kdump for ARM Linux?

[robherring2@gmail.com (Rob Herring)]

On Tue, Apr 22, 2014 at 7:18 PM, Andy Nicholas
<andy.nicholas.work@gmail.com> wrote:
> Hello, hopefully this is the correct mailing list for this question:

Yes.

> I'm working on a derivative of the mainline Linux 3.8 kernel and I'm trying
> to enable ARM kdump functionality to collect crashdumps when we have kernel
> problems. But, there's some kind of unresolved issue in
> arch/arm/mm/ioremap.c which prevents /proc/vmcore from being able to
> function properly. I checked and this appears to still be in kernel.org
> Linux 3.13. The target platform is a regular, recent ARMv7 compatible chip.

There are only irregular platforms.

> Has this issue been addressed in any version of an ARM Linux kernel? Or, is
> there a work-around? If not, how does a crashdump file get generated when/if
> the kernel crashes or a driver faults?
>
>>> WARNING: at arch/arm/mm/ioremap.c:244
> __arm_ioremap_pfn_caller+0x1d8/0x1f0()
>>> Modules linked in:
>>> [<c0016660>] (unwind_backtrace+0x0/0x138) from [<c0020480>]
> (warn_slowpath_common+0x4c/0x64)
>>> [<c0020480>] (warn_slowpath_common+0x4c/0x64) from [<c00204b4>]
> (warn_slowpath_null+0x1c/0x24)
>>> [<c00204b4>] (warn_slowpath_null+0x1c/0x24) from [<c0018e9c>]
> (__arm_ioremap_pfn_caller+0x1d8/0x1f0)
>>> [<c0018e9c>] (__arm_ioremap_pfn_caller+0x1d8/0x1f0) from [<c0018f20>]
> (__arm_ioremap_caller+0x54/0x5c)
>>> [<c0018f20>] (__arm_ioremap_caller+0x54/0x5c) from [<c0018c58>]
> (__arm_ioremap+0x18/0x1c)
>>> [<c0018c58>] (__arm_ioremap+0x18/0x1c) from [<c00168fc>]
> (copy_oldmem_page+0x34/0xc0)
>>> [<c00168fc>] (copy_oldmem_page+0x34/0xc0) from [<c010350c>]
> (read_from_oldmem+0xb8/0xe4)
>>> [<c010350c>] (read_from_oldmem+0xb8/0xe4) from [<c05c37e0>]
> (parse_crash_elf32_headers+0x184/0x43c)
>>> [<c05c37e0>] (parse_crash_elf32_headers+0x184/0x43c) from [<c05c3b64>]
> (vmcore_init+0xcc/0x198)
>>> [<c05c3b64>] (vmcore_init+0xcc/0x198) from [<c000863c>]
> (do_one_initcall+0x34/0x184)
>>> [<c000863c>] (do_one_initcall+0x34/0x184) from [<c05b28dc>]
> (kernel_init_freeable+0xfc/0x1c8)
>>> [<c05b28dc>] (kernel_init_freeable+0xfc/0x1c8) from [<c04326d4>]
> (kernel_init+0x8/0xe4)
>>> [<c04326d4>] (kernel_init+0x8/0xe4) from [<c000ec58>]
> (ret_from_fork+0x14/0x3c)
>>>
>>> arch/arm/mm/ioremap.c:244
>>>        /*
>>>          * Don't allow RAM to be mapped - this causes problems with
> ARMv6+
>>>          */
>>>         if (WARN_ON(pfn_valid(pfn)))
>>>                 return NULL;
>
> Thanks in advance for any help.

This is on purpose as the comment says because 2 mappings of different
types (of memory attributes) is undefined behavior on ARMv6+. RAM is
normal memory type and mmio devices are device memory. This check
could possibly be relaxed in the case the memory type is compatible
with normal RAM mappings. Then perhaps ioremap_cache could be used on
RAM for kdump.

Rob