[Crash-utility] [PATCH v5] arm64: fix kernel memory map handling for kaslr-enabled

[AKASHI Takahiro]

On Tue, Jun 14, 2016 at 10:08:25AM -0400, Dave Anderson wrote:
> 
> > Dave,
> > 
> > On Mon, Jun 13, 2016 at 04:06:22PM -0400, Dave Anderson wrote:
> > > 
> > > 
> > > ----- Original Message -----
> > > > In my next version of kdump patch, the following VMCOREINFO will be
> > > > added:
> > > >     NUMBER(VA_BITS)
> > > >     NUMBER(kimage_voffset)
> > > >     NUMBER(PHYS_OFFSET)
> > > >     KERNELOFFSET
> > 
> > Is "KERNELOFFSET" useful for arm64?
> > As you said, this value is used as a trigger of KASLR_CHECK,
> > but the current kernel doesn't have "randomize_modules".
> 
> Right, so AIUI, this crash-7.0.7 update is not valid for Linux 4.1 and later
> kernels where "randomize_modules" was removed:
> 
>          - Linux 3.15 and later kernels configured with CONFIG_RANDOMIZE_BASE
>            can be now be readily identified because of new kernel symbols that
>            have been added.  For those kernels, the new "--kaslr=<offset>" 
>            and/or "--kaslr=auto" options are not necessary for ELF or compressed
>            kdump vmcores, or for live systems that have /proc/kallsyms showing
>            the relocated symbol values.  A new KASLR initialization function
>            called kaslr_init() is now called by symtab_init() prior to the 
>            initial symbol-sorting operation.  If kaslr_init() determines that
>            KASLR may be in effect, it will trigger a search for the relevant
>            vmlinux symbols during the sorting operation, which in turn will
>            cause the relocation value to be automatically calculated.
>            (anderson at redhat.com)
> 
> So for those kernels, --kaslr would be required.  
> 
> But it appears that I can just change crash to look for "module_load_offset", 
> which went into the kernel at the same time as "randomize_modules":
> 
>   +#ifdef CONFIG_RANDOMIZE_BASE
>   +static unsigned long module_load_offset;
>   +static int randomize_modules = 1;
>   +
> 
> and still exists after it was removed from arch/x86/kernel/module.c:
> 
>   #ifdef CONFIG_RANDOMIZE_BASE
>   static unsigned long module_load_offset;
>   -static int randomize_modules = 1;
> 
> For arm64 in the crash utility, I see that you would never use KASLR_CHECK because 
> (RELOC_AUTO|KASLR) gets set by default if NEW_VMEMMAP is set.    
> 
> Would it be useful for makedumpfile?

Given that makedumpfile doesn't reply on vmlinux, "kaslr_offset" is
not useful.
        -> Pratyush, any comments?

So I'd like to exclude KERNELOFFSET in my kexec/dkump patch, v19,
hopefully posted this weekend.
(This won't hurt crash utility anyway.)

Thanks,
-Takahiro AKASHI

> And while I understand you're dead set against
> adding anything that could be considered superfluous, but since it does exist for x86_64,
> it seems natural to include it for arm64. 
> 
> Dave
>  
> 
> 
> 
>  
> > Thanks,
> > -Takahiro AKASHI
> > 
> 
> --
> Crash-utility mailing list
> Crash-utility at redhat.com
> https://www.redhat.com/mailman/listinfo/crash-utility