[PATCH 1/3] x86/EFI: Fix detection of buildid

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

The format of the buildid is a property of the binary, not a property of how
it was loaded.  This fixes buildid recognition when starting xen.efi from it's
MB2 entrypoint.

Suggested-by: Ross Lagerwall <ross.lagerwall@xxxxxxxxxx>
Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
---
CC: Jan Beulich <JBeulich@xxxxxxxx>
CC: Roger Pau Monné <roger.pau@xxxxxxxxxx>
CC: Ross Lagerwall <ross.lagerwall@xxxxxxxxxx>
CC: Marek Marczykowski-Górecki <marmarek@xxxxxxxxxxxxxxxxxxxxxx>
CC: Daniel P. Smith <dpsmith@xxxxxxxxxxxxxxxxxxxx>

I don't like this patch and tried hard to do it in a better way, but the EFI
aspects of the build system are too intractable.

While on x86 I can in principle pull the same common-stubs.o trick, split on
XEN_BUILD_PE rather than XEN_BUILD_EFI, that doesn't work on ARM which
hand-codes it's PE-ness.  Also, it's really not EFI related, other than as a
consequence of that being the only reason we use PE32+ binaries.

Binutils 2.25 is now the minimum, and the makefiles can be cleaned up
somewhat, but I need to backport this patch, internally at least.
---
 xen/common/version.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/xen/common/version.c b/xen/common/version.c
index 5474b8e385be..56b51c81d2fc 100644
--- a/xen/common/version.c
+++ b/xen/common/version.c
@@ -203,8 +203,11 @@ void __init xen_build_init(void)
     rc = xen_build_id_check(n, sz, &build_id_p, &build_id_len);
 
 #ifdef CONFIG_X86
-    /* Alternatively we may have a CodeView record from an EFI build. */
-    if ( rc && efi_enabled(EFI_LOADER) )
+    /*
+     * xen.efi built with a new enough toolchain will have a CodeView record,
+     * not an ELF note.
+     */
+    if ( rc )
     {
         const struct pe_external_debug_directory *dir = (const void *)n;
 
-- 
2.39.5