Re: [PATCH bpf] bpf: fix panic in stack_map_get_build_id() on i386 and arm32

[Song Liu]

> On Jan 8, 2019, at 2:20 PM, Song Liu <songliubrav...@fb.com> wrote:
> 
> As Naresh reported, test_stacktrace_build_id() causes panic on i386 and
> arm32 systems. This is caused by page_address() returns NULL in certain
> cases.
> 
> This patch fixes this error by using kmap_atomic/kunmap_atomic instead
> of page_address.
> 
> Fixes: 615755a77b24 (" bpf: extend stackmap to save binary_build_id+offset 
> instead of address")
> Reported-by: Naresh Kamboju <naresh.kamb...@linaro.org>
> Signed-off-by: Song Liu <songliubrav...@fb.com>
> ---
> kernel/bpf/stackmap.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/bpf/stackmap.c b/kernel/bpf/stackmap.c
> index 90daf285de03..d9e2483669d0 100644
> --- a/kernel/bpf/stackmap.c
> +++ b/kernel/bpf/stackmap.c
> @@ -260,7 +260,7 @@ static int stack_map_get_build_id(struct vm_area_struct 
> *vma,
>               return -EFAULT; /* page not mapped */
> 
>       ret = -EINVAL;
> -     page_addr = page_address(page);
> +     page_addr = kmap_atomic(page);
>       ehdr = (Elf32_Ehdr *)page_addr;
> 
>       /* compare magic x7f "ELF" */
> @@ -276,6 +276,7 @@ static int stack_map_get_build_id(struct vm_area_struct 
> *vma,
>       else if (ehdr->e_ident[EI_CLASS] == ELFCLASS64)
>               ret = stack_map_get_build_id_64(page_addr, build_id);
> out:
> +     kunmap_atomic(page_addr);
>       put_page(page);
>       return ret;
> }
> -- 
> 2.17.1
> 

Hi Naresh,

I only tested this patch on i386 vm. Could you please help test it on
ARM32 systems?

Thanks,
Song