On Thu, 10 Nov 2022 at 15:47, Ilias Apalodimas <ilias.apalodi...@linaro.org> wrote: > > [...] > > > > > + goto out; > > > > + > > > > + ret = efi_open_volume_int(file_info.current_volume, &root); > > > > + if (ret != EFI_SUCCESS) > > > > + goto out; > > > > + > > > > + ret = efi_file_open_int(root, &f, file_info.current_path, > > > > EFI_FILE_MODE_READ, 0); > > > > + if (ret != EFI_SUCCESS) > > > > + goto out; > > > > > > I think it would be better here if we could use efi_file_from_path(). > > > I think we can't easily do that atm since we can't convert the filename to > > > a device path with efi_dp_from_file() since we don't have the block info. > > > > Here we have a device path of volume(file_info.current_volume) and > > filename(file_info.current_path), so we can create a full device path to > > call > > efi_file_from_path(). > > # cmd/eficonfig.c::create_selected_device_path() create the full device > > path, > > we can reuse it. > > > > > > > > Since that requires a further clean up, I am fine keeping it as-is for > > > now, > > > but add a comment saying we should replace that with efi_file_from_path() > > > eventually. > > > > Probably I don't understand what is improved when we replace current code > > with efi_file_from_path(). > > I just prefer using common functions to open a file, rather than open > coding open_volume + file_open. OTOH efi_file_from_path() just > converts that DP into a filepath and reads the file. So on a second > thought leave this as is, we don't need a comment.
Thank you for the confirmation. Regards, Masahisa Kojima > > [...] > > Regards > /Ilias