Re: [crash] Re: Latest brk patchset

[Ingo Molnar <mingo@elte.hu>]

* Jeremy Fitzhardinge <jeremy@goop.org> wrote:

> Ingo Molnar wrote:
>> * Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>>
>>   
>>> Ingo Molnar wrote:
>>>     
>>>> * Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>>>>
>>>>         
>>>>> H. Peter Anvin wrote:
>>>>>             
>>>>>> Well, the semantics are different; the .bss section is zeroed while the
>>>>>> brk isn't, and the brk symbols don't necessarily point to the data
>>>>>> associated with those particular symbols, unlike (of course) the bss.
>>>>>>
>>>>>> It's not a big issue, obviously, it just seems cleaner to me that way.
>>>>>>                   
>>>>> OK, I just added a couple of changes to:
>>>>>
>>>>>    * make the brk reservation symbols have the form ".brk.NAME" to make
>>>>>      them inaccessible from C, and to make them look obviously
>>>>>      different from normal symbols (more like sections, since it is
>>>>>      their size that's more important than their address)
>>>>>    * Put all the brk stuff in a .brk section
>>>>>    * Mention alignment in the comment for the slop space
>>>>>
>>>>>    J
>>>>>
>>>>> The following changes since commit 1e08816af0bc345995c3f26ce4eaba1171ffb531:
>>>>>  Ingo Molnar (1):
>>>>>        Merge branch 'linus'
>>>>>
>>>>> are available in the git repository at:
>>>>>
>>>>>  git://git.kernel.org/pub/scm/linux/kernel/git/jeremy/xen.git push/x86/brk
>>>>>             
>>>> the previous kit in tip:x86/setup-memory is causing crashes. One of 
>>>> them is:
>>>>
>>>> init_memory_mapping: 0000000000000000-00000000377fe000              
>>>>                  0000000000 - 00377fe000 page 4k                    
>>>>                                 Kernel panic - not syncing: Cannot 
>>>> find space for the kernel page tables            Pid: 0, comm: 
>>>> swapper Not tainted 2.6.29-rc8-tip-02516-g83219b0-dirty #35476      
>>>>   Call Trace:                                                       
>>>>                    [<c0128b7b>] panic+0x4b/0x100                    
>>>>                                    [<c074a989>] 
>>>> init_memory_mapping+0x429/0x430                                     
>>>>    [<c0cde790>] setup_arch+0x430/0x890                              
>>>>                    [<c0148f4e>] ? lockdep_init_map+0x2e/0x150       
>>>>                                    [<c036f392>] ? 
>>>> __spin_lock_init+0x32/0x60                                          
>>>>  [<c01298d0>] ? printk+0x20/0x30                                    
>>>>                  [<c0cdc966>] start_kernel+0xc6/0x330               
>>>>                                  [<c0cdc321>]  
>>>> i386_start_kernel+0x41/0x50               
>>>>
>>>> full crashlog below, config attached.
>>>>         
>>> What branch is this?  I'm trying to build with your config with 
>>> current tip/master w/ tip/x86/setup-memory merged into it, but "make 
>>> ARCH=i386 oldconfig" is asking me about CONFIG_SMP, which makes me 
>>> think the config file is incomplete.
>>>     
>>
>> just accept the defaults in 'make oldconfig' and you'll get the right 
>> config.
>>   
>
> Got some build failures due to warnings treated as errors.  

Sorry - you need CONFIG_ALLOW_WARNINGS=y. Indeed my tests can 
emit !CONFIG_ALLOW_WARNINGS configs.

> And the resulting kernel booted fine under qemu with 1Gbyte of 
> memory.  I'll try on real hardware a bit later, but it doesn't 
> seem like something that should be affected by qemu vs native, 
> unless it has something to do with the specific e820 map.

Note that the crash was reproducible and it very clearly went 
away when i excluded those commits from tip:master.

	Ingo