On Wed, 3 Jun 2020 at 12:05, Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:
>
> Hello Arvind Sankar,
>
> The patch f61900fd0ebf: "efi/libstub: Unify initrd loading across
> architectures" from Apr 30, 2020, leads to the following static
> checker warning:
>
> drivers/firmware/efi/libstub/efi-stub-helper.c:481 efi_load_initrd_cmdline()
> error: we previously assumed 'image' could be null (see line 474)
>
> drivers/firmware/efi/libstub/efi-stub-helper.c
> 466 static
> 467 efi_status_t efi_load_initrd_cmdline(efi_loaded_image_t *image,
> 468 unsigned long *load_addr,
> 469 unsigned long *load_size,
> 470 unsigned long soft_limit,
> 471 unsigned long hard_limit)
> 472 {
> 473 if (!IS_ENABLED(CONFIG_EFI_GENERIC_STUB_INITRD_CMDLINE_LOADER) ||
> 474 (IS_ENABLED(CONFIG_X86) && (!efi_is_native() || image == NULL))) {
> ^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^
> Assume X86 is not configured but "image" is NULL.
>
> 475 *load_addr = *load_size = 0;
> 476 return EFI_SUCCESS;
> 477 }
> 478
> 479 return handle_cmdline_files(image, L"initrd=", sizeof(L"initrd=") - 2,
> ^^^^^
> Dereferenced inside the function.
image is guaranteed to be non-null on !x86 architectures.
>
> 480 soft_limit, hard_limit,
> 481 load_addr, load_size);
> 482 }
>
> regards,
> dan carpenter
