On Wed, Feb 29, 2012 at 09:21:23AM -0800, Eugene Surovegin wrote:
> Per-CPU allocations are not guaranteed to be physically contiguous.
> However, kdump kernel and user-space code assumes that per-CPU
> memory, used for saving CPU registers on crash, is.
> This can cause corrupted /proc/vmcore in some cases - the main
> symptom being huge ELF note section.
>
> Force page alignment for note_buf_t to ensure that this assumption holds.
>
Hi Eugene,
Where do we make assumption that crash_notes address is page aligned?
In first kernel we save notes using virtual addresses as returned by
per_cpu_ptr(). So first kernel should be fine (crash_save_cpu()).
In second kernel, fs/proc/vmcore.c code does not seem to be assuming
that notes are stored as page aligned address.
merge_note_headers_elf64()
read_from_oldmem()
offset = (unsigned long)(*ppos % PAGE_SIZE);
pfn = (unsigned long)(*ppos / PAGE_SIZE);
We see to calculate pfn and offset inside the page. Map the pfn and then
access offset.
So as long as whole of the note section is stored on a single physical
page it is not a problem.
Are you referring to the fact that note could be stored on two different
physical pages and then kexec-tools does not know about the physical
address of second page, hence second kernel does not know about it
and we never read that data. That sounds like a problem.
So it make sense to force the page size alignment and if notes ever
grow beyond 1 page, we need to come up with more complex ways of
communicating more than 1 physical address to second kernel.
Thanks
Vivek
> Signed-off-by: Eugene Surovegin <surovegin at google.com>
> CC: Eric Biederman <ebiederm at xmission.com>
> CC: Vivek Goyal <vgoyal at redhat.com>
> CC: kexec-list <kexec at lists.infradead.org>
> ---
> kernel/kexec.c | 9 +++++++--
> 1 files changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/kexec.c b/kernel/kexec.c
> index 7b08867..e641b5c 100644
> --- a/kernel/kexec.c
> +++ b/kernel/kexec.c
> @@ -1232,8 +1232,13 @@ void crash_save_cpu(struct pt_regs *regs, int cpu)
>
> static int __init crash_notes_memory_init(void)
> {
> - /* Allocate memory for saving cpu registers. */
> - crash_notes = alloc_percpu(note_buf_t);
> + /* Allocate memory for saving cpu registers.
> + * Force page alignment to avoid crossing physical page boundary -
> + * kexec-tools and kernel /proc/vmcore handler assume these per-CPU
> + * chunks are physically contiguous.
> + */
> + crash_notes = (note_buf_t __percpu *)__alloc_percpu(sizeof(note_buf_t),
> + PAGE_SIZE);
> if (!crash_notes) {
> printk("Kexec: Memory allocation for saving cpu register"
> " states failed\n");
> --
> 1.7.9.1
