[Crash-utility] Re: crash-4.0-4.3 and linux-2.6.22.1-20.fc7

[Dave Anderson]

Andi Kleen wrote:
 > > > Using the latest F7 kernel and crash I'm getting the following
 > > > error:
 > > > WARNING: cannot read linux_banner string
 > > > crash: /usr/lib/debug/lib/modules/2.6.22.1-20.fc7/vmlinux and
 > > > /dev/crash do not match!
 > > > I've tried both crash and crash -f with no luck.
 > > > I've search the mailing list and was unable to find any
 > > > thing talked about this problem... So I'm thinking this
 > > > may not be a known problem... If it is, please let me know
 > > > otherwise does anybody have idea what the problem could be?
 > > > tia,
 > > > steved.
 > > >
 > >
 > > This elephant still remains in the room...
 > >
 > > I can reproduce it with a 2.6.21-1.3194.fc7 kernel, which
 > > I believe has 2.6.22 stuff backported to it.
 >
 > Don't know about the Fedora kernel, but at least in mainline
 > __pa/__va still work as before for i386/x86-64. It's a single
 > add/sub on i386 and on x86-64 similar, but with different ranges
 > for kernel text vs the other kernel memory.
 >
 > The load address is variable now, but still has a default
 > set at compile time in CONFIG.
 >
 > What might have changed when you enable the relocatable kernel
 > is that the ELF addresses in the vmlinux are different. But
 > at runtime there is no difference.
 >
 > -Andi

Vivek Goyal wrote:
 > Hi Dave,
 >
 > Sorry, could not reply on Friday.
 >
 > This looks weared. Only one possible theory comes to mind. In FC7 kernels
 > following should be config variables.
 >
 > CONFIG_PHYSICAL_START=16MB
 > CONFIG_PHYSICAL_ALIGN=4MB.
 >
 > That means kernel has been compiled for 16MB physical address. That's
 > why linux_banner symbol address is in higher range. But at run time it
 > has been loaded by grub at 1MB location. Because alignment required is
 > 4MB kernel will move itself to physical address 4MB and that would
 > probably put linux banner somewhere at 6MB location.
 >
 > So these kind of problems will occur with first kernel is not run at
 > the address it has been compiled for. Can you please confirm this by
 > looking at FC7 config file.
 >
 > An easy way to fix this would be to change CONFIG_PHYSICAL_START=1MB.
 >
 > Thanks
 > Vivek

Thanks Vivek, that restores sanity.

Upon rebuilding my FC7 kernel with CONFIG_PHYSICAL_START changed
from 16MB to 1MB, with CONFIG_PHYSICAL_ALIGN left at 4MB, i.e.:

   CONFIG_PHYSICAL_START=0x100000
   CONFIG_PHYSICAL_ALIGN=0x400000

The kernel gets compiled for a 4MB physical address:

   $ nm -Bn vmlinux | grep "^c04"
   c0400000 T _text
   c0400000 T startup_32
   c0401000 T startup_32_smp
   c0401080 t checkCPUtype
   c0401101 t is486
   c0401108 t is386
   c0401175 t check_x87
   c04011a0 T setup_pda
   c04011c2 t setup_idt
   c04011df t rp_sidt
   c0401262 t early_divide_err
   c0401268 t early_illegal_opcode
   c0401271 t early_protection_fault
   c0401278 t early_page_fault
   c040127f t early_fault
   c04012a7 t hlt_loop
   c04012ac t ignore_int
   c04012f0 T _stext
   c04012f0 t run_init_process
   c04012f0 T stext
   ...

and so the same virtual-to-physical translation can be used for
both non-kernel unity-mapped kernel virtual addresses as well as
for kernel symbol addresses.

Thanks again,
   Dave