On 11/18/23 08:35, Ilias Apalodimas wrote:
Hi Heinrich

On Fri, 17 Nov 2023 at 01:08, Heinrich Schuchardt
<heinrich.schucha...@canonical.com> wrote:

The boot options created by eficonfig should use shortened device-paths to
avoid problems if drives are enumerated in a different sequence.

Signed-off-by: Heinrich Schuchardt <heinrich.schucha...@canonical.com>
---
  cmd/eficonfig.c | 1 +
  1 file changed, 1 insertion(+)

diff --git a/cmd/eficonfig.c b/cmd/eficonfig.c
index e6e8a0a488..3d95092109 100644
--- a/cmd/eficonfig.c
+++ b/cmd/eficonfig.c
@@ -528,6 +528,7 @@ struct efi_device_path *eficonfig_create_device_path(struct 
efi_device_path *dp_
         p += fp_size;
         *((struct efi_device_path *)p) = END;

+       dp_volume = efi_dp_shorten(dp_volume);

efi_dp_shorten can return null.  Dont we need to check that here?

Thank you for reviewing.

In the current usage we always have a dp_volume selected in eficonfig_select_volume. Only if we had an EFI_SIMPLE_FILE_PROTOCOL on a device that is not a block device we could receive NULL here.

Thinking of future support of network files it is probably wiser to catch this case too.

Best regards

Heinrich


Thanks
/Ilias
         dp = efi_dp_append(dp_volume, (struct efi_device_path *)buf);
         free(buf);

--
2.40.1


Reply via email to