore.h:119nel /boot/bzImage2.6test8bugme root=/dev/hda3
earlyprintk=vga
in_atomic():1, irqs_disabled():1size=0x1ab92d]
Call Trace:
[<c0120eb2>] __might_sleep+0x9a/0xa4e kernel.
[<c0123e39>] acquire_console_sem+0x2d/0x50
[<c01240f5>] register_console+0x115/0x194
[<c049dff3>] setup_early_printk+0x163/0x17c
[<c0105000>] Letext+0x0/0x60
[<c04906d3>] start_kernel+0x4f/0x1b0
Comment 6 Zwane Mwaikambo 2003-12-14 21:59:48 UTC 
ACPI shouldn't be getting enabled for that system, for my edification could you
send a booting dmesg?